Specialize map[string]iface when unmarshaling kvs

name                               old time/op    new time/op    delta
UnmarshalDataset/config-8            12.3ms ± 0%    10.9ms ± 0%  -11.36%  (p=0.008 n=5+5)
UnmarshalDataset/canada-8            55.2ms ± 0%    55.4ms ± 1%     ~     (p=0.690 n=5+5)
UnmarshalDataset/citm_catalog-8      16.5ms ± 1%    15.7ms ± 0%   -4.43%  (p=0.008 n=5+5)
UnmarshalDataset/twitter-8           7.01ms ± 1%    6.63ms ± 0%   -5.34%  (p=0.008 n=5+5)
UnmarshalDataset/code-8              52.0ms ± 0%    48.7ms ± 2%   -6.43%  (p=0.008 n=5+5)
UnmarshalDataset/example-8            119µs ± 0%     110µs ± 3%   -6.81%  (p=0.008 n=5+5)
Unmarshal/SimpleDocument/struct-8     432ns ± 1%     431ns ± 1%     ~     (p=0.222 n=5+5)
Unmarshal/SimpleDocument/map-8        573ns ± 1%     540ns ± 1%   -5.64%  (p=0.008 n=5+5)
Unmarshal/ReferenceFile/struct-8     33.7µs ± 0%    33.6µs ± 0%     ~     (p=0.310 n=5+5)
Unmarshal/ReferenceFile/map-8        44.2µs ± 1%    41.7µs ± 1%   -5.63%  (p=0.008 n=5+5)
Unmarshal/HugoFrontMatter-8          7.28µs ± 1%    6.74µs ± 1%   -7.42%  (p=0.008 n=5+5)
[Geo mean]                            292µs          277µs        -4.91%

name                               old speed      new speed      delta
UnmarshalDataset/config-8          85.2MB/s ± 0%  96.1MB/s ± 0%  +12.82%  (p=0.008 n=5+5)
UnmarshalDataset/canada-8          39.9MB/s ± 0%  39.8MB/s ± 1%     ~     (p=0.690 n=5+5)
UnmarshalDataset/citm_catalog-8    33.9MB/s ± 1%  35.5MB/s ± 0%   +4.64%  (p=0.008 n=5+5)
UnmarshalDataset/twitter-8         63.1MB/s ± 1%  66.6MB/s ± 0%   +5.65%  (p=0.008 n=5+5)
UnmarshalDataset/code-8            51.6MB/s ± 0%  55.1MB/s ± 2%   +6.88%  (p=0.008 n=5+5)
UnmarshalDataset/example-8         68.3MB/s ± 0%  73.4MB/s ± 3%   +7.34%  (p=0.008 n=5+5)
Unmarshal/SimpleDocument/struct-8  25.4MB/s ± 1%  25.5MB/s ± 1%     ~     (p=0.246 n=5+5)
Unmarshal/SimpleDocument/map-8     19.2MB/s ± 1%  20.4MB/s ± 1%   +5.99%  (p=0.008 n=5+5)
Unmarshal/ReferenceFile/struct-8    156MB/s ± 0%   156MB/s ± 0%     ~     (p=0.310 n=5+5)
Unmarshal/ReferenceFile/map-8       119MB/s ± 1%   126MB/s ± 1%   +5.97%  (p=0.008 n=5+5)
Unmarshal/HugoFrontMatter-8        75.0MB/s ± 1%  81.0MB/s ± 1%   +8.01%  (p=0.008 n=5+5)
[Geo mean]                         56.1MB/s       59.0MB/s        +5.17%

name                               old alloc/op   new alloc/op   delta
UnmarshalDataset/config-8            5.26MB ± 0%    4.75MB ± 0%   -9.66%  (p=0.008 n=5+5)
UnmarshalDataset/canada-8            83.0MB ± 0%    83.0MB ± 0%   -0.00%  (p=0.008 n=5+5)
UnmarshalDataset/citm_catalog-8      34.7MB ± 0%    34.3MB ± 0%   -1.13%  (p=0.008 n=5+5)
UnmarshalDataset/twitter-8           12.7MB ± 0%    12.5MB ± 0%   -1.44%  (p=0.008 n=5+5)
UnmarshalDataset/code-8              15.3MB ± 0%    13.9MB ± 0%   -9.27%  (p=0.008 n=5+5)
UnmarshalDataset/example-8            186kB ± 0%     182kB ± 0%   -2.20%  (p=0.008 n=5+5)
Unmarshal/SimpleDocument/struct-8      805B ± 0%      805B ± 0%     ~     (all equal)
Unmarshal/SimpleDocument/map-8       1.13kB ± 0%    1.12kB ± 0%   -1.41%  (p=0.008 n=5+5)
Unmarshal/ReferenceFile/struct-8     20.9kB ± 0%    20.9kB ± 0%     ~     (all equal)
Unmarshal/ReferenceFile/map-8        36.4kB ± 0%    35.4kB ± 0%     ~     (p=0.079 n=4+5)
Unmarshal/HugoFrontMatter-8          7.20kB ± 0%    6.98kB ± 0%   -3.11%  (p=0.008 n=5+5)
[Geo mean]                            312kB          303kB        -2.86%

name                               old allocs/op  new allocs/op  delta
UnmarshalDataset/config-8              189k ± 0%      157k ± 0%  -16.80%  (p=0.029 n=4+4)
UnmarshalDataset/canada-8              782k ± 0%      782k ± 0%   -0.00%  (p=0.008 n=5+5)
UnmarshalDataset/citm_catalog-8        191k ± 0%      167k ± 0%  -12.75%  (p=0.000 n=4+5)
UnmarshalDataset/twitter-8            56.9k ± 0%     45.5k ± 0%  -20.02%  (p=0.016 n=5+4)
UnmarshalDataset/code-8                626k ± 0%      537k ± 0%  -14.22%  (p=0.008 n=5+5)
UnmarshalDataset/example-8            1.36k ± 0%     1.11k ± 0%  -18.53%  (p=0.008 n=5+5)
Unmarshal/SimpleDocument/struct-8      9.00 ± 0%      9.00 ± 0%     ~     (all equal)
Unmarshal/SimpleDocument/map-8         13.0 ± 0%      12.0 ± 0%   -7.69%  (p=0.008 n=5+5)
Unmarshal/ReferenceFile/struct-8        183 ± 0%       183 ± 0%     ~     (all equal)
Unmarshal/ReferenceFile/map-8           526 ± 0%       466 ± 0%  -11.41%  (p=0.008 n=5+5)
Unmarshal/HugoFrontMatter-8             126 ± 0%       112 ± 0%  -11.11%  (p=0.008 n=5+5)
[Geo mean]                            3.73k          3.34k       -10.51%

.gitattributes

@@ -1,3 +1,4 @@
 * text=auto
 
 benchmark/benchmark.toml text eol=lf
+testdata/** text eol=lf

.github/release.yml

@@ -2,6 +2,7 @@ changelog:
   exclude:
     labels:
       - build
+      - testing
   categories:
     - title: What's new
       labels:

.github/workflows/coverage.yml

@@ -15,6 +15,6 @@ jobs:
       - name: Setup go
         uses: actions/setup-go@master
         with:
-          go-version: 1.16
+          go-version: 1.18
       - name: Run tests with coverage
         run: ./ci.sh coverage -d "${GITHUB_BASE_REF-HEAD}"

.github/workflows/release.yml

@@ -0,0 +1,39 @@
+name: release
+on:
+  push:
+    tags:
+      - "v2.*"
+  workflow_call:
+    inputs:
+      args:
+        description: "Extra arguments to pass goreleaser"
+        default: ""
+        required: false
+        type: string
+
+jobs:
+  release:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v2
+        with:
+          fetch-depth: 0
+      - name: Set up Go
+        uses: actions/setup-go@v2
+        with:
+          go-version: 1.18
+      - name: Login to GitHub Container Registry
+        uses: docker/login-action@v1
+        with:
+          registry: ghcr.io
+          username: ${{ github.actor }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+      - name: Run GoReleaser
+        uses: goreleaser/goreleaser-action@v2
+        with:
+          distribution: goreleaser
+          version: latest
+          args: release ${{ inputs.args }} --rm-dist
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

.github/workflows/workflow.yml

@@ -12,14 +12,21 @@ jobs:
     strategy:
       matrix:
         os: [ 'ubuntu-latest', 'windows-latest', 'macos-latest']
-        go: [ '1.16', '1.17' ]
+        go: [ '1.17', '1.18' ]
     runs-on: ${{ matrix.os }}
     name: ${{ matrix.go }}/${{ matrix.os }}
     steps:
       - uses: actions/checkout@master
+        with:
+          fetch-depth: 0
       - name: Setup go ${{ matrix.go }}
         uses: actions/setup-go@master
         with:
           go-version: ${{ matrix.go }}
       - name: Run unit tests
         run: go test -race ./...
+  release-check:
+    if: ${{ github.ref != 'refs/heads/v2' }}
+    uses: pelletier/go-toml/.github/workflows/release.yml@v2
+    with:
+      args: --snapshot

.gitignore

@@ -3,3 +3,4 @@ fuzz/
 cmd/tomll/tomll
 cmd/tomljson/tomljson
 cmd/tomltestgen/tomltestgen
+dist

.goreleaser.yaml

@@ -0,0 +1,111 @@
+before:
+  hooks:
+    - go mod tidy
+    - go fmt ./...
+    - go test ./...
+builds:
+  - id: tomll
+    main: ./cmd/tomll
+    binary: tomll
+    env:
+      - CGO_ENABLED=0
+    flags:
+      - -trimpath
+    ldflags:
+      - -X main.version={{.Version}} -X main.commit={{.Commit}} -X main.date={{.CommitDate}}
+    mod_timestamp: '{{ .CommitTimestamp }}'
+    targets:
+      - linux_amd64
+      - windows_amd64
+      - darwin_amd64
+      - darwin_arm64
+  - id: tomljson
+    main: ./cmd/tomljson
+    binary: tomljson
+    env:
+      - CGO_ENABLED=0
+    flags:
+      - -trimpath
+    ldflags:
+      - -X main.version={{.Version}} -X main.commit={{.Commit}} -X main.date={{.CommitDate}}
+    mod_timestamp: '{{ .CommitTimestamp }}'
+    targets:
+      - linux_amd64
+      - windows_amd64
+      - darwin_amd64
+      - darwin_arm64
+  - id: jsontoml
+    main: ./cmd/jsontoml
+    binary: jsontoml
+    env:
+      - CGO_ENABLED=0
+    flags:
+      - -trimpath
+    ldflags:
+      - -X main.version={{.Version}} -X main.commit={{.Commit}} -X main.date={{.CommitDate}}
+    mod_timestamp: '{{ .CommitTimestamp }}'
+    targets:
+      - linux_amd64
+      - windows_amd64
+      - darwin_amd64
+      - darwin_arm64
+universal_binaries:
+  - id: tomll
+    replace: true
+    name_template: tomll
+  - id: tomljson
+    replace: true
+    name_template: tomljson
+  - id: jsontoml
+    replace: true
+    name_template: jsontoml
+archives:
+- id: jsontoml
+  format: tar.xz
+  builds:
+    - jsontoml
+  files:
+  - none*
+  name_template: "{{ .Binary }}_{{.Version}}_{{ .Os }}_{{ .Arch }}"
+- id: tomljson
+  format: tar.xz
+  builds:
+    - tomljson
+  files:
+  - none*
+  name_template: "{{ .Binary }}_{{.Version}}_{{ .Os }}_{{ .Arch }}"
+- id: tomll
+  format: tar.xz
+  builds:
+    - tomll
+  files:
+  - none*
+  name_template: "{{ .Binary }}_{{.Version}}_{{ .Os }}_{{ .Arch }}"
+dockers:
+  - id: tools
+    goos: linux
+    goarch: amd64
+    ids:
+      - jsontoml
+      - tomljson
+      - tomll
+    image_templates:
+      - "ghcr.io/pelletier/go-toml:latest"
+      - "ghcr.io/pelletier/go-toml:{{ .Tag }}"
+      - "ghcr.io/pelletier/go-toml:v{{ .Major }}"
+    skip_push: false
+checksum:
+  name_template: 'sha256sums.txt'
+snapshot:
+  name_template: "{{ incpatch .Version }}-next"
+release:
+  github:
+    owner: pelletier
+    name: go-toml
+  draft: true
+  prerelease: auto
+  mode: replace
+changelog:
+  use: github-native
+announce:
+  skip: true

CONTRIBUTING.md

@@ -155,6 +155,8 @@ Checklist:
 - Does not introduce backward-incompatible changes (unless discussed).
 - Has relevant doc changes.
 - Benchstat does not show performance regression.
+- Pull request is [labeled appropriately][pr-labels].
+- Title will be understandable in the changelog.
 
 1. Merge using "squash and merge".
 2. Make sure to edit the commit message to keep all the useful information
@@ -163,13 +165,25 @@ Checklist:
 
 ### New release
 
-1. Go to [releases][releases]. Click on "X commits to master since this
-   release".
-2. Make note of all the changes. Look for backward incompatible changes,
-   new features, and bug fixes.
-3. Pick the new version using the above and semver.
-4. Create a [new release][new-release].
-5. Follow the same format as [1.1.0][release-110].
+1. Decide on the next version number. Use semver.
+2. Generate release notes using [`gh`][gh]. Example:
+```
+$ gh api -X POST \
+  -F tag_name='v2.0.0-beta.5' \
+  -F target_commitish='v2' \
+  -F previous_tag_name='v2.0.0-beta.4' \
+  --jq '.body' \
+  repos/pelletier/go-toml/releases/generate-notes
+```
+3. Look for "Other changes". That would indicate a pull request not labeled
+   properly. Tweak labels and pull request titles until changelog looks good for
+   users.
+4. [Draft new release][new-release].
+5. Fill tag and target with the same value used to generate the changelog.
+6. Set title to the new tag value.
+7. Paste the generated changelog.
+8. Check "create discussion", in the "Releases" category.
+9. Check pre-release if new version is an alpha or beta.
 
 [issues-tracker]: https://github.com/pelletier/go-toml/issues
 [bug-report]: https://github.com/pelletier/go-toml/issues/new?template=bug_report.md
@@ -177,6 +191,6 @@ Checklist:
 [readme]: ./README.md
 [fork]: https://help.github.com/articles/fork-a-repo
 [pull-request]: https://help.github.com/en/articles/creating-a-pull-request
-[releases]: https://github.com/pelletier/go-toml/releases
 [new-release]: https://github.com/pelletier/go-toml/releases/new
-[release-110]: https://github.com/pelletier/go-toml/releases/tag/v1.1.0
+[gh]: https://github.com/cli/cli
+[pr-labels]: https://github.com/pelletier/go-toml/blob/v2/.github/release.yml

Dockerfile

@@ -0,0 +1,5 @@
+FROM scratch
+ENV PATH "$PATH:/bin"
+COPY tomll /bin/tomll
+COPY tomljson /bin/tomljson
+COPY jsontoml /bin/jsontoml

README.md

@@ -21,7 +21,8 @@ encouraged to try out this version.
 
 ## Documentation
 
-Full API, examples, and implementation notes are available in the Go documentation.
+Full API, examples, and implementation notes are available in the Go
+documentation.
 
 [![Go Reference](https://pkg.go.dev/badge/github.com/pelletier/go-toml/v2.svg)](https://pkg.go.dev/github.com/pelletier/go-toml/v2)
 
@@ -51,12 +52,13 @@ operations should not be shockingly slow. See [benchmarks](#benchmarks).
 the TOML document was not prevent in the target structure. This is a great way
 to check for typos. [See example in the documentation][strict].
 
-[strict]: https://pkg.go.dev/github.com/pelletier/go-toml/v2#example-Decoder.SetStrict
+[strict]: https://pkg.go.dev/github.com/pelletier/go-toml/v2#example-Decoder.DisallowUnknownFields
 
 ### Contextualized errors
 
-When decoding errors occur, go-toml returns [`DecodeError`][decode-err]), which
-contains a human readable contextualized version of the error. For example:
+When most decoding errors occur, go-toml returns [`DecodeError`][decode-err]),
+which contains a human readable contextualized version of the error. For
+example:
 
 ```
 2| key1 = "value1"
@@ -159,11 +161,11 @@ Execution time speedup compared to other Go TOML libraries:
     </thead>
     <tbody>
         <tr><td>Marshal/HugoFrontMatter-2</td><td>1.9x</td><td>1.9x</td></tr>
-        <tr><td>Marshal/ReferenceFile/map-2</td><td>1.7x</td><td>1.9x</td></tr>
-        <tr><td>Marshal/ReferenceFile/struct-2</td><td>2.4x</td><td>2.6x</td></tr>
-        <tr><td>Unmarshal/HugoFrontMatter-2</td><td>2.9x</td><td>2.5x</td></tr>
-        <tr><td>Unmarshal/ReferenceFile/map-2</td><td>2.7x</td><td>2.6x</td></tr>
-        <tr><td>Unmarshal/ReferenceFile/struct-2</td><td>4.8x</td><td>5.1x</td></tr>
+        <tr><td>Marshal/ReferenceFile/map-2</td><td>1.7x</td><td>1.8x</td></tr>
+        <tr><td>Marshal/ReferenceFile/struct-2</td><td>2.2x</td><td>2.5x</td></tr>
+        <tr><td>Unmarshal/HugoFrontMatter-2</td><td>2.9x</td><td>2.9x</td></tr>
+        <tr><td>Unmarshal/ReferenceFile/map-2</td><td>2.6x</td><td>2.9x</td></tr>
+        <tr><td>Unmarshal/ReferenceFile/struct-2</td><td>4.4x</td><td>5.3x</td></tr>
      </tbody>
 </table>
 <details><summary>See more</summary>
@@ -176,17 +178,17 @@ provided for completeness.</p>
         <tr><th>Benchmark</th><th>go-toml v1</th><th>BurntSushi/toml</th></tr>
     </thead>
     <tbody>
-        <tr><td>Marshal/SimpleDocument/map-2</td><td>1.7x</td><td>2.1x</td></tr>
-        <tr><td>Marshal/SimpleDocument/struct-2</td><td>2.5x</td><td>2.8x</td></tr>
-        <tr><td>Unmarshal/SimpleDocument/map-2</td><td>4.1x</td><td>3.1x</td></tr>
-        <tr><td>Unmarshal/SimpleDocument/struct-2</td><td>6.4x</td><td>4.3x</td></tr>
-        <tr><td>UnmarshalDataset/example-2</td><td>3.4x</td><td>3.2x</td></tr>
-        <tr><td>UnmarshalDataset/code-2</td><td>2.2x</td><td>2.5x</td></tr>
-        <tr><td>UnmarshalDataset/twitter-2</td><td>2.8x</td><td>2.7x</td></tr>
-        <tr><td>UnmarshalDataset/citm_catalog-2</td><td>2.2x</td><td>2.0x</td></tr>
-        <tr><td>UnmarshalDataset/canada-2</td><td>1.8x</td><td>1.4x</td></tr>
-        <tr><td>UnmarshalDataset/config-2</td><td>4.4x</td><td>2.9x</td></tr>
-        <tr><td>[Geo mean]</td><td>2.8x</td><td>2.6x</td></tr>
+        <tr><td>Marshal/SimpleDocument/map-2</td><td>1.8x</td><td>2.9x</td></tr>
+        <tr><td>Marshal/SimpleDocument/struct-2</td><td>2.7x</td><td>4.2x</td></tr>
+        <tr><td>Unmarshal/SimpleDocument/map-2</td><td>4.5x</td><td>3.1x</td></tr>
+        <tr><td>Unmarshal/SimpleDocument/struct-2</td><td>6.2x</td><td>3.9x</td></tr>
+        <tr><td>UnmarshalDataset/example-2</td><td>3.1x</td><td>3.5x</td></tr>
+        <tr><td>UnmarshalDataset/code-2</td><td>2.3x</td><td>3.1x</td></tr>
+        <tr><td>UnmarshalDataset/twitter-2</td><td>2.5x</td><td>2.6x</td></tr>
+        <tr><td>UnmarshalDataset/citm_catalog-2</td><td>2.1x</td><td>2.2x</td></tr>
+        <tr><td>UnmarshalDataset/canada-2</td><td>1.6x</td><td>1.3x</td></tr>
+        <tr><td>UnmarshalDataset/config-2</td><td>4.3x</td><td>3.2x</td></tr>
+        <tr><td>[Geo mean]</td><td>2.7x</td><td>2.8x</td></tr>
      </tbody>
 </table>
 <p>This table can be generated with <code>./ci.sh benchmark -a -html</code>.</p>
@@ -206,6 +208,44 @@ In case of trouble: [Go Modules FAQ][mod-faq].
 
 [mod-faq]: https://github.com/golang/go/wiki/Modules#why-does-installing-a-tool-via-go-get-fail-with-error-cannot-find-main-module
 
+## Tools
+
+Go-toml provides three handy command line tools:
+
+ * `tomljson`: Reads a TOML file and outputs its JSON representation.
+
+    ```
+    $ go install github.com/pelletier/go-toml/v2/cmd/tomljson@latest
+    $ tomljson --help
+    ```
+
+ * `jsontoml`: Reads a JSON file and outputs a TOML representation.
+
+    ```
+    $ go install github.com/pelletier/go-toml/v2/cmd/jsontoml@latest
+    $ jsontoml --help
+    ```
+
+ * `tomll`: Lints and reformats a TOML file.
+
+    ```
+    $ go install github.com/pelletier/go-toml/v2/cmd/tomll@latest
+    $ tomll --help
+    ```
+
+### Docker image
+
+Those tools are also available as a [Docker image][docker]. For example, to use
+`tomljson`:
+
+```
+docker run -i ghcr.io/pelletier/go-toml:v2 tomljson < example.toml
+```
+
+Multiple versions are availble on [ghcr.io][docker].
+
+[docker]: https://github.com/pelletier/go-toml/pkgs/container/go-toml
+
 ## Migrating from v1
 
 This section describes the differences between v1 and v2, with some pointers on
@@ -324,6 +364,29 @@ The recommended replacement is pre-filling the struct before unmarshaling.
 
 [go-defaults]: https://github.com/mcuadros/go-defaults
 
+#### `toml.Tree` replacement
+
+This structure was the initial attempt at providing a document model for
+go-toml. It allows manipulating the structure of any document, encoding and
+decoding from their TOML representation. While a more robust feature was
+initially planned in go-toml v2, this has been ultimately [removed from
+scope][nodoc] of this library, with no plan to add it back at the moment. The
+closest equivalent at the moment would be to unmarshal into an `interface{}` and
+use type assertions and/or reflection to manipulate the arbitrary
+structure. However this would fall short of providing all of the TOML features
+such as adding comments and be specific about whitespace.
+
+
+#### `toml.Position` are not retrievable anymore
+
+The API for retrieving the position (line, column) of a specific TOML element do
+not exist anymore. This was done to minimize the amount of concepts introduced
+by the library (query path), and avoid the performance hit related to storing
+positions in the absence of a document model, for a feature that seemed to have
+little use. Errors however have gained more detailed position
+information. Position retrieval seems better fitted for a document model, which
+has been [removed from the scope][nodoc] of go-toml v2 at the moment.
+
 ### Encoding / Marshal
 
 #### Default struct fields order
@@ -359,7 +422,8 @@ fmt.Println("v2:\n" + string(b))
 ```
 
 There is no way to make v2 encoder behave like v1. A workaround could be to
-manually sort the fields alphabetically in the struct definition.
+manually sort the fields alphabetically in the struct definition, or generate
+struct types using `reflect.StructOf`.
 
 #### No indentation by default
 
@@ -407,7 +471,9 @@ fmt.Println("v2 Encoder:\n" + string(buf.Bytes()))
 
 V1 always uses double quotes (`"`) around strings and keys that cannot be
 represented bare (unquoted). V2 uses single quotes instead by default (`'`),
-unless a character cannot be represented, then falls back to double quotes.
+unless a character cannot be represented, then falls back to double quotes. As a
+result of this change, `Encoder.QuoteMapKeys` has been removed, as it is not
+useful anymore.
 
 There is no way to make v2 encoder behave like v1.
 
@@ -422,6 +488,69 @@ There is no way to make v2 encoder behave like v1.
 
 [tm]: https://golang.org/pkg/encoding/#TextMarshaler
 
+#### `Encoder.CompactComments` has been removed
+
+Emitting compact comments is now the default behavior of go-toml. This option
+is not necessary anymore.
+
+#### Struct tags have been merged
+
+V1 used to provide multiple struct tags: `comment`, `commented`, `multiline`,
+`toml`, and `omitempty`. To behave more like the standard library, v2 has merged
+`toml`, `multiline`, and `omitempty`. For example:
+
+```go
+type doc struct {
+	// v1
+	F string `toml:"field" multiline:"true" omitempty:"true"`
+	// v2
+	F string `toml:"field,multiline,omitempty"`
+}
+```
+
+Has a result, the `Encoder.SetTag*` methods have been removed, as there is just
+one tag now.
+
+
+#### `commented` tag has been removed
+
+There is no replacement for the `commented` tag. This feature would be better
+suited in a proper document model for go-toml v2, which has been [cut from
+scope][nodoc] at the moment.
+
+#### `Encoder.ArraysWithOneElementPerLine` has been renamed
+
+The new name is `Encoder.SetArraysMultiline`. The behavior should be the same.
+
+#### `Encoder.Indentation` has been renamed
+
+The new name is `Encoder.SetIndentSymbol`. The behavior should be the same.
+
+
+#### Embedded structs behave like stdlib
+
+V1 defaults to merging embedded struct fields into the embedding struct. This
+behavior was unexpected because it does not follow the standard library. To
+avoid breaking backward compatibility, the `Encoder.PromoteAnonymous` method was
+added to make the encoder behave correctly. Given backward compatibility is not
+a problem anymore, v2 does the right thing by default: it follows the behavior
+of `encoding/json`. `Encoder.PromoteAnonymous` has been removed.
+
+[nodoc]: https://github.com/pelletier/go-toml/discussions/506#discussioncomment-1526038
+
+### `query`
+
+go-toml v1 provided the [`go-toml/query`][query] package. It allowed to run
+JSONPath-style queries on TOML files. This feature is not available in v2. For a
+replacement, check out [dasel][dasel].
+
+This package has been removed because it was essentially not supported anymore
+(last commit May 2020), increased the complexity of the code base, and more
+complete solutions exist out there.
+
+[query]: https://github.com/pelletier/go-toml/tree/f99d6bbca119636aeafcf351ee52b3d202782627/query
+[dasel]: https://github.com/TomWright/dasel
+
 ## License
 
 The MIT License (MIT). Read [LICENSE](LICENSE).

ci.sh

@@ -76,7 +76,8 @@ cover() {
     fi
 
     pushd "$dir"
-    go test -covermode=atomic -coverprofile=coverage.out ./...
+    go test -covermode=atomic  -coverpkg=./... -coverprofile=coverage.out.tmp ./...
+    cat coverage.out.tmp | grep -v testsuite | grep -v tomltestgen | grep -v gotoml-test-decoder > coverage.out
     go tool cover -func=coverage.out
     popd
 
@@ -103,16 +104,23 @@ coverage() {
 
 	    echo ""
 
-	    target_pct="$(cat ${target_out} |sed -E 's/.*total.*\t([0-9.]+)%/\1/;t;d')"
-	    head_pct="$(cat ${head_out} |sed -E 's/.*total.*\t([0-9.]+)%/\1/;t;d')"
+	    target_pct="$(tail -n2 ${target_out} | head -n1 | sed -E 's/.*total.*\t([0-9.]+)%.*/\1/')"
+	    head_pct="$(tail -n2 ${head_out} | head -n1 | sed -E 's/.*total.*\t([0-9.]+)%/\1/')"
 	    echo "Results: ${target} ${target_pct}% HEAD ${head_pct}%"
 
 	    delta_pct=$(echo "$head_pct - $target_pct" | bc -l)
 	    echo "Delta: ${delta_pct}"
 
 	    if [[ $delta_pct = \-* ]]; then
-		echo "Regression!";
-		return 1
+		    echo "Regression!";
+
+            target_diff="${output_dir}/target.diff.txt"
+            head_diff="${output_dir}/head.diff.txt"
+            cat "${target_out}" | grep -E '^github.com/pelletier/go-toml' | tr -s "\t " | cut -f 2,3 | sort > "${target_diff}"
+            cat "${head_out}" | grep -E '^github.com/pelletier/go-toml' | tr -s "\t " | cut -f 2,3 | sort > "${head_diff}"
+
+            diff --side-by-side --suppress-common-lines "${target_diff}" "${head_diff}"
+		    return 1
 	    fi
 	    return 0
 	    ;;

cmd/gotoml-test-decoder/main.go

@@ -6,7 +6,7 @@ import (
 	"os"
 	"path"
 
-	"github.com/pelletier/go-toml/v2/testsuite"
+	"github.com/pelletier/go-toml/v2/internal/testsuite"
 )
 
 func main() {

cmd/jsontoml/main.go

@@ -0,0 +1,55 @@
+// Package jsontoml is a program that converts JSON to TOML.
+//
+// Usage
+//
+// Reading from stdin:
+//
+//   cat file.json | jsontoml > file.toml
+//
+// Reading from a file:
+//
+//   jsontoml file.json > file.toml
+//
+// Installation
+//
+// Using Go:
+//
+//   go install github.com/pelletier/go-toml/v2/cmd/jsontoml@latest
+package main
+
+import (
+	"encoding/json"
+	"io"
+
+	"github.com/pelletier/go-toml/v2"
+	"github.com/pelletier/go-toml/v2/internal/cli"
+)
+
+const usage = `jsontoml can be used in two ways:
+Reading from stdin:
+  cat file.json | jsontoml > file.toml
+
+Reading from a file:
+  jsontoml file.json > file.toml
+`
+
+func main() {
+	p := cli.Program{
+		Usage: usage,
+		Fn:    convert,
+	}
+	p.Execute()
+}
+
+func convert(r io.Reader, w io.Writer) error {
+	var v interface{}
+
+	d := json.NewDecoder(r)
+	err := d.Decode(&v)
+	if err != nil {
+		return err
+	}
+
+	e := toml.NewEncoder(w)
+	return e.Encode(v)
+}

cmd/jsontoml/main_test.go

@@ -0,0 +1,49 @@
+package main
+
+import (
+	"bytes"
+	"strings"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func TestConvert(t *testing.T) {
+	examples := []struct {
+		name     string
+		input    string
+		expected string
+		errors   bool
+	}{
+		{
+			name: "valid json",
+			input: `
+{
+  "mytoml": {
+    "a": 42
+  }
+}`,
+			expected: `[mytoml]
+a = 42.0
+
+`,
+		},
+		{
+			name:   "invalid json",
+			input:  `{ foo`,
+			errors: true,
+		},
+	}
+
+	for _, e := range examples {
+		b := new(bytes.Buffer)
+		err := convert(strings.NewReader(e.input), b)
+		if e.errors {
+			require.Error(t, err)
+		} else {
+			assert.NoError(t, err)
+			assert.Equal(t, e.expected, b.String())
+		}
+	}
+}

cmd/tomljson/main.go

@@ -0,0 +1,63 @@
+// Package tomljson is a program that converts TOML to JSON.
+//
+// Usage
+//
+// Reading from stdin:
+//
+//   cat file.toml | tomljson > file.json
+//
+// Reading from a file:
+//
+//   tomljson file.toml > file.json
+//
+// Installation
+//
+// Using Go:
+//
+//   go install github.com/pelletier/go-toml/v2/cmd/tomljson@latest
+package main
+
+import (
+	"encoding/json"
+	"errors"
+	"fmt"
+	"io"
+
+	"github.com/pelletier/go-toml/v2"
+	"github.com/pelletier/go-toml/v2/internal/cli"
+)
+
+const usage = `tomljson can be used in two ways:
+Reading from stdin:
+  cat file.toml | tomljson > file.json
+
+Reading from a file:
+  tomljson file.toml > file.json
+`
+
+func main() {
+	p := cli.Program{
+		Usage: usage,
+		Fn:    convert,
+	}
+	p.Execute()
+}
+
+func convert(r io.Reader, w io.Writer) error {
+	var v interface{}
+
+	d := toml.NewDecoder(r)
+	err := d.Decode(&v)
+	if err != nil {
+		var derr *toml.DecodeError
+		if errors.As(err, &derr) {
+			row, col := derr.Position()
+			return fmt.Errorf("%s\nerror occurred at row %d column %d", derr.String(), row, col)
+		}
+		return err
+	}
+
+	e := json.NewEncoder(w)
+	e.SetIndent("", "  ")
+	return e.Encode(v)
+}

cmd/tomljson/main_test.go

@@ -0,0 +1,61 @@
+package main
+
+import (
+	"bytes"
+	"fmt"
+	"io"
+	"strings"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func TestConvert(t *testing.T) {
+	examples := []struct {
+		name     string
+		input    io.Reader
+		expected string
+		errors   bool
+	}{
+		{
+			name: "valid toml",
+			input: strings.NewReader(`
+[mytoml]
+a = 42`),
+			expected: `{
+  "mytoml": {
+    "a": 42
+  }
+}
+`,
+		},
+		{
+			name:   "invalid toml",
+			input:  strings.NewReader(`bad = []]`),
+			errors: true,
+		},
+		{
+			name:   "bad reader",
+			input:  &badReader{},
+			errors: true,
+		},
+	}
+
+	for _, e := range examples {
+		b := new(bytes.Buffer)
+		err := convert(e.input, b)
+		if e.errors {
+			require.Error(t, err)
+		} else {
+			assert.NoError(t, err)
+			assert.Equal(t, e.expected, b.String())
+		}
+	}
+}
+
+type badReader struct{}
+
+func (r *badReader) Read([]byte) (int, error) {
+	return 0, fmt.Errorf("reader failed on purpose")
+}

cmd/tomll/main.go

@@ -0,0 +1,58 @@
+// Package tomll is a linter program for TOML.
+//
+// Usage
+//
+// Reading from stdin, writing to stdout:
+//
+//   cat file.toml | tomll
+//
+// Reading and updating a list of files in place:
+//
+//   tomll a.toml b.toml c.toml
+//
+// Installation
+//
+// Using Go:
+//
+//   go install github.com/pelletier/go-toml/v2/cmd/tomll@latest
+package main
+
+import (
+	"io"
+
+	"github.com/pelletier/go-toml/v2"
+	"github.com/pelletier/go-toml/v2/internal/cli"
+)
+
+const usage = `tomll can be used in two ways:
+
+Reading from stdin, writing to stdout:
+  cat file.toml | tomll > file.toml
+
+Reading and updating a list of files in place:
+  tomll a.toml b.toml c.toml
+
+When given a list of files, tomll will modify all files in place without asking.
+`
+
+func main() {
+	p := cli.Program{
+		Usage:   usage,
+		Fn:      convert,
+		Inplace: true,
+	}
+	p.Execute()
+}
+
+func convert(r io.Reader, w io.Writer) error {
+	var v interface{}
+
+	d := toml.NewDecoder(r)
+	err := d.Decode(&v)
+	if err != nil {
+		return err
+	}
+
+	e := toml.NewEncoder(w)
+	return e.Encode(v)
+}

cmd/tomll/main_test.go

@@ -0,0 +1,46 @@
+package main
+
+import (
+	"bytes"
+	"strings"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func TestConvert(t *testing.T) {
+	examples := []struct {
+		name     string
+		input    string
+		expected string
+		errors   bool
+	}{
+		{
+			name: "valid toml",
+			input: `
+mytoml.a = 42.0
+`,
+			expected: `[mytoml]
+a = 42.0
+
+`,
+		},
+		{
+			name:   "invalid toml",
+			input:  `[what`,
+			errors: true,
+		},
+	}
+
+	for _, e := range examples {
+		b := new(bytes.Buffer)
+		err := convert(strings.NewReader(e.input), b)
+		if e.errors {
+			require.Error(t, err)
+		} else {
+			assert.NoError(t, err)
+			assert.Equal(t, e.expected, b.String())
+		}
+	}
+}

cmd/tomltestgen/main.go

@@ -43,8 +43,7 @@ type testsCollection struct {
 	Count     int
 }
 
-const srcTemplate = "// +build testsuite\n\n" +
-	"// Generated by tomltestgen for toml-test ref {{.Ref}} on {{.Timestamp}}\n" +
+const srcTemplate = "// Generated by tomltestgen for toml-test ref {{.Ref}} on {{.Timestamp}}\n" +
 	"package toml_test\n" +
 	" import (\n" +
 	"	\"testing\"\n" +

decode.go

@@ -35,13 +35,22 @@ func parseLocalDate(b []byte) (LocalDate, error) {
 		return date, newDecodeError(b, "dates are expected to have the format YYYY-MM-DD")
 	}
 
-	date.Year = parseDecimalDigits(b[0:4])
+	var err error
 
-	v := parseDecimalDigits(b[5:7])
+	date.Year, err = parseDecimalDigits(b[0:4])
+	if err != nil {
+		return LocalDate{}, err
+	}
 
-	date.Month = v
+	date.Month, err = parseDecimalDigits(b[5:7])
+	if err != nil {
+		return LocalDate{}, err
+	}
 
-	date.Day = parseDecimalDigits(b[8:10])
+	date.Day, err = parseDecimalDigits(b[8:10])
+	if err != nil {
+		return LocalDate{}, err
+	}
 
 	if !isValidDate(date.Year, date.Month, date.Day) {
 		return LocalDate{}, newDecodeError(b, "impossible date")
@@ -50,15 +59,18 @@ func parseLocalDate(b []byte) (LocalDate, error) {
 	return date, nil
 }
 
-func parseDecimalDigits(b []byte) int {
+func parseDecimalDigits(b []byte) (int, error) {
 	v := 0
 
-	for _, c := range b {
+	for i, c := range b {
+		if c < '0' || c > '9' {
+			return 0, newDecodeError(b[i:i+1], "expected digit (0-9)")
+		}
 		v *= 10
 		v += int(c - '0')
 	}
 
-	return v
+	return v, nil
 }
 
 func parseDateTime(b []byte) (time.Time, error) {
@@ -87,15 +99,42 @@ func parseDateTime(b []byte) (time.Time, error) {
 		if len(b) != dateTimeByteLen {
 			return time.Time{}, newDecodeError(b, "invalid date-time timezone")
 		}
-		direction := 1
-		if b[0] == '-' {
+		var direction int
+		switch b[0] {
+		case '-':
 			direction = -1
+		case '+':
+			direction = +1
+		default:
+			return time.Time{}, newDecodeError(b[:1], "invalid timezone offset character")
+		}
+
+		if b[3] != ':' {
+			return time.Time{}, newDecodeError(b[3:4], "expected a : separator")
+		}
+
+		hours, err := parseDecimalDigits(b[1:3])
+		if err != nil {
+			return time.Time{}, err
+		}
+		if hours > 23 {
+			return time.Time{}, newDecodeError(b[:1], "invalid timezone offset hours")
+		}
+
+		minutes, err := parseDecimalDigits(b[4:6])
+		if err != nil {
+			return time.Time{}, err
+		}
+		if minutes > 59 {
+			return time.Time{}, newDecodeError(b[:1], "invalid timezone offset minutes")
 		}
 
-		hours := digitsToInt(b[1:3])
-		minutes := digitsToInt(b[4:6])
 		seconds := direction * (hours*3600 + minutes*60)
-		zone = time.FixedZone("", seconds)
+		if seconds == 0 {
+			zone = time.UTC
+		} else {
+			zone = time.FixedZone("", seconds)
+		}
 		b = b[dateTimeByteLen:]
 	}
 
@@ -159,7 +198,13 @@ func parseLocalTime(b []byte) (LocalTime, []byte, error) {
 		return t, nil, newDecodeError(b, "times are expected to have the format HH:MM:SS[.NNNNNN]")
 	}
 
-	t.Hour = parseDecimalDigits(b[0:2])
+	var err error
+
+	t.Hour, err = parseDecimalDigits(b[0:2])
+	if err != nil {
+		return t, nil, err
+	}
+
 	if t.Hour > 23 {
 		return t, nil, newDecodeError(b[0:2], "hour cannot be greater 23")
 	}
@@ -167,7 +212,10 @@ func parseLocalTime(b []byte) (LocalTime, []byte, error) {
 		return t, nil, newDecodeError(b[2:3], "expecting colon between hours and minutes")
 	}
 
-	t.Minute = parseDecimalDigits(b[3:5])
+	t.Minute, err = parseDecimalDigits(b[3:5])
+	if err != nil {
+		return t, nil, err
+	}
 	if t.Minute > 59 {
 		return t, nil, newDecodeError(b[3:5], "minutes cannot be greater 59")
 	}
@@ -175,42 +223,58 @@ func parseLocalTime(b []byte) (LocalTime, []byte, error) {
 		return t, nil, newDecodeError(b[5:6], "expecting colon between minutes and seconds")
 	}
 
-	t.Second = parseDecimalDigits(b[6:8])
-	if t.Second > 59 {
-		return t, nil, newDecodeError(b[3:5], "seconds cannot be greater 59")
+	t.Second, err = parseDecimalDigits(b[6:8])
+	if err != nil {
+		return t, nil, err
 	}
 
-	const minLengthWithFrac = 9
-	if len(b) >= minLengthWithFrac && b[minLengthWithFrac-1] == '.' {
+	if t.Second > 60 {
+		return t, nil, newDecodeError(b[6:8], "seconds cannot be greater 60")
+	}
+
+	b = b[8:]
+
+	if len(b) >= 1 && b[0] == '.' {
 		frac := 0
+		precision := 0
 		digits := 0
 
-		for i, c := range b[minLengthWithFrac:] {
+		for i, c := range b[1:] {
 			if !isDigit(c) {
 				if i == 0 {
-					return t, nil, newDecodeError(b[i:i+1], "need at least one digit after fraction point")
+					return t, nil, newDecodeError(b[0:1], "need at least one digit after fraction point")
 				}
-
 				break
 			}
+			digits++
 
 			const maxFracPrecision = 9
 			if i >= maxFracPrecision {
-				return t, nil, newDecodeError(b[i:i+1], "maximum precision for date time is nanosecond")
+				// go-toml allows decoding fractional seconds
+				// beyond the supported precision of 9
+				// digits. It truncates the fractional component
+				// to the supported precision and ignores the
+				// remaining digits.
+				//
+				// https://github.com/pelletier/go-toml/discussions/707
+				continue
 			}
 
 			frac *= 10
 			frac += int(c - '0')
-			digits++
+			precision++
 		}
 
-		t.Nanosecond = frac * nspow[digits]
-		t.Precision = digits
+		if precision == 0 {
+			return t, nil, newDecodeError(b[:1], "nanoseconds need at least one digit")
+		}
 
-		return t, b[9+digits:], nil
+		t.Nanosecond = frac * nspow[precision]
+		t.Precision = precision
+
+		return t, b[1+digits:], nil
 	}
-
-	return t, b[8:], nil
+	return t, b, nil
 }
 
 //nolint:cyclop
@@ -249,10 +313,10 @@ func parseFloat(b []byte) (float64, error) {
 	}
 
 	start := 0
-	if b[0] == '+' || b[0] == '-' {
+	if cleaned[0] == '+' || cleaned[0] == '-' {
 		start = 1
 	}
-	if b[start] == '0' && isDigit(b[start+1]) {
+	if cleaned[start] == '0' && isDigit(cleaned[start+1]) {
 		return 0, newDecodeError(b, "float integer part cannot have leading zeroes")
 	}
 
@@ -335,8 +399,17 @@ func parseIntDec(b []byte) (int64, error) {
 }
 
 func checkAndRemoveUnderscoresIntegers(b []byte) ([]byte, error) {
-	if b[0] == '_' {
-		return nil, newDecodeError(b[0:1], "number cannot start with underscore")
+	start := 0
+	if b[start] == '+' || b[start] == '-' {
+		start++
+	}
+
+	if len(b) == start {
+		return b, nil
+	}
+
+	if b[start] == '_' {
+		return nil, newDecodeError(b[start:start+1], "number cannot start with underscore")
 	}
 
 	if b[len(b)-1] == '_' {
@@ -405,6 +478,13 @@ func checkAndRemoveUnderscoresFloats(b []byte) ([]byte, error) {
 			if !before {
 				return nil, newDecodeError(b[i-1:i+1], "number must have at least one digit between underscores")
 			}
+			if i < len(b)-1 && (b[i+1] == 'e' || b[i+1] == 'E') {
+				return nil, newDecodeError(b[i+1:i+2], "cannot have underscore before exponent")
+			}
+			before = false
+		case '+', '-':
+			// signed exponents
+			cleaned = append(cleaned, c)
 			before = false
 		case 'e', 'E':
 			if i < len(b)-1 && b[i+1] == '_' {
@@ -430,7 +510,7 @@ func checkAndRemoveUnderscoresFloats(b []byte) ([]byte, error) {
 
 // isValidDate checks if a provided date is a date that exists.
 func isValidDate(year int, month int, day int) bool {
-	return day <= daysIn(month, year)
+	return month > 0 && month < 13 && day > 0 && day <= daysIn(month, year)
 }
 
 // daysBefore[m] counts the number of days in a non-leap year

errors.go

@@ -27,7 +27,7 @@ type DecodeError struct {
 // corresponding field in the target value. It contains all the missing fields
 // in Errors.
 //
-// Emitted by Decoder when SetStrict(true) was called.
+// Emitted by Decoder when DisallowUnknownFields() was called.
 type StrictMissingError struct {
 	// One error per field that could not be found.
 	Errors []DecodeError

errors_test.go

@@ -212,12 +212,12 @@ func ExampleDecodeError() {
 
 	fmt.Println(err)
 
-	//nolint:errorlint
-	de := err.(*DecodeError)
-	fmt.Println(de.String())
-
-	row, col := de.Position()
-	fmt.Println("error occurred at row", row, "column", col)
+	var derr *DecodeError
+	if errors.As(err, &derr) {
+		fmt.Println(derr.String())
+		row, col := derr.Position()
+		fmt.Println("error occurred at row", row, "column", col)
+	}
 	// Output:
 	// toml: number must have at least one digit between underscores
 	// 1| name = 123__456

fuzz_test.go

@@ -0,0 +1,56 @@
+//go:build go1.18
+// +build go1.18
+
+package toml_test
+
+import (
+	"io/ioutil"
+	"strings"
+	"testing"
+
+	"github.com/pelletier/go-toml/v2"
+	"github.com/stretchr/testify/require"
+)
+
+func FuzzUnmarshal(f *testing.F) {
+	file, err := ioutil.ReadFile("benchmark/benchmark.toml")
+	if err != nil {
+		panic(err)
+	}
+	f.Add(file)
+
+	f.Fuzz(func(t *testing.T, b []byte) {
+		if strings.Contains(string(b), "nan") {
+			// Current limitation of testify.
+			// https://github.com/stretchr/testify/issues/624
+			t.Skip("can't compare NaNs")
+		}
+
+		t.Log("INITIAL DOCUMENT ===========================")
+		t.Log(string(b))
+
+		var v interface{}
+		err := toml.Unmarshal(b, &v)
+		if err != nil {
+			return
+		}
+
+		t.Log("DECODED VALUE ===========================")
+		t.Logf("%#+v", v)
+
+		encoded, err := toml.Marshal(v)
+		if err != nil {
+			t.Fatalf("cannot marshal unmarshaled document: %s", err)
+		}
+
+		t.Log("ENCODED DOCUMENT ===========================")
+		t.Log(string(encoded))
+
+		var v2 interface{}
+		err = toml.Unmarshal(encoded, &v2)
+		if err != nil {
+			t.Fatalf("failed round trip: %s", err)
+		}
+		require.Equal(t, v, v2)
+	})
+}

go.mod

@@ -2,5 +2,4 @@ module github.com/pelletier/go-toml/v2
 
 go 1.16
 
-// latest (v1.7.0) doesn't have the fix for time.Time
-require github.com/stretchr/testify v1.7.1-0.20210427113832-6241f9ab9942
+require github.com/stretchr/testify v1.7.1

go.sum

@@ -3,8 +3,8 @@ github.com/davecgh/go-spew v1.1.0/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSs
 github.com/pmezard/go-difflib v1.0.0 h1:4DBwDE0NGyQoBHbLQYPwSUPoCMWR5BEzIk/f1lZbAQM=
 github.com/pmezard/go-difflib v1.0.0/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
 github.com/stretchr/objx v0.1.0/go.mod h1:HFkY916IF+rwdDfMAkV7OtwuqBVzrE8GR6GFx+wExME=
-github.com/stretchr/testify v1.7.1-0.20210427113832-6241f9ab9942 h1:t0lM6y/M5IiUZyvbBTcngso8SZEZICH7is9B6g/obVU=
-github.com/stretchr/testify v1.7.1-0.20210427113832-6241f9ab9942/go.mod h1:6Fq8oRcR53rry900zMqJjRRixrwX3KX962/h/Wwjteg=
+github.com/stretchr/testify v1.7.1 h1:5TQK59W5E3v0r2duFAb7P95B6hEeOyEnHRa8MjYSMTY=
+github.com/stretchr/testify v1.7.1/go.mod h1:6Fq8oRcR53rry900zMqJjRRixrwX3KX962/h/Wwjteg=
 gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405 h1:yhCVgyC4o1eVCa2tZl7eS0r+SDo693bJlVdllGtEeKM=
 gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
 gopkg.in/yaml.v3 v3.0.0-20200313102051-9f266ea9e77c h1:dUUwHk2QECo/6vqA44rthZ8ie2QXMNeKRTHCNY2nXvo=

internal/ast/ast.go

@@ -20,8 +20,8 @@ type Iterator struct {
 	node    *Node
 }
 
-// Next moves the iterator forward and returns true if points to a node, false
-// otherwise.
+// Next moves the iterator forward and returns true if points to a
+// node, false otherwise.
 func (c *Iterator) Next() bool {
 	if !c.started {
 		c.started = true
@@ -31,8 +31,8 @@ func (c *Iterator) Next() bool {
 	return c.node.Valid()
 }
 
-// IsLast returns true if the current node of the iterator is the last one.
-// Subsequent call to Next() will return false.
+// IsLast returns true if the current node of the iterator is the last
+// one.  Subsequent call to Next() will return false.
 func (c *Iterator) IsLast() bool {
 	return c.node.next == 0
 }
@@ -62,20 +62,20 @@ func (r *Root) at(idx Reference) *Node {
 	return &r.nodes[idx]
 }
 
-// Arrays have one child per element in the array.
-// InlineTables have one child per key-value pair in the table.
-// KeyValues have at least two children. The first one is the value. The
-// rest make a potentially dotted key.
-// Table and Array table have one child per element of the key they
-// represent (same as KeyValue, but without the last node being the value).
-// children []Node
+// Arrays have one child per element in the array.  InlineTables have
+// one child per key-value pair in the table.  KeyValues have at least
+// two children. The first one is the value. The rest make a
+// potentially dotted key.  Table and Array table have one child per
+// element of the key they represent (same as KeyValue, but without
+// the last node being the value).
 type Node struct {
 	Kind Kind
 	Raw  Range  // Raw bytes from the input.
-	Data []byte // Node value (could be either allocated or referencing the input).
+	Data []byte // Node value (either allocated or referencing the input).
 
-	// References to other nodes, as offsets in the backing array from this
-	// node. References can go backward, so those can be negative.
+	// References to other nodes, as offsets in the backing array
+	// from this node. References can go backward, so those can be
+	// negative.
 	next  int // 0 if last element
 	child int // 0 if no child
 }
@@ -85,8 +85,8 @@ type Range struct {
 	Length uint32
 }
 
-// Next returns a copy of the next node, or an invalid Node if there is no
-// next node.
+// Next returns a copy of the next node, or an invalid Node if there
+// is no next node.
 func (n *Node) Next() *Node {
 	if n.next == 0 {
 		return nil
@@ -96,9 +96,9 @@ func (n *Node) Next() *Node {
 	return (*Node)(danger.Stride(ptr, size, n.next))
 }
 
-// Child returns a copy of the first child node of this node. Other children
-// can be accessed calling Next on the first child.
-// Returns an invalid Node if there is none.
+// Child returns a copy of the first child node of this node. Other
+// children can be accessed calling Next on the first child.  Returns
+// an invalid Node if there is none.
 func (n *Node) Child() *Node {
 	if n.child == 0 {
 		return nil
@@ -113,10 +113,9 @@ func (n *Node) Valid() bool {
 	return n != nil
 }
 
-// Key returns the child nodes making the Key on a supported node. Panics
-// otherwise.
-// They are guaranteed to be all be of the Kind Key. A simple key would return
-// just one element.
+// Key returns the child nodes making the Key on a supported
+// node. Panics otherwise.  They are guaranteed to be all be of the
+// Kind Key. A simple key would return just one element.
 func (n *Node) Key() Iterator {
 	switch n.Kind {
 	case KeyValue:
@@ -133,10 +132,9 @@ func (n *Node) Key() Iterator {
 }
 
 // Value returns a pointer to the value node of a KeyValue.
-// Guaranteed to be non-nil.
-// Panics if not called on a KeyValue node, or if the Children are malformed.
+// Guaranteed to be non-nil.  Panics if not called on a KeyValue node,
+// or if the Children are malformed.
 func (n *Node) Value() *Node {
-	assertKind(KeyValue, *n)
 	return n.Child()
 }
 
@@ -144,9 +142,3 @@ func (n *Node) Value() *Node {
 func (n *Node) Children() Iterator {
 	return Iterator{node: n.Child()}
 }
-
-func assertKind(k Kind, n Node) {
-	if n.Kind != k {
-		panic(fmt.Errorf("method was expecting a %s, not a %s", k, n.Kind))
-	}
-}

internal/cli/cli.go

@@ -0,0 +1,88 @@
+package cli
+
+import (
+	"bytes"
+	"errors"
+	"flag"
+	"fmt"
+	"io"
+	"io/ioutil"
+	"os"
+
+	"github.com/pelletier/go-toml/v2"
+)
+
+type ConvertFn func(r io.Reader, w io.Writer) error
+
+type Program struct {
+	Usage string
+	Fn    ConvertFn
+	// Inplace allows the command to take more than one file as argument and
+	// perform convertion in place on each provided file.
+	Inplace bool
+}
+
+func (p *Program) Execute() {
+	flag.Usage = func() { fmt.Fprintf(os.Stderr, p.Usage) }
+	flag.Parse()
+	os.Exit(p.main(flag.Args(), os.Stdin, os.Stdout, os.Stderr))
+}
+
+func (p *Program) main(files []string, input io.Reader, output, error io.Writer) int {
+	err := p.run(files, input, output)
+	if err != nil {
+
+		var derr *toml.DecodeError
+		if errors.As(err, &derr) {
+			fmt.Fprintln(error, derr.String())
+			row, col := derr.Position()
+			fmt.Fprintln(error, "error occurred at row", row, "column", col)
+		} else {
+			fmt.Fprintln(error, err.Error())
+		}
+
+		return -1
+	}
+	return 0
+}
+
+func (p *Program) run(files []string, input io.Reader, output io.Writer) error {
+	if len(files) > 0 {
+		if p.Inplace {
+			return p.runAllFilesInPlace(files)
+		}
+		f, err := os.Open(files[0])
+		if err != nil {
+			return err
+		}
+		defer f.Close()
+		input = f
+	}
+	return p.Fn(input, output)
+}
+
+func (p *Program) runAllFilesInPlace(files []string) error {
+	for _, path := range files {
+		err := p.runFileInPlace(path)
+		if err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+func (p *Program) runFileInPlace(path string) error {
+	in, err := ioutil.ReadFile(path)
+	if err != nil {
+		return err
+	}
+
+	out := new(bytes.Buffer)
+
+	err = p.Fn(bytes.NewReader(in), out)
+	if err != nil {
+		return err
+	}
+
+	return ioutil.WriteFile(path, out.Bytes(), 0600)
+}

internal/cli/cli_test.go

@@ -0,0 +1,172 @@
+package cli
+
+import (
+	"bytes"
+	"fmt"
+	"io"
+	"io/ioutil"
+	"os"
+	"path"
+	"strings"
+	"testing"
+
+	"github.com/pelletier/go-toml/v2"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func processMain(args []string, input io.Reader, stdout, stderr io.Writer, f ConvertFn) int {
+	p := Program{Fn: f}
+	return p.main(args, input, stdout, stderr)
+}
+
+func TestProcessMainStdin(t *testing.T) {
+	stdout := new(bytes.Buffer)
+	stderr := new(bytes.Buffer)
+	input := strings.NewReader("this is the input")
+
+	exit := processMain([]string{}, input, stdout, stderr, func(r io.Reader, w io.Writer) error {
+		return nil
+	})
+
+	assert.Equal(t, 0, exit)
+	assert.Empty(t, stdout.String())
+	assert.Empty(t, stderr.String())
+}
+
+func TestProcessMainStdinErr(t *testing.T) {
+	stdout := new(bytes.Buffer)
+	stderr := new(bytes.Buffer)
+	input := strings.NewReader("this is the input")
+
+	exit := processMain([]string{}, input, stdout, stderr, func(r io.Reader, w io.Writer) error {
+		return fmt.Errorf("something bad")
+	})
+
+	assert.Equal(t, -1, exit)
+	assert.Empty(t, stdout.String())
+	assert.NotEmpty(t, stderr.String())
+}
+
+func TestProcessMainStdinDecodeErr(t *testing.T) {
+	stdout := new(bytes.Buffer)
+	stderr := new(bytes.Buffer)
+	input := strings.NewReader("this is the input")
+
+	exit := processMain([]string{}, input, stdout, stderr, func(r io.Reader, w io.Writer) error {
+		var v interface{}
+		return toml.Unmarshal([]byte(`qwe = 001`), &v)
+	})
+
+	assert.Equal(t, -1, exit)
+	assert.Empty(t, stdout.String())
+	assert.Contains(t, stderr.String(), "error occurred at")
+}
+
+func TestProcessMainFileExists(t *testing.T) {
+	tmpfile, err := ioutil.TempFile("", "example")
+	require.NoError(t, err)
+	defer os.Remove(tmpfile.Name())
+	_, err = tmpfile.Write([]byte(`some data`))
+	require.NoError(t, err)
+
+	stdout := new(bytes.Buffer)
+	stderr := new(bytes.Buffer)
+
+	exit := processMain([]string{tmpfile.Name()}, nil, stdout, stderr, func(r io.Reader, w io.Writer) error {
+		return nil
+	})
+
+	assert.Equal(t, 0, exit)
+	assert.Empty(t, stdout.String())
+	assert.Empty(t, stderr.String())
+}
+
+func TestProcessMainFileDoesNotExist(t *testing.T) {
+	stdout := new(bytes.Buffer)
+	stderr := new(bytes.Buffer)
+
+	exit := processMain([]string{"/lets/hope/this/does/not/exist"}, nil, stdout, stderr, func(r io.Reader, w io.Writer) error {
+		return nil
+	})
+
+	assert.Equal(t, -1, exit)
+	assert.Empty(t, stdout.String())
+	assert.NotEmpty(t, stderr.String())
+}
+
+func TestProcessMainFilesInPlace(t *testing.T) {
+	dir, err := ioutil.TempDir("", "")
+	require.NoError(t, err)
+	defer os.RemoveAll(dir)
+
+	path1 := path.Join(dir, "file1")
+	path2 := path.Join(dir, "file2")
+
+	err = ioutil.WriteFile(path1, []byte("content 1"), 0600)
+	require.NoError(t, err)
+	err = ioutil.WriteFile(path2, []byte("content 2"), 0600)
+	require.NoError(t, err)
+
+	p := Program{
+		Fn:      dummyFileFn,
+		Inplace: true,
+	}
+
+	exit := p.main([]string{path1, path2}, os.Stdin, os.Stdout, os.Stderr)
+
+	require.Equal(t, 0, exit)
+
+	v1, err := ioutil.ReadFile(path1)
+	require.NoError(t, err)
+	require.Equal(t, "1", string(v1))
+
+	v2, err := ioutil.ReadFile(path2)
+	require.NoError(t, err)
+	require.Equal(t, "2", string(v2))
+}
+
+func TestProcessMainFilesInPlaceErrRead(t *testing.T) {
+	p := Program{
+		Fn:      dummyFileFn,
+		Inplace: true,
+	}
+
+	exit := p.main([]string{"/this/path/is/invalid"}, os.Stdin, os.Stdout, os.Stderr)
+
+	require.Equal(t, -1, exit)
+}
+
+func TestProcessMainFilesInPlaceFailFn(t *testing.T) {
+	dir, err := ioutil.TempDir("", "")
+	require.NoError(t, err)
+	defer os.RemoveAll(dir)
+
+	path1 := path.Join(dir, "file1")
+
+	err = ioutil.WriteFile(path1, []byte("content 1"), 0600)
+	require.NoError(t, err)
+
+	p := Program{
+		Fn:      func(io.Reader, io.Writer) error { return fmt.Errorf("oh no") },
+		Inplace: true,
+	}
+
+	exit := p.main([]string{path1}, os.Stdin, os.Stdout, os.Stderr)
+
+	require.Equal(t, -1, exit)
+
+	v1, err := ioutil.ReadFile(path1)
+	require.NoError(t, err)
+	require.Equal(t, "content 1", string(v1))
+}
+
+func dummyFileFn(r io.Reader, w io.Writer) error {
+	b, err := ioutil.ReadAll(r)
+	if err != nil {
+		return err
+	}
+	v := strings.SplitN(string(b), " ", 2)[1]
+	_, err = w.Write([]byte(v))
+	return err
+}

internal/danger/typeid.go

@@ -0,0 +1,23 @@
+package danger
+
+import (
+	"reflect"
+	"unsafe"
+)
+
+// typeID is used as key in encoder and decoder caches to enable using
+// the optimize runtime.mapaccess2_fast64 function instead of the more
+// expensive lookup if we were to use reflect.Type as map key.
+//
+// typeID holds the pointer to the reflect.Type value, which is unique
+// in the program.
+//
+// https://github.com/segmentio/encoding/blob/master/json/codec.go#L59-L61
+type TypeID unsafe.Pointer
+
+func MakeTypeID(t reflect.Type) TypeID {
+	// reflect.Type has the fields:
+	// typ unsafe.Pointer
+	// ptr unsafe.Pointer
+	return TypeID((*[2]unsafe.Pointer)(unsafe.Pointer(&t))[1])
+}

internal/imported_tests/unmarshal_imported_test.go

@@ -457,35 +457,6 @@ func TestEmptytomlUnmarshal(t *testing.T) {
 	assert.Equal(t, emptyTestData, result)
 }
 
-func TestEmptyUnmarshalOmit(t *testing.T) {
-	t.Skipf("Have not figured yet if omitempty is a good idea")
-
-	type emptyMarshalTestStruct2 struct {
-		Title      string                  `toml:"title"`
-		Bool       bool                    `toml:"bool,omitempty"`
-		Int        int                     `toml:"int, omitempty"`
-		String     string                  `toml:"string,omitempty "`
-		StringList []string                `toml:"stringlist,omitempty"`
-		Ptr        *basicMarshalTestStruct `toml:"ptr,omitempty"`
-		Map        map[string]string       `toml:"map,omitempty"`
-	}
-
-	emptyTestData2 := emptyMarshalTestStruct2{
-		Title:      "Placeholder",
-		Bool:       false,
-		Int:        0,
-		String:     "",
-		StringList: []string{},
-		Ptr:        nil,
-		Map:        map[string]string{},
-	}
-
-	result := emptyMarshalTestStruct2{}
-	err := toml.Unmarshal(emptyTestToml, &result)
-	require.NoError(t, err)
-	assert.Equal(t, emptyTestData2, result)
-}
-
 type pointerMarshalTestStruct struct {
 	Str       *string
 	List      *[]string
@@ -956,6 +927,29 @@ func TestUnmarshalMapWithTypedKey(t *testing.T) {
 	}
 }
 
+func TestUnmarshalTypeTableHeader(t *testing.T) {
+	testToml := []byte(`
+		[test]
+		a = 1
+		`)
+
+	type header string
+	var result map[header]map[string]int
+	err := toml.Unmarshal(testToml, &result)
+	if err != nil {
+		t.Errorf("Received unexpected error: %s", err)
+		return
+	}
+
+	expected := map[header]map[string]int{
+		"test": map[string]int{"a": 1},
+	}
+
+	if !reflect.DeepEqual(result, expected) {
+		t.Errorf("Bad unmarshal: expected %v, got %v", expected, result)
+	}
+}
+
 func TestUnmarshalNonPointer(t *testing.T) {
 	a := 1
 	err := toml.Unmarshal([]byte{}, a)
@@ -1954,7 +1948,7 @@ func decoder(doc string) *toml.Decoder {
 
 func strictDecoder(doc string) *toml.Decoder {
 	d := decoder(doc)
-	d.SetStrict(true)
+	d.DisallowUnknownFields()
 	return d
 }
 

internal/testsuite/testsuite.go

@@ -43,7 +43,7 @@ func DecodeStdin() error {
 	j := json.NewEncoder(os.Stdout)
 	j.SetIndent("", "  ")
 	if err := j.Encode(addTag("", decoded)); err != nil {
-		fmt.Errorf("Error encoding JSON: %s", err)
+		return fmt.Errorf("Error encoding JSON: %s", err)
 	}
 
 	return nil

internal/tracker/seen.go

@@ -3,6 +3,7 @@ package tracker
 import (
 	"bytes"
 	"fmt"
+	"sync"
 
 	"github.com/pelletier/go-toml/v2/internal/ast"
 )
@@ -54,63 +55,104 @@ func (k keyKind) String() string {
 type SeenTracker struct {
 	entries    []entry
 	currentIdx int
-	nextID     int
+}
+
+var pool sync.Pool
+
+func (s *SeenTracker) reset() {
+	// Always contains a root element at index 0.
+	s.currentIdx = 0
+	if len(s.entries) == 0 {
+		s.entries = make([]entry, 1, 2)
+	} else {
+		s.entries = s.entries[:1]
+	}
+	s.entries[0].child = -1
+	s.entries[0].next = -1
 }
 
 type entry struct {
-	id       int
-	parent   int
+	// Use -1 to indicate no child or no sibling.
+	child int
+	next  int
+
 	name     []byte
 	kind     keyKind
 	explicit bool
+	kv       bool
 }
 
-// Remove all descendent of node at position idx.
-func (s *SeenTracker) clear(idx int) {
-	p := s.entries[idx].id
-	rest := clear(p, s.entries[idx+1:])
-	s.entries = s.entries[:idx+1+len(rest)]
-}
-
-func clear(parentID int, entries []entry) []entry {
-	for i := 0; i < len(entries); {
-		if entries[i].parent == parentID {
-			id := entries[i].id
-			copy(entries[i:], entries[i+1:])
-			entries = entries[:len(entries)-1]
-			rest := clear(id, entries[i:])
-			entries = entries[:i+len(rest)]
-		} else {
-			i++
+// Find the index of the child of parentIdx with key k. Returns -1 if
+// it does not exist.
+func (s *SeenTracker) find(parentIdx int, k []byte) int {
+	for i := s.entries[parentIdx].child; i >= 0; i = s.entries[i].next {
+		if bytes.Equal(s.entries[i].name, k) {
+			return i
 		}
 	}
-	return entries
+	return -1
 }
 
-func (s *SeenTracker) create(parentIdx int, name []byte, kind keyKind, explicit bool) int {
-	parentID := s.id(parentIdx)
+// Remove all descendants of node at position idx.
+func (s *SeenTracker) clear(idx int) {
+	if idx >= len(s.entries) {
+		return
+	}
+
+	for i := s.entries[idx].child; i >= 0; {
+		next := s.entries[i].next
+		n := s.entries[0].next
+		s.entries[0].next = i
+		s.entries[i].next = n
+		s.entries[i].name = nil
+		s.clear(i)
+		i = next
+	}
+
+	s.entries[idx].child = -1
+}
+
+func (s *SeenTracker) create(parentIdx int, name []byte, kind keyKind, explicit bool, kv bool) int {
+	e := entry{
+		child: -1,
+		next:  s.entries[parentIdx].child,
 
-	idx := len(s.entries)
-	s.entries = append(s.entries, entry{
-		id:       s.nextID,
-		parent:   parentID,
 		name:     name,
 		kind:     kind,
 		explicit: explicit,
-	})
-	s.nextID++
+		kv:       kv,
+	}
+	var idx int
+	if s.entries[0].next >= 0 {
+		idx = s.entries[0].next
+		s.entries[0].next = s.entries[idx].next
+		s.entries[idx] = e
+	} else {
+		idx = len(s.entries)
+		s.entries = append(s.entries, e)
+	}
+
+	s.entries[parentIdx].child = idx
+
 	return idx
 }
 
-// CheckExpression takes a top-level node and checks that it does not contain keys
-// that have been seen in previous calls, and validates that types are consistent.
+func (s *SeenTracker) setExplicitFlag(parentIdx int) {
+	for i := s.entries[parentIdx].child; i >= 0; i = s.entries[i].next {
+		if s.entries[i].kv {
+			s.entries[i].explicit = true
+			s.entries[i].kv = false
+		}
+		s.setExplicitFlag(i)
+	}
+}
+
+// CheckExpression takes a top-level node and checks that it does not contain
+// keys that have been seen in previous calls, and validates that types are
+// consistent.
 func (s *SeenTracker) CheckExpression(node *ast.Node) error {
 	if s.entries == nil {
-		// Skip ID = 0 to remove the confusion between nodes whose parent has
-		// id 0 and root nodes (parent id is 0 because it's the zero value).
-		s.nextID = 1
-		// Start unscoped, so idx is negative.
-		s.currentIdx = -1
+		s.reset()
 	}
 	switch node.Kind {
 	case ast.KeyValue:
@@ -125,9 +167,13 @@ func (s *SeenTracker) CheckExpression(node *ast.Node) error {
 }
 
 func (s *SeenTracker) checkTable(node *ast.Node) error {
+	if s.currentIdx >= 0 {
+		s.setExplicitFlag(s.currentIdx)
+	}
+
 	it := node.Key()
 
-	parentIdx := -1
+	parentIdx := 0
 
 	// This code is duplicated in checkArrayTable. This is because factoring
 	// it in a function requires to copy the iterator, or allocate it to the
@@ -142,7 +188,12 @@ func (s *SeenTracker) checkTable(node *ast.Node) error {
 		idx := s.find(parentIdx, k)
 
 		if idx < 0 {
-			idx = s.create(parentIdx, k, tableKind, false)
+			idx = s.create(parentIdx, k, tableKind, false, false)
+		} else {
+			entry := s.entries[idx]
+			if entry.kind == valueKind {
+				return fmt.Errorf("toml: expected %s to be a table, not a %s", string(k), entry.kind)
+			}
 		}
 		parentIdx = idx
 	}
@@ -160,7 +211,7 @@ func (s *SeenTracker) checkTable(node *ast.Node) error {
 		}
 		s.entries[idx].explicit = true
 	} else {
-		idx = s.create(parentIdx, k, tableKind, true)
+		idx = s.create(parentIdx, k, tableKind, true, false)
 	}
 
 	s.currentIdx = idx
@@ -169,9 +220,13 @@ func (s *SeenTracker) checkTable(node *ast.Node) error {
 }
 
 func (s *SeenTracker) checkArrayTable(node *ast.Node) error {
+	if s.currentIdx >= 0 {
+		s.setExplicitFlag(s.currentIdx)
+	}
+
 	it := node.Key()
 
-	parentIdx := -1
+	parentIdx := 0
 
 	for it.Next() {
 		if it.IsLast() {
@@ -183,8 +238,14 @@ func (s *SeenTracker) checkArrayTable(node *ast.Node) error {
 		idx := s.find(parentIdx, k)
 
 		if idx < 0 {
-			idx = s.create(parentIdx, k, tableKind, false)
+			idx = s.create(parentIdx, k, tableKind, false, false)
+		} else {
+			entry := s.entries[idx]
+			if entry.kind == valueKind {
+				return fmt.Errorf("toml: expected %s to be a table, not a %s", string(k), entry.kind)
+			}
 		}
+
 		parentIdx = idx
 	}
 
@@ -198,7 +259,7 @@ func (s *SeenTracker) checkArrayTable(node *ast.Node) error {
 		}
 		s.clear(idx)
 	} else {
-		idx = s.create(parentIdx, k, arrayTableKind, true)
+		idx = s.create(parentIdx, k, arrayTableKind, true, false)
 	}
 
 	s.currentIdx = idx
@@ -207,53 +268,89 @@ func (s *SeenTracker) checkArrayTable(node *ast.Node) error {
 }
 
 func (s *SeenTracker) checkKeyValue(node *ast.Node) error {
-	it := node.Key()
-
 	parentIdx := s.currentIdx
+	it := node.Key()
 
 	for it.Next() {
 		k := it.Node().Data
 
 		idx := s.find(parentIdx, k)
 
-		if idx >= 0 {
-			if s.entries[idx].kind != tableKind {
-				return fmt.Errorf("toml: expected %s to be a table, not a %s", string(k), s.entries[idx].kind)
-			}
-			if s.entries[idx].explicit {
+		if idx < 0 {
+			idx = s.create(parentIdx, k, tableKind, false, true)
+		} else {
+			entry := s.entries[idx]
+			if it.IsLast() {
+				return fmt.Errorf("toml: key %s is already defined", string(k))
+			} else if entry.kind != tableKind {
+				return fmt.Errorf("toml: expected %s to be a table, not a %s", string(k), entry.kind)
+			} else if entry.explicit {
 				return fmt.Errorf("toml: cannot redefine table %s that has already been explicitly defined", string(k))
 			}
-		} else {
-			idx = s.create(parentIdx, k, tableKind, false)
 		}
+
 		parentIdx = idx
 	}
 
-	kind := valueKind
+	s.entries[parentIdx].kind = valueKind
 
-	if node.Value().Kind == ast.InlineTable {
-		kind = tableKind
+	value := node.Value()
+
+	switch value.Kind {
+	case ast.InlineTable:
+		return s.checkInlineTable(value)
+	case ast.Array:
+		return s.checkArray(value)
 	}
-	s.entries[parentIdx].kind = kind
 
 	return nil
 }
 
-func (s *SeenTracker) id(idx int) int {
-	if idx >= 0 {
-		return s.entries[idx].id
+func (s *SeenTracker) checkArray(node *ast.Node) error {
+	it := node.Children()
+	for it.Next() {
+		n := it.Node()
+		switch n.Kind {
+		case ast.InlineTable:
+			err := s.checkInlineTable(n)
+			if err != nil {
+				return err
+			}
+		case ast.Array:
+			err := s.checkArray(n)
+			if err != nil {
+				return err
+			}
+		}
 	}
-	return 0
+	return nil
 }
 
-func (s *SeenTracker) find(parentIdx int, k []byte) int {
-	parentID := s.id(parentIdx)
-
-	for i := parentIdx + 1; i < len(s.entries); i++ {
-		if s.entries[i].parent == parentID && bytes.Equal(s.entries[i].name, k) {
-			return i
+func (s *SeenTracker) checkInlineTable(node *ast.Node) error {
+	if pool.New == nil {
+		pool.New = func() interface{} {
+			return &SeenTracker{}
 		}
 	}
 
-	return -1
+	s = pool.Get().(*SeenTracker)
+	s.reset()
+
+	it := node.Children()
+	for it.Next() {
+		n := it.Node()
+		err := s.checkKeyValue(n)
+		if err != nil {
+			return err
+		}
+	}
+
+	// As inline tables are self-contained, the tracker does not
+	// need to retain the details of what they contain. The
+	// keyValue element that creates the inline table is kept to
+	// mark the presence of the inline table and prevent
+	// redefinition of its keys: check* functions cannot walk into
+	// a value.
+	pool.Put(s)
+	return nil
 }

internal/tracker/seen_test.go

@@ -0,0 +1,15 @@
+package tracker
+
+import (
+	"testing"
+	"unsafe"
+
+	"github.com/stretchr/testify/require"
+)
+
+func TestEntrySize(t *testing.T) {
+	// Validate no regression on the size of entry{}. This is a critical bit for
+	// performance of unmarshaling documents. Should only be increased with care
+	// and a very good reason.
+	require.LessOrEqual(t, 48, int(unsafe.Sizeof(entry{})))
+}

marshaler.go

@@ -11,6 +11,7 @@ import (
 	"strconv"
 	"strings"
 	"time"
+	"unicode"
 )
 
 // Marshal serializes a Go value as a TOML document.
@@ -54,8 +55,9 @@ func NewEncoder(w io.Writer) *Encoder {
 // inline tag:
 //
 //   MyField `inline:"true"`
-func (enc *Encoder) SetTablesInline(inline bool) {
+func (enc *Encoder) SetTablesInline(inline bool) *Encoder {
 	enc.tablesInline = inline
+	return enc
 }
 
 // SetArraysMultiline forces the encoder to emit all arrays with one element per
@@ -64,20 +66,23 @@ func (enc *Encoder) SetTablesInline(inline bool) {
 // This behavior can be controlled on an individual struct field basis with the multiline tag:
 //
 //   MyField `multiline:"true"`
-func (enc *Encoder) SetArraysMultiline(multiline bool) {
+func (enc *Encoder) SetArraysMultiline(multiline bool) *Encoder {
 	enc.arraysMultiline = multiline
+	return enc
 }
 
 // SetIndentSymbol defines the string that should be used for indentation. The
 // provided string is repeated for each indentation level. Defaults to two
 // spaces.
-func (enc *Encoder) SetIndentSymbol(s string) {
+func (enc *Encoder) SetIndentSymbol(s string) *Encoder {
 	enc.indentSymbol = s
+	return enc
 }
 
 // SetIndentTables forces the encoder to intent tables and array tables.
-func (enc *Encoder) SetIndentTables(indent bool) {
+func (enc *Encoder) SetIndentTables(indent bool) *Encoder {
 	enc.indentTables = indent
+	return enc
 }
 
 // Encode writes a TOML representation of v to the stream.
@@ -99,27 +104,31 @@ func (enc *Encoder) SetIndentTables(indent bool) {
 // Intermediate tables are always printed.
 //
 // By default, strings are encoded as literal string, unless they contain either
-// a newline character or a single quote. In that case they are emitted as quoted
-// strings.
+// a newline character or a single quote. In that case they are emitted as
+// quoted strings.
 //
 // When encoding structs, fields are encoded in order of definition, with their
 // exact name.
 //
 // Struct tags
 //
-// The following struct tags are available to tweak encoding on a per-field
-// basis:
+// The encoding of each public struct field can be customized by the format
+// string in the "toml" key of the struct field's tag. This follows
+// encoding/json's convention. The format string starts with the name of the
+// field, optionally followed by a comma-separated list of options. The name may
+// be empty in order to provide options without overriding the default name.
 //
-//   toml:"foo"
-//     Changes the name of the key to use for the field to foo.
+// The "multiline" option emits strings as quoted multi-line TOML strings. It
+// has no effect on fields that would not be encoded as strings.
 //
-//   multiline:"true"
-//     When the field contains a string, it will be emitted as a quoted
-//     multi-line TOML string.
+// The "inline" option turns fields that would be emitted as tables into inline
+// tables instead. It has no effect on other fields.
 //
-//   inline:"true"
-//     When the field would normally be encoded as a table, it is instead
-//     encoded as an inline table.
+// The "omitempty" option prevents empty values or groups from being emitted.
+//
+// In addition to the "toml" tag struct tag, a "comment" tag can be used to emit
+// a TOML comment before the value being annotated. Comments are ignored inside
+// inline tables.
 func (enc *Encoder) Encode(v interface{}) error {
 	var (
 		b   []byte
@@ -147,6 +156,8 @@ func (enc *Encoder) Encode(v interface{}) error {
 
 type valueOptions struct {
 	multiline bool
+	omitempty bool
+	comment   string
 }
 
 type encoderCtx struct {
@@ -196,16 +207,29 @@ func (ctx *encoderCtx) isRoot() bool {
 	return len(ctx.parentKey) == 0 && !ctx.hasKey
 }
 
-//nolint:cyclop,funlen
 func (enc *Encoder) encode(b []byte, ctx encoderCtx, v reflect.Value) ([]byte, error) {
-	if !v.IsZero() {
-		i, ok := v.Interface().(time.Time)
-		if ok {
-			return i.AppendFormat(b, time.RFC3339), nil
+	i := v.Interface()
+
+	switch x := i.(type) {
+	case time.Time:
+		if x.Nanosecond() > 0 {
+			return x.AppendFormat(b, time.RFC3339Nano), nil
 		}
+		return x.AppendFormat(b, time.RFC3339), nil
+	case LocalTime:
+		return append(b, x.String()...), nil
+	case LocalDate:
+		return append(b, x.String()...), nil
+	case LocalDateTime:
+		return append(b, x.String()...), nil
 	}
 
-	if v.Type().Implements(textMarshalerType) {
+	hasTextMarshaler := v.Type().Implements(textMarshalerType)
+	if hasTextMarshaler || (v.CanAddr() && reflect.PtrTo(v.Type()).Implements(textMarshalerType)) {
+		if !hasTextMarshaler {
+			v = v.Addr()
+		}
+
 		if ctx.isRoot() {
 			return nil, fmt.Errorf("toml: type %s implementing the TextMarshaler interface cannot be a root element", v.Type())
 		}
@@ -245,16 +269,31 @@ func (enc *Encoder) encode(b []byte, ctx encoderCtx, v reflect.Value) ([]byte, e
 	case reflect.String:
 		b = enc.encodeString(b, v.String(), ctx.options)
 	case reflect.Float32:
-		if math.Trunc(v.Float()) == v.Float() {
-			b = strconv.AppendFloat(b, v.Float(), 'f', 1, 32)
+		f := v.Float()
+
+		if math.IsNaN(f) {
+			b = append(b, "nan"...)
+		} else if f > math.MaxFloat32 {
+			b = append(b, "inf"...)
+		} else if f < -math.MaxFloat32 {
+			b = append(b, "-inf"...)
+		} else if math.Trunc(f) == f {
+			b = strconv.AppendFloat(b, f, 'f', 1, 32)
 		} else {
-			b = strconv.AppendFloat(b, v.Float(), 'f', -1, 32)
+			b = strconv.AppendFloat(b, f, 'f', -1, 32)
 		}
 	case reflect.Float64:
-		if math.Trunc(v.Float()) == v.Float() {
-			b = strconv.AppendFloat(b, v.Float(), 'f', 1, 64)
+		f := v.Float()
+		if math.IsNaN(f) {
+			b = append(b, "nan"...)
+		} else if f > math.MaxFloat64 {
+			b = append(b, "inf"...)
+		} else if f < -math.MaxFloat64 {
+			b = append(b, "-inf"...)
+		} else if math.Trunc(f) == f {
+			b = strconv.AppendFloat(b, f, 'f', 1, 64)
 		} else {
-			b = strconv.AppendFloat(b, v.Float(), 'f', -1, 64)
+			b = strconv.AppendFloat(b, f, 'f', -1, 64)
 		}
 	case reflect.Bool:
 		if v.Bool() {
@@ -285,16 +324,16 @@ func isNil(v reflect.Value) bool {
 func (enc *Encoder) encodeKv(b []byte, ctx encoderCtx, options valueOptions, v reflect.Value) ([]byte, error) {
 	var err error
 
-	if !ctx.hasKey {
-		panic("caller of encodeKv should have set the key in the context")
+	if (ctx.options.omitempty || options.omitempty) && isEmptyValue(v) {
+		return b, nil
 	}
+
+	if !ctx.inline {
+		b = enc.encodeComment(ctx.indent, options.comment, b)
+	}
+
 	b = enc.indent(ctx.indent, b)
-
-	b, err = enc.encodeKey(b, ctx.key)
-	if err != nil {
-		return nil, err
-	}
-
+	b = enc.encodeKey(b, ctx.key)
 	b = append(b, " = "...)
 
 	// create a copy of the context because the value of a KV shouldn't
@@ -312,6 +351,24 @@ func (enc *Encoder) encodeKv(b []byte, ctx encoderCtx, options valueOptions, v r
 	return b, nil
 }
 
+func isEmptyValue(v reflect.Value) bool {
+	switch v.Kind() {
+	case reflect.Array, reflect.Map, reflect.Slice, reflect.String:
+		return v.Len() == 0
+	case reflect.Bool:
+		return !v.Bool()
+	case reflect.Int, reflect.Int8, reflect.Int16, reflect.Int32, reflect.Int64:
+		return v.Int() == 0
+	case reflect.Uint, reflect.Uint8, reflect.Uint16, reflect.Uint32, reflect.Uint64, reflect.Uintptr:
+		return v.Uint() == 0
+	case reflect.Float32, reflect.Float64:
+		return v.Float() == 0
+	case reflect.Interface, reflect.Ptr:
+		return v.IsNil()
+	}
+	return false
+}
+
 const literalQuote = '\''
 
 func (enc *Encoder) encodeString(b []byte, v string, options valueOptions) []byte {
@@ -323,7 +380,13 @@ func (enc *Encoder) encodeString(b []byte, v string, options valueOptions) []byt
 }
 
 func needsQuoting(v string) bool {
-	return strings.ContainsAny(v, "'\b\f\n\r\t")
+	// TODO: vectorize
+	for _, b := range []byte(v) {
+		if b == '\'' || b == '\r' || b == '\n' || invalidAscii(b) {
+			return true
+		}
+	}
+	return false
 }
 
 // caller should have checked that the string does not contain new lines or ' .
@@ -395,7 +458,7 @@ func (enc *Encoder) encodeQuotedString(multiline bool, b []byte, v string) []byt
 	return b
 }
 
-// called should have checked that the string is in A-Z / a-z / 0-9 / - / _ .
+// caller should have checked that the string is in A-Z / a-z / 0-9 / - / _ .
 func (enc *Encoder) encodeUnquotedKey(b []byte, v string) []byte {
 	return append(b, v...)
 }
@@ -405,24 +468,17 @@ func (enc *Encoder) encodeTableHeader(ctx encoderCtx, b []byte) ([]byte, error)
 		return b, nil
 	}
 
+	b = enc.encodeComment(ctx.indent, ctx.options.comment, b)
+
 	b = enc.indent(ctx.indent, b)
 
 	b = append(b, '[')
 
-	var err error
-
-	b, err = enc.encodeKey(b, ctx.parentKey[0])
-	if err != nil {
-		return nil, err
-	}
+	b = enc.encodeKey(b, ctx.parentKey[0])
 
 	for _, k := range ctx.parentKey[1:] {
 		b = append(b, '.')
-
-		b, err = enc.encodeKey(b, k)
-		if err != nil {
-			return nil, err
-		}
+		b = enc.encodeKey(b, k)
 	}
 
 	b = append(b, "]\n"...)
@@ -431,19 +487,19 @@ func (enc *Encoder) encodeTableHeader(ctx encoderCtx, b []byte) ([]byte, error)
 }
 
 //nolint:cyclop
-func (enc *Encoder) encodeKey(b []byte, k string) ([]byte, error) {
+func (enc *Encoder) encodeKey(b []byte, k string) []byte {
 	needsQuotation := false
 	cannotUseLiteral := false
 
+	if len(k) == 0 {
+		return append(b, "''"...)
+	}
+
 	for _, c := range k {
 		if (c >= 'A' && c <= 'Z') || (c >= 'a' && c <= 'z') || (c >= '0' && c <= '9') || c == '-' || c == '_' {
 			continue
 		}
 
-		if c == '\n' {
-			return nil, fmt.Errorf("toml: new line characters in keys are not supported")
-		}
-
 		if c == literalQuote {
 			cannotUseLiteral = true
 		}
@@ -451,13 +507,17 @@ func (enc *Encoder) encodeKey(b []byte, k string) ([]byte, error) {
 		needsQuotation = true
 	}
 
+	if needsQuotation && needsQuoting(k) {
+		cannotUseLiteral = true
+	}
+
 	switch {
 	case cannotUseLiteral:
-		return enc.encodeQuotedString(false, b, k), nil
+		return enc.encodeQuotedString(false, b, k)
 	case needsQuotation:
-		return enc.encodeLiteralString(b, k), nil
+		return enc.encodeLiteralString(b, k)
 	default:
-		return enc.encodeUnquotedKey(b, k), nil
+		return enc.encodeUnquotedKey(b, k)
 	}
 }
 
@@ -511,18 +571,26 @@ type table struct {
 }
 
 func (t *table) pushKV(k string, v reflect.Value, options valueOptions) {
+	for _, e := range t.kvs {
+		if e.Key == k {
+			return
+		}
+	}
+
 	t.kvs = append(t.kvs, entry{Key: k, Value: v, Options: options})
 }
 
 func (t *table) pushTable(k string, v reflect.Value, options valueOptions) {
+	for _, e := range t.tables {
+		if e.Key == k {
+			return
+		}
+	}
 	t.tables = append(t.tables, entry{Key: k, Value: v, Options: options})
 }
 
-func (enc *Encoder) encodeStruct(b []byte, ctx encoderCtx, v reflect.Value) ([]byte, error) {
-	var t table
-
-	//nolint:godox
-	// TODO: cache this?
+func walkStruct(ctx encoderCtx, t *table, v reflect.Value) {
+	// TODO: cache this
 	typ := v.Type()
 	for i := 0; i < typ.NumField(); i++ {
 		fieldType := typ.Field(i)
@@ -532,45 +600,121 @@ func (enc *Encoder) encodeStruct(b []byte, ctx encoderCtx, v reflect.Value) ([]b
 			continue
 		}
 
-		k, ok := fieldType.Tag.Lookup("toml")
-		if !ok {
-			k = fieldType.Name
-		}
+		tag := fieldType.Tag.Get("toml")
 
 		// special field name to skip field
-		if k == "-" {
+		if tag == "-" {
 			continue
 		}
 
+		k, opts := parseTag(tag)
+		if !isValidName(k) {
+			k = ""
+		}
+
 		f := v.Field(i)
 
+		if k == "" {
+			if fieldType.Anonymous {
+				if fieldType.Type.Kind() == reflect.Struct {
+					walkStruct(ctx, t, f)
+				}
+				continue
+			} else {
+				k = fieldType.Name
+			}
+		}
+
 		if isNil(f) {
 			continue
 		}
 
 		options := valueOptions{
-			multiline: fieldBoolTag(fieldType, "multiline"),
+			multiline: opts.multiline,
+			omitempty: opts.omitempty,
+			comment:   fieldType.Tag.Get("comment"),
 		}
 
-		inline := fieldBoolTag(fieldType, "inline")
-
-		if inline || !willConvertToTableOrArrayTable(ctx, f) {
+		if opts.inline || !willConvertToTableOrArrayTable(ctx, f) {
 			t.pushKV(k, f, options)
 		} else {
 			t.pushTable(k, f, options)
 		}
 	}
+}
+
+func (enc *Encoder) encodeStruct(b []byte, ctx encoderCtx, v reflect.Value) ([]byte, error) {
+	var t table
+
+	walkStruct(ctx, &t, v)
 
 	return enc.encodeTable(b, ctx, t)
 }
 
-func fieldBoolTag(field reflect.StructField, tag string) bool {
-	x, ok := field.Tag.Lookup(tag)
-
-	return ok && x == "true"
+func (enc *Encoder) encodeComment(indent int, comment string, b []byte) []byte {
+	if comment != "" {
+		b = enc.indent(indent, b)
+		b = append(b, "# "...)
+		b = append(b, comment...)
+		b = append(b, '\n')
+	}
+	return b
+}
+
+func isValidName(s string) bool {
+	if s == "" {
+		return false
+	}
+	for _, c := range s {
+		switch {
+		case strings.ContainsRune("!#$%&()*+-./:;<=>?@[]^_{|}~ ", c):
+			// Backslash and quote chars are reserved, but
+			// otherwise any punctuation chars are allowed
+			// in a tag name.
+		case !unicode.IsLetter(c) && !unicode.IsDigit(c):
+			return false
+		}
+	}
+	return true
+}
+
+type tagOptions struct {
+	multiline bool
+	inline    bool
+	omitempty bool
+}
+
+func parseTag(tag string) (string, tagOptions) {
+	opts := tagOptions{}
+
+	idx := strings.Index(tag, ",")
+	if idx == -1 {
+		return tag, opts
+	}
+
+	raw := tag[idx+1:]
+	tag = string(tag[:idx])
+	for raw != "" {
+		var o string
+		i := strings.Index(raw, ",")
+		if i >= 0 {
+			o, raw = raw[:i], raw[i+1:]
+		} else {
+			o, raw = raw, ""
+		}
+		switch o {
+		case "multiline":
+			opts.multiline = true
+		case "inline":
+			opts.inline = true
+		case "omitempty":
+			opts.omitempty = true
+		}
+	}
+
+	return tag, opts
 }
 
-//nolint:cyclop
 func (enc *Encoder) encodeTable(b []byte, ctx encoderCtx, t table) ([]byte, error) {
 	var err error
 
@@ -653,7 +797,7 @@ func willConvertToTable(ctx encoderCtx, v reflect.Value) bool {
 	if !v.IsValid() {
 		return false
 	}
-	if v.Type() == timeType || v.Type().Implements(textMarshalerType) {
+	if v.Type() == timeType || v.Type().Implements(textMarshalerType) || (v.Kind() != reflect.Ptr && v.CanAddr() && reflect.PtrTo(v.Type()).Implements(textMarshalerType)) {
 		return false
 	}
 
@@ -675,6 +819,9 @@ func willConvertToTable(ctx encoderCtx, v reflect.Value) bool {
 }
 
 func willConvertToTableOrArrayTable(ctx encoderCtx, v reflect.Value) bool {
+	if ctx.insideKv {
+		return false
+	}
 	t := v.Type()
 
 	if t.Kind() == reflect.Interface {
@@ -720,7 +867,6 @@ func (enc *Encoder) encodeSlice(b []byte, ctx encoderCtx, v reflect.Value) ([]by
 func (enc *Encoder) encodeSliceAsArrayTable(b []byte, ctx encoderCtx, v reflect.Value) ([]byte, error) {
 	ctx.shiftKey()
 
-	var err error
 	scratch := make([]byte, 0, 64)
 	scratch = append(scratch, "[["...)
 
@@ -729,10 +875,7 @@ func (enc *Encoder) encodeSliceAsArrayTable(b []byte, ctx encoderCtx, v reflect.
 			scratch = append(scratch, '.')
 		}
 
-		scratch, err = enc.encodeKey(scratch, k)
-		if err != nil {
-			return nil, err
-		}
+		scratch = enc.encodeKey(scratch, k)
 	}
 
 	scratch = append(scratch, "]]\n"...)
@@ -741,6 +884,7 @@ func (enc *Encoder) encodeSliceAsArrayTable(b []byte, ctx encoderCtx, v reflect.
 	for i := 0; i < v.Len(); i++ {
 		b = append(b, scratch...)
 
+		var err error
 		b, err = enc.encode(b, ctx, v.Index(i))
 		if err != nil {
 			return nil, err

marshaler_test.go

@@ -4,20 +4,28 @@ import (
 	"bytes"
 	"encoding/json"
 	"fmt"
+	"math"
+	"math/big"
 	"strings"
 	"testing"
+	"time"
 
 	"github.com/pelletier/go-toml/v2"
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
 )
 
-//nolint:funlen
 func TestMarshal(t *testing.T) {
 	someInt := 42
 
 	type structInline struct {
-		A interface{} `inline:"true"`
+		A interface{} `toml:",inline"`
+	}
+
+	type comments struct {
+		One   int
+		Two   int   `comment:"Before kv"`
+		Three []int `comment:"Before array"`
 	}
 
 	examples := []struct {
@@ -38,7 +46,7 @@ func TestMarshal(t *testing.T) {
 			v: map[string]string{
 				"hel\nlo": "world",
 			},
-			err: true,
+			expected: `"hel\nlo" = 'world'`,
 		},
 		{
 			desc: `map with " in key`,
@@ -193,9 +201,9 @@ name = 'Alice'
 		{
 			desc: "string escapes",
 			v: map[string]interface{}{
-				"a": `'"\`,
+				"a": "'\b\f\r\t\"\\",
 			},
-			expected: `a = "'\"\\"`,
+			expected: `a = "'\b\f\r\t\"\\"`,
 		},
 		{
 			desc: "string utf8 low",
@@ -242,7 +250,7 @@ name = 'Alice'
 		{
 			desc: "multi-line forced",
 			v: struct {
-				A string `multiline:"true"`
+				A string `toml:",multiline"`
 			}{
 				A: "hello\nworld",
 			},
@@ -253,7 +261,7 @@ world"""`,
 		{
 			desc: "inline field",
 			v: struct {
-				A map[string]string `inline:"true"`
+				A map[string]string `toml:",inline"`
 				B map[string]string
 			}{
 				A: map[string]string{
@@ -272,7 +280,7 @@ isinline = 'no'
 		{
 			desc: "mutiline array int",
 			v: struct {
-				A []int `multiline:"true"`
+				A []int `toml:",multiline"`
 				B []int
 			}{
 				A: []int{1, 2, 3, 4},
@@ -291,7 +299,7 @@ B = [1, 2, 3, 4]
 		{
 			desc: "mutiline array in array",
 			v: struct {
-				A [][]int `multiline:"true"`
+				A [][]int `toml:",multiline"`
 			}{
 				A: [][]int{{1, 2}, {3, 4}},
 			},
@@ -373,7 +381,8 @@ hello = 'world'`,
 			v: map[string][]map[string]string{
 				"a\n": {{"hello": "world"}},
 			},
-			err: true,
+			expected: `[["a\n"]]
+hello = 'world'`,
 		},
 		{
 			desc: "newline in map in slice",
@@ -433,7 +442,7 @@ hello = 'world'`,
 			v: map[string]interface{}{
 				"hello\nworld": 42,
 			},
-			err: true,
+			expected: `"hello\nworld" = 42`,
 		},
 		{
 			desc: "new line in parent of nested table key",
@@ -442,7 +451,8 @@ hello = 'world'`,
 					"inner": 42,
 				},
 			},
-			err: true,
+			expected: `["hello\nworld"]
+inner = 42`,
 		},
 		{
 			desc: "new line in nested table key",
@@ -453,7 +463,9 @@ hello = 'world'`,
 					},
 				},
 			},
-			err: true,
+			expected: `[parent]
+[parent."in\ner"]
+foo = 42`,
 		},
 		{
 			desc: "invalid map key",
@@ -469,6 +481,37 @@ hello = 'world'`,
 			},
 			err: true,
 		},
+		{
+			desc: "time",
+			v: struct {
+				T time.Time
+			}{
+				T: time.Time{},
+			},
+			expected: `T = 0001-01-01T00:00:00Z`,
+		},
+		{
+			desc: "time nano",
+			v: struct {
+				T time.Time
+			}{
+				T: time.Date(1979, time.May, 27, 0, 32, 0, 999999000, time.UTC),
+			},
+			expected: `T = 1979-05-27T00:32:00.999999Z`,
+		},
+		{
+			desc: "bool",
+			v: struct {
+				A bool
+				B bool
+			}{
+				A: false,
+				B: true,
+			},
+			expected: `
+A = false
+B = true`,
+		},
 		{
 			desc: "numbers",
 			v: struct {
@@ -483,6 +526,7 @@ hello = 'world'`,
 				I int16
 				J int8
 				K int
+				L float64
 			}{
 				A: 1.1,
 				B: 42,
@@ -495,6 +539,7 @@ hello = 'world'`,
 				I: 42,
 				J: 42,
 				K: 42,
+				L: 2.2,
 			},
 			expected: `
 A = 1.1
@@ -507,7 +552,29 @@ G = 42
 H = 42
 I = 42
 J = 42
-K = 42`,
+K = 42
+L = 2.2`,
+		},
+		{
+			desc: "comments",
+			v: struct {
+				Table comments `comment:"Before table"`
+			}{
+				Table: comments{
+					One:   1,
+					Two:   2,
+					Three: []int{1, 2, 3},
+				},
+			},
+			expected: `
+# Before table
+[Table]
+One = 1
+# Before kv
+Two = 2
+# Before array
+Three = [1, 2, 3]
+`,
 		},
 	}
 
@@ -551,7 +618,7 @@ K = 42`,
 
 type flagsSetters []struct {
 	name string
-	f    func(enc *toml.Encoder, flag bool)
+	f    func(enc *toml.Encoder, flag bool) *toml.Encoder
 }
 
 var allFlags = flagsSetters{
@@ -603,6 +670,33 @@ func equalStringsIgnoreNewlines(t *testing.T, expected string, actual string) {
 	assert.Equal(t, strings.Trim(expected, cutset), strings.Trim(actual, cutset))
 }
 
+func TestMarshalFloats(t *testing.T) {
+	v := map[string]float32{
+		"nan":  float32(math.NaN()),
+		"+inf": float32(math.Inf(1)),
+		"-inf": float32(math.Inf(-1)),
+	}
+
+	expected := `'+inf' = inf
+-inf = -inf
+nan = nan
+`
+
+	actual, err := toml.Marshal(v)
+	require.NoError(t, err)
+	require.Equal(t, expected, string(actual))
+
+	v64 := map[string]float64{
+		"nan":  math.NaN(),
+		"+inf": math.Inf(1),
+		"-inf": math.Inf(-1),
+	}
+
+	actual, err = toml.Marshal(v64)
+	require.NoError(t, err)
+	require.Equal(t, expected, string(actual))
+}
+
 //nolint:funlen
 func TestMarshalIndentTables(t *testing.T) {
 	examples := []struct {
@@ -734,6 +828,60 @@ func TestEncoderSetIndentSymbol(t *testing.T) {
 	equalStringsIgnoreNewlines(t, expected, w.String())
 }
 
+func TestEncoderOmitempty(t *testing.T) {
+	type doc struct {
+		String  string            `toml:",omitempty,multiline"`
+		Bool    bool              `toml:",omitempty,multiline"`
+		Int     int               `toml:",omitempty,multiline"`
+		Int8    int8              `toml:",omitempty,multiline"`
+		Int16   int16             `toml:",omitempty,multiline"`
+		Int32   int32             `toml:",omitempty,multiline"`
+		Int64   int64             `toml:",omitempty,multiline"`
+		Uint    uint              `toml:",omitempty,multiline"`
+		Uint8   uint8             `toml:",omitempty,multiline"`
+		Uint16  uint16            `toml:",omitempty,multiline"`
+		Uint32  uint32            `toml:",omitempty,multiline"`
+		Uint64  uint64            `toml:",omitempty,multiline"`
+		Float32 float32           `toml:",omitempty,multiline"`
+		Float64 float64           `toml:",omitempty,multiline"`
+		MapNil  map[string]string `toml:",omitempty,multiline"`
+		Slice   []string          `toml:",omitempty,multiline"`
+		Ptr     *string           `toml:",omitempty,multiline"`
+		Iface   interface{}       `toml:",omitempty,multiline"`
+		Struct  struct{}          `toml:",omitempty,multiline"`
+	}
+
+	d := doc{}
+
+	b, err := toml.Marshal(d)
+	require.NoError(t, err)
+
+	expected := `[Struct]`
+
+	equalStringsIgnoreNewlines(t, expected, string(b))
+}
+
+func TestEncoderTagFieldName(t *testing.T) {
+	type doc struct {
+		String string `toml:"hello"`
+		OkSym  string `toml:"#"`
+		Bad    string `toml:"\"`
+	}
+
+	d := doc{String: "world"}
+
+	b, err := toml.Marshal(d)
+	require.NoError(t, err)
+
+	expected := `
+hello = 'world'
+'#' = ''
+Bad = ''
+`
+
+	equalStringsIgnoreNewlines(t, expected, string(b))
+}
+
 func TestIssue436(t *testing.T) {
 	data := []byte(`{"a": [ { "b": { "c": "d" } } ]}`)
 
@@ -798,6 +946,146 @@ func TestIssue590(t *testing.T) {
 	require.NoError(t, err)
 }
 
+func TestIssue571(t *testing.T) {
+	type Foo struct {
+		Float32 float32
+		Float64 float64
+	}
+
+	const closeEnough = 1e-9
+
+	foo := Foo{
+		Float32: 42,
+		Float64: 43,
+	}
+	b, err := toml.Marshal(foo)
+	require.NoError(t, err)
+
+	var foo2 Foo
+	err = toml.Unmarshal(b, &foo2)
+	require.NoError(t, err)
+
+	assert.InDelta(t, 42, foo2.Float32, closeEnough)
+	assert.InDelta(t, 43, foo2.Float64, closeEnough)
+}
+
+func TestIssue678(t *testing.T) {
+	type Config struct {
+		BigInt big.Int
+	}
+
+	cfg := &Config{
+		BigInt: *big.NewInt(123),
+	}
+
+	out, err := toml.Marshal(cfg)
+	require.NoError(t, err)
+	equalStringsIgnoreNewlines(t, "BigInt = '123'", string(out))
+
+	cfg2 := &Config{}
+	err = toml.Unmarshal(out, cfg2)
+	require.NoError(t, err)
+	require.Equal(t, cfg, cfg2)
+}
+
+func TestIssue752(t *testing.T) {
+	type Fooer interface {
+		Foo() string
+	}
+
+	type Container struct {
+		Fooer
+	}
+
+	c := Container{}
+
+	out, err := toml.Marshal(c)
+	require.NoError(t, err)
+	require.Equal(t, "", string(out))
+}
+
+func TestMarshalNestedAnonymousStructs(t *testing.T) {
+	type Embedded struct {
+		Value string `toml:"value" json:"value"`
+		Top   struct {
+			Value string `toml:"value" json:"value"`
+		} `toml:"top" json:"top"`
+	}
+
+	type Named struct {
+		Value string `toml:"value" json:"value"`
+	}
+
+	var doc struct {
+		Embedded
+		Named     `toml:"named" json:"named"`
+		Anonymous struct {
+			Value string `toml:"value" json:"value"`
+		} `toml:"anonymous" json:"anonymous"`
+	}
+
+	expected := `value = ''
+[top]
+value = ''
+
+[named]
+value = ''
+
+[anonymous]
+value = ''
+
+`
+
+	result, err := toml.Marshal(doc)
+	require.NoError(t, err)
+	require.Equal(t, expected, string(result))
+}
+
+func TestMarshalNestedAnonymousStructs_DuplicateField(t *testing.T) {
+	type Embedded struct {
+		Value string `toml:"value" json:"value"`
+		Top   struct {
+			Value string `toml:"value" json:"value"`
+		} `toml:"top" json:"top"`
+	}
+
+	var doc struct {
+		Value string `toml:"value" json:"value"`
+		Embedded
+	}
+	doc.Embedded.Value = "shadowed"
+	doc.Value = "shadows"
+
+	expected := `value = 'shadows'
+[top]
+value = ''
+
+`
+
+	result, err := toml.Marshal(doc)
+	require.NoError(t, err)
+	require.NoError(t, err)
+	require.Equal(t, expected, string(result))
+}
+
+func TestLocalTime(t *testing.T) {
+	v := map[string]toml.LocalTime{
+		"a": toml.LocalTime{
+			Hour:       1,
+			Minute:     2,
+			Second:     3,
+			Nanosecond: 4,
+		},
+	}
+
+	expected := `a = 01:02:03.000000004
+`
+
+	out, err := toml.Marshal(v)
+	require.NoError(t, err)
+	require.Equal(t, expected, string(out))
+}
+
 func ExampleMarshal() {
 	type MyConfig struct {
 		Version int
@@ -822,26 +1110,3 @@ func ExampleMarshal() {
 	// Name = 'go-toml'
 	// Tags = ['go', 'toml']
 }
-
-func TestIssue571(t *testing.T) {
-	type Foo struct {
-		Float32 float32
-		Float64 float64
-	}
-
-	const closeEnough = 1e-9
-
-	foo := Foo{
-		Float32: 42,
-		Float64: 43,
-	}
-	b, err := toml.Marshal(foo)
-	require.NoError(t, err)
-
-	var foo2 Foo
-	err = toml.Unmarshal(b, &foo2)
-	require.NoError(t, err)
-
-	assert.InDelta(t, 42, foo2.Float32, closeEnough)
-	assert.InDelta(t, 43, foo2.Float64, closeEnough)
-}

parser.go

@@ -549,7 +549,7 @@ func (p *parser) parseMultilineBasicString(b []byte) ([]byte, []byte, []byte, er
 	startIdx := i
 	endIdx := len(token) - len(`"""`)
 
-	if escaped < 0 {
+	if !escaped {
 		str := token[startIdx:endIdx]
 		verr := utf8TomlValidAlreadyEscaped(str)
 		if verr.Zero() {
@@ -578,6 +578,10 @@ func (p *parser) parseMultilineBasicString(b []byte) ([]byte, []byte, []byte, er
 				switch token[i+j] {
 				case ' ', '\t':
 					continue
+				case '\r':
+					if token[i+j+1] == '\n' {
+						continue
+					}
 				case '\n':
 					isLastNonWhitespaceOnLine = true
 				}
@@ -613,6 +617,8 @@ func (p *parser) parseMultilineBasicString(b []byte) ([]byte, []byte, []byte, er
 				builder.WriteByte('\r')
 			case 't':
 				builder.WriteByte('\t')
+			case 'e':
+				builder.WriteByte(0x1B)
 			case 'u':
 				x, err := hexToRune(atmost(token[i+1:], 4), 4)
 				if err != nil {
@@ -689,13 +695,13 @@ func (p *parser) parseKey(b []byte) (ast.Reference, []byte, error) {
 }
 
 func (p *parser) parseSimpleKey(b []byte) (raw, key, rest []byte, err error) {
+	if len(b) == 0 {
+		return nil, nil, nil, newDecodeError(b, "expected key but found none")
+	}
+
 	// simple-key = quoted-key / unquoted-key
 	// unquoted-key = 1*( ALPHA / DIGIT / %x2D / %x5F ) ; A-Z / a-z / 0-9 / - / _
 	// quoted-key = basic-string / literal-string
-	if len(b) == 0 {
-		return nil, nil, nil, newDecodeError(b, "key is incomplete")
-	}
-
 	switch {
 	case b[0] == '\'':
 		return p.parseLiteralString(b)
@@ -736,7 +742,7 @@ func (p *parser) parseBasicString(b []byte) ([]byte, []byte, []byte, error) {
 	// Fast path. If there is no escape sequence, the string should just be
 	// an UTF-8 encoded string, which is the same as Go. In that case,
 	// validate the string and return a direct reference to the buffer.
-	if escaped < 0 {
+	if !escaped {
 		str := token[startIdx:endIdx]
 		verr := utf8TomlValidAlreadyEscaped(str)
 		if verr.Zero() {
@@ -770,6 +776,8 @@ func (p *parser) parseBasicString(b []byte) ([]byte, []byte, []byte, error) {
 				builder.WriteByte('\r')
 			case 't':
 				builder.WriteByte('\t')
+			case 'e':
+				builder.WriteByte(0x1B)
 			case 'u':
 				x, err := hexToRune(token[i+1:len(token)-1], 4)
 				if err != nil {
@@ -866,7 +874,6 @@ func (p *parser) parseIntOrFloatOrDateTime(b []byte) (ast.Reference, []byte, err
 		return p.scanIntOrFloat(b)
 	}
 
-	//nolint:gomnd
 	if len(b) < 3 {
 		return p.scanIntOrFloat(b)
 	}
@@ -884,23 +891,13 @@ func (p *parser) parseIntOrFloatOrDateTime(b []byte) (ast.Reference, []byte, err
 		if idx == 2 && c == ':' || (idx == 4 && c == '-') {
 			return p.scanDateTime(b)
 		}
+
+		break
 	}
 
 	return p.scanIntOrFloat(b)
 }
 
-func digitsToInt(b []byte) int {
-	x := 0
-
-	for _, d := range b {
-		x *= 10
-		x += int(d - '0')
-	}
-
-	return x
-}
-
-//nolint:gocognit,cyclop
 func (p *parser) scanDateTime(b []byte) (ast.Reference, []byte, error) {
 	// scans for contiguous characters in [0-9T:Z.+-], and up to one space if
 	// followed by a digit.
@@ -970,7 +967,7 @@ byteLoop:
 func (p *parser) scanIntOrFloat(b []byte) (ast.Reference, []byte, error) {
 	i := 0
 
-	if len(b) > 2 && b[0] == '0' && b[1] != '.' && b[1] != 'e' {
+	if len(b) > 2 && b[0] == '0' && b[1] != '.' && b[1] != 'e' && b[1] != 'E' {
 		var isValidRune validRuneFn
 
 		switch b[1] {

parser_test.go

@@ -1,6 +1,8 @@
 package toml
 
 import (
+	"strconv"
+	"strings"
 	"testing"
 
 	"github.com/pelletier/go-toml/v2/internal/ast"
@@ -371,6 +373,23 @@ func BenchmarkParseBasicStringWithUnicode(b *testing.B) {
 	})
 }
 
+func BenchmarkParseBasicStringsEasy(b *testing.B) {
+	p := &parser{}
+
+	for _, size := range []int{1, 4, 8, 16, 21} {
+		b.Run(strconv.Itoa(size), func(b *testing.B) {
+			input := []byte(`"` + strings.Repeat("A", size) + `"`)
+
+			b.ReportAllocs()
+			b.SetBytes(int64(len(input)))
+
+			for i := 0; i < b.N; i++ {
+				p.parseBasicString(input)
+			}
+		})
+	}
+}
+
 func TestParser_AST_DateTimes(t *testing.T) {
 	examples := []struct {
 		desc  string

scanner.go

@@ -53,7 +53,7 @@ func scanLiteralString(b []byte) ([]byte, []byte, error) {
 		switch b[i] {
 		case '\'':
 			return b[:i+1], b[i+1:], nil
-		case '\n':
+		case '\n', '\r':
 			return nil, nil, newDecodeError(b[i:i+1], "literal strings cannot have new lines")
 		}
 		size := utf8ValidNext(b[i:])
@@ -76,30 +76,42 @@ func scanMultilineLiteralString(b []byte) ([]byte, []byte, error) {
 	// mll-char = %x09 / %x20-26 / %x28-7E / non-ascii
 	// mll-quotes = 1*2apostrophe
 	for i := 3; i < len(b); {
-		if scanFollowsMultilineLiteralStringDelimiter(b[i:]) {
-			i += 3
+		switch b[i] {
+		case '\'':
+			if scanFollowsMultilineLiteralStringDelimiter(b[i:]) {
+				i += 3
 
-			// At that point we found 3 apostrophe, and i is the
-			// index of the byte after the third one. The scanner
-			// needs to be eager, because there can be an extra 2
-			// apostrophe that can be accepted at the end of the
-			// string.
+				// At that point we found 3 apostrophe, and i is the
+				// index of the byte after the third one. The scanner
+				// needs to be eager, because there can be an extra 2
+				// apostrophe that can be accepted at the end of the
+				// string.
+
+				if i >= len(b) || b[i] != '\'' {
+					return b[:i], b[i:], nil
+				}
+				i++
+
+				if i >= len(b) || b[i] != '\'' {
+					return b[:i], b[i:], nil
+				}
+				i++
+
+				if i < len(b) && b[i] == '\'' {
+					return nil, nil, newDecodeError(b[i-3:i+1], "''' not allowed in multiline literal string")
+				}
 
-			if i >= len(b) || b[i] != '\'' {
 				return b[:i], b[i:], nil
 			}
-			i++
-
-			if i >= len(b) || b[i] != '\'' {
-				return b[:i], b[i:], nil
+		case '\r':
+			if len(b) < i+2 {
+				return nil, nil, newDecodeError(b[len(b):], `need a \n after \r`)
 			}
-			i++
-
-			if i < len(b) && b[i] == '\'' {
-				return nil, nil, newDecodeError(b[i-3:i+1], "''' not allowed in multiline literal string")
+			if b[i+1] != '\n' {
+				return nil, nil, newDecodeError(b[i:i+2], `need a \n after \r`)
 			}
-
-			return b[:i], b[i:], nil
+			i += 2 // skip the \n
+			continue
 		}
 		size := utf8ValidNext(b[i:])
 		if size == 0 {
@@ -149,6 +161,12 @@ func scanComment(b []byte) ([]byte, []byte, error) {
 		if b[i] == '\n' {
 			return b[:i], b[i:], nil
 		}
+		if b[i] == '\r' {
+			if i+1 < len(b) && b[i+1] == '\n' {
+				return b[:i+1], b[i+1:], nil
+			}
+			return nil, nil, newDecodeError(b[i:i+1], "invalid character in comment")
+		}
 		size := utf8ValidNext(b[i:])
 		if size == 0 {
 			return nil, nil, newDecodeError(b[i:i+1], "invalid character in comment")
@@ -160,42 +178,26 @@ func scanComment(b []byte) ([]byte, []byte, error) {
 	return b, b[len(b):], nil
 }
 
-func scanBasicString(b []byte) ([]byte, int, []byte, error) {
+func scanBasicString(b []byte) ([]byte, bool, []byte, error) {
 	// basic-string = quotation-mark *basic-char quotation-mark
 	// quotation-mark = %x22            ; "
 	// basic-char = basic-unescaped / escaped
 	// basic-unescaped = wschar / %x21 / %x23-5B / %x5D-7E / non-ascii
 	// escaped = escape escape-seq-char
-	escaped := -1 // index of the first \. -1 means no escape character in there.
+	escaped := false
 	i := 1
 
-loop:
 	for ; i < len(b); i++ {
 		switch b[i] {
 		case '"':
 			return b[:i+1], escaped, b[i+1:], nil
-		case '\n':
-			return nil, escaped, nil, newDecodeError(b[i:i+1], "basic strings cannot have new lines")
-		case '\\':
-			if len(b) < i+2 {
-				return nil, escaped, nil, newDecodeError(b[i:i+1], "need a character after \\")
-			}
-			escaped = i
-			i += 2 // skip the next character
-			break loop
-		}
-	}
-
-	for ; i < len(b); i++ {
-		switch b[i] {
-		case '"':
-			return b[:i+1], escaped, b[i+1:], nil
-		case '\n':
+		case '\n', '\r':
 			return nil, escaped, nil, newDecodeError(b[i:i+1], "basic strings cannot have new lines")
 		case '\\':
 			if len(b) < i+2 {
 				return nil, escaped, nil, newDecodeError(b[i:i+1], "need a character after \\")
 			}
+			escaped = true
 			i++ // skip the next character
 		}
 	}
@@ -203,7 +205,7 @@ loop:
 	return nil, escaped, nil, newDecodeError(b[len(b):], `basic string not terminated by "`)
 }
 
-func scanMultilineBasicString(b []byte) ([]byte, int, []byte, error) {
+func scanMultilineBasicString(b []byte) ([]byte, bool, []byte, error) {
 	// ml-basic-string = ml-basic-string-delim [ newline ] ml-basic-body
 	// ml-basic-string-delim
 	// ml-basic-string-delim = 3quotation-mark
@@ -215,10 +217,9 @@ func scanMultilineBasicString(b []byte) ([]byte, int, []byte, error) {
 	// mlb-unescaped = wschar / %x21 / %x23-5B / %x5D-7E / non-ascii
 	// mlb-escaped-nl = escape ws newline *( wschar / newline )
 
-	escaped := -1
+	escaped := false
 	i := 3
 
-loop:
 	for ; i < len(b); i++ {
 		switch b[i] {
 		case '"':
@@ -251,23 +252,16 @@ loop:
 			if len(b) < i+2 {
 				return nil, escaped, nil, newDecodeError(b[len(b):], "need a character after \\")
 			}
-			escaped = i
-			i += 2 // skip the next character
-			break loop
-		}
-	}
-
-	for ; i < len(b); i++ {
-		switch b[i] {
-		case '"':
-			if scanFollowsMultilineBasicStringDelimiter(b[i:]) {
-				return b[:i+3], escaped, b[i+3:], nil
-			}
-		case '\\':
-			if len(b) < i+2 {
-				return nil, escaped, nil, newDecodeError(b[len(b):], "need a character after \\")
-			}
+			escaped = true
 			i++ // skip the next character
+		case '\r':
+			if len(b) < i+2 {
+				return nil, escaped, nil, newDecodeError(b[len(b):], `need a \n after \r`)
+			}
+			if b[i+1] != '\n' {
+				return nil, escaped, nil, newDecodeError(b[i:i+2], `need a \n after \r`)
+			}
+			i++ // skip the \n
 		}
 	}
 

testdata/fuzz/FuzzUnmarshal/0c08b924aa26af23d16b77029b4f5567a96af93402e0472fef7b837b37580d7c

@@ -0,0 +1,2 @@
+go test fuzz v1
+[]byte("0=0000-01-01 00:00:00")

testdata/fuzz/FuzzUnmarshal/249a017d8ffa0d88d41c594ada5399f764833f64050180cb39f106d12666853f

@@ -0,0 +1,2 @@
+go test fuzz v1
+[]byte("\"\\n\"=\"\"")

testdata/fuzz/FuzzUnmarshal/391ce67b866d58464d067c5981e99951fd499896453a2fe2779a6bfe7df11bf5

@@ -0,0 +1,2 @@
+go test fuzz v1
+[]byte("''=0")

testdata/fuzz/FuzzUnmarshal/49ae83283c5cfbed874bcacbfb342de107b69c2a3d3c35c2d99f67345fe89946

@@ -0,0 +1,2 @@
+go test fuzz v1
+[]byte("0=0000-01-01")

testdata/fuzz/FuzzUnmarshal/69ea82e85bb4ad966ae86bb05eea3af69ed1a826c846f1bf2282f329b4d8fb36

@@ -0,0 +1,2 @@
+go test fuzz v1
+[]byte("0=\"\"\"\\U00000000\"\"\"")

testdata/fuzz/FuzzUnmarshal/746910fac6deb42e7e62dce60b333c81c39d844a122c9075894c023400e8fdbf

@@ -0,0 +1,2 @@
+go test fuzz v1
+[]byte("0=[[{}]]")

testdata/fuzz/FuzzUnmarshal/844d1171a3adc95e51627f0e9378bbaa193fe2d0284f4ba8ed19a5226a8cbc74

@@ -0,0 +1,2 @@
+go test fuzz v1
+[]byte("\"\\b\"=\"\"")

testdata/fuzz/FuzzUnmarshal/900c54b5071113ae2a5365b3156377ddd775b9d38add61629544903cb9c94b00

@@ -0,0 +1,2 @@
+go test fuzz v1
+[]byte("0=inf")

testdata/fuzz/FuzzUnmarshal/c87e8eb4e8fa40fd3c89d8ed16a049ce81592ded710965ee29a4e506eb5427b3

@@ -0,0 +1,2 @@
+go test fuzz v1
+[]byte("0=0000-01-01 00:00:00+00:00")

testdata/fuzz/FuzzUnmarshal/d0e5d6fc3b865cfae5480c01f301b87c932722ae0c1d692deea413349ea028be

@@ -0,0 +1,2 @@
+go test fuzz v1
+[]byte("0=[{}]")

testdata/fuzz/FuzzUnmarshal/d2cecea5b8ee5f148419671cef0644178cdba1ee91d3f295be38da576e89ef40

@@ -0,0 +1,2 @@
+go test fuzz v1
+[]byte("0=nan")

toml_testgen_support_test.go

@@ -8,7 +8,7 @@ import (
 	"testing"
 
 	"github.com/pelletier/go-toml/v2"
-	"github.com/pelletier/go-toml/v2/testsuite"
+	"github.com/pelletier/go-toml/v2/internal/testsuite"
 	"github.com/stretchr/testify/require"
 )
 

toml_testgen_test.go

@@ -1,4 +1,4 @@
-// Generated by tomltestgen for toml-test ref master on 2021-09-30T20:29:36-05:00
+// Generated by tomltestgen for toml-test ref master on 2022-04-07T20:09:42-04:00
 package toml_test
 
 import (
@@ -55,11 +55,51 @@ func TestTOMLTest_Invalid_Array_TextInArray(t *testing.T) {
 	testgenInvalid(t, input)
 }
 
+func TestTOMLTest_Invalid_Bool_AlmostFalseWithExtra(t *testing.T) {
+	input := "a = falsify\n"
+	testgenInvalid(t, input)
+}
+
+func TestTOMLTest_Invalid_Bool_AlmostFalse(t *testing.T) {
+	input := "a = fals\n"
+	testgenInvalid(t, input)
+}
+
+func TestTOMLTest_Invalid_Bool_AlmostTrueWithExtra(t *testing.T) {
+	input := "a = truthy\n"
+	testgenInvalid(t, input)
+}
+
+func TestTOMLTest_Invalid_Bool_AlmostTrue(t *testing.T) {
+	input := "a = tru\n"
+	testgenInvalid(t, input)
+}
+
+func TestTOMLTest_Invalid_Bool_JustF(t *testing.T) {
+	input := "a = f\n"
+	testgenInvalid(t, input)
+}
+
+func TestTOMLTest_Invalid_Bool_JustT(t *testing.T) {
+	input := "a = t\n"
+	testgenInvalid(t, input)
+}
+
 func TestTOMLTest_Invalid_Bool_MixedCase(t *testing.T) {
 	input := "valid = False\n"
 	testgenInvalid(t, input)
 }
 
+func TestTOMLTest_Invalid_Bool_StartingSameFalse(t *testing.T) {
+	input := "a = falsey\n"
+	testgenInvalid(t, input)
+}
+
+func TestTOMLTest_Invalid_Bool_StartingSameTrue(t *testing.T) {
+	input := "a = truer\n"
+	testgenInvalid(t, input)
+}
+
 func TestTOMLTest_Invalid_Bool_WrongCaseFalse(t *testing.T) {
 	input := "b = FALSE\n"
 	testgenInvalid(t, input)
@@ -70,6 +110,31 @@ func TestTOMLTest_Invalid_Bool_WrongCaseTrue(t *testing.T) {
 	testgenInvalid(t, input)
 }
 
+func TestTOMLTest_Invalid_Control_BareCr(t *testing.T) {
+	input := "# The following line contains a single carriage return control character\r\n\r"
+	testgenInvalid(t, input)
+}
+
+func TestTOMLTest_Invalid_Control_BareFormfeed(t *testing.T) {
+	input := "bare-formfeed = \f\n"
+	testgenInvalid(t, input)
+}
+
+func TestTOMLTest_Invalid_Control_BareNull(t *testing.T) {
+	input := "bare-null = \"some value\" \x00\n"
+	testgenInvalid(t, input)
+}
+
+func TestTOMLTest_Invalid_Control_BareVerticalTab(t *testing.T) {
+	input := "bare-vertical-tab = \v\n"
+	testgenInvalid(t, input)
+}
+
+func TestTOMLTest_Invalid_Control_CommentCr(t *testing.T) {
+	input := "comment-cr = \"Carriage return in comment\" # \ra=1\n"
+	testgenInvalid(t, input)
+}
+
 func TestTOMLTest_Invalid_Control_CommentDel(t *testing.T) {
 	input := "comment-del = \"0x7f\" # \u007f\n"
 	testgenInvalid(t, input)
@@ -175,33 +240,73 @@ func TestTOMLTest_Invalid_Control_StringUs(t *testing.T) {
 	testgenInvalid(t, input)
 }
 
-func TestTOMLTest_Invalid_Datetime_ImpossibleDate(t *testing.T) {
-	input := "d = 2006-01-50T00:00:00Z\n"
+func TestTOMLTest_Invalid_Datetime_HourOver(t *testing.T) {
+	input := "# time-hour       = 2DIGIT  ; 00-23\nd = 2006-01-01T24:00:00-00:00\n"
+	testgenInvalid(t, input)
+}
+
+func TestTOMLTest_Invalid_Datetime_MdayOver(t *testing.T) {
+	input := "# date-mday       = 2DIGIT  ; 01-28, 01-29, 01-30, 01-31 based on\n#                           ; month/year\nd = 2006-01-32T00:00:00-00:00\n"
+	testgenInvalid(t, input)
+}
+
+func TestTOMLTest_Invalid_Datetime_MdayUnder(t *testing.T) {
+	input := "# date-mday       = 2DIGIT  ; 01-28, 01-29, 01-30, 01-31 based on\n#                           ; month/year\nd = 2006-01-00T00:00:00-00:00\n"
+	testgenInvalid(t, input)
+}
+
+func TestTOMLTest_Invalid_Datetime_MinuteOver(t *testing.T) {
+	input := "# time-minute     = 2DIGIT  ; 00-59\nd = 2006-01-01T00:60:00-00:00\n"
+	testgenInvalid(t, input)
+}
+
+func TestTOMLTest_Invalid_Datetime_MonthOver(t *testing.T) {
+	input := "# date-month      = 2DIGIT  ; 01-12\nd = 2006-13-01T00:00:00-00:00\n"
+	testgenInvalid(t, input)
+}
+
+func TestTOMLTest_Invalid_Datetime_MonthUnder(t *testing.T) {
+	input := "# date-month      = 2DIGIT  ; 01-12\nd = 2007-00-01T00:00:00-00:00\n"
 	testgenInvalid(t, input)
 }
 
 func TestTOMLTest_Invalid_Datetime_NoLeadsWithMilli(t *testing.T) {
-	input := "with-milli = 1987-07-5T17:45:00.12Z\n"
+	input := "# Day \"5\" instead of \"05\"; the leading zero is required.\nwith-milli = 1987-07-5T17:45:00.12Z\n"
 	testgenInvalid(t, input)
 }
 
 func TestTOMLTest_Invalid_Datetime_NoLeads(t *testing.T) {
-	input := "no-leads = 1987-7-05T17:45:00Z\n"
+	input := "# Month \"7\" instead of \"07\"; the leading zero is required.\nno-leads = 1987-7-05T17:45:00Z\n"
 	testgenInvalid(t, input)
 }
 
 func TestTOMLTest_Invalid_Datetime_NoSecs(t *testing.T) {
-	input := "no-secs = 1987-07-05T17:45Z\n"
+	input := "# No seconds in time.\nno-secs = 1987-07-05T17:45Z\n"
 	testgenInvalid(t, input)
 }
 
 func TestTOMLTest_Invalid_Datetime_NoT(t *testing.T) {
-	input := "no-t = 1987-07-0517:45:00Z\n"
+	input := "# No \"t\" or \"T\" between the date and time.\nno-t = 1987-07-0517:45:00Z\n"
+	testgenInvalid(t, input)
+}
+
+func TestTOMLTest_Invalid_Datetime_SecondOver(t *testing.T) {
+	input := "# time-second     = 2DIGIT  ; 00-58, 00-59, 00-60 based on leap second\n#                           ; rules\nd = 2006-01-01T00:00:61-00:00\n"
+	testgenInvalid(t, input)
+}
+
+func TestTOMLTest_Invalid_Datetime_TimeNoLeads2(t *testing.T) {
+	input := "# Leading 0 is always required.\nd = 01:32:0\n"
+	testgenInvalid(t, input)
+}
+
+func TestTOMLTest_Invalid_Datetime_TimeNoLeads(t *testing.T) {
+	input := "# Leading 0 is always required.\nd = 1:32:00\n"
 	testgenInvalid(t, input)
 }
 
 func TestTOMLTest_Invalid_Datetime_TrailingT(t *testing.T) {
-	input := "d = 2006-01-30T\n"
+	input := "# Date cannot end with trailing T\nd = 2006-01-30T\n"
 	testgenInvalid(t, input)
 }
 
@@ -375,6 +480,11 @@ func TestTOMLTest_Invalid_Float_TrailingPoint(t *testing.T) {
 	testgenInvalid(t, input)
 }
 
+func TestTOMLTest_Invalid_Float_TrailingUsExp(t *testing.T) {
+	input := "# trailing underscore in integer part is not allowed\ntrailing-us-exp = 1_e2\n# trailing underscore in float part is not allowed\ntrailing-us-exp2 = 1.2_e2\n"
+	testgenInvalid(t, input)
+}
+
 func TestTOMLTest_Invalid_Float_TrailingUs(t *testing.T) {
 	input := "trailing-us = 1.2_\n"
 	testgenInvalid(t, input)
@@ -390,11 +500,21 @@ func TestTOMLTest_Invalid_Float_UsBeforePoint(t *testing.T) {
 	testgenInvalid(t, input)
 }
 
+func TestTOMLTest_Invalid_InlineTable_Add(t *testing.T) {
+	input := "a={}\n# Inline tables are immutable and can't be extended\n[a.b]\n"
+	testgenInvalid(t, input)
+}
+
 func TestTOMLTest_Invalid_InlineTable_DoubleComma(t *testing.T) {
 	input := "t = {x=3,,y=4}\n"
 	testgenInvalid(t, input)
 }
 
+func TestTOMLTest_Invalid_InlineTable_DuplicateKey(t *testing.T) {
+	input := "# Duplicate keys within an inline table are invalid\na={b=1, b=2}\n"
+	testgenInvalid(t, input)
+}
+
 func TestTOMLTest_Invalid_InlineTable_Empty(t *testing.T) {
 	input := "t = {,}\n"
 	testgenInvalid(t, input)
@@ -425,6 +545,11 @@ func TestTOMLTest_Invalid_InlineTable_NoComma(t *testing.T) {
 	testgenInvalid(t, input)
 }
 
+func TestTOMLTest_Invalid_InlineTable_Overwrite(t *testing.T) {
+	input := "a.b=0\n# Since table \"a\" is already defined, it can't be replaced by an inline table.\na={}\n"
+	testgenInvalid(t, input)
+}
+
 func TestTOMLTest_Invalid_InlineTable_TrailingComma(t *testing.T) {
 	input := "# A terminating comma (also called trailing comma) is not permitted after the\n# last key/value pair in an inline table\nabc = { abc = 123, }\n"
 	testgenInvalid(t, input)
@@ -460,6 +585,21 @@ func TestTOMLTest_Invalid_Integer_DoubleUs(t *testing.T) {
 	testgenInvalid(t, input)
 }
 
+func TestTOMLTest_Invalid_Integer_IncompleteBin(t *testing.T) {
+	input := "incomplete-bin = 0b\n"
+	testgenInvalid(t, input)
+}
+
+func TestTOMLTest_Invalid_Integer_IncompleteHex(t *testing.T) {
+	input := "incomplete-hex = 0x\n"
+	testgenInvalid(t, input)
+}
+
+func TestTOMLTest_Invalid_Integer_IncompleteOct(t *testing.T) {
+	input := "incomplete-oct = 0o\n"
+	testgenInvalid(t, input)
+}
+
 func TestTOMLTest_Invalid_Integer_InvalidBin(t *testing.T) {
 	input := "invalid-bin = 0b0012\n"
 	testgenInvalid(t, input)
@@ -505,6 +645,11 @@ func TestTOMLTest_Invalid_Integer_LeadingZero2(t *testing.T) {
 	testgenInvalid(t, input)
 }
 
+func TestTOMLTest_Invalid_Integer_LeadingZero3(t *testing.T) {
+	input := "leading-zero-3 = 0_0\n"
+	testgenInvalid(t, input)
+}
+
 func TestTOMLTest_Invalid_Integer_LeadingZeroSign1(t *testing.T) {
 	input := "leading-zero-sign-1 = -01\n"
 	testgenInvalid(t, input)
@@ -515,6 +660,11 @@ func TestTOMLTest_Invalid_Integer_LeadingZeroSign2(t *testing.T) {
 	testgenInvalid(t, input)
 }
 
+func TestTOMLTest_Invalid_Integer_LeadingZeroSign3(t *testing.T) {
+	input := "leading-zero-sign-3 = +0_1\n"
+	testgenInvalid(t, input)
+}
+
 func TestTOMLTest_Invalid_Integer_NegativeBin(t *testing.T) {
 	input := "negative-bin = -0b11010110\n"
 	testgenInvalid(t, input)
@@ -720,11 +870,16 @@ func TestTOMLTest_Invalid_String_BadConcat(t *testing.T) {
 	testgenInvalid(t, input)
 }
 
-func TestTOMLTest_Invalid_String_BadEscape(t *testing.T) {
+func TestTOMLTest_Invalid_String_BadEscape1(t *testing.T) {
 	input := "invalid-escape = \"This string has a bad \\a escape character.\"\n"
 	testgenInvalid(t, input)
 }
 
+func TestTOMLTest_Invalid_String_BadEscape2(t *testing.T) {
+	input := "invalid-escape = \"This string has a bad \\  escape character.\"\n\n"
+	testgenInvalid(t, input)
+}
+
 func TestTOMLTest_Invalid_String_BadMultiline(t *testing.T) {
 	input := "multi = \"first line\nsecond line\"\n"
 	testgenInvalid(t, input)
@@ -795,6 +950,21 @@ func TestTOMLTest_Invalid_String_MissingQuotes(t *testing.T) {
 	testgenInvalid(t, input)
 }
 
+func TestTOMLTest_Invalid_String_MultilineBadEscape1(t *testing.T) {
+	input := "k = \"\"\"t\\a\"\"\"\n\n"
+	testgenInvalid(t, input)
+}
+
+func TestTOMLTest_Invalid_String_MultilineBadEscape2(t *testing.T) {
+	input := "# \\<Space> is not a valid escape.\nk = \"\"\"t\\ t\"\"\"\n"
+	testgenInvalid(t, input)
+}
+
+func TestTOMLTest_Invalid_String_MultilineBadEscape3(t *testing.T) {
+	input := "# \\<Space> is not a valid escape.\nk = \"\"\"t\\ \"\"\"\n\n"
+	testgenInvalid(t, input)
+}
+
 func TestTOMLTest_Invalid_String_MultilineEscapeSpace(t *testing.T) {
 	input := "a = \"\"\"\n  foo \\ \\n\n  bar\"\"\"\n"
 	testgenInvalid(t, input)
@@ -815,11 +985,6 @@ func TestTOMLTest_Invalid_String_MultilineQuotes1(t *testing.T) {
 	testgenInvalid(t, input)
 }
 
-func TestTOMLTest_Invalid_String_MultilineQuotes2(t *testing.T) {
-	input := "a = \"\"\"6 quotes: \"\"\"\"\"\"\n"
-	testgenInvalid(t, input)
-}
-
 func TestTOMLTest_Invalid_String_NoClose(t *testing.T) {
 	input := "no-ending-quote = \"One time, at band camp\n"
 	testgenInvalid(t, input)
@@ -835,6 +1000,16 @@ func TestTOMLTest_Invalid_String_WrongClose(t *testing.T) {
 	testgenInvalid(t, input)
 }
 
+func TestTOMLTest_Invalid_Table_AppendWithDottedKeys1(t *testing.T) {
+	input := "# First a.b.c defines a table: a.b.c = {z=9}\n#\n# Then we define a.b.c.t = \"str\" to add a str to the above table, making it:\n#\n#   a.b.c = {z=9, t=\"...\"}\n#\n# While this makes sense, logically, it was decided this is not valid TOML as\n# it's too confusing/convoluted.\n# \n# See: https://github.com/toml-lang/toml/issues/846\n#      https://github.com/toml-lang/toml/pull/859\n\n[a.b.c]\n  z = 9\n\n[a]\n  b.c.t = \"Using dotted keys to add to [a.b.c] after explicitly defining it above is not allowed\"\n"
+	testgenInvalid(t, input)
+}
+
+func TestTOMLTest_Invalid_Table_AppendWithDottedKeys2(t *testing.T) {
+	input := "# This is the same issue as in injection-1.toml, except that nests one level\n# deeper. See that file for a more complete description.\n\n[a.b.c.d]\n  z = 9\n\n[a]\n  b.c.d.k.t = \"Using dotted keys to add to [a.b.c.d] after explicitly defining it above is not allowed\"\n"
+	testgenInvalid(t, input)
+}
+
 func TestTOMLTest_Invalid_Table_ArrayEmpty(t *testing.T) {
 	input := "[[]]\nname = \"Born to Run\"\n"
 	testgenInvalid(t, input)
@@ -850,6 +1025,16 @@ func TestTOMLTest_Invalid_Table_ArrayMissingBracket(t *testing.T) {
 	testgenInvalid(t, input)
 }
 
+func TestTOMLTest_Invalid_Table_DuplicateKeyDottedTable(t *testing.T) {
+	input := "[fruit]\napple.color = \"red\"\n\n[fruit.apple] # INVALID\n"
+	testgenInvalid(t, input)
+}
+
+func TestTOMLTest_Invalid_Table_DuplicateKeyDottedTable2(t *testing.T) {
+	input := "[fruit]\napple.taste.sweet = true\n\n[fruit.apple.taste] # INVALID\n"
+	testgenInvalid(t, input)
+}
+
 func TestTOMLTest_Invalid_Table_DuplicateKeyTable(t *testing.T) {
 	input := "[fruit]\ntype = \"apple\"\n\n[fruit.type]\napple = \"yes\"\n"
 	testgenInvalid(t, input)
@@ -885,16 +1070,6 @@ func TestTOMLTest_Invalid_Table_EqualsSign(t *testing.T) {
 	testgenInvalid(t, input)
 }
 
-func TestTOMLTest_Invalid_Table_Injection1(t *testing.T) {
-	input := "[a.b.c]\n  z = 9\n[a]\n  b.c.t = \"Using dotted keys to add to [a.b.c] after explicitly defining it above is not allowed\"\n  \n# see https://github.com/toml-lang/toml/issues/846\n"
-	testgenInvalid(t, input)
-}
-
-func TestTOMLTest_Invalid_Table_Injection2(t *testing.T) {
-	input := "[a.b.c.d]\n  z = 9\n[a]\n  b.c.d.k.t = \"Using dotted keys to add to [a.b.c.d] after explicitly defining it above is not allowed\"\n  \n# see https://github.com/toml-lang/toml/issues/846\n"
-	testgenInvalid(t, input)
-}
-
 func TestTOMLTest_Invalid_Table_Llbrace(t *testing.T) {
 	input := "[ [table]]\n"
 	testgenInvalid(t, input)
@@ -1061,8 +1236,14 @@ func TestTOMLTest_Valid_Comment_AtEof2(t *testing.T) {
 }
 
 func TestTOMLTest_Valid_Comment_Everywhere(t *testing.T) {
-	input := "# Top comment.\n  # Top comment.\n# Top comment.\n\n# [no-extraneous-groups-please]\n\n[group] # Comment\nanswer = 42 # Comment\n# no-extraneous-keys-please = 999\n# Inbetween comment.\nmore = [ # Comment\n  # What about multiple # comments?\n  # Can you handle it?\n  #\n          # Evil.\n# Evil.\n  42, 42, # Comments within arrays are fun.\n  # What about multiple # comments?\n  # Can you handle it?\n  #\n          # Evil.\n# Evil.\n# ] Did I fool you?\n] # Hopefully not.\n\n# Make sure the space between the datetime and \"#\" isn't lexed.\nd = 1979-05-27T07:32:12-07:00  # c\n"
-	jsonRef := "{\n  \"group\": {\n    \"answer\": {\n      \"type\": \"integer\",\n      \"value\": \"42\"\n    },\n    \"d\": {\n      \"type\": \"datetime\",\n      \"value\": \"1979-05-27T07:32:12-07:00\"\n    },\n    \"more\": [\n      {\n        \"type\": \"integer\",\n        \"value\": \"42\"\n      },\n      {\n        \"type\": \"integer\",\n        \"value\": \"42\"\n      }\n    ]\n  }\n}\n"
+	input := "# Top comment.\n  # Top comment.\n# Top comment.\n\n# [no-extraneous-groups-please]\n\n[group] # Comment\nanswer = 42 # Comment\n# no-extraneous-keys-please = 999\n# Inbetween comment.\nmore = [ # Comment\n  # What about multiple # comments?\n  # Can you handle it?\n  #\n          # Evil.\n# Evil.\n  42, 42, # Comments within arrays are fun.\n  # What about multiple # comments?\n  # Can you handle it?\n  #\n          # Evil.\n# Evil.\n# ] Did I fool you?\n] # Hopefully not.\n\n# Make sure the space between the datetime and \"#\" isn't lexed.\ndt = 1979-05-27T07:32:12-07:00  # c\nd = 1979-05-27 # Comment\n"
+	jsonRef := "{\n  \"group\": {\n    \"answer\": {\n      \"type\": \"integer\",\n      \"value\": \"42\"\n    },\n    \"dt\": {\n      \"type\": \"datetime\",\n      \"value\": \"1979-05-27T07:32:12-07:00\"\n    },\n    \"d\": {\n      \"type\": \"date-local\",\n      \"value\": \"1979-05-27\"\n    },\n    \"more\": [\n      {\n        \"type\": \"integer\",\n        \"value\": \"42\"\n      },\n      {\n        \"type\": \"integer\",\n        \"value\": \"42\"\n      }\n    ]\n  }\n}\n"
+	testgenValid(t, input, jsonRef)
+}
+
+func TestTOMLTest_Valid_Comment_Noeol(t *testing.T) {
+	input := "# single comment without any eol characters"
+	jsonRef := "{}\n"
 	testgenValid(t, input, jsonRef)
 }
 
@@ -1097,8 +1278,8 @@ func TestTOMLTest_Valid_Datetime_Local(t *testing.T) {
 }
 
 func TestTOMLTest_Valid_Datetime_Milliseconds(t *testing.T) {
-	input := "utc1  = 1987-07-05T17:45:56.123456Z\nutc2  = 1987-07-05T17:45:56.6Z\nwita1 = 1987-07-05T17:45:56.123456+08:00\nwita2 = 1987-07-05T17:45:56.6+08:00\n"
-	jsonRef := "{\n  \"utc1\": {\n    \"type\": \"datetime\",\n    \"value\": \"1987-07-05T17:45:56.123456Z\"\n  },\n  \"utc2\": {\n    \"type\": \"datetime\",\n    \"value\": \"1987-07-05T17:45:56.600000Z\"\n  },\n  \"wita1\": {\n    \"type\": \"datetime\",\n    \"value\": \"1987-07-05T17:45:56.123456+08:00\"\n  },\n  \"wita2\": {\n    \"type\": \"datetime\",\n    \"value\": \"1987-07-05T17:45:56.600000+08:00\"\n  }\n}\n"
+	input := "utc1  = 1987-07-05T17:45:56.1234Z\nutc2  = 1987-07-05T17:45:56.6Z\nwita1 = 1987-07-05T17:45:56.1234+08:00\nwita2 = 1987-07-05T17:45:56.6+08:00\n"
+	jsonRef := "{\n  \"utc1\": {\n    \"type\": \"datetime\",\n    \"value\": \"1987-07-05T17:45:56.1234Z\"\n  },\n  \"utc2\": {\n    \"type\": \"datetime\",\n    \"value\": \"1987-07-05T17:45:56.6000Z\"\n  },\n  \"wita1\": {\n    \"type\": \"datetime\",\n    \"value\": \"1987-07-05T17:45:56.1234+08:00\"\n  },\n  \"wita2\": {\n    \"type\": \"datetime\",\n    \"value\": \"1987-07-05T17:45:56.6000+08:00\"\n  }\n}\n"
 	testgenValid(t, input, jsonRef)
 }
 
@@ -1360,6 +1541,12 @@ func TestTOMLTest_Valid_String_Empty(t *testing.T) {
 	testgenValid(t, input, jsonRef)
 }
 
+func TestTOMLTest_Valid_String_EscapeEsc(t *testing.T) {
+	input := "esc = \"\\e There is no escape! \\e\"\n"
+	jsonRef := "{\n  \"esc\": {\n    \"type\": \"string\",\n    \"value\": \"\\u001b There is no escape! \\u001b\"\n  }\n}\n"
+	testgenValid(t, input, jsonRef)
+}
+
 func TestTOMLTest_Valid_String_EscapeTricky(t *testing.T) {
 	input := "end_esc = \"String does not end here\\\" but ends here\\\\\"\nlit_end_esc = 'String ends here\\'\n\nmultiline_unicode = \"\"\"\n\\u00a0\"\"\"\n\nmultiline_not_unicode = \"\"\"\n\\\\u0041\"\"\"\n\nmultiline_end_esc = \"\"\"When will it end? \\\"\"\"...\"\"\\\" should be here\\\"\"\"\"\n\nlit_multiline_not_unicode = '''\n\\u007f'''\n\nlit_multiline_end = '''There is no escape\\'''\n"
 	jsonRef := "{\n  \"end_esc\": {\n    \"type\": \"string\",\n    \"value\": \"String does not end here\\\" but ends here\\\\\"\n  },\n  \"lit_end_esc\": {\n    \"type\": \"string\",\n    \"value\": \"String ends here\\\\\"\n  },\n  \"lit_multiline_end\": {\n    \"type\": \"string\",\n    \"value\": \"There is no escape\\\\\"\n  },\n  \"lit_multiline_not_unicode\": {\n    \"type\": \"string\",\n    \"value\": \"\\\\u007f\"\n  },\n  \"multiline_end_esc\": {\n    \"type\": \"string\",\n    \"value\": \"When will it end? \\\"\\\"\\\"...\\\"\\\"\\\" should be here\\\"\"\n  },\n  \"multiline_not_unicode\": {\n    \"type\": \"string\",\n    \"value\": \"\\\\u0041\"\n  },\n  \"multiline_unicode\": {\n    \"type\": \"string\",\n    \"value\": \"\u00a0\"\n  }\n}\n"
@@ -1378,14 +1565,20 @@ func TestTOMLTest_Valid_String_Escapes(t *testing.T) {
 	testgenValid(t, input, jsonRef)
 }
 
+func TestTOMLTest_Valid_String_MultilineEscapedCrlf(t *testing.T) {
+	input := "# The following line should be an unescaped backslash followed by a Windows\r\n# newline sequence (\"\\r\\n\")\r\n0=\"\"\"\\\r\n\"\"\"\r\n"
+	jsonRef := "{\n  \"0\": {\n    \"type\": \"string\",\n    \"value\": \"\"\n  }\n}\n"
+	testgenValid(t, input, jsonRef)
+}
+
 func TestTOMLTest_Valid_String_MultilineQuotes(t *testing.T) {
-	input := "# Make sure that quotes inside multiline strings are allowed, including right\n# after the opening '''/\"\"\" and before the closing '''/\"\"\"\n\nlit_one = ''''one quote''''\nlit_two = '''''two quotes'''''\nlit_one_space = ''' 'one quote' '''\nlit_two_space = ''' ''two quotes'' '''\n\none = \"\"\"\"one quote\"\"\"\"\ntwo = \"\"\"\"\"two quotes\"\"\"\"\"\none_space = \"\"\" \"one quote\" \"\"\"\ntwo_space = \"\"\" \"\"two quotes\"\" \"\"\"\n\nmismatch1 = \"\"\"aaa'''bbb\"\"\"\nmismatch2 = '''aaa\"\"\"bbb'''\n"
-	jsonRef := "{\n  \"lit_one\": {\n    \"type\": \"string\",\n    \"value\": \"'one quote'\"\n  },\n  \"lit_one_space\": {\n    \"type\": \"string\",\n    \"value\": \" 'one quote' \"\n  },\n  \"lit_two\": {\n    \"type\": \"string\",\n    \"value\": \"''two quotes''\"\n  },\n  \"lit_two_space\": {\n    \"type\": \"string\",\n    \"value\": \" ''two quotes'' \"\n  },\n  \"mismatch1\": {\n    \"type\": \"string\",\n    \"value\": \"aaa'''bbb\"\n  },\n  \"mismatch2\": {\n    \"type\": \"string\",\n    \"value\": \"aaa\\\"\\\"\\\"bbb\"\n  },\n  \"one\": {\n    \"type\": \"string\",\n    \"value\": \"\\\"one quote\\\"\"\n  },\n  \"one_space\": {\n    \"type\": \"string\",\n    \"value\": \" \\\"one quote\\\" \"\n  },\n  \"two\": {\n    \"type\": \"string\",\n    \"value\": \"\\\"\\\"two quotes\\\"\\\"\"\n  },\n  \"two_space\": {\n    \"type\": \"string\",\n    \"value\": \" \\\"\\\"two quotes\\\"\\\" \"\n  }\n}\n"
+	input := "# Make sure that quotes inside multiline strings are allowed, including right\n# after the opening '''/\"\"\" and before the closing '''/\"\"\"\n\nlit_one = ''''one quote''''\nlit_two = '''''two quotes'''''\nlit_one_space = ''' 'one quote' '''\nlit_two_space = ''' ''two quotes'' '''\n\none = \"\"\"\"one quote\"\"\"\"\ntwo = \"\"\"\"\"two quotes\"\"\"\"\"\none_space = \"\"\" \"one quote\" \"\"\"\ntwo_space = \"\"\" \"\"two quotes\"\" \"\"\"\n\nmismatch1 = \"\"\"aaa'''bbb\"\"\"\nmismatch2 = '''aaa\"\"\"bbb'''\n\n# Three opening \"\"\", then one escaped \", then two \"\" (allowed), and then three\n# closing \"\"\"\nescaped = \"\"\"lol\\\"\"\"\"\"\"\n"
+	jsonRef := "{\n  \"escaped\": {\n    \"type\": \"string\",\n    \"value\": \"lol\\\"\\\"\\\"\"\n  },\n  \"lit_one\": {\n    \"type\": \"string\",\n    \"value\": \"'one quote'\"\n  },\n  \"lit_one_space\": {\n    \"type\": \"string\",\n    \"value\": \" 'one quote' \"\n  },\n  \"lit_two\": {\n    \"type\": \"string\",\n    \"value\": \"''two quotes''\"\n  },\n  \"lit_two_space\": {\n    \"type\": \"string\",\n    \"value\": \" ''two quotes'' \"\n  },\n  \"mismatch1\": {\n    \"type\": \"string\",\n    \"value\": \"aaa'''bbb\"\n  },\n  \"mismatch2\": {\n    \"type\": \"string\",\n    \"value\": \"aaa\\\"\\\"\\\"bbb\"\n  },\n  \"one\": {\n    \"type\": \"string\",\n    \"value\": \"\\\"one quote\\\"\"\n  },\n  \"one_space\": {\n    \"type\": \"string\",\n    \"value\": \" \\\"one quote\\\" \"\n  },\n  \"two\": {\n    \"type\": \"string\",\n    \"value\": \"\\\"\\\"two quotes\\\"\\\"\"\n  },\n  \"two_space\": {\n    \"type\": \"string\",\n    \"value\": \" \\\"\\\"two quotes\\\"\\\" \"\n  }\n}\n"
 	testgenValid(t, input, jsonRef)
 }
 
 func TestTOMLTest_Valid_String_Multiline(t *testing.T) {
-	input := "# NOTE: this file includes some literal tab characters.\n\nmultiline_empty_one = \"\"\"\"\"\"\nmultiline_empty_two = \"\"\"\n\"\"\"\nmultiline_empty_three = \"\"\"\\\n    \"\"\"\nmultiline_empty_four = \"\"\"\\\n   \\\n   \\  \n   \"\"\"\n\nequivalent_one = \"The quick brown fox jumps over the lazy dog.\"\nequivalent_two = \"\"\"\nThe quick brown \\\n\n\n  fox jumps over \\\n    the lazy dog.\"\"\"\n\nequivalent_three = \"\"\"\\\n       The quick brown \\\n       fox jumps over \\\n       the lazy dog.\\\n       \"\"\"\n\nwhitespace-after-bs = \"\"\"\\\n       The quick brown \\\n       fox jumps over \\   \n       the lazy dog.\\\t\n       \"\"\"\n\nno-space = \"\"\"a\\\n    b\"\"\"\n\nkeep-ws-before = \"\"\"a   \t\\\n   b\"\"\"\n\nescape-bs-1 = \"\"\"a \\\\\nb\"\"\"\n\nescape-bs-2 = \"\"\"a \\\\\\\nb\"\"\"\n\nescape-bs-3 = \"\"\"a \\\\\\\\\n  b\"\"\"\n"
+	input := "# NOTE: this file includes some literal tab characters.\n\nmultiline_empty_one = \"\"\"\"\"\"\n\n# A newline immediately following the opening delimiter will be trimmed.\nmultiline_empty_two = \"\"\"\n\"\"\"\n\n# \\ at the end of line trims newlines as well; note that last \\ is followed by\n# two spaces, which are ignored.\nmultiline_empty_three = \"\"\"\\\n    \"\"\"\nmultiline_empty_four = \"\"\"\\\n   \\\n   \\  \n   \"\"\"\n\nequivalent_one = \"The quick brown fox jumps over the lazy dog.\"\nequivalent_two = \"\"\"\nThe quick brown \\\n\n\n  fox jumps over \\\n    the lazy dog.\"\"\"\n\nequivalent_three = \"\"\"\\\n       The quick brown \\\n       fox jumps over \\\n       the lazy dog.\\\n       \"\"\"\n\nwhitespace-after-bs = \"\"\"\\\n       The quick brown \\\n       fox jumps over \\   \n       the lazy dog.\\\t\n       \"\"\"\n\nno-space = \"\"\"a\\\n    b\"\"\"\n\n# Has tab character.\nkeep-ws-before = \"\"\"a   \t\\\n   b\"\"\"\n\nescape-bs-1 = \"\"\"a \\\\\nb\"\"\"\n\nescape-bs-2 = \"\"\"a \\\\\\\nb\"\"\"\n\nescape-bs-3 = \"\"\"a \\\\\\\\\n  b\"\"\"\n"
 	jsonRef := "{\n  \"equivalent_one\": {\n    \"type\": \"string\",\n    \"value\": \"The quick brown fox jumps over the lazy dog.\"\n  },\n  \"equivalent_three\": {\n    \"type\": \"string\",\n    \"value\": \"The quick brown fox jumps over the lazy dog.\"\n  },\n  \"equivalent_two\": {\n    \"type\": \"string\",\n    \"value\": \"The quick brown fox jumps over the lazy dog.\"\n  },\n  \"escape-bs-1\": {\n    \"type\": \"string\",\n    \"value\": \"a \\\\\\nb\"\n  },\n  \"escape-bs-2\": {\n    \"type\": \"string\",\n    \"value\": \"a \\\\b\"\n  },\n  \"escape-bs-3\": {\n    \"type\": \"string\",\n    \"value\": \"a \\\\\\\\\\n  b\"\n  },\n  \"keep-ws-before\": {\n    \"type\": \"string\",\n    \"value\": \"a   \\tb\"\n  },\n  \"multiline_empty_four\": {\n    \"type\": \"string\",\n    \"value\": \"\"\n  },\n  \"multiline_empty_one\": {\n    \"type\": \"string\",\n    \"value\": \"\"\n  },\n  \"multiline_empty_three\": {\n    \"type\": \"string\",\n    \"value\": \"\"\n  },\n  \"multiline_empty_two\": {\n    \"type\": \"string\",\n    \"value\": \"\"\n  },\n  \"no-space\": {\n    \"type\": \"string\",\n    \"value\": \"ab\"\n  },\n  \"whitespace-after-bs\": {\n    \"type\": \"string\",\n    \"value\": \"The quick brown fox jumps over the lazy dog.\"\n  }\n}\n"
 	testgenValid(t, input, jsonRef)
 }
@@ -1397,7 +1590,7 @@ func TestTOMLTest_Valid_String_Nl(t *testing.T) {
 }
 
 func TestTOMLTest_Valid_String_RawMultiline(t *testing.T) {
-	input := "oneline = '''This string has a ' quote character.'''\nfirstnl = '''\nThis string has a ' quote character.'''\nmultiline = '''\nThis string\nhas ' a quote character\nand more than\none newline\nin it.'''\n"
+	input := "# Single ' should be allowed.\noneline = '''This string has a ' quote character.'''\n\n# A newline immediately following the opening delimiter will be trimmed.\nfirstnl = '''\nThis string has a ' quote character.'''\n\n# All other whitespace and newline characters remain intact.\nmultiline = '''\nThis string\nhas ' a quote character\nand more than\none newline\nin it.'''\n"
 	jsonRef := "{\n  \"firstnl\": {\n    \"type\": \"string\",\n    \"value\": \"This string has a ' quote character.\"\n  },\n  \"multiline\": {\n    \"type\": \"string\",\n    \"value\": \"This string\\nhas ' a quote character\\nand more than\\none newline\\nin it.\"\n  },\n  \"oneline\": {\n    \"type\": \"string\",\n    \"value\": \"This string has a ' quote character.\"\n  }\n}\n"
 	testgenValid(t, input, jsonRef)
 }

types.go

@@ -11,3 +11,4 @@ var textMarshalerType = reflect.TypeOf(new(encoding.TextMarshaler)).Elem()
 var textUnmarshalerType = reflect.TypeOf(new(encoding.TextUnmarshaler)).Elem()
 var mapStringInterfaceType = reflect.TypeOf(map[string]interface{}{})
 var sliceInterfaceType = reflect.TypeOf([]interface{}{})
+var stringType = reflect.TypeOf("")

unmarshaler.go

@@ -9,10 +9,11 @@ import (
 	"math"
 	"reflect"
 	"strings"
-	"sync"
+	"sync/atomic"
 	"time"
 
 	"github.com/pelletier/go-toml/v2/internal/ast"
+	"github.com/pelletier/go-toml/v2/internal/danger"
 	"github.com/pelletier/go-toml/v2/internal/tracker"
 )
 
@@ -41,24 +42,27 @@ func NewDecoder(r io.Reader) *Decoder {
 	return &Decoder{r: r}
 }
 
-// SetStrict toggles decoding in stict mode.
+// DisallowUnknownFields causes the Decoder to return an error when the
+// destination is a struct and the input contains a key that does not match a
+// non-ignored field.
 //
-// When the decoder is in strict mode, it will record fields from the document
-// that could not be set on the target value. In that case, the decoder returns
-// a StrictMissingError that can be used to retrieve the individual errors as
-// well as generate a human readable description of the missing fields.
-func (d *Decoder) SetStrict(strict bool) {
-	d.strict = strict
+// In that case, the Decoder returns a StrictMissingError that can be used to
+// retrieve the individual errors as well as generate a human readable
+// description of the missing fields.
+func (d *Decoder) DisallowUnknownFields() *Decoder {
+	d.strict = true
+	return d
 }
 
 // Decode the whole content of r into v.
 //
 // By default, values in the document that don't exist in the target Go value
-// are ignored. See Decoder.SetStrict() to change this behavior.
+// are ignored. See Decoder.DisallowUnknownFields() to change this behavior.
 //
 // When a TOML local date, time, or date-time is decoded into a time.Time, its
 // value is represented in time.Local timezone. Otherwise the approriate Local*
-// structure is used.
+// structure is used. For time values, precision up to the nanosecond is
+// supported by truncating extra digits.
 //
 // Empty tables decoded in an interface{} create an empty initialized
 // map[string]interface{}.
@@ -70,6 +74,11 @@ func (d *Decoder) SetStrict(strict bool) {
 // bounds for the target type (which includes negative numbers when decoding
 // into an unsigned int).
 //
+// If an error occurs while decoding the content of the document, this function
+// returns a toml.DecodeError, providing context about the issue. When using
+// strict mode and a field is missing, a `toml.StrictMissingError` is
+// returned. In any other case, this function returns a standard Go error.
+//
 // Type mapping
 //
 // List of supported TOML types and their associated accepted Go types:
@@ -129,6 +138,23 @@ type decoder struct {
 
 	// Strict mode
 	strict strict
+
+	// Current context for the error.
+	errorContext *errorContext
+}
+
+type errorContext struct {
+	Struct reflect.Type
+	Field  []int
+}
+
+func (d *decoder) typeMismatchError(toml string, target reflect.Type) error {
+	if d.errorContext != nil && d.errorContext.Struct != nil {
+		ctx := d.errorContext
+		f := ctx.Struct.FieldByIndex(ctx.Field)
+		return fmt.Errorf("toml: cannot decode TOML %s into struct field %s.%s of type %s", toml, ctx.Struct, f.Name, f.Type)
+	}
+	return fmt.Errorf("toml: cannot decode TOML %s into a Go value of type %s", toml, target)
 }
 
 func (d *decoder) expr() *ast.Node {
@@ -204,15 +230,6 @@ func (d *decoder) fromParser(root reflect.Value) error {
 	return d.p.Error()
 }
 
-/*
-Rules for the unmarshal code:
-
-- The stack is used to keep track of which values need to be set where.
-- handle* functions <=> switch on a given ast.Kind.
-- unmarshalX* functions need to unmarshal a node of kind X.
-- An "object" is either a struct or a map.
-*/
-
 func (d *decoder) handleRootExpression(expr *ast.Node, v reflect.Value) error {
 	var x reflect.Value
 	var err error
@@ -296,10 +313,12 @@ func (d *decoder) handleArrayTableCollectionLast(key ast.Iterator, v reflect.Val
 		return v, nil
 	case reflect.Slice:
 		elemType := v.Type().Elem()
+		var elem reflect.Value
 		if elemType.Kind() == reflect.Interface {
-			elemType = mapStringInterfaceType
+			elem = makeMapStringInterface()
+		} else {
+			elem = reflect.New(elemType).Elem()
 		}
-		elem := reflect.New(elemType).Elem()
 		elem2, err := d.handleArrayTable(key, elem)
 		if err != nil {
 			return reflect.Value{}, err
@@ -343,7 +362,9 @@ func (d *decoder) handleArrayTableCollection(key ast.Iterator, v reflect.Value)
 		if err != nil {
 			return reflect.Value{}, err
 		}
-		v.Elem().Set(elem)
+		if elem.IsValid() {
+			v.Elem().Set(elem)
+		}
 
 		return v, nil
 	case reflect.Slice:
@@ -370,6 +391,84 @@ func (d *decoder) handleArrayTableCollection(key ast.Iterator, v reflect.Value)
 	return d.handleArrayTable(key, v)
 }
 
+func (d *decoder) handleKeyValuePartMapStringInterface(key ast.Iterator, value *ast.Node, m map[string]interface{}) (reflect.Value, error) {
+	k := string(key.Node().Data)
+
+	newMap := false
+	if m == nil {
+		newMap = true
+		m = make(map[string]interface{}, 8)
+	}
+
+	set := false
+	v, ok := m[k]
+	if !ok || key.IsLast() {
+		set = true
+		v = nil
+	}
+
+	mv := reflect.ValueOf(&v).Elem()
+
+	x, err := d.handleKeyValueInner(key, value, mv)
+	if err != nil {
+		return reflect.Value{}, err
+	}
+	if x.IsValid() {
+		mv = x
+		set = true
+	}
+
+	if set {
+		m[k] = mv.Interface()
+	}
+
+	if newMap {
+		return reflect.ValueOf(m), nil
+	}
+
+	return reflect.Value{}, nil
+}
+
+func (d *decoder) handleKeyPartMapStringInterface(key ast.Iterator, m map[string]interface{}, nextFn handlerFn, makeFn valueMakerFn) (reflect.Value, error) {
+	newMap := false
+
+	k := string(key.Node().Data)
+	if m == nil {
+		newMap = true
+		m = make(map[string]interface{}, 8)
+	}
+
+	v, ok := m[k]
+	set := false
+
+	if !ok || v == nil {
+		set = true
+		v = makeFn().Interface()
+	}
+
+	mv := reflect.ValueOf(v)
+
+	x, err := nextFn(key, mv)
+	if err != nil {
+		return reflect.Value{}, err
+	}
+
+	if x.IsValid() {
+		mv = x
+		set = true
+	}
+
+	if set {
+		m[k] = mv.Interface()
+	}
+
+	if newMap {
+		return reflect.ValueOf(m), nil
+	}
+
+	return reflect.Value{}, nil
+}
+
 func (d *decoder) handleKeyPart(key ast.Iterator, v reflect.Value, nextFn handlerFn, makeFn valueMakerFn) (reflect.Value, error) {
 	var rv reflect.Value
 
@@ -384,12 +483,20 @@ func (d *decoder) handleKeyPart(key ast.Iterator, v reflect.Value, nextFn handle
 		elem = v.Elem()
 		return d.handleKeyPart(key, elem, nextFn, makeFn)
 	case reflect.Map:
-		// Create the key for the map element. For now assume it's a string.
-		mk := reflect.ValueOf(string(key.Node().Data))
+		vt := v.Type()
+
+		if vt == mapStringInterfaceType {
+			m := v.Interface().(map[string]interface{})
+			return d.handleKeyPartMapStringInterface(key, m, nextFn, makeFn)
+		}
+
+		// Create the key for the map element. Convert to key type.
+		mk := reflect.ValueOf(string(key.Node().Data)).Convert(vt.Key())
 
 		// If the map does not exist, create it.
 		if v.IsNil() {
-			v = reflect.MakeMap(v.Type())
+			vt := v.Type()
+			v = reflect.MakeMap(vt)
 			rv = v
 		}
 
@@ -401,7 +508,7 @@ func (d *decoder) handleKeyPart(key ast.Iterator, v reflect.Value, nextFn handle
 			// map[string]interface{} or a []interface{} depending on whether
 			// this is the last part of the array table key.
 
-			t := v.Type().Elem()
+			t := vt.Elem()
 			if t.Kind() == reflect.Interface {
 				mv = makeFn()
 			} else {
@@ -415,7 +522,8 @@ func (d *decoder) handleKeyPart(key ast.Iterator, v reflect.Value, nextFn handle
 			}
 			set = true
 		} else if !mv.CanAddr() {
-			t := v.Type().Elem()
+			vt := v.Type()
+			t := vt.Elem()
 			oldmv := mv
 			mv = reflect.New(t).Elem()
 			mv.Set(oldmv)
@@ -436,12 +544,20 @@ func (d *decoder) handleKeyPart(key ast.Iterator, v reflect.Value, nextFn handle
 			v.SetMapIndex(mk, mv)
 		}
 	case reflect.Struct:
-		f, found := structField(v, string(key.Node().Data))
+		path, found := structFieldPath(v, string(key.Node().Data))
 		if !found {
 			d.skipUntilTable = true
 			return reflect.Value{}, nil
 		}
 
+		if d.errorContext == nil {
+			d.errorContext = new(errorContext)
+		}
+		t := v.Type()
+		d.errorContext.Struct = t
+		d.errorContext.Field = path
+
+		f := v.FieldByIndex(path)
 		x, err := nextFn(key, f)
 		if err != nil || d.skipUntilTable {
 			return reflect.Value{}, err
@@ -449,11 +565,13 @@ func (d *decoder) handleKeyPart(key ast.Iterator, v reflect.Value, nextFn handle
 		if x.IsValid() {
 			f.Set(x)
 		}
+		d.errorContext.Field = nil
+		d.errorContext.Struct = nil
 	case reflect.Interface:
 		if v.Elem().IsValid() {
 			v = v.Elem()
 		} else {
-			v = reflect.MakeMap(mapStringInterfaceType)
+			v = makeMapStringInterface()
 		}
 
 		x, err := d.handleKeyPart(key, v, nextFn, makeFn)
@@ -569,7 +687,7 @@ func (d *decoder) tryTextUnmarshaler(node *ast.Node, v reflect.Value) (bool, err
 	if v.CanAddr() && v.Addr().Type().Implements(textUnmarshalerType) {
 		err := v.Addr().Interface().(encoding.TextUnmarshaler).UnmarshalText(node.Data)
 		if err != nil {
-			return false, newDecodeError(d.p.Raw(node.Raw), "error calling UnmarshalText: %w", err)
+			return false, newDecodeError(d.p.Raw(node.Raw), "%w", err)
 		}
 
 		return true, nil
@@ -649,7 +767,7 @@ func (d *decoder) unmarshalArray(array *ast.Node, v reflect.Value) error {
 	default:
 		// TODO: use newDecodeError, but first the parser needs to fill
 		//   array.Data.
-		return fmt.Errorf("toml: cannot store array in Go type %s", v.Kind())
+		return d.typeMismatchError("array", v.Type())
 	}
 
 	elemType := v.Type().Elem()
@@ -697,7 +815,7 @@ func (d *decoder) unmarshalInlineTable(itable *ast.Node, v reflect.Value) error
 	case reflect.Interface:
 		elem := v.Elem()
 		if !elem.IsValid() {
-			elem = reflect.MakeMap(mapStringInterfaceType)
+			elem = makeMapStringInterface()
 			v.Set(elem)
 		}
 		return d.unmarshalInlineTable(itable, elem)
@@ -896,7 +1014,7 @@ func (d *decoder) unmarshalInteger(value *ast.Node, v reflect.Value) error {
 	case reflect.Interface:
 		r = reflect.ValueOf(i)
 	default:
-		return fmt.Errorf("toml: cannot store TOML integer into a Go %s", v.Kind())
+		return d.typeMismatchError("integer", v.Type())
 	}
 
 	if !r.Type().AssignableTo(v.Type()) {
@@ -953,12 +1071,20 @@ func (d *decoder) handleKeyValuePart(key ast.Iterator, value *ast.Node, v reflec
 	// There is no guarantee over what it could be.
 	switch v.Kind() {
 	case reflect.Map:
-		mk := reflect.ValueOf(string(key.Node().Data))
+		vt := v.Type()
 
-		keyType := v.Type().Key()
-		if !mk.Type().AssignableTo(keyType) {
-			if !mk.Type().ConvertibleTo(keyType) {
-				return reflect.Value{}, fmt.Errorf("toml: cannot convert map key of type %s to expected type %s", mk.Type(), keyType)
+		if vt == mapStringInterfaceType {
+			m := v.Interface().(map[string]interface{})
+			return d.handleKeyValuePartMapStringInterface(key, value, m)
+		}
+
+		mk := reflect.ValueOf(string(key.Node().Data))
+		mkt := stringType
+
+		keyType := vt.Key()
+		if !mkt.AssignableTo(keyType) {
+			if !mkt.ConvertibleTo(keyType) {
+				return reflect.Value{}, fmt.Errorf("toml: cannot convert map key of type %s to expected type %s", mkt, keyType)
 			}
 
 			mk = mk.Convert(keyType)
@@ -966,7 +1092,7 @@ func (d *decoder) handleKeyValuePart(key ast.Iterator, value *ast.Node, v reflec
 
 		// If the map does not exist, create it.
 		if v.IsNil() {
-			v = reflect.MakeMap(v.Type())
+			v = reflect.MakeMap(vt)
 			rv = v
 		}
 
@@ -975,12 +1101,10 @@ func (d *decoder) handleKeyValuePart(key ast.Iterator, value *ast.Node, v reflec
 		if !mv.IsValid() {
 			set = true
 			mv = reflect.New(v.Type().Elem()).Elem()
-		} else {
-			if key.IsLast() {
-				var x interface{}
-				mv = reflect.ValueOf(&x).Elem()
-				set = true
-			}
+		} else if key.IsLast() {
+			var x interface{}
+			mv = reflect.ValueOf(&x).Elem()
+			set = true
 		}
 
 		nv, err := d.handleKeyValueInner(key, value, mv)
@@ -996,12 +1120,20 @@ func (d *decoder) handleKeyValuePart(key ast.Iterator, value *ast.Node, v reflec
 			v.SetMapIndex(mk, mv)
 		}
 	case reflect.Struct:
-		f, found := structField(v, string(key.Node().Data))
+		path, found := structFieldPath(v, string(key.Node().Data))
 		if !found {
 			d.skipUntilTable = true
 			break
 		}
 
+		if d.errorContext == nil {
+			d.errorContext = new(errorContext)
+		}
+		t := v.Type()
+		d.errorContext.Struct = t
+		d.errorContext.Field = path
+
+		f := v.FieldByIndex(path)
 		x, err := d.handleKeyValueInner(key, value, f)
 		if err != nil {
 			return reflect.Value{}, err
@@ -1010,14 +1142,17 @@ func (d *decoder) handleKeyValuePart(key ast.Iterator, value *ast.Node, v reflec
 		if x.IsValid() {
 			f.Set(x)
 		}
+		d.errorContext.Struct = nil
+		d.errorContext.Field = nil
 	case reflect.Interface:
 		v = v.Elem()
 
-		// Following encoding/toml: decoding an object into an interface{}, it
-		// needs to always hold a map[string]interface{}. This is for the types
-		// to be consistent whether a previous value was set or not.
+		// Following encoding/json: decoding an object into an
+		// interface{}, it needs to always hold a
+		// map[string]interface{}. This is for the types to be
+		// consistent whether a previous value was set or not.
 		if !v.IsValid() || v.Type() != mapStringInterfaceType {
-			v = reflect.MakeMap(mapStringInterfaceType)
+			v = makeMapStringInterface()
 		}
 
 		x, err := d.handleKeyValuePart(key, value, v)
@@ -1064,80 +1199,68 @@ func initAndDereferencePointer(v reflect.Value) reflect.Value {
 
 type fieldPathsMap = map[string][]int
 
-type fieldPathsCache struct {
-	m map[reflect.Type]fieldPathsMap
-	l sync.RWMutex
-}
+var globalFieldPathsCache atomic.Value // map[danger.TypeID]fieldPathsMap
 
-func (c *fieldPathsCache) get(t reflect.Type) (fieldPathsMap, bool) {
-	c.l.RLock()
-	paths, ok := c.m[t]
-	c.l.RUnlock()
+func structFieldPath(v reflect.Value, name string) ([]int, bool) {
+	t := v.Type()
 
-	return paths, ok
-}
+	cache, _ := globalFieldPathsCache.Load().(map[danger.TypeID]fieldPathsMap)
+	fieldPaths, ok := cache[danger.MakeTypeID(t)]
 
-func (c *fieldPathsCache) set(t reflect.Type, m fieldPathsMap) {
-	c.l.Lock()
-	c.m[t] = m
-	c.l.Unlock()
-}
-
-var globalFieldPathsCache = fieldPathsCache{
-	m: map[reflect.Type]fieldPathsMap{},
-	l: sync.RWMutex{},
-}
-
-func structField(v reflect.Value, name string) (reflect.Value, bool) {
-	//nolint:godox
-	// TODO: cache this, and reduce allocations
-	fieldPaths, ok := globalFieldPathsCache.get(v.Type())
 	if !ok {
 		fieldPaths = map[string][]int{}
 
-		path := make([]int, 0, 16)
+		forEachField(t, nil, func(name string, path []int) {
+			fieldPaths[name] = path
+			// extra copy for the case-insensitive match
+			fieldPaths[strings.ToLower(name)] = path
+		})
 
-		var walk func(reflect.Value)
-		walk = func(v reflect.Value) {
-			t := v.Type()
-			for i := 0; i < t.NumField(); i++ {
-				l := len(path)
-				path = append(path, i)
-				f := t.Field(i)
-
-				if f.Anonymous {
-					walk(v.Field(i))
-				} else if f.PkgPath == "" {
-					// only consider exported fields
-					fieldName, ok := f.Tag.Lookup("toml")
-					if !ok {
-						fieldName = f.Name
-					}
-
-					pathCopy := make([]int, len(path))
-					copy(pathCopy, path)
-
-					fieldPaths[fieldName] = pathCopy
-					// extra copy for the case-insensitive match
-					fieldPaths[strings.ToLower(fieldName)] = pathCopy
-				}
-				path = path[:l]
-			}
+		newCache := make(map[danger.TypeID]fieldPathsMap, len(cache)+1)
+		newCache[danger.MakeTypeID(t)] = fieldPaths
+		for k, v := range cache {
+			newCache[k] = v
 		}
-
-		walk(v)
-
-		globalFieldPathsCache.set(v.Type(), fieldPaths)
+		globalFieldPathsCache.Store(newCache)
 	}
 
 	path, ok := fieldPaths[name]
 	if !ok {
 		path, ok = fieldPaths[strings.ToLower(name)]
 	}
-
-	if !ok {
-		return reflect.Value{}, false
-	}
-
-	return v.FieldByIndex(path), true
+	return path, ok
+}
+
+func forEachField(t reflect.Type, path []int, do func(name string, path []int)) {
+	n := t.NumField()
+	for i := 0; i < n; i++ {
+		f := t.Field(i)
+
+		if !f.Anonymous && f.PkgPath != "" {
+			// only consider exported fields.
+			continue
+		}
+
+		fieldPath := append(path, i)
+		fieldPath = fieldPath[:len(fieldPath):len(fieldPath)]
+
+		if f.Anonymous {
+			forEachField(f.Type, fieldPath, do)
+			continue
+		}
+
+		name := f.Tag.Get("toml")
+		if name == "-" {
+			continue
+		}
+
+		if i := strings.IndexByte(name, ','); i >= 0 {
+			name = name[:i]
+		}
+		if name == "" {
+			name = f.Name
+		}
+
+		do(name, fieldPath)
+	}
 }

utf8.go

@@ -140,8 +140,45 @@ func utf8ValidNext(p []byte) int {
 	return size
 }
 
+var invalidAsciiTable = [256]bool{
+	0x00: true,
+	0x01: true,
+	0x02: true,
+	0x03: true,
+	0x04: true,
+	0x05: true,
+	0x06: true,
+	0x07: true,
+	0x08: true,
+	// 0x09 TAB
+	// 0x0A LF
+	0x0B: true,
+	0x0C: true,
+	// 0x0D CR
+	0x0E: true,
+	0x0F: true,
+	0x10: true,
+	0x11: true,
+	0x12: true,
+	0x13: true,
+	0x14: true,
+	0x15: true,
+	0x16: true,
+	0x17: true,
+	0x18: true,
+	0x19: true,
+	0x1A: true,
+	0x1B: true,
+	0x1C: true,
+	0x1D: true,
+	0x1E: true,
+	0x1F: true,
+	// 0x20 - 0x7E Printable ASCII characters
+	0x7F: true,
+}
+
 func invalidAscii(b byte) bool {
-	return b <= 0x08 || (b > 0x0A && b < 0x0D) || (b > 0x0D && b <= 0x1F) || b == 0x7F
+	return invalidAsciiTable[b]
 }
 
 // acceptRange gives the range of valid values for the second byte in a UTF-8