Linear array storage for AST

This commit is contained in:
Thomas Pelletier
2021-03-25 19:56:02 -04:00
parent 8a8d1233bb
commit 1bae751a45
7 changed files with 659 additions and 460 deletions
+100 -177
View File
@@ -2,182 +2,103 @@ package ast
import (
"fmt"
"strings"
)
type Kind int
const (
// meta
Invalid Kind = iota
Comment
Key
// top level structures
Table
ArrayTable
KeyValue
// containers values
Array
InlineTable
// values
String
Bool
Float
Integer
LocalDate
LocalDateTime
DateTime
Time
)
func (k Kind) String() string {
switch k {
case Invalid:
return "Invalid"
case Comment:
return "Comment"
case Key:
return "Key"
case Table:
return "Table"
case ArrayTable:
return "ArrayTable"
case KeyValue:
return "KeyValue"
case Array:
return "Array"
case InlineTable:
return "InlineTable"
case String:
return "String"
case Bool:
return "Bool"
case Float:
return "Float"
case Integer:
return "Integer"
case LocalDate:
return "LocalDate"
case LocalDateTime:
return "LocalDateTime"
case DateTime:
return "DateTime"
case Time:
return "Time"
}
panic(fmt.Errorf("Kind.String() not implemented for '%d'", k))
// Iterator starts uninitialized, you need to call Next() first.
//
// For example:
//
// it := n.Children()
// for it.Next() {
// it.Node()
// }
type Iterator interface {
// Next moves the iterator forward and returns true if points to a node, false
// otherwise.
Next() bool
// Node returns a copy of the node pointed at by the iterator.
Node() Node
}
type Root []Node
// Dot returns a dot representation of the AST for debugging.
func (r Root) Sdot() string {
type edge struct {
from int
childIdx int
to int
}
var nodes []Node
var edges []edge // indexes into nodes
nodes = append(nodes, Node{
Kind: Invalid,
Data: []byte(`ROOT`),
Children: r,
})
var processNode func(int, int, *Node)
processNode = func(parentIdx int, childIdx int, node *Node) {
idx := len(nodes)
nodes = append(nodes, *node)
edges = append(edges, edge{
from: parentIdx,
childIdx: childIdx,
to: idx,
})
for i, c := range node.Children {
processNode(idx, i, &c)
}
}
for i, n := range r {
processNode(0, i, &n)
}
var b strings.Builder
b.WriteString("digraph tree {\n")
b.WriteString("\tnode [shape=record];\n")
for i, node := range nodes {
label := ""
attrs := map[string]string{}
if i == 0 {
var ports []string
for i := 0; i < len(node.Children); i++ {
ports = append(ports, fmt.Sprintf("<f%d> %d", i, i))
}
joinedPorts := strings.Join(ports, "|")
label = fmt.Sprintf("{ROOT|{%s}}", joinedPorts)
} else {
fields := []string{node.Kind.String()}
if len(node.Data) > 0 {
fields = append(fields, string(node.Data))
}
var ports []string
for i := 0; i < len(node.Children); i++ {
ports = append(ports, fmt.Sprintf("<f%d> %d", i, i))
}
joinedPorts := strings.Join(ports, "|")
joinedFields := strings.Join(fields, "|")
label = fmt.Sprintf("{{%s}", joinedFields)
if len(ports) > 0 {
label += fmt.Sprintf("|{%s}", joinedPorts)
}
label += "}"
if node.Kind == Invalid {
attrs["style"] = "filled"
attrs["fillcolor"] = "red"
}
}
_, _ = fmt.Fprintf(&b, "\tnode%d [label=\"%s\"", i, label)
for k, v := range attrs {
_, _ = fmt.Fprintf(&b, ", %s=\"%s\"", k, v)
}
_, _ = fmt.Fprintf(&b, "];\n")
}
b.WriteString("\n")
for _, e := range edges {
_, _ = fmt.Fprintf(&b, "\tnode%d:f%d -> node%d;\n", e.from, e.childIdx, e.to)
}
b.WriteString("}")
return b.String()
type chainIterator struct {
started bool
node Node
}
func (c *chainIterator) Next() bool {
if !c.started {
c.started = true
} else if c.node.Valid() {
c.node = c.node.Next()
}
return c.node.Valid()
}
func (c *chainIterator) Node() Node {
return c.node
}
type Root struct {
nodes []Node
}
func (r *Root) Iterator() Iterator {
it := &chainIterator{}
if len(r.nodes) > 0 {
it.node = r.nodes[0]
}
return it
}
func (r *Root) Reset() {
r.nodes = r.nodes[:0]
}
func (r *Root) at(idx int) Node {
// TODO: unsafe to point to the node directly
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
type Node struct {
Kind Kind
Data []byte // Raw bytes from the input
// 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 last 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
// next idx (in the root array). 0 if last of the collection.
next int
// child idx (in the root array). 0 if no child.
child int
// pointer to the root array
root *Root
}
// 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 NoNode
}
return n.root.at(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.
func (n Node) Child() Node {
if n.child <= 0 {
return NoNode
}
return n.root.at(n.child)
}
func (n Node) Valid() bool {
return n.Kind != Invalid
}
var NoNode = Node{}
@@ -186,15 +107,16 @@ var NoNode = Node{}
// otherwise.
// They are guaranteed to be all be of the Kind Key. A simple key would return
// just one element.
func (n *Node) Key() []Node {
func (n *Node) Key() Iterator {
switch n.Kind {
case KeyValue:
if len(n.Children) < 2 {
panic(fmt.Errorf("KeyValue should have at least two children, not %d", len(n.Children)))
value := n.Child()
if !value.Valid() {
panic(fmt.Errorf("KeyValue should have at least two children"))
}
return n.Children[:len(n.Children)-1]
return &chainIterator{node: value.Next()}
case Table, ArrayTable:
return n.Children
return &chainIterator{node: n.Child()}
default:
panic(fmt.Errorf("Key() is not supported on a %s", n.Kind))
}
@@ -203,15 +125,16 @@ func (n *Node) Key() []Node {
// 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.
func (n *Node) Value() *Node {
func (n Node) Value() Node {
assertKind(KeyValue, n)
if len(n.Children) < 2 {
panic(fmt.Errorf("KeyValue should have at least two children, not %d", len(n.Children)))
}
return &n.Children[len(n.Children)-1]
return n.Child()
}
func assertKind(k Kind, n *Node) {
func (n Node) Children() Iterator {
return &chainIterator{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))
}
+58
View File
@@ -0,0 +1,58 @@
package ast
type Builder struct {
nodes []Node
lastIdx int
}
type Reference struct {
idx int
set bool
}
func (r Reference) Valid() bool {
return r.set
}
func (b *Builder) Finish() *Root {
r := &Root{
nodes: b.nodes,
}
b.nodes = nil
for i := range r.nodes {
r.nodes[i].root = r
}
return r
}
func (b *Builder) Push(n Node) Reference {
b.lastIdx = len(b.nodes)
b.nodes = append(b.nodes, n)
return Reference{
idx: b.lastIdx,
set: true,
}
}
func (b *Builder) PushAndChain(n Node) Reference {
newIdx := len(b.nodes)
b.nodes = append(b.nodes, n)
if b.lastIdx >= 0 {
b.nodes[b.lastIdx].next = newIdx
}
b.lastIdx = newIdx
return Reference{
idx: b.lastIdx,
set: true,
}
}
func (b *Builder) AttachChild(parent Reference, child Reference) {
b.nodes[parent.idx].child = child.idx
}
func (b *Builder) Chain(from Reference, to Reference) {
b.nodes[from.idx].next = to.idx
}
+69
View File
@@ -0,0 +1,69 @@
package ast
import "fmt"
type Kind int
const (
// meta
Invalid Kind = iota
Comment
Key
// top level structures
Table
ArrayTable
KeyValue
// containers values
Array
InlineTable
// values
String
Bool
Float
Integer
LocalDate
LocalDateTime
DateTime
Time
)
func (k Kind) String() string {
switch k {
case Invalid:
return "Invalid"
case Comment:
return "Comment"
case Key:
return "Key"
case Table:
return "Table"
case ArrayTable:
return "ArrayTable"
case KeyValue:
return "KeyValue"
case Array:
return "Array"
case InlineTable:
return "InlineTable"
case String:
return "String"
case Bool:
return "Bool"
case Float:
return "Float"
case Integer:
return "Integer"
case LocalDate:
return "LocalDate"
case LocalDateTime:
return "LocalDateTime"
case DateTime:
return "DateTime"
case Time:
return "Time"
}
panic(fmt.Errorf("Kind.String() not implemented for '%d'", k))
}
+179 -129
View File
@@ -10,11 +10,11 @@ import (
)
type parser struct {
tree ast.Root
builder ast.Builder
}
func (p *parser) parse(b []byte) error {
b, err := p.parseExpression(b)
last, b, err := p.parseExpression(b)
if err != nil {
return err
}
@@ -24,10 +24,15 @@ func (p *parser) parse(b []byte) error {
return err
}
b, err = p.parseExpression(b)
var next ast.Reference
next, b, err = p.parseExpression(b)
if err != nil {
return err
}
if next.Valid() {
p.builder.Chain(last, next)
last = next
}
}
return nil
}
@@ -43,49 +48,48 @@ func (p *parser) parseNewline(b []byte) ([]byte, error) {
return nil, fmt.Errorf("expected newline but got %#U", b[0])
}
func (p *parser) parseExpression(b []byte) ([]byte, error) {
func (p *parser) parseExpression(b []byte) (ast.Reference, []byte, error) {
//expression = ws [ comment ]
//expression =/ ws keyval ws [ comment ]
//expression =/ ws table ws [ comment ]
var ref ast.Reference
b = p.parseWhitespace(b)
if len(b) == 0 {
return b, nil
return ref, b, nil
}
if b[0] == '#' {
_, rest, err := scanComment(b)
return rest, err
return ref, rest, err
}
if b[0] == '\n' || b[0] == '\r' {
return b, nil
return ref, b, nil
}
var err error
var node ast.Node
if b[0] == '[' {
node, b, err = p.parseTable(b)
ref, b, err = p.parseTable(b)
} else {
node, b, err = p.parseKeyval(b)
ref, b, err = p.parseKeyval(b)
}
if err != nil {
return nil, err
return ref, nil, err
}
p.tree = append(p.tree, node)
b = p.parseWhitespace(b)
if len(b) > 0 && b[0] == '#' {
_, rest, err := scanComment(b)
return rest, err
return ref, rest, err
}
return b, nil
return ref, b, nil
}
func (p *parser) parseTable(b []byte) (ast.Node, []byte, error) {
func (p *parser) parseTable(b []byte) (ast.Reference, []byte, error) {
//table = std-table / array-table
if len(b) > 1 && b[1] == '[' {
return p.parseArrayTable(b)
@@ -93,90 +97,95 @@ func (p *parser) parseTable(b []byte) (ast.Node, []byte, error) {
return p.parseStdTable(b)
}
func (p *parser) parseArrayTable(b []byte) (ast.Node, []byte, error) {
func (p *parser) parseArrayTable(b []byte) (ast.Reference, []byte, error) {
//array-table = array-table-open key array-table-close
//array-table-open = %x5B.5B ws ; [[ Double left square bracket
//array-table-close = ws %x5D.5D ; ]] Double right square bracket
node := ast.Node{
ref := p.builder.Push(ast.Node{
Kind: ast.ArrayTable,
}
})
b = b[2:]
b = p.parseWhitespace(b)
k, b, err := p.parseKey(b)
if err != nil {
return node, nil, err
return ref, nil, err
}
node.Children = k
p.builder.AttachChild(ref, k)
b = p.parseWhitespace(b)
b, err = expect(']', b)
if err != nil {
return node, nil, err
return ref, nil, err
}
b, err = expect(']', b)
return node, b, err
return ref, b, err
}
func (p *parser) parseStdTable(b []byte) (ast.Node, []byte, error) {
func (p *parser) parseStdTable(b []byte) (ast.Reference, []byte, error) {
//std-table = std-table-open key std-table-close
//std-table-open = %x5B ws ; [ Left square bracket
//std-table-close = ws %x5D ; ] Right square bracket
node := ast.Node{
ref := p.builder.Push(ast.Node{
Kind: ast.Table,
}
})
b = b[1:]
b = p.parseWhitespace(b)
key, b, err := p.parseKey(b)
if err != nil {
return ast.NoNode, nil, err
return ref, nil, err
}
node.Children = key
p.builder.AttachChild(ref, key)
b = p.parseWhitespace(b)
b, err = expect(']', b)
return node, b, err
return ref, b, err
}
func (p *parser) parseKeyval(b []byte) (ast.Node, []byte, error) {
func (p *parser) parseKeyval(b []byte) (ast.Reference, []byte, error) {
//keyval = key keyval-sep val
node := ast.Node{
ref := p.builder.Push(ast.Node{
Kind: ast.KeyValue,
}
})
key, b, err := p.parseKey(b)
if err != nil {
return ast.NoNode, nil, err
return ast.Reference{}, nil, err
}
node.Children = append(node.Children, key...)
//keyval-sep = ws %x3D ws ; =
b = p.parseWhitespace(b)
b, err = expect('=', b)
if err != nil {
return ast.NoNode, nil, err
return ast.Reference{}, nil, err
}
b = p.parseWhitespace(b)
valNode, b, err := p.parseVal(b)
if err == nil {
node.Children = append(node.Children, valNode)
valRef, b, err := p.parseVal(b)
if err != nil {
return ref, b, err
}
return node, b, err
p.builder.Chain(valRef, key)
p.builder.AttachChild(ref, valRef)
return ref, b, err
}
func (p *parser) parseVal(b []byte) (ast.Node, []byte, error) {
func (p *parser) parseVal(b []byte) (ast.Reference, []byte, error) {
// val = string / boolean / array / inline-table / date-time / float / integer
var ref ast.Reference
if len(b) == 0 {
return ast.NoNode, nil, fmt.Errorf("expected value, not eof")
return ref, nil, fmt.Errorf("expected value, not eof")
}
node := ast.Node{}
var err error
c := b[0]
@@ -189,10 +198,12 @@ func (p *parser) parseVal(b []byte) (ast.Node, []byte, error) {
v, b, err = p.parseBasicString(b)
}
if err == nil {
node.Kind = ast.String
node.Data = v
ref = p.builder.Push(ast.Node{
Kind: ast.String,
Data: v,
})
}
return node, b, err
return ref, b, err
case '\'':
var v []byte
if scanFollowsMultilineLiteralStringDelimiter(b) {
@@ -201,35 +212,36 @@ func (p *parser) parseVal(b []byte) (ast.Node, []byte, error) {
v, b, err = p.parseLiteralString(b)
}
if err == nil {
node.Kind = ast.String
node.Data = v
ref = p.builder.Push(ast.Node{
Kind: ast.String,
Data: v,
})
}
return node, b, err
return ref, b, err
case 't':
if !scanFollowsTrue(b) {
return node, nil, fmt.Errorf("expected 'true'")
return ref, nil, fmt.Errorf("expected 'true'")
}
node.Kind = ast.Bool
node.Data = b[:4]
return node, b[4:], nil
ref = p.builder.Push(ast.Node{
Kind: ast.Bool,
Data: b[:4],
})
return ref, b[4:], nil
case 'f':
if !scanFollowsFalse(b) {
return node, nil, fmt.Errorf("expected 'false'")
return ast.Reference{}, nil, fmt.Errorf("expected 'false'")
}
node.Kind = ast.Bool
node.Data = b[:5]
return node, b[5:], nil
ref = p.builder.Push(ast.Node{
Kind: ast.Bool,
Data: b[:5],
})
return ref, b[5:], nil
case '[':
node.Kind = ast.Array
b, err := p.parseValArray(&node, b)
return node, b, err
return p.parseValArray(b)
case '{':
node.Kind = ast.InlineTable
b, err := p.parseInlineTable(&node, b)
return node, b, err
return p.parseInlineTable(b)
default:
b, err = p.parseIntOrFloatOrDateTime(&node, b)
return node, b, err
return p.parseIntOrFloatOrDateTime(b)
}
}
@@ -241,16 +253,22 @@ func (p *parser) parseLiteralString(b []byte) ([]byte, []byte, error) {
return v[1 : len(v)-1], rest, nil
}
func (p *parser) parseInlineTable(node *ast.Node, b []byte) ([]byte, error) {
func (p *parser) parseInlineTable(b []byte) (ast.Reference, []byte, error) {
//inline-table = inline-table-open [ inline-table-keyvals ] inline-table-close
//inline-table-open = %x7B ws ; {
//inline-table-close = ws %x7D ; }
//inline-table-sep = ws %x2C ws ; , Comma
//inline-table-keyvals = keyval [ inline-table-sep inline-table-keyvals ]
b = b[1:]
parent := p.builder.Push(ast.Node{
Kind: ast.InlineTable,
})
first := true
var child ast.Reference
b = b[1:]
var err error
for len(b) > 0 {
b = p.parseWhitespace(b)
@@ -261,24 +279,32 @@ func (p *parser) parseInlineTable(node *ast.Node, b []byte) ([]byte, error) {
if !first {
b, err = expect(',', b)
if err != nil {
return nil, err
return parent, nil, err
}
b = p.parseWhitespace(b)
}
var kv ast.Node
var kv ast.Reference
kv, b, err = p.parseKeyval(b)
if err != nil {
return nil, err
return parent, nil, err
}
node.Children = append(node.Children, kv)
if first {
p.builder.AttachChild(parent, kv)
first = false
} else {
p.builder.Chain(child, kv)
}
child = kv
first = false
}
return expect('}', b)
rest, err := expect('}', b)
return parent, rest, err
}
func (p *parser) parseValArray(node *ast.Node, b []byte) ([]byte, error) {
func (p *parser) parseValArray(b []byte) (ast.Reference, []byte, error) {
//array = array-open [ array-values ] ws-comment-newline array-close
//array-open = %x5B ; [
//array-close = %x5D ; ]
@@ -289,16 +315,22 @@ func (p *parser) parseValArray(node *ast.Node, b []byte) ([]byte, error) {
b = b[1:]
parent := p.builder.Push(ast.Node{
Kind: ast.Array,
})
first := true
var lastChild ast.Reference
var err error
for len(b) > 0 {
b, err = p.parseOptionalWhitespaceCommentNewline(b)
if err != nil {
return nil, err
return parent, nil, err
}
if len(b) == 0 {
return nil, unexpectedCharacter{b: b}
return parent, nil, unexpectedCharacter{b: b}
}
if b[0] == ']' {
@@ -306,29 +338,38 @@ func (p *parser) parseValArray(node *ast.Node, b []byte) ([]byte, error) {
}
if b[0] == ',' {
if first {
return nil, fmt.Errorf("array cannot start with comma")
return parent, nil, fmt.Errorf("array cannot start with comma")
}
b = b[1:]
b, err = p.parseOptionalWhitespaceCommentNewline(b)
if err != nil {
return nil, err
return parent, nil, err
}
}
var valueNode ast.Node
valueNode, b, err = p.parseVal(b)
var valueRef ast.Reference
valueRef, b, err = p.parseVal(b)
if err != nil {
return nil, err
return parent, nil, err
}
node.Children = append(node.Children, valueNode)
if first {
p.builder.AttachChild(parent, valueRef)
first = false
} else {
p.builder.Chain(lastChild, valueRef)
}
lastChild = valueRef
b, err = p.parseOptionalWhitespaceCommentNewline(b)
if err != nil {
return nil, err
return parent, nil, err
}
first = false
}
return expect(']', b)
rest, err := expect(']', b)
return parent, rest, err
}
func (p *parser) parseOptionalWhitespaceCommentNewline(b []byte) ([]byte, error) {
@@ -454,7 +495,7 @@ func (p *parser) parseMultilineBasicString(b []byte) ([]byte, []byte, error) {
return builder.Bytes(), rest, nil
}
func (p *parser) parseKey(b []byte) ([]ast.Node, []byte, error) {
func (p *parser) parseKey(b []byte) (ast.Reference, []byte, error) {
//key = simple-key / dotted-key
//simple-key = quoted-key / unquoted-key
//
@@ -464,14 +505,12 @@ func (p *parser) parseKey(b []byte) ([]ast.Node, []byte, error) {
//
//dot-sep = ws %x2E ws ; . Period
var nodes []ast.Node
key, b, err := p.parseSimpleKey(b)
if err != nil {
return nodes, nil, err
return ast.Reference{}, nil, err
}
nodes = append(nodes, ast.Node{
ref := p.builder.Push(ast.Node{
Kind: ast.Key,
Data: key,
})
@@ -481,14 +520,14 @@ func (p *parser) parseKey(b []byte) ([]ast.Node, []byte, error) {
if len(b) > 0 && b[0] == '.' {
b, err = expect('.', b)
if err != nil {
return nodes, nil, err
return ref, nil, err
}
b = p.parseWhitespace(b)
key, b, err = p.parseSimpleKey(b)
if err != nil {
return nodes, nil, err
return ref, nil, err
}
nodes = append(nodes, ast.Node{
p.builder.PushAndChain(ast.Node{
Kind: ast.Key,
Data: key,
})
@@ -497,7 +536,7 @@ func (p *parser) parseKey(b []byte) ([]ast.Node, []byte, error) {
}
}
return nodes, b, nil
return ref, b, nil
}
func (p *parser) parseSimpleKey(b []byte) (key, rest []byte, err error) {
@@ -609,28 +648,30 @@ func (p *parser) parseWhitespace(b []byte) []byte {
return rest
}
func (p *parser) parseIntOrFloatOrDateTime(node *ast.Node, b []byte) ([]byte, error) {
func (p *parser) parseIntOrFloatOrDateTime(b []byte) (ast.Reference, []byte, error) {
switch b[0] {
case 'i':
if !scanFollowsInf(b) {
return nil, fmt.Errorf("expected 'inf'")
return ast.Reference{}, nil, fmt.Errorf("expected 'inf'")
}
node.Kind = ast.Float
node.Data = b[:3]
return b[3:], nil
return p.builder.Push(ast.Node{
Kind: ast.Float,
Data: b[:3],
}), b[3:], nil
case 'n':
if !scanFollowsNan(b) {
return nil, fmt.Errorf("expected 'nan'")
return ast.Reference{}, nil, fmt.Errorf("expected 'nan'")
}
node.Kind = ast.Float
node.Data = b[:3]
return b[3:], nil
return p.builder.Push(ast.Node{
Kind: ast.Float,
Data: b[:3],
}), b[3:], nil
case '+', '-':
return p.scanIntOrFloat(node, b)
return p.scanIntOrFloat(b)
}
if len(b) < 3 {
return p.scanIntOrFloat(node, b)
return p.scanIntOrFloat(b)
}
s := 5
if len(b) < s {
@@ -641,10 +682,10 @@ func (p *parser) parseIntOrFloatOrDateTime(node *ast.Node, b []byte) ([]byte, er
continue
}
if idx == 2 && c == ':' || (idx == 4 && c == '-') {
return p.scanDateTime(node, b)
return p.scanDateTime(b)
}
}
return p.scanIntOrFloat(node, b)
return p.scanIntOrFloat(b)
}
func digitsToInt(b []byte) int {
@@ -656,7 +697,7 @@ func digitsToInt(b []byte) int {
return x
}
func (p *parser) scanDateTime(node *ast.Node, b []byte) ([]byte, error) {
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.
@@ -686,22 +727,25 @@ func (p *parser) scanDateTime(node *ast.Node, b []byte) ([]byte, error) {
}
}
var kind ast.Kind
if hasTime {
if hasTz {
node.Kind = ast.DateTime
kind = ast.DateTime
} else {
node.Kind = ast.LocalDateTime
kind = ast.LocalDateTime
}
} else {
if hasTz {
return nil, fmt.Errorf("possible DateTime cannot have a timezone but no time component")
return ast.Reference{}, nil, fmt.Errorf("possible DateTime cannot have a timezone but no time component")
}
node.Kind = ast.LocalDate
kind = ast.LocalDate
}
node.Data = b[:i]
return b[i:], nil
return p.builder.Push(ast.Node{
Kind: kind,
Data: b[:i],
}), b[i:], nil
}
func (p *parser) parseDateTime(b []byte) ([]byte, error) {
@@ -964,7 +1008,7 @@ func (p *parser) parseTime(b []byte) ([]byte, error) {
return b[idx:], nil
}
func (p *parser) scanIntOrFloat(node *ast.Node, b []byte) ([]byte, error) {
func (p *parser) scanIntOrFloat(b []byte) (ast.Reference, []byte, error) {
i := 0
if len(b) > 2 && b[0] == '0' {
@@ -989,9 +1033,10 @@ func (p *parser) scanIntOrFloat(node *ast.Node, b []byte) ([]byte, error) {
}
}
node.Kind = ast.Integer
node.Data = b[:i]
return b[i:], nil
return p.builder.Push(ast.Node{
Kind: ast.Integer,
Data: b[:i],
}), b[i:], nil
}
isFloat := false
@@ -1010,31 +1055,36 @@ func (p *parser) scanIntOrFloat(node *ast.Node, b []byte) ([]byte, error) {
if c == 'i' {
if scanFollowsInf(b[i:]) {
node.Kind = ast.Float
node.Data = b[:i+3]
return b[i+3:], nil
return p.builder.Push(ast.Node{
Kind: ast.Float,
Data: b[:i+3],
}), b[i+3:], nil
}
return nil, fmt.Errorf("unexpected character i while scanning for a number")
return ast.Reference{}, nil, fmt.Errorf("unexpected character i while scanning for a number")
}
if c == 'n' {
if scanFollowsNan(b[i:]) {
node.Kind = ast.Float
node.Data = b[:i+3]
return b[i+3:], nil
return p.builder.Push(ast.Node{
Kind: ast.Float,
Data: b[:i+3],
}), b[i+3:], nil
}
return nil, fmt.Errorf("unexpected character n while scanning for a number")
return ast.Reference{}, nil, fmt.Errorf("unexpected character n while scanning for a number")
}
break
}
kind := ast.Integer
if isFloat {
node.Kind = ast.Float
} else {
node.Kind = ast.Integer
kind = ast.Float
}
node.Data = b[:i]
return b[i:], nil
return p.builder.Push(ast.Node{
Kind: kind,
Data: b[:i],
}), b[i:], nil
}
func isDigit(r byte) bool {
+134 -50
View File
@@ -125,44 +125,128 @@ func TestParser_AST_Numbers(t *testing.T) {
} else {
require.NoError(t, err)
expected := ast.Root{
ast.Node{
expected := astRoot{
astNode{
Kind: ast.KeyValue,
Children: []ast.Node{
{Kind: ast.Key, Data: []byte(`A`)},
Children: []astNode{
{Kind: e.kind, Data: []byte(e.input)},
{Kind: ast.Key, Data: []byte(`A`)},
},
},
}
require.Equal(t, expected, p.tree)
compareAST(t, expected, p.builder.Finish())
}
})
}
}
type astRoot []astNode
type astNode struct {
Kind ast.Kind
Data []byte
Children []astNode
}
func compareAST(t *testing.T, expected astRoot, actual *ast.Root) {
it := actual.Iterator()
compareIterator(t, expected, it)
}
func compareIterator(t *testing.T, expected []astNode, actual ast.Iterator) {
idx := 0
for actual.Next() {
n := actual.Node()
if idx >= len(expected) {
t.Fatal("extra child in actual tree")
}
e := expected[idx]
require.Equal(t, e.Kind, n.Kind)
require.Equal(t, e.Data, n.Data)
compareIterator(t, e.Children, n.Children())
idx++
}
if idx < len(expected) {
t.Fatal("missing children in actual", "idx =", idx, "expected =", len(expected))
}
}
func (r astRoot) toOrig() *ast.Root {
builder := &ast.Builder{}
var last ast.Reference
for i, n := range r {
ref := builder.Push(ast.Node{
Kind: n.Kind,
Data: n.Data,
})
if i > 0 {
builder.Chain(last, ref)
}
last = ref
if len(n.Children) > 0 {
c := childrenToOrig(builder, n.Children)
builder.AttachChild(ref, c)
}
}
return builder.Finish()
}
func childrenToOrig(b *ast.Builder, nodes []astNode) ast.Reference {
var first ast.Reference
var last ast.Reference
for i, n := range nodes {
ref := b.Push(ast.Node{
Kind: n.Kind,
Data: n.Data,
})
if i == 0 {
first = ref
} else {
b.Chain(last, ref)
}
last = ref
if len(n.Children) > 0 {
c := childrenToOrig(b, n.Children)
b.AttachChild(ref, c)
}
}
return first
}
func TestParser_AST(t *testing.T) {
examples := []struct {
desc string
input string
ast ast.Root
ast astRoot
err bool
}{
{
desc: "simple string assignment",
input: `A = "hello"`,
ast: ast.Root{
ast.Node{
ast: astRoot{
astNode{
Kind: ast.KeyValue,
Children: []ast.Node{
{
Kind: ast.Key,
Data: []byte(`A`),
},
Children: []astNode{
{
Kind: ast.String,
Data: []byte(`hello`),
},
{
Kind: ast.Key,
Data: []byte(`A`),
},
},
},
},
@@ -170,18 +254,18 @@ func TestParser_AST(t *testing.T) {
{
desc: "simple bool assignment",
input: `A = true`,
ast: ast.Root{
ast.Node{
ast: astRoot{
astNode{
Kind: ast.KeyValue,
Children: []ast.Node{
{
Kind: ast.Key,
Data: []byte(`A`),
},
Children: []astNode{
{
Kind: ast.Bool,
Data: []byte(`true`),
},
{
Kind: ast.Key,
Data: []byte(`A`),
},
},
},
},
@@ -189,24 +273,20 @@ func TestParser_AST(t *testing.T) {
{
desc: "array of strings",
input: `A = ["hello", ["world", "again"]]`,
ast: ast.Root{
ast.Node{
ast: astRoot{
astNode{
Kind: ast.KeyValue,
Children: []ast.Node{
{
Kind: ast.Key,
Data: []byte(`A`),
},
Children: []astNode{
{
Kind: ast.Array,
Children: []ast.Node{
Children: []astNode{
{
Kind: ast.String,
Data: []byte(`hello`),
},
{
Kind: ast.Array,
Children: []ast.Node{
Children: []astNode{
{
Kind: ast.String,
Data: []byte(`world`),
@@ -219,6 +299,10 @@ func TestParser_AST(t *testing.T) {
},
},
},
{
Kind: ast.Key,
Data: []byte(`A`),
},
},
},
},
@@ -226,17 +310,13 @@ func TestParser_AST(t *testing.T) {
{
desc: "array of arrays of strings",
input: `A = ["hello", "world"]`,
ast: ast.Root{
ast.Node{
ast: astRoot{
astNode{
Kind: ast.KeyValue,
Children: []ast.Node{
{
Kind: ast.Key,
Data: []byte(`A`),
},
Children: []astNode{
{
Kind: ast.Array,
Children: []ast.Node{
Children: []astNode{
{
Kind: ast.String,
Data: []byte(`hello`),
@@ -247,6 +327,10 @@ func TestParser_AST(t *testing.T) {
},
},
},
{
Kind: ast.Key,
Data: []byte(`A`),
},
},
},
},
@@ -254,33 +338,33 @@ func TestParser_AST(t *testing.T) {
{
desc: "inline table",
input: `name = { first = "Tom", last = "Preston-Werner" }`,
ast: ast.Root{
ast.Node{
ast: astRoot{
astNode{
Kind: ast.KeyValue,
Children: []ast.Node{
{
Kind: ast.Key,
Data: []byte(`name`),
},
Children: []astNode{
{
Kind: ast.InlineTable,
Children: []ast.Node{
Children: []astNode{
{
Kind: ast.KeyValue,
Children: []ast.Node{
{Kind: ast.Key, Data: []byte(`first`)},
Children: []astNode{
{Kind: ast.String, Data: []byte(`Tom`)},
{Kind: ast.Key, Data: []byte(`first`)},
},
},
{
Kind: ast.KeyValue,
Children: []ast.Node{
{Kind: ast.Key, Data: []byte(`last`)},
Children: []astNode{
{Kind: ast.String, Data: []byte(`Preston-Werner`)},
{Kind: ast.Key, Data: []byte(`last`)},
},
},
},
},
{
Kind: ast.Key,
Data: []byte(`name`),
},
},
},
},
@@ -295,7 +379,7 @@ func TestParser_AST(t *testing.T) {
require.Error(t, err)
} else {
require.NoError(t, err)
require.Equal(t, e.ast, p.tree)
compareAST(t, e.ast, p.builder.Finish())
}
})
}
+44 -30
View File
@@ -17,7 +17,7 @@ func Unmarshal(data []byte, v interface{}) error {
}
d := decoder{}
return d.fromAst(p.tree, v)
return d.fromAst(p.builder.Finish(), v)
}
type decoder struct {
@@ -41,7 +41,7 @@ func (d *decoder) arrayIndex(append bool, v reflect.Value) int {
return idx
}
func (d *decoder) fromAst(tree ast.Root, v interface{}) error {
func (d *decoder) fromAst(tree *ast.Root, v interface{}) error {
r := reflect.ValueOf(v)
if r.Kind() != reflect.Ptr {
return fmt.Errorf("need to target a pointer, not %s", r.Kind())
@@ -54,14 +54,17 @@ func (d *decoder) fromAst(tree ast.Root, v interface{}) error {
var skipUntilTable bool
var root target = valueTarget(r.Elem())
current := root
for _, node := range tree {
it := tree.Iterator()
for it.Next() {
node := it.Node()
var found bool
switch node.Kind {
case ast.KeyValue:
if skipUntilTable {
continue
}
err = d.unmarshalKeyValue(current, &node)
err = d.unmarshalKeyValue(current, node)
found = true
case ast.Table:
current, found, err = d.scopeWithKey(root, node.Key())
@@ -91,10 +94,12 @@ func (d *decoder) fromAst(tree ast.Root, v interface{}) error {
//
// When encountering slices, it should always use its last element, and error
// if the slice does not have any.
func (d *decoder) scopeWithKey(x target, key []ast.Node) (target, bool, error) {
func (d *decoder) scopeWithKey(x target, key ast.Iterator) (target, bool, error) {
var err error
found := true
for _, n := range key {
for key.Next() {
n := key.Node()
x, found, err = d.scopeTableTarget(false, x, string(n.Data))
if err != nil || !found {
return nil, found, err
@@ -108,18 +113,21 @@ func (d *decoder) scopeWithKey(x target, key []ast.Node) (target, bool, error) {
//
// It is the same as scopeWithKey, but when scoping the last part of the key
// it creates a new element in the array instead of using the last one.
func (d *decoder) scopeWithArrayTable(x target, key []ast.Node) (target, bool, error) {
func (d *decoder) scopeWithArrayTable(x target, key ast.Iterator) (target, bool, error) {
var err error
found := true
if len(key) > 1 {
for _, n := range key[:len(key)-1] {
x, found, err = d.scopeTableTarget(false, x, string(n.Data))
if err != nil || !found {
return nil, found, err
}
for key.Next() {
n := key.Node()
if !n.Next().Valid() { // want to stop at one before last
break
}
x, found, err = d.scopeTableTarget(false, x, string(n.Data))
if err != nil || !found {
return nil, found, err
}
}
x, found, err = d.scopeTableTarget(false, x, string(key[len(key)-1].Data))
n := key.Node()
x, found, err = d.scopeTableTarget(false, x, string(n.Data))
if err != nil || !found {
return x, found, err
}
@@ -152,7 +160,7 @@ func (d *decoder) scopeWithArrayTable(x target, key []ast.Node) (target, bool, e
return x, found, err
}
func (d *decoder) unmarshalKeyValue(x target, node *ast.Node) error {
func (d *decoder) unmarshalKeyValue(x target, node ast.Node) error {
assertNode(ast.KeyValue, node)
x, found, err := d.scopeWithKey(x, node.Key())
@@ -170,7 +178,7 @@ func (d *decoder) unmarshalKeyValue(x target, node *ast.Node) error {
var textUnmarshalerType = reflect.TypeOf(new(encoding.TextUnmarshaler)).Elem()
func tryTextUnmarshaler(x target, node *ast.Node) (bool, error) {
func tryTextUnmarshaler(x target, node ast.Node) (bool, error) {
v := x.get()
if v.Kind() != reflect.Struct {
@@ -185,7 +193,7 @@ func tryTextUnmarshaler(x target, node *ast.Node) (bool, error) {
return false, nil
}
func (d *decoder) unmarshalValue(x target, node *ast.Node) error {
func (d *decoder) unmarshalValue(x target, node ast.Node) error {
v := x.get()
if v.Kind() == reflect.Ptr {
if !v.Elem().IsValid() {
@@ -225,7 +233,7 @@ func (d *decoder) unmarshalValue(x target, node *ast.Node) error {
}
}
func unmarshalLocalDateTime(x target, node *ast.Node) error {
func unmarshalLocalDateTime(x target, node ast.Node) error {
assertNode(ast.LocalDateTime, node)
v, rest, err := parseLocalDateTime(node.Data)
if err != nil {
@@ -237,7 +245,7 @@ func unmarshalLocalDateTime(x target, node *ast.Node) error {
return setLocalDateTime(x, v)
}
func unmarshalDateTime(x target, node *ast.Node) error {
func unmarshalDateTime(x target, node ast.Node) error {
assertNode(ast.DateTime, node)
v, err := parseDateTime(node.Data)
if err != nil {
@@ -254,18 +262,18 @@ func setDateTime(x target, v time.Time) error {
return x.set(reflect.ValueOf(v))
}
func unmarshalString(x target, node *ast.Node) error {
func unmarshalString(x target, node ast.Node) error {
assertNode(ast.String, node)
return setString(x, string(node.Data))
}
func unmarshalBool(x target, node *ast.Node) error {
func unmarshalBool(x target, node ast.Node) error {
assertNode(ast.Bool, node)
v := node.Data[0] == 't'
return setBool(x, v)
}
func unmarshalInteger(x target, node *ast.Node) error {
func unmarshalInteger(x target, node ast.Node) error {
assertNode(ast.Integer, node)
v, err := parseInteger(node.Data)
if err != nil {
@@ -274,7 +282,7 @@ func unmarshalInteger(x target, node *ast.Node) error {
return setInt64(x, v)
}
func unmarshalFloat(x target, node *ast.Node) error {
func unmarshalFloat(x target, node ast.Node) error {
assertNode(ast.Float, node)
v, err := parseFloat(node.Data)
if err != nil {
@@ -283,11 +291,13 @@ func unmarshalFloat(x target, node *ast.Node) error {
return setFloat64(x, v)
}
func (d *decoder) unmarshalInlineTable(x target, node *ast.Node) error {
func (d *decoder) unmarshalInlineTable(x target, node ast.Node) error {
assertNode(ast.InlineTable, node)
for _, kv := range node.Children {
err := d.unmarshalKeyValue(x, &kv)
it := node.Children()
for it.Next() {
n := it.Node()
err := d.unmarshalKeyValue(x, n)
if err != nil {
return err
}
@@ -295,7 +305,7 @@ func (d *decoder) unmarshalInlineTable(x target, node *ast.Node) error {
return nil
}
func (d *decoder) unmarshalArray(x target, node *ast.Node) error {
func (d *decoder) unmarshalArray(x target, node ast.Node) error {
assertNode(ast.Array, node)
err := ensureValueIndexable(x)
@@ -303,7 +313,10 @@ func (d *decoder) unmarshalArray(x target, node *ast.Node) error {
return err
}
for idx, n := range node.Children {
it := node.Children()
idx := 0
for it.Next() {
n := it.Node()
v, err := elementAt(x, idx)
if err != nil {
return err
@@ -313,15 +326,16 @@ func (d *decoder) unmarshalArray(x target, node *ast.Node) error {
// mimic encoding/json
break
}
err = d.unmarshalValue(v, &n)
err = d.unmarshalValue(v, n)
if err != nil {
return err
}
idx++
}
return nil
}
func assertNode(expected ast.Kind, node *ast.Node) {
func assertNode(expected ast.Kind, node ast.Node) {
if node.Kind != expected {
panic(fmt.Errorf("expected node of kind %s, not %s", expected, node.Kind))
}
+75 -74
View File
@@ -660,18 +660,18 @@ B = "data"`,
}
func TestFromAst_KV(t *testing.T) {
root := ast.Root{
ast.Node{
root := astRoot{
astNode{
Kind: ast.KeyValue,
Children: []ast.Node{
{
Kind: ast.Key,
Data: []byte(`Foo`),
},
Children: []astNode{
{
Kind: ast.String,
Data: []byte(`hello`),
},
{
Kind: ast.Key,
Data: []byte(`Foo`),
},
},
},
}
@@ -682,44 +682,44 @@ func TestFromAst_KV(t *testing.T) {
x := Doc{}
d := decoder{}
err := d.fromAst(root, &x)
err := d.fromAst(root.toOrig(), &x)
require.NoError(t, err)
assert.Equal(t, Doc{Foo: "hello"}, x)
}
func TestFromAst_Table(t *testing.T) {
t.Run("one level table on struct", func(t *testing.T) {
root := ast.Root{
ast.Node{
root := astRoot{
astNode{
Kind: ast.Table,
Children: []ast.Node{
Children: []astNode{
{Kind: ast.Key, Data: []byte(`Level1`)},
},
},
ast.Node{
astNode{
Kind: ast.KeyValue,
Children: []ast.Node{
{
Kind: ast.Key,
Data: []byte(`A`),
},
Children: []astNode{
{
Kind: ast.String,
Data: []byte(`hello`),
},
},
},
ast.Node{
Kind: ast.KeyValue,
Children: []ast.Node{
{
Kind: ast.Key,
Data: []byte(`B`),
Data: []byte(`A`),
},
},
},
astNode{
Kind: ast.KeyValue,
Children: []astNode{
{
Kind: ast.String,
Data: []byte(`world`),
},
{
Kind: ast.Key,
Data: []byte(`B`),
},
},
},
}
@@ -735,7 +735,7 @@ func TestFromAst_Table(t *testing.T) {
x := Doc{}
d := decoder{}
err := d.fromAst(root, &x)
err := d.fromAst(root.toOrig(), &x)
require.NoError(t, err)
assert.Equal(t, Doc{
Level1: Level1{
@@ -745,25 +745,25 @@ func TestFromAst_Table(t *testing.T) {
}, x)
})
t.Run("one level table on struct", func(t *testing.T) {
root := ast.Root{
ast.Node{
root := astRoot{
astNode{
Kind: ast.Table,
Children: []ast.Node{
Children: []astNode{
{Kind: ast.Key, Data: []byte(`A`)},
{Kind: ast.Key, Data: []byte(`B`)},
},
},
ast.Node{
astNode{
Kind: ast.KeyValue,
Children: []ast.Node{
{
Kind: ast.Key,
Data: []byte(`C`),
},
Children: []astNode{
{
Kind: ast.String,
Data: []byte(`value`),
},
{
Kind: ast.Key,
Data: []byte(`C`),
},
},
},
}
@@ -782,7 +782,7 @@ func TestFromAst_Table(t *testing.T) {
x := Doc{}
d := decoder{}
err := d.fromAst(root, &x)
err := d.fromAst(root.toOrig(), &x)
require.NoError(t, err)
assert.Equal(t, Doc{
A: A{B: B{C: "value"}},
@@ -792,32 +792,33 @@ func TestFromAst_Table(t *testing.T) {
func TestFromAst_InlineTable(t *testing.T) {
t.Run("one level of strings", func(t *testing.T) {
root := ast.Root{
ast.Node{
root := astRoot{
astNode{
Kind: ast.KeyValue,
Children: []ast.Node{
{
Kind: ast.Key,
Data: []byte(`Name`)},
Children: []astNode{
{
Kind: ast.InlineTable,
Children: []ast.Node{
Children: []astNode{
{
Kind: ast.KeyValue,
Children: []ast.Node{
{Kind: ast.Key, Data: []byte(`First`)},
Children: []astNode{
{Kind: ast.String, Data: []byte(`Tom`)},
{Kind: ast.Key, Data: []byte(`First`)},
},
},
{
Kind: ast.KeyValue,
Children: []ast.Node{
{Kind: ast.Key, Data: []byte(`Last`)},
Children: []astNode{
{Kind: ast.String, Data: []byte(`Preston-Werner`)},
{Kind: ast.Key, Data: []byte(`Last`)},
},
},
},
},
{
Kind: ast.Key,
Data: []byte(`Name`),
},
},
},
}
@@ -833,7 +834,7 @@ func TestFromAst_InlineTable(t *testing.T) {
x := Doc{}
d := decoder{}
err := d.fromAst(root, &x)
err := d.fromAst(root.toOrig(), &x)
require.NoError(t, err)
assert.Equal(t, Doc{
Name: Name{
@@ -847,17 +848,13 @@ func TestFromAst_InlineTable(t *testing.T) {
func TestFromAst_Slice(t *testing.T) {
t.Run("slice of string", func(t *testing.T) {
root := ast.Root{
ast.Node{
root := astRoot{
astNode{
Kind: ast.KeyValue,
Children: []ast.Node{
{
Kind: ast.Key,
Data: []byte(`Foo`),
},
Children: []astNode{
{
Kind: ast.Array,
Children: []ast.Node{
Children: []astNode{
{
Kind: ast.String,
Data: []byte(`hello`),
@@ -868,6 +865,10 @@ func TestFromAst_Slice(t *testing.T) {
},
},
},
{
Kind: ast.Key,
Data: []byte(`Foo`),
},
},
},
}
@@ -878,23 +879,19 @@ func TestFromAst_Slice(t *testing.T) {
x := Doc{}
d := decoder{}
err := d.fromAst(root, &x)
err := d.fromAst(root.toOrig(), &x)
require.NoError(t, err)
assert.Equal(t, Doc{Foo: []string{"hello", "world"}}, x)
})
t.Run("slice of interfaces for strings", func(t *testing.T) {
root := ast.Root{
ast.Node{
root := astRoot{
astNode{
Kind: ast.KeyValue,
Children: []ast.Node{
{
Kind: ast.Key,
Data: []byte(`Foo`),
},
Children: []astNode{
{
Kind: ast.Array,
Children: []ast.Node{
Children: []astNode{
{
Kind: ast.String,
Data: []byte(`hello`),
@@ -905,6 +902,10 @@ func TestFromAst_Slice(t *testing.T) {
},
},
},
{
Kind: ast.Key,
Data: []byte(`Foo`),
},
},
},
}
@@ -915,30 +916,26 @@ func TestFromAst_Slice(t *testing.T) {
x := Doc{}
d := decoder{}
err := d.fromAst(root, &x)
err := d.fromAst(root.toOrig(), &x)
require.NoError(t, err)
assert.Equal(t, Doc{Foo: []interface{}{"hello", "world"}}, x)
})
t.Run("slice of interfaces with slices", func(t *testing.T) {
root := ast.Root{
ast.Node{
root := astRoot{
astNode{
Kind: ast.KeyValue,
Children: []ast.Node{
{
Kind: ast.Key,
Data: []byte(`Foo`),
},
Children: []astNode{
{
Kind: ast.Array,
Children: []ast.Node{
Children: []astNode{
{
Kind: ast.String,
Data: []byte(`hello`),
},
{
Kind: ast.Array,
Children: []ast.Node{
Children: []astNode{
{
Kind: ast.String,
Data: []byte(`inner1`),
@@ -951,6 +948,10 @@ func TestFromAst_Slice(t *testing.T) {
},
},
},
{
Kind: ast.Key,
Data: []byte(`Foo`),
},
},
},
}
@@ -961,7 +962,7 @@ func TestFromAst_Slice(t *testing.T) {
x := Doc{}
d := decoder{}
err := d.fromAst(root, &x)
err := d.fromAst(root.toOrig(), &x)
require.NoError(t, err)
assert.Equal(t, Doc{Foo: []interface{}{"hello", []interface{}{"inner1", "inner2"}}}, x)
})