Compare commits

...

2 Commits

Author SHA1 Message Date
Claude b7ffaf15eb Move regression tests to public Unmarshal API with full error string assertions
Remove the unstable package test and consolidate all test cases into
TestDecodeError_PositionAfterComment, which exercises toml.Unmarshal
and validates the complete human-readable error output including
context lines and tilde markers.

https://claude.ai/code/session_01EXYfFXc3DDGpQ27sWdXTKq
2026-04-12 11:59:46 +00:00
Claude 0248fc4c8c Fix incorrect error positions for non-suffix subslices in unstable.Parser.Range (#1047)
The unsafe removal (#1021) replaced danger.SubsliceOffset (pointer
arithmetic) with len(p.data)-len(b), which only works for suffix
slices. Parser.Range is called with arbitrary interior subslices
(e.g. ParserError.Highlight), so the offset was wrong whenever the
error occurred after previously scanned content like comments.

Fix by using reflect.ValueOf().Pointer() to recover the actual data
pointer, matching the approach already used in errors.go.

https://claude.ai/code/session_01EXYfFXc3DDGpQ27sWdXTKq
2026-04-12 11:52:16 +00:00
2 changed files with 75 additions and 3 deletions
+67
View File
@@ -301,6 +301,73 @@ OtherMissing = 1
assert.Equal(t, 2, len(strictErr.Unwrap())) assert.Equal(t, 2, len(strictErr.Unwrap()))
} }
func TestDecodeError_PositionAfterComment(t *testing.T) {
// Regression test for https://github.com/pelletier/go-toml/issues/1047
// Error positions must be correct when the error occurs after comments or
// other content that was already scanned past.
examples := []struct {
desc string
doc string
expectedRow int
expectedCol int
expectedStr string
}{
{
desc: "invalid key after comment",
doc: "# comment\n= \"value\"",
expectedRow: 2,
expectedCol: 1,
expectedStr: "1| # comment\n2| = \"value\"\n | ~ invalid character at start of key: =",
},
{
desc: "invalid key after two comments",
doc: "# one\n# two\n= \"value\"",
expectedRow: 3,
expectedCol: 1,
expectedStr: "1| # one\n2| # two\n3| = \"value\"\n | ~ invalid character at start of key: =",
},
{
desc: "invalid key after key-value pair",
doc: "a = 1\n= 2",
expectedRow: 2,
expectedCol: 1,
expectedStr: "1| a = 1\n2| = 2\n | ~ invalid character at start of key: =",
},
{
desc: "invalid key after blank line",
doc: "a = 1\n\n= 2",
expectedRow: 3,
expectedCol: 1,
expectedStr: "1| a = 1\n2|\n3| = 2\n | ~ invalid character at start of key: =",
},
}
for _, e := range examples {
t.Run(e.desc, func(t *testing.T) {
var v interface{}
err := Unmarshal([]byte(e.doc), &v)
if err == nil {
t.Fatal("expected an error")
}
var derr *DecodeError
if !errors.As(err, &derr) {
t.Fatalf("error not a *DecodeError: %T: %v", err, err)
}
row, col := derr.Position()
if row != e.expectedRow {
t.Errorf("row: got %d, want %d (error: %s)", row, e.expectedRow, derr.String())
}
if col != e.expectedCol {
t.Errorf("col: got %d, want %d (error: %s)", col, e.expectedCol, derr.String())
}
assert.Equal(t, e.expectedStr, derr.String())
})
}
}
func ExampleDecodeError() { func ExampleDecodeError() {
doc := `name = 123__456` doc := `name = 123__456`
+8 -3
View File
@@ -3,6 +3,7 @@ package unstable
import ( import (
"bytes" "bytes"
"fmt" "fmt"
"reflect"
"unicode" "unicode"
"github.com/pelletier/go-toml/v2/internal/characters" "github.com/pelletier/go-toml/v2/internal/characters"
@@ -83,10 +84,14 @@ func (p *Parser) rangeOfToken(token, rest []byte) Range {
} }
// subsliceOffset returns the byte offset of subslice b within p.data. // subsliceOffset returns the byte offset of subslice b within p.data.
// b must be a suffix (tail) of p.data. // b must be a subslice of p.data (sharing the same backing array).
func (p *Parser) subsliceOffset(b []byte) int { func (p *Parser) subsliceOffset(b []byte) int {
// b is a suffix of p.data, so its offset is len(p.data) - len(b) if len(b) == 0 {
return len(p.data) - len(b) return 0
}
dataPtr := reflect.ValueOf(p.data).Pointer()
subPtr := reflect.ValueOf(b).Pointer()
return int(subPtr - dataPtr)
} }
// Raw returns the slice corresponding to the bytes in the given range. // Raw returns the slice corresponding to the bytes in the given range.