aboutsummaryrefslogtreecommitdiff
path: root/src/urubu/driver/lexer.go
diff options
context:
space:
mode:
Diffstat (limited to 'src/urubu/driver/lexer.go')
-rw-r--r--src/urubu/driver/lexer.go1158
1 files changed, 1158 insertions, 0 deletions
diff --git a/src/urubu/driver/lexer.go b/src/urubu/driver/lexer.go
new file mode 100644
index 0000000..7423668
--- /dev/null
+++ b/src/urubu/driver/lexer.go
@@ -0,0 +1,1158 @@
+package lexer
+
+import (
+ "bytes"
+ _ "embed"
+ "fmt"
+ "go/ast"
+ "go/format"
+ "go/parser"
+ "go/token"
+ "io"
+ "strings"
+ "text/template"
+
+ "urubu/grammar/lexical"
+ spec "urubu/spec/grammar"
+)
+
+type ModeID int
+
+func (id ModeID) Int() int {
+ return int(id)
+}
+
+type StateID int
+
+func (id StateID) Int() int {
+ return int(id)
+}
+
+type KindID int
+
+func (id KindID) Int() int {
+ return int(id)
+}
+
+type ModeKindID int
+
+func (id ModeKindID) Int() int {
+ return int(id)
+}
+
+type LexSpec interface {
+ InitialMode() ModeID
+ Pop(mode ModeID, modeKind ModeKindID) bool
+ Push(mode ModeID, modeKind ModeKindID) (ModeID, bool)
+ ModeName(mode ModeID) string
+ InitialState(mode ModeID) StateID
+ NextState(mode ModeID, state StateID, v int) (StateID, bool)
+ Accept(mode ModeID, state StateID) (ModeKindID, bool)
+ KindIDAndName(mode ModeID, modeKind ModeKindID) (KindID, string)
+}
+
+// Token representes a token.
+type Token struct {
+ // ModeID is an ID of a lex mode.
+ ModeID ModeID
+
+ // KindID is an ID of a kind. This is unique among all modes.
+ KindID KindID
+
+ // ModeKindID is an ID of a lexical kind. This is unique only within a mode.
+ // Note that you need to use KindID field if you want to identify a kind across all modes.
+ ModeKindID ModeKindID
+
+ // BytePos is a byte position where a token appears.
+ BytePos int
+
+ // ByteLen is a length of a token.
+ ByteLen int
+
+ // Row is a row number where a token appears.
+ Row int
+
+ // Col is a column number where a token appears.
+ // Note that Col is counted in code points, not bytes.
+ Col int
+
+ // Lexeme is a byte sequence matched a pattern of a lexical specification.
+ Lexeme []byte
+
+ // When this field is true, it means the token is the EOF token.
+ EOF bool
+
+ // When this field is true, it means the token is an error token.
+ Invalid bool
+}
+
+type LexerOption func(l *Lexer) error
+
+// DisableModeTransition disables the active mode transition. Thus, even if the lexical specification has the push and pop
+// operations, the lexer doesn't perform these operations. When the lexical specification has multiple modes, and this option is
+// enabled, you need to call the Lexer.Push and Lexer.Pop methods to perform the mode transition. You can use the Lexer.Mode method
+// to know the current lex mode.
+func DisableModeTransition() LexerOption {
+ return func(l *Lexer) error {
+ l.passiveModeTran = true
+ return nil
+ }
+}
+
+type lexerState struct {
+ srcPtr int
+ row int
+ col int
+}
+
+type Lexer struct {
+ spec LexSpec
+ src []byte
+ state lexerState
+ lastAcceptedState lexerState
+ tokBuf []*Token
+ modeStack []ModeID
+ passiveModeTran bool
+}
+
+// NewLexer returns a new lexer.
+func NewLexer(spec LexSpec, src io.Reader, opts ...LexerOption) (*Lexer, error) {
+ b, err := io.ReadAll(src)
+ if err != nil {
+ return nil, err
+ }
+ l := &Lexer{
+ spec: spec,
+ src: b,
+ state: lexerState{
+ srcPtr: 0,
+ row: 0,
+ col: 0,
+ },
+ lastAcceptedState: lexerState{
+ srcPtr: 0,
+ row: 0,
+ col: 0,
+ },
+ modeStack: []ModeID{
+ spec.InitialMode(),
+ },
+ passiveModeTran: false,
+ }
+ for _, opt := range opts {
+ err := opt(l)
+ if err != nil {
+ return nil, err
+ }
+ }
+
+ return l, nil
+}
+
+// Next returns a next token.
+func (l *Lexer) Next() (*Token, error) {
+ if len(l.tokBuf) > 0 {
+ tok := l.tokBuf[0]
+ l.tokBuf = l.tokBuf[1:]
+ return tok, nil
+ }
+
+ tok, err := l.nextAndTransition()
+ if err != nil {
+ return nil, err
+ }
+ if !tok.Invalid {
+ return tok, nil
+ }
+ errTok := tok
+ for {
+ tok, err = l.nextAndTransition()
+ if err != nil {
+ return nil, err
+ }
+ if !tok.Invalid {
+ break
+ }
+ errTok.ByteLen += tok.ByteLen
+ errTok.Lexeme = append(errTok.Lexeme, tok.Lexeme...)
+ }
+ l.tokBuf = append(l.tokBuf, tok)
+
+ return errTok, nil
+}
+
+func (l *Lexer) nextAndTransition() (*Token, error) {
+ tok, err := l.next()
+ if err != nil {
+ return nil, err
+ }
+ if tok.EOF || tok.Invalid {
+ return tok, nil
+ }
+ if l.passiveModeTran {
+ return tok, nil
+ }
+ mode := l.Mode()
+ if l.spec.Pop(mode, tok.ModeKindID) {
+ err := l.PopMode()
+ if err != nil {
+ return nil, err
+ }
+ }
+ if mode, ok := l.spec.Push(mode, tok.ModeKindID); ok {
+ l.PushMode(mode)
+ }
+ // The checking length of the mode stack must be at after pop and push operations because those operations can be performed
+ // at the same time. When the mode stack has just one element and popped it, the mode stack will be temporarily emptied.
+ // However, since a push operation may be performed immediately after it, the lexer allows the stack to be temporarily empty.
+ if len(l.modeStack) == 0 {
+ return nil, fmt.Errorf("a mode stack must have at least one element")
+ }
+ return tok, nil
+}
+
+func (l *Lexer) next() (*Token, error) {
+ mode := l.Mode()
+ state := l.spec.InitialState(mode)
+ buf := []byte{}
+ startPos := l.state.srcPtr
+ row := l.state.row
+ col := l.state.col
+ var tok *Token
+ for {
+ v, eof := l.read()
+ if eof {
+ if tok != nil {
+ l.revert()
+ return tok, nil
+ }
+ // When `buf` has unaccepted data and reads the EOF, the lexer treats the buffered data as an invalid token.
+ if len(buf) > 0 {
+ return &Token{
+ ModeID: mode,
+ ModeKindID: 0,
+ BytePos: startPos,
+ ByteLen: l.state.srcPtr - startPos,
+ Lexeme: buf,
+ Row: row,
+ Col: col,
+ Invalid: true,
+ }, nil
+ }
+ return &Token{
+ ModeID: mode,
+ ModeKindID: 0,
+ BytePos: startPos,
+ Row: row,
+ Col: col,
+ EOF: true,
+ }, nil
+ }
+ buf = append(buf, v)
+ nextState, ok := l.spec.NextState(mode, state, int(v))
+ if !ok {
+ if tok != nil {
+ l.revert()
+ return tok, nil
+ }
+ return &Token{
+ ModeID: mode,
+ ModeKindID: 0,
+ BytePos: startPos,
+ ByteLen: l.state.srcPtr - startPos,
+ Lexeme: buf,
+ Row: row,
+ Col: col,
+ Invalid: true,
+ }, nil
+ }
+ state = nextState
+ if modeKindID, ok := l.spec.Accept(mode, state); ok {
+ kindID, _ := l.spec.KindIDAndName(mode, modeKindID)
+ tok = &Token{
+ ModeID: mode,
+ KindID: kindID,
+ ModeKindID: modeKindID,
+ BytePos: startPos,
+ ByteLen: l.state.srcPtr - startPos,
+ Lexeme: buf,
+ Row: row,
+ Col: col,
+ }
+ l.accept()
+ }
+ }
+}
+
+// Mode returns the current lex mode.
+func (l *Lexer) Mode() ModeID {
+ return l.modeStack[len(l.modeStack)-1]
+}
+
+// PushMode adds a lex mode onto the mode stack.
+func (l *Lexer) PushMode(mode ModeID) {
+ l.modeStack = append(l.modeStack, mode)
+}
+
+// PopMode removes a lex mode from the top of the mode stack.
+func (l *Lexer) PopMode() error {
+ sLen := len(l.modeStack)
+ if sLen == 0 {
+ return fmt.Errorf("cannot pop a lex mode from a lex mode stack any more")
+ }
+ l.modeStack = l.modeStack[:sLen-1]
+ return nil
+}
+
+func (l *Lexer) read() (byte, bool) {
+ if l.state.srcPtr >= len(l.src) {
+ return 0, true
+ }
+
+ b := l.src[l.state.srcPtr]
+ l.state.srcPtr++
+
+ // Count the token positions.
+ // The driver treats LF as the end of lines and counts columns in code points, not bytes.
+ // To count in code points, we refer to the First Byte column in the Table 3-6.
+ //
+ // Reference:
+ // - [Table 3-6] https://www.unicode.org/versions/Unicode13.0.0/ch03.pdf > Table 3-6. UTF-8 Bit Distribution
+ if b < 128 {
+ // 0x0A is LF.
+ if b == 0x0A {
+ l.state.row++
+ l.state.col = 0
+ } else {
+ l.state.col++
+ }
+ } else if b>>5 == 6 || b>>4 == 14 || b>>3 == 30 {
+ l.state.col++
+ }
+
+ return b, false
+}
+
+// accept saves the current state.
+func (l *Lexer) accept() {
+ l.lastAcceptedState = l.state
+}
+
+// revert reverts the lexer state to the last accepted state.
+//
+// We must not call this function consecutively.
+func (l *Lexer) revert() {
+ l.state = l.lastAcceptedState
+}
+
+type lexSpec struct {
+ spec *spec.LexicalSpec
+}
+
+func NewLexSpec(spec *spec.LexicalSpec) *lexSpec {
+ return &lexSpec{
+ spec: spec,
+ }
+}
+
+func (s *lexSpec) InitialMode() ModeID {
+ return ModeID(s.spec.InitialModeID.Int())
+}
+
+func (s *lexSpec) Pop(mode ModeID, modeKind ModeKindID) bool {
+ return s.spec.Specs[mode].Pop[modeKind] == 1
+}
+
+func (s *lexSpec) Push(mode ModeID, modeKind ModeKindID) (ModeID, bool) {
+ modeID := s.spec.Specs[mode].Push[modeKind]
+ return ModeID(modeID.Int()), !modeID.IsNil()
+}
+
+func (s *lexSpec) ModeName(mode ModeID) string {
+ return s.spec.ModeNames[mode].String()
+}
+
+func (s *lexSpec) InitialState(mode ModeID) StateID {
+ return StateID(s.spec.Specs[mode].DFA.InitialStateID.Int())
+}
+
+func (s *lexSpec) NextState(mode ModeID, state StateID, v int) (StateID, bool) {
+ switch s.spec.CompressionLevel {
+ case 2:
+ tran := s.spec.Specs[mode].DFA.Transition
+ rowNum := tran.RowNums[state]
+ d := tran.UniqueEntries.RowDisplacement[rowNum]
+ if tran.UniqueEntries.Bounds[d+v] != rowNum {
+ return StateID(tran.UniqueEntries.EmptyValue.Int()), false
+ }
+ return StateID(tran.UniqueEntries.Entries[d+v].Int()), true
+ case 1:
+ tran := s.spec.Specs[mode].DFA.Transition
+ next := tran.UncompressedUniqueEntries[tran.RowNums[state]*tran.OriginalColCount+v]
+ if next == spec.StateIDNil {
+ return StateID(spec.StateIDNil.Int()), false
+ }
+ return StateID(next.Int()), true
+ }
+
+ modeSpec := s.spec.Specs[mode]
+ next := modeSpec.DFA.UncompressedTransition[state.Int()*modeSpec.DFA.ColCount+v]
+ if next == spec.StateIDNil {
+ return StateID(spec.StateIDNil), false
+ }
+ return StateID(next.Int()), true
+}
+
+func (s *lexSpec) Accept(mode ModeID, state StateID) (ModeKindID, bool) {
+ modeKindID := s.spec.Specs[mode].DFA.AcceptingStates[state]
+ return ModeKindID(modeKindID.Int()), modeKindID != spec.LexModeKindIDNil
+}
+
+func (s *lexSpec) KindIDAndName(mode ModeID, modeKind ModeKindID) (KindID, string) {
+ kindID := s.spec.KindIDs[mode][modeKind]
+ return KindID(kindID.Int()), s.spec.KindNames[kindID].String()
+}
+
+// go:embed lexer.go
+var lexerCoreSrc string
+
+func GenLexer(lexSpec *spec.LexicalSpec, pkgName string) ([]byte, error) {
+ var lexerSrc string
+ {
+ fset := token.NewFileSet()
+ f, err := parser.ParseFile(fset, "lexer.go", lexerCoreSrc, parser.ParseComments)
+ if err != nil {
+ return nil, err
+ }
+
+ var b strings.Builder
+ err = format.Node(&b, fset, f)
+ if err != nil {
+ return nil, err
+ }
+
+ lexerSrc = b.String()
+ }
+
+ var modeIDsSrc string
+ {
+ var b strings.Builder
+ fmt.Fprintf(&b, "const (\n")
+ for i, k := range lexSpec.ModeNames {
+ if i == spec.LexModeIDNil.Int() {
+ fmt.Fprintf(&b, " ModeIDNil ModeID = %v\n", i)
+ continue
+ }
+ fmt.Fprintf(&b, " ModeID%v ModeID = %v\n", lexical.SnakeCaseToUpperCamelCase(k.String()), i)
+ }
+ fmt.Fprintf(&b, ")")
+
+ modeIDsSrc = b.String()
+ }
+
+ var modeNamesSrc string
+ {
+ var b strings.Builder
+ fmt.Fprintf(&b, "const (\n")
+ for i, k := range lexSpec.ModeNames {
+ if i == spec.LexModeIDNil.Int() {
+ fmt.Fprintf(&b, " ModeNameNil = %#v\n", "")
+ continue
+ }
+ fmt.Fprintf(&b, " ModeName%v = %#v\n", lexical.SnakeCaseToUpperCamelCase(k.String()), k)
+ }
+ fmt.Fprintf(&b, ")")
+
+ modeNamesSrc = b.String()
+ }
+
+ var modeIDToNameSrc string
+ {
+ var b strings.Builder
+ fmt.Fprintf(&b, `
+// ModeIDToName converts a mode ID to a name.
+func ModeIDToName(id ModeID) string {
+ switch id {`)
+ for i, k := range lexSpec.ModeNames {
+ if i == spec.LexModeIDNil.Int() {
+ fmt.Fprintf(&b, `
+ case ModeIDNil:
+ return ModeNameNil`)
+ continue
+ }
+ name := lexical.SnakeCaseToUpperCamelCase(k.String())
+ fmt.Fprintf(&b, `
+ case ModeID%v:
+ return ModeName%v`, name, name)
+ }
+ fmt.Fprintf(&b, `
+ }
+ return ""
+}
+`)
+
+ modeIDToNameSrc = b.String()
+ }
+
+ var kindIDsSrc string
+ {
+ var b strings.Builder
+ fmt.Fprintf(&b, "const (\n")
+ for i, k := range lexSpec.KindNames {
+ if i == spec.LexKindIDNil.Int() {
+ fmt.Fprintf(&b, " KindIDNil KindID = %v\n", i)
+ continue
+ }
+ fmt.Fprintf(&b, " KindID%v KindID = %v\n", lexical.SnakeCaseToUpperCamelCase(k.String()), i)
+ }
+ fmt.Fprintf(&b, ")")
+
+ kindIDsSrc = b.String()
+ }
+
+ var kindNamesSrc string
+ {
+ var b strings.Builder
+ fmt.Fprintf(&b, "const (\n")
+ fmt.Fprintf(&b, " KindNameNil = %#v\n", "")
+ for _, k := range lexSpec.KindNames[1:] {
+ fmt.Fprintf(&b, " KindName%v = %#v\n", lexical.SnakeCaseToUpperCamelCase(k.String()), k)
+ }
+ fmt.Fprintf(&b, ")")
+
+ kindNamesSrc = b.String()
+ }
+
+ var kindIDToNameSrc string
+ {
+ var b strings.Builder
+ fmt.Fprintf(&b, `
+// KindIDToName converts a kind ID to a name.
+func KindIDToName(id KindID) string {
+ switch id {`)
+ for i, k := range lexSpec.KindNames {
+ if i == spec.LexModeIDNil.Int() {
+ fmt.Fprintf(&b, `
+ case KindIDNil:
+ return KindNameNil`)
+ continue
+ }
+ name := lexical.SnakeCaseToUpperCamelCase(k.String())
+ fmt.Fprintf(&b, `
+ case KindID%v:
+ return KindName%v`, name, name)
+ }
+ fmt.Fprintf(&b, `
+ }
+ return ""
+}
+`)
+
+ kindIDToNameSrc = b.String()
+ }
+
+ var specSrc string
+ {
+ t, err := template.New("").Funcs(genTemplateFuncs(lexSpec)).Parse(lexSpecTemplate)
+ if err != nil {
+ return nil, err
+ }
+
+ var b strings.Builder
+ err = t.Execute(&b, map[string]interface{}{
+ "initialModeID": "ModeID" + lexical.SnakeCaseToUpperCamelCase(lexSpec.ModeNames[lexSpec.InitialModeID].String()),
+ "modeIDNil": "ModeIDNil",
+ "modeKindIDNil": spec.LexModeKindIDNil,
+ "stateIDNil": spec.StateIDNil,
+ "compressionLevel": lexSpec.CompressionLevel,
+ })
+ if err != nil {
+ return nil, err
+ }
+
+ specSrc = b.String()
+ }
+
+ var src string
+ {
+ tmpl := `// Code generated by vartan-go. DO NOT EDIT.
+{{ .lexerSrc }}
+
+{{ .modeIDsSrc }}
+
+{{ .modeNamesSrc }}
+
+{{ .modeIDToNameSrc }}
+
+{{ .kindIDsSrc }}
+
+{{ .kindNamesSrc }}
+
+{{ .kindIDToNameSrc }}
+
+{{ .specSrc }}
+`
+
+ t, err := template.New("").Parse(tmpl)
+ if err != nil {
+ return nil, err
+ }
+
+ var b strings.Builder
+ err = t.Execute(&b, map[string]string{
+ "lexerSrc": lexerSrc,
+ "modeIDsSrc": modeIDsSrc,
+ "modeNamesSrc": modeNamesSrc,
+ "modeIDToNameSrc": modeIDToNameSrc,
+ "kindIDsSrc": kindIDsSrc,
+ "kindNamesSrc": kindNamesSrc,
+ "kindIDToNameSrc": kindIDToNameSrc,
+ "specSrc": specSrc,
+ })
+ if err != nil {
+ return nil, err
+ }
+
+ src = b.String()
+ }
+
+ fset := token.NewFileSet()
+ f, err := parser.ParseFile(fset, "", src, parser.ParseComments)
+ if err != nil {
+ return nil, err
+ }
+
+ f.Name = ast.NewIdent(pkgName)
+
+ var b bytes.Buffer
+ err = format.Node(&b, fset, f)
+ if err != nil {
+ return nil, err
+ }
+
+ return b.Bytes(), nil
+}
+
+const lexSpecTemplate = `
+type lexSpec struct {
+ pop [][]bool
+ push [][]ModeID
+ modeNames []string
+ initialStates []StateID
+ acceptances [][]ModeKindID
+ kindIDs [][]KindID
+ kindNames []string
+ initialModeID ModeID
+ modeIDNil ModeID
+ modeKindIDNil ModeKindID
+ stateIDNil StateID
+
+ rowNums [][]int
+ rowDisplacements [][]int
+ bounds [][]int
+ entries [][]StateID
+ originalColCounts []int
+}
+
+func NewLexSpec() *lexSpec {
+ return &lexSpec{
+ pop: {{ genPopTable }},
+ push: {{ genPushTable }},
+ modeNames: {{ genModeNameTable }},
+ initialStates: {{ genInitialStateTable }},
+ acceptances: {{ genAcceptTable }},
+ kindIDs: {{ genKindIDTable }},
+ kindNames: {{ genKindNameTable }},
+ initialModeID: {{ .initialModeID }},
+ modeIDNil: {{ .modeIDNil }},
+ modeKindIDNil: {{ .modeKindIDNil }},
+ stateIDNil: {{ .stateIDNil }},
+
+ rowNums: {{ genRowNums }},
+ rowDisplacements: {{ genRowDisplacements }},
+ bounds: {{ genBounds }},
+ entries: {{ genEntries }},
+ originalColCounts: {{ genOriginalColCounts }},
+ }
+}
+
+func (s *lexSpec) InitialMode() ModeID {
+ return s.initialModeID
+}
+
+func (s *lexSpec) Pop(mode ModeID, modeKind ModeKindID) bool {
+ return s.pop[mode][modeKind]
+}
+
+func (s *lexSpec) Push(mode ModeID, modeKind ModeKindID) (ModeID, bool) {
+ id := s.push[mode][modeKind]
+ return id, id != s.modeIDNil
+}
+
+func (s *lexSpec) ModeName(mode ModeID) string {
+ return s.modeNames[mode]
+}
+
+func (s *lexSpec) InitialState(mode ModeID) StateID {
+ return s.initialStates[mode]
+}
+
+func (s *lexSpec) NextState(mode ModeID, state StateID, v int) (StateID, bool) {
+{{ if eq .compressionLevel 2 -}}
+ rowNum := s.rowNums[mode][state]
+ d := s.rowDisplacements[mode][rowNum]
+ if s.bounds[mode][d+v] != rowNum {
+ return s.stateIDNil, false
+ }
+ return s.entries[mode][d+v], true
+{{ else if eq .compressionLevel 1 -}}
+ rowNum := s.rowNums[mode][state]
+ colCount := s.originalColCounts[mode]
+ next := s.entries[mode][rowNum*colCount+v]
+ if next == s.stateIDNil {
+ return s.stateIDNil, false
+ }
+ return next, true
+{{ else -}}
+ colCount := s.originalColCounts[mode]
+ next := s.entries[mode][int(state)*colCount+v]
+ if next == s.stateIDNil {
+ return s.stateIDNil, false
+ }
+ return next, true
+{{ end -}}
+}
+
+func (s *lexSpec) Accept(mode ModeID, state StateID) (ModeKindID, bool) {
+ id := s.acceptances[mode][state]
+ return id, id != s.modeKindIDNil
+}
+
+func (s *lexSpec) KindIDAndName(mode ModeID, modeKind ModeKindID) (KindID, string) {
+ id := s.kindIDs[mode][modeKind]
+ return id, s.kindNames[id]
+}
+`
+
+func genTemplateFuncs(lexSpec *spec.LexicalSpec) template.FuncMap {
+ fns := template.FuncMap{
+ "genPopTable": func() string {
+ var b strings.Builder
+ fmt.Fprintf(&b, "[][]bool{\n")
+ for i, s := range lexSpec.Specs {
+ if i == spec.LexModeIDNil.Int() {
+ fmt.Fprintf(&b, "nil,\n")
+ continue
+ }
+
+ c := 1
+ fmt.Fprintf(&b, "{\n")
+ for _, v := range s.Pop {
+ fmt.Fprintf(&b, "%v, ", v != 0)
+
+ if c == 20 {
+ fmt.Fprintf(&b, "\n")
+ c = 1
+ } else {
+ c++
+ }
+ }
+ if c > 1 {
+ fmt.Fprintf(&b, "\n")
+ }
+ fmt.Fprintf(&b, "},\n")
+ }
+ fmt.Fprintf(&b, "}")
+ return b.String()
+ },
+ "genPushTable": func() string {
+ var b strings.Builder
+ fmt.Fprintf(&b, "[][]ModeID{\n")
+ for i, s := range lexSpec.Specs {
+ if i == spec.LexModeIDNil.Int() {
+ fmt.Fprintf(&b, "nil,\n")
+ continue
+ }
+
+ c := 1
+ fmt.Fprintf(&b, "{\n")
+ for _, v := range s.Push {
+ fmt.Fprintf(&b, "%v,", v)
+
+ if c == 20 {
+ fmt.Fprintf(&b, "\n")
+ c = 1
+ } else {
+ c++
+ }
+ }
+ if c > 1 {
+ fmt.Fprintf(&b, "\n")
+ }
+ fmt.Fprintf(&b, "},\n")
+ }
+ fmt.Fprintf(&b, "}")
+ return b.String()
+ },
+ "genModeNameTable": func() string {
+ var b strings.Builder
+ fmt.Fprintf(&b, "[]string{\n")
+ for i, name := range lexSpec.ModeNames {
+ if i == spec.LexModeIDNil.Int() {
+ fmt.Fprintf(&b, "ModeNameNil,\n")
+ continue
+ }
+ fmt.Fprintf(&b, "ModeName%v,\n", lexical.SnakeCaseToUpperCamelCase(name.String()))
+ }
+ fmt.Fprintf(&b, "}")
+ return b.String()
+ },
+ "genInitialStateTable": func() string {
+ var b strings.Builder
+ fmt.Fprintf(&b, "[]StateID{\n")
+ for i, s := range lexSpec.Specs {
+ if i == spec.LexModeIDNil.Int() {
+ fmt.Fprintf(&b, "%v,\n", spec.StateIDNil)
+ continue
+ }
+
+ fmt.Fprintf(&b, "%v,\n", s.DFA.InitialStateID)
+ }
+ fmt.Fprintf(&b, "}")
+ return b.String()
+ },
+ "genAcceptTable": func() string {
+ var b strings.Builder
+ fmt.Fprintf(&b, "[][]ModeKindID{\n")
+ for i, s := range lexSpec.Specs {
+ if i == spec.LexModeIDNil.Int() {
+ fmt.Fprintf(&b, "nil,\n")
+ continue
+ }
+
+ c := 1
+ fmt.Fprintf(&b, "{\n")
+ for _, v := range s.DFA.AcceptingStates {
+ fmt.Fprintf(&b, "%v,", v)
+
+ if c == 20 {
+ fmt.Fprintf(&b, "\n")
+ c = 1
+ } else {
+ c++
+ }
+ }
+ if c > 1 {
+ fmt.Fprintf(&b, "\n")
+ }
+ fmt.Fprintf(&b, "},\n")
+ }
+ fmt.Fprintf(&b, "}")
+ return b.String()
+ },
+ "genKindIDTable": func() string {
+ var b strings.Builder
+ fmt.Fprintf(&b, "[][]KindID{\n")
+ for i, ids := range lexSpec.KindIDs {
+ if i == spec.LexModeIDNil.Int() {
+ fmt.Fprintf(&b, "nil,\n")
+ continue
+ }
+
+ fmt.Fprintf(&b, "{\n")
+ for j, id := range ids {
+ if j == spec.LexModeKindIDNil.Int() {
+ fmt.Fprintf(&b, "KindIDNil,\n")
+ continue
+ }
+ fmt.Fprintf(&b, "KindID%v,\n", lexical.SnakeCaseToUpperCamelCase(string(lexSpec.KindNames[id].String())))
+ }
+ fmt.Fprintf(&b, "},\n")
+ }
+ fmt.Fprintf(&b, "}")
+ return b.String()
+ },
+ "genKindNameTable": func() string {
+ var b strings.Builder
+ fmt.Fprintf(&b, "[]string{\n")
+ for i, name := range lexSpec.KindNames {
+ if i == spec.LexKindIDNil.Int() {
+ fmt.Fprintf(&b, "KindNameNil,\n")
+ continue
+ }
+ fmt.Fprintf(&b, "KindName%v,\n", lexical.SnakeCaseToUpperCamelCase(name.String()))
+ }
+ fmt.Fprintf(&b, "}")
+ return b.String()
+ },
+ }
+
+ switch lexSpec.CompressionLevel {
+ case 2:
+ fns["genRowNums"] = func() string {
+ var b strings.Builder
+ fmt.Fprintf(&b, "[][]int{\n")
+ for i, s := range lexSpec.Specs {
+ if i == spec.LexModeIDNil.Int() {
+ fmt.Fprintf(&b, "nil,\n")
+ continue
+ }
+
+ c := 1
+ fmt.Fprintf(&b, "{\n")
+ for _, v := range s.DFA.Transition.RowNums {
+ fmt.Fprintf(&b, "%v,", v)
+
+ if c == 20 {
+ fmt.Fprintf(&b, "\n")
+ c = 1
+ } else {
+ c++
+ }
+ }
+ if c > 1 {
+ fmt.Fprintf(&b, "\n")
+ }
+ fmt.Fprintf(&b, "},\n")
+ }
+ fmt.Fprintf(&b, "}")
+ return b.String()
+ }
+
+ fns["genRowDisplacements"] = func() string {
+ var b strings.Builder
+ fmt.Fprintf(&b, "[][]int{\n")
+ for i, s := range lexSpec.Specs {
+ if i == spec.LexModeIDNil.Int() {
+ fmt.Fprintf(&b, "nil,\n")
+ continue
+ }
+
+ c := 1
+ fmt.Fprintf(&b, "{\n")
+ for _, d := range s.DFA.Transition.UniqueEntries.RowDisplacement {
+ fmt.Fprintf(&b, "%v,", d)
+
+ if c == 20 {
+ fmt.Fprintf(&b, "\n")
+ c = 1
+ } else {
+ c++
+ }
+ }
+ if c > 1 {
+ fmt.Fprintf(&b, "\n")
+ }
+ fmt.Fprintf(&b, "},\n")
+ }
+ fmt.Fprintf(&b, "}")
+ return b.String()
+ }
+
+ fns["genBounds"] = func() string {
+ var b strings.Builder
+ fmt.Fprintf(&b, "[][]int{\n")
+ for i, s := range lexSpec.Specs {
+ if i == spec.LexModeIDNil.Int() {
+ fmt.Fprintf(&b, "nil,\n")
+ continue
+ }
+
+ c := 1
+ fmt.Fprintf(&b, "{\n")
+ for _, v := range s.DFA.Transition.UniqueEntries.Bounds {
+ fmt.Fprintf(&b, "%v,", v)
+
+ if c == 20 {
+ fmt.Fprintf(&b, "\n")
+ c = 1
+ } else {
+ c++
+ }
+ }
+ if c > 1 {
+ fmt.Fprintf(&b, "\n")
+ }
+ fmt.Fprintf(&b, "},\n")
+ }
+ fmt.Fprintf(&b, "}")
+ return b.String()
+ }
+
+ fns["genEntries"] = func() string {
+ var b strings.Builder
+ fmt.Fprintf(&b, "[][]StateID{\n")
+ for i, s := range lexSpec.Specs {
+ if i == spec.LexModeIDNil.Int() {
+ fmt.Fprintf(&b, "nil,\n")
+ continue
+ }
+
+ c := 1
+ fmt.Fprintf(&b, "{\n")
+ for _, v := range s.DFA.Transition.UniqueEntries.Entries {
+ fmt.Fprintf(&b, "%v,", v)
+
+ if c == 20 {
+ fmt.Fprintf(&b, "\n")
+ c = 1
+ } else {
+ c++
+ }
+ }
+ if c > 1 {
+ fmt.Fprintf(&b, "\n")
+ }
+ fmt.Fprintf(&b, "},\n")
+ }
+ fmt.Fprintf(&b, "}")
+ return b.String()
+ }
+
+ fns["genOriginalColCounts"] = func() string {
+ return "nil"
+ }
+ case 1:
+ fns["genRowNums"] = func() string {
+ var b strings.Builder
+ fmt.Fprintf(&b, "[][]int{\n")
+ for i, s := range lexSpec.Specs {
+ if i == spec.LexModeIDNil.Int() {
+ fmt.Fprintf(&b, "nil,\n")
+ continue
+ }
+
+ c := 1
+ fmt.Fprintf(&b, "{\n")
+ for _, v := range s.DFA.Transition.RowNums {
+ fmt.Fprintf(&b, "%v,", v)
+
+ if c == 20 {
+ fmt.Fprintf(&b, "\n")
+ c = 1
+ } else {
+ c++
+ }
+ }
+ if c > 1 {
+ fmt.Fprintf(&b, "\n")
+ }
+ fmt.Fprintf(&b, "},\n")
+ }
+ fmt.Fprintf(&b, "}")
+ return b.String()
+ }
+
+ fns["genRowDisplacements"] = func() string {
+ return "nil"
+ }
+
+ fns["genBounds"] = func() string {
+ return "nil"
+ }
+
+ fns["genEntries"] = func() string {
+ var b strings.Builder
+ fmt.Fprintf(&b, "[][]StateID{\n")
+ for i, s := range lexSpec.Specs {
+ if i == spec.LexModeIDNil.Int() {
+ fmt.Fprintf(&b, "nil,\n")
+ continue
+ }
+
+ c := 1
+ fmt.Fprintf(&b, "{\n")
+ for _, v := range s.DFA.Transition.UncompressedUniqueEntries {
+ fmt.Fprintf(&b, "%v,", v)
+
+ if c == 20 {
+ fmt.Fprintf(&b, "\n")
+ c = 1
+ } else {
+ c++
+ }
+ }
+ if c > 1 {
+ fmt.Fprintf(&b, "\n")
+ }
+ fmt.Fprintf(&b, "},\n")
+ }
+ fmt.Fprintf(&b, "}")
+ return b.String()
+ }
+
+ fns["genOriginalColCounts"] = func() string {
+ var b strings.Builder
+ fmt.Fprintf(&b, "[]int{\n")
+ for i, s := range lexSpec.Specs {
+ if i == spec.LexModeIDNil.Int() {
+ fmt.Fprintf(&b, "0,\n")
+ continue
+ }
+
+ fmt.Fprintf(&b, "%v,\n", s.DFA.Transition.OriginalColCount)
+ }
+ fmt.Fprintf(&b, "}")
+ return b.String()
+ }
+ default:
+ fns["genRowNums"] = func() string {
+ return "nil"
+ }
+
+ fns["genRowDisplacements"] = func() string {
+ return "nil"
+ }
+
+ fns["genBounds"] = func() string {
+ return "nil"
+ }
+
+ fns["genEntries"] = func() string {
+ var b strings.Builder
+ fmt.Fprintf(&b, "[][]StateID{\n")
+ for i, s := range lexSpec.Specs {
+ if i == spec.LexModeIDNil.Int() {
+ fmt.Fprintf(&b, "nil,\n")
+ continue
+ }
+
+ c := 1
+ fmt.Fprintf(&b, "{\n")
+ for _, v := range s.DFA.UncompressedTransition {
+ fmt.Fprintf(&b, "%v,", v)
+
+ if c == 20 {
+ fmt.Fprintf(&b, "\n")
+ c = 1
+ } else {
+ c++
+ }
+ }
+ if c > 1 {
+ fmt.Fprintf(&b, "\n")
+ }
+ fmt.Fprintf(&b, "},\n")
+ }
+ fmt.Fprintf(&b, "}")
+ return b.String()
+ }
+
+ fns["genOriginalColCounts"] = func() string {
+ var b strings.Builder
+ fmt.Fprintf(&b, "[]int{\n")
+ for i, s := range lexSpec.Specs {
+ if i == spec.LexModeIDNil.Int() {
+ fmt.Fprintf(&b, "0,\n")
+ continue
+ }
+
+ fmt.Fprintf(&b, "%v,\n", s.DFA.ColCount)
+ }
+ fmt.Fprintf(&b, "}")
+ return b.String()
+ }
+ }
+
+ return fns
+}
generated by cgit v1.2.3 (git 2.49.0) at 2025-07-06 10:33:47 +0000