aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
Diffstat
-rw-r--r--driver/parser.go42
-rw-r--r--grammar/grammar.go47
-rw-r--r--spec/grammar.go1
-rw-r--r--spec/lexer.go13
-rw-r--r--spec/lexer_test.go6
-rw-r--r--spec/parser.go27
6 files changed, 92 insertions, 44 deletions
diff --git a/driver/parser.go b/driver/parser.go
index 08120da..c13413a 100644
--- a/driver/parser.go
+++ b/driver/parser.go
@@ -5,7 +5,6 @@ import (
"io"
mldriver "github.com/nihei9/maleeni/driver"
- mlspec "github.com/nihei9/maleeni/spec"
"github.com/nihei9/vartan/spec"
)
@@ -292,23 +291,12 @@ ACTION_LOOP:
continue ACTION_LOOP
}
- var eKindNames []string
- {
- eKinds, eof := p.expectedKinds(p.top())
- for _, k := range eKinds {
- eKindNames = append(eKindNames, k.String())
- }
- if eof {
- eKindNames = append(eKindNames, "<EOF>")
- }
- }
-
p.synErrs = append(p.synErrs, &SyntaxError{
Row: tok.Row,
Col: tok.Col,
Message: "unexpected token",
Token: tok,
- ExpectedTerminals: eKindNames,
+ ExpectedTerminals: p.expectedTerms(p.top()),
})
for {
@@ -418,26 +406,30 @@ func (p *Parser) SyntaxErrors() []*SyntaxError {
return p.synErrs
}
-func (p *Parser) expectedKinds(state int) ([]mlspec.LexKindName, bool) {
- kinds := []mlspec.LexKindName{}
- eof := false
+func (p *Parser) expectedTerms(state int) []string {
+ kinds := []string{}
terms := p.gram.ParsingTable.ExpectedTerminals[state]
+ aliases := p.gram.LexicalSpecification.Maleeni.KindAliases
for _, tsym := range terms {
- if tsym == p.gram.ParsingTable.EOFSymbol {
- eof = true
- continue
- }
-
// We don't add the error symbol to the look-ahead symbols because users cannot input the error symbol
// intentionally.
if tsym == p.gram.ParsingTable.ErrorSymbol {
continue
}
- kindID := p.gram.LexicalSpecification.Maleeni.TerminalToKind[tsym]
- kindName := p.gram.LexicalSpecification.Maleeni.Spec.KindNames[kindID]
- kinds = append(kinds, kindName)
+ if tsym == p.gram.ParsingTable.EOFSymbol {
+ kinds = append(kinds, "<eof>")
+ continue
+ }
+
+ if alias := aliases[tsym]; alias != "" {
+ kinds = append(kinds, alias)
+ } else {
+ term2Kind := p.gram.LexicalSpecification.Maleeni.TerminalToKind
+ kindNames := p.gram.LexicalSpecification.Maleeni.Spec.KindNames
+ kinds = append(kinds, kindNames[term2Kind[tsym]].String())
+ }
}
- return kinds, eof
+ return kinds
}
diff --git a/grammar/grammar.go b/grammar/grammar.go
index 27fafc9..b911ee2 100644
--- a/grammar/grammar.go
+++ b/grammar/grammar.go
@@ -73,6 +73,7 @@ const reservedSymbolNameError = "error"
type Grammar struct {
lexSpec *mlspec.LexSpec
skipLexKinds []mlspec.LexKindName
+ kindAliases map[symbol]string
sym2AnonPat map[symbol]string
productionSet *productionSet
augmentedStartSymbol symbol
@@ -158,6 +159,7 @@ func (b *GrammarBuilder) Build() (*Grammar, error) {
return &Grammar{
lexSpec: symTabAndLexSpec.lexSpec,
skipLexKinds: symTabAndLexSpec.skip,
+ kindAliases: symTabAndLexSpec.aliases,
sym2AnonPat: symTabAndLexSpec.sym2AnonPat,
productionSet: prodsAndActs.prods,
augmentedStartSymbol: prodsAndActs.augStartSym,
@@ -269,6 +271,7 @@ type symbolTableAndLexSpec struct {
errSym symbol
skip []mlspec.LexKindName
skipSyms []string
+ aliases map[symbol]string
}
func (b *GrammarBuilder) genSymbolTableAndLexSpec(root *spec.RootNode) (*symbolTableAndLexSpec, error) {
@@ -332,6 +335,7 @@ func (b *GrammarBuilder) genSymbolTableAndLexSpec(root *spec.RootNode) (*symbolT
skipKinds := []mlspec.LexKindName{}
skipSyms := []string{}
+ aliases := map[symbol]string{}
for _, prod := range root.LexProductions {
if sym, exist := symTab.toSymbol(prod.LHS); exist {
if sym == errSym {
@@ -352,12 +356,12 @@ func (b *GrammarBuilder) genSymbolTableAndLexSpec(root *spec.RootNode) (*symbolT
continue
}
- _, err := symTab.registerTerminalSymbol(prod.LHS)
+ lhsSym, err := symTab.registerTerminalSymbol(prod.LHS)
if err != nil {
return nil, err
}
- entry, skip, specErr, err := genLexEntry(prod)
+ entry, skip, alias, specErr, err := genLexEntry(prod)
if err != nil {
return nil, err
}
@@ -369,6 +373,9 @@ func (b *GrammarBuilder) genSymbolTableAndLexSpec(root *spec.RootNode) (*symbolT
skipKinds = append(skipKinds, mlspec.LexKindName(prod.LHS))
skipSyms = append(skipSyms, prod.LHS)
}
+ if alias != "" {
+ aliases[lhsSym] = alias
+ }
entries = append(entries, entry)
}
@@ -402,17 +409,18 @@ func (b *GrammarBuilder) genSymbolTableAndLexSpec(root *spec.RootNode) (*symbolT
errSym: errSym,
skip: skipKinds,
skipSyms: skipSyms,
+ aliases: aliases,
}, nil
}
-func genLexEntry(prod *spec.ProductionNode) (*mlspec.LexEntry, bool, *verr.SpecError, error) {
+func genLexEntry(prod *spec.ProductionNode) (*mlspec.LexEntry, bool, string, *verr.SpecError, error) {
var modes []mlspec.LexModeName
if prod.Directive != nil {
dir := prod.Directive
switch dir.Name {
case "mode":
if len(dir.Parameters) == 0 {
- return nil, false, &verr.SpecError{
+ return nil, false, "", &verr.SpecError{
Cause: semErrDirInvalidParam,
Detail: fmt.Sprintf("'mode' directive needs an ID parameter"),
Row: dir.Pos.Row,
@@ -421,7 +429,7 @@ func genLexEntry(prod *spec.ProductionNode) (*mlspec.LexEntry, bool, *verr.SpecE
}
for _, param := range dir.Parameters {
if param.ID == "" {
- return nil, false, &verr.SpecError{
+ return nil, false, "", &verr.SpecError{
Cause: semErrDirInvalidParam,
Detail: fmt.Sprintf("'mode' directive needs an ID parameter"),
Row: param.Pos.Row,
@@ -431,7 +439,7 @@ func genLexEntry(prod *spec.ProductionNode) (*mlspec.LexEntry, bool, *verr.SpecE
modes = append(modes, mlspec.LexModeName(param.ID))
}
default:
- return nil, false, &verr.SpecError{
+ return nil, false, "", &verr.SpecError{
Cause: semErrDirInvalidName,
Detail: dir.Name,
Row: dir.Pos.Row,
@@ -444,12 +452,13 @@ func genLexEntry(prod *spec.ProductionNode) (*mlspec.LexEntry, bool, *verr.SpecE
var skip bool
var push mlspec.LexModeName
var pop bool
+ var alias string
if alt.Directive != nil {
dir := alt.Directive
switch dir.Name {
case "skip":
if len(dir.Parameters) > 0 {
- return nil, false, &verr.SpecError{
+ return nil, false, "", &verr.SpecError{
Cause: semErrDirInvalidParam,
Detail: fmt.Sprintf("'skip' directive needs no parameter"),
Row: dir.Pos.Row,
@@ -459,7 +468,7 @@ func genLexEntry(prod *spec.ProductionNode) (*mlspec.LexEntry, bool, *verr.SpecE
skip = true
case "push":
if len(dir.Parameters) != 1 || dir.Parameters[0].ID == "" {
- return nil, false, &verr.SpecError{
+ return nil, false, "", &verr.SpecError{
Cause: semErrDirInvalidParam,
Detail: fmt.Sprintf("'push' directive needs an ID parameter"),
Row: dir.Pos.Row,
@@ -469,7 +478,7 @@ func genLexEntry(prod *spec.ProductionNode) (*mlspec.LexEntry, bool, *verr.SpecE
push = mlspec.LexModeName(dir.Parameters[0].ID)
case "pop":
if len(dir.Parameters) > 0 {
- return nil, false, &verr.SpecError{
+ return nil, false, "", &verr.SpecError{
Cause: semErrDirInvalidParam,
Detail: fmt.Sprintf("'pop' directive needs no parameter"),
Row: dir.Pos.Row,
@@ -477,8 +486,18 @@ func genLexEntry(prod *spec.ProductionNode) (*mlspec.LexEntry, bool, *verr.SpecE
}, nil
}
pop = true
+ case "alias":
+ if len(dir.Parameters) != 1 || dir.Parameters[0].String == "" {
+ return nil, false, "", &verr.SpecError{
+ Cause: semErrDirInvalidParam,
+ Detail: fmt.Sprintf("'alias' directive needs a string parameter"),
+ Row: dir.Pos.Row,
+ Col: dir.Pos.Col,
+ }, nil
+ }
+ alias = dir.Parameters[0].String
default:
- return nil, false, &verr.SpecError{
+ return nil, false, "", &verr.SpecError{
Cause: semErrDirInvalidName,
Detail: dir.Name,
Row: dir.Pos.Row,
@@ -493,7 +512,7 @@ func genLexEntry(prod *spec.ProductionNode) (*mlspec.LexEntry, bool, *verr.SpecE
Pattern: mlspec.LexPattern(alt.Elements[0].Pattern),
Push: push,
Pop: pop,
- }, skip, nil, nil
+ }, skip, alias, nil, nil
}
type productionsAndActions struct {
@@ -920,6 +939,11 @@ func Compile(gram *Grammar, opts ...CompileOption) (*spec.CompiledGrammar, error
return nil, err
}
+ kindAliases := make([]string, gram.symbolTable.termNum.Int())
+ for _, sym := range gram.symbolTable.terminalSymbols() {
+ kindAliases[sym.num().Int()] = gram.kindAliases[sym]
+ }
+
nonTerms, err := gram.symbolTable.nonTerminalTexts()
if err != nil {
return nil, err
@@ -1033,6 +1057,7 @@ func Compile(gram *Grammar, opts ...CompileOption) (*spec.CompiledGrammar, error
KindToTerminal: kind2Term,
TerminalToKind: term2Kind,
Skip: skip,
+ KindAliases: kindAliases,
},
},
ParsingTable: &spec.ParsingTable{
diff --git a/spec/grammar.go b/spec/grammar.go
index 7901957..42e6dc2 100644
--- a/spec/grammar.go
+++ b/spec/grammar.go
@@ -18,6 +18,7 @@ type Maleeni struct {
KindToTerminal []int `json:"kind_to_terminal"`
TerminalToKind []int `json:"terminal_to_kind"`
Skip []int `json:"skip"`
+ KindAliases []string `json:"kind_aliases"`
}
type ParsingTable struct {
diff --git a/spec/lexer.go b/spec/lexer.go
index 258faae..d513d7d 100644
--- a/spec/lexer.go
+++ b/spec/lexer.go
@@ -21,11 +21,12 @@ const (
tokenKindKWFragment = tokenKind("fragment")
tokenKindID = tokenKind("id")
tokenKindTerminalPattern = tokenKind("terminal pattern")
+ tokenKindStringLiteral = tokenKind("string")
tokenKindColon = tokenKind(":")
tokenKindOr = tokenKind("|")
tokenKindSemicolon = tokenKind(";")
tokenKindDirectiveMarker = tokenKind("#")
- tokenKindTreeNodeOpen = tokenKind("'(")
+ tokenKindTreeNodeOpen = tokenKind("#(")
tokenKindTreeNodeClose = tokenKind(")")
tokenKindPosition = tokenKind("$")
tokenKindExpantion = tokenKind("...")
@@ -77,6 +78,14 @@ func newTerminalPatternToken(text string, pos Position) *token {
}
}
+func newStringLiteralToken(text string, pos Position) *token {
+ return &token{
+ kind: tokenKindStringLiteral,
+ text: text,
+ pos: pos,
+ }
+}
+
func newPositionToken(num int, pos Position) *token {
return &token{
kind: tokenKindPosition,
@@ -234,7 +243,7 @@ func (l *lexer) lexAndSkipWSs() (*token, error) {
Col: tok.Col + 1,
}
}
- return newTerminalPatternToken(mlspec.EscapePattern(pat), newPosition(tok.Row+1, tok.Col+1)), nil
+ return newStringLiteralToken(pat, newPosition(tok.Row+1, tok.Col+1)), nil
case "colon":
return newSymbolToken(tokenKindColon, newPosition(tok.Row+1, tok.Col+1)), nil
case "or":
diff --git a/spec/lexer_test.go b/spec/lexer_test.go
index 51e5f59..c3540f6 100644
--- a/spec/lexer_test.go
+++ b/spec/lexer_test.go
@@ -16,6 +16,10 @@ func TestLexer_Run(t *testing.T) {
return newTerminalPatternToken(text, newPosition(1, 0))
}
+ strTok := func(text string) *token {
+ return newStringLiteralToken(text, newPosition(1, 0))
+ }
+
symTok := func(kind tokenKind) *token {
return newSymbolToken(kind, newPosition(1, 0))
}
@@ -40,7 +44,7 @@ func TestLexer_Run(t *testing.T) {
tokens: []*token{
idTok("id"),
termPatTok("terminal"),
- termPatTok(`\.\*\+\?\|\(\)\[\\`),
+ strTok(`.*+?|()[\`),
symTok(tokenKindColon),
symTok(tokenKindOr),
symTok(tokenKindSemicolon),
diff --git a/spec/parser.go b/spec/parser.go
index 53bf491..9778521 100644
--- a/spec/parser.go
+++ b/spec/parser.go
@@ -4,6 +4,7 @@ import (
"fmt"
"io"
+ mlspec "github.com/nihei9/maleeni/spec"
verr "github.com/nihei9/vartan/error"
)
@@ -47,9 +48,10 @@ type DirectiveNode struct {
}
type ParameterNode struct {
- ID string
- Tree *TreeStructNode
- Pos Position
+ ID string
+ String string
+ Tree *TreeStructNode
+ Pos Position
}
type TreeStructNode struct {
@@ -262,10 +264,15 @@ func (p *parser) parseFragment() *FragmentNode {
raiseSyntaxError(p.pos.Row, synErrNoColon)
}
- if !p.consume(tokenKindTerminalPattern) {
+ var rhs string
+ switch {
+ case p.consume(tokenKindTerminalPattern):
+ rhs = p.lastTok.text
+ case p.consume(tokenKindStringLiteral):
+ rhs = mlspec.EscapePattern(p.lastTok.text)
+ default:
raiseSyntaxError(p.pos.Row, synErrFragmentNoPattern)
}
- rhs := p.lastTok.text
p.consume(tokenKindNewline)
@@ -398,6 +405,11 @@ func (p *parser) parseElement() *ElementNode {
Pattern: p.lastTok.text,
Pos: p.lastTok.pos,
}
+ case p.consume(tokenKindStringLiteral):
+ return &ElementNode{
+ Pattern: mlspec.EscapePattern(p.lastTok.text),
+ Pos: p.lastTok.pos,
+ }
}
return nil
}
@@ -436,6 +448,11 @@ func (p *parser) parseParameter() *ParameterNode {
ID: p.lastTok.text,
Pos: p.lastTok.pos,
}
+ case p.consume(tokenKindStringLiteral):
+ return &ParameterNode{
+ String: p.lastTok.text,
+ Pos: p.lastTok.pos,
+ }
case p.consume(tokenKindTreeNodeOpen):
if !p.consume(tokenKindID) {
raiseSyntaxError(p.pos.Row, synErrTreeInvalidFirstElem)
generated by cgit v1.2.3 (git 2.49.0) at 2025-07-12 23:42:05 +0000