Merge pull request 'parser: implement basic dosini parsing' (#2)

Reviewed-on: #2

go.mod

@@ -2,8 +2,6 @@ module asciigoat.org/ini
 
 go 1.19
 
-replace asciigoat.org/core => ../core
-
 require (
 	asciigoat.org/core v0.3.6
 	github.com/mgechev/revive v1.3.3

go.sum

@@ -1,3 +1,5 @@
+asciigoat.org/core v0.3.6 h1:b1vL090OxylmSOwLQryjrmC8FhhCtktMyeJSy1e6LwI=
+asciigoat.org/core v0.3.6/go.mod h1:tXj+JUutxRbcO40ZQRuUVaZ4rnYz1kAZ0nblisV8u74=
 github.com/BurntSushi/toml v1.3.2 h1:o7IhLm0Msx3BaB+n3Ag7L8EVlByGnpq14C4YWiu/gL8=
 github.com/BurntSushi/toml v1.3.2/go.mod h1:CxXYINrC8qIiEnFrOxCa7Jy5BFHlXnUU2pbicEuybxQ=
 github.com/chavacava/garif v0.0.0-20230608123814-4bd63c2919ab h1:5JxePczlyGAtj6R1MUEFZ/UFud6FfsOejq7xLC2ZIb0=

parser/lexer.go

@@ -1,8 +1,6 @@
 package parser
 
-import (
-	"asciigoat.org/core/lexer"
-)
+import "asciigoat.org/core/lexer"
 
 // Run parses the source
 func (p *Parser) Run() error {
@@ -17,46 +15,33 @@ func (p *Parser) lexStart() (lexer.StateFn, error) {
 		r, _, err := p.src.ReadRune()
 		switch {
 		case err != nil:
-			// read error
 			return p.emitError("", err)
 		case IsNewLine(r):
 			// new line
-			p.lexNewLine(r)
+			p.lexMoreNewLine(r)
 			p.stepLine()
 		case IsSpace(r):
 			// whitespace
 			p.stepRune()
-		default:
-			// token
+		case IsCommentStart(r):
+			// switch to comment lexer
 			p.src.UnreadRune()
-			return p.lexToken, nil
+			return p.lexComment, nil
+		case IsSectionStart(r):
+			// section
+			return p.lexSectionStart, nil
+		default:
+			// entry
+			p.src.UnreadRune()
+			return p.lexEntryStart, nil
 		}
 	}
 }
 
-func (p *Parser) lexToken() (lexer.StateFn, error) {
-	p.src.AcceptAll(IsNotSpace)
-
-	p.pushString(TokenUnknown)
-
-	return p.lexStart, nil
-}
-
-func (p *Parser) lexNewLine(r1 rune) {
-	// r1 is warrantied to be either \n or \r
+func (p *Parser) lexMoreNewLine(r1 rune) {
+	// r1 is warrantied to be either '\r' or '\n'
 	r2, _, err := p.src.ReadRune()
-
 	switch r1 {
-	case '\r':
-		switch {
-		case r2 == '\n':
-			// CR LN
-		case err == nil:
-			// CR
-			p.src.UnreadRune()
-		default:
-			// CR EOF
-		}
 	case '\n':
 		switch {
 		case r2 == '\r':
@@ -67,7 +52,93 @@ func (p *Parser) lexNewLine(r1 rune) {
 		default:
 			// LN EOF
 		}
+	case '\r':
+		switch {
+		case r2 == '\n':
+			// CR LN
+		case err == nil:
+			// CR
+			p.src.UnreadRune()
+		default:
+			// CR EOF
+		}
 	default:
 		panic("unreachable")
 	}
 }
+
+func (p *Parser) lexComment() (lexer.StateFn, error) {
+	// until the end of the line
+	p.src.AcceptAll(IsNotNewLine)
+
+	err := p.emitString(TokenComment)
+	return p.lexStart, err
+}
+
+func (p *Parser) lexSectionStart() (lexer.StateFn, error) {
+	if err := p.emitString(TokenSectionStart); err != nil {
+		return nil, err
+	}
+
+	// remove whitespace between `[` and the name
+	if p.src.AcceptAll(IsSpaceNotNewLine) {
+		p.stepString()
+	}
+
+	if !p.src.AcceptAll(IsName) {
+		// no name
+		return p.emitError("section name missing", lexer.ErrUnacceptableRune)
+	}
+
+	if err := p.emitString(TokenSectionName); err != nil {
+		return nil, err
+	}
+
+	// remove whitespace between the name andthe closing `]`
+	if p.src.AcceptAll(IsSpaceNotNewLine) {
+		p.stepString()
+	}
+
+	r, _, err := p.src.ReadRune()
+	switch {
+	case err != nil:
+		return p.emitError("", err)
+	case IsSectionEnd(r):
+		err := p.emitString(TokenSectionEnd)
+		return p.lexStart, err
+	default:
+		return p.emitInvalidRune(r)
+	}
+}
+
+func (p *Parser) lexEntryStart() (lexer.StateFn, error) {
+	p.src.AcceptAll(IsName)
+	if err := p.emitString(TokenFieldKey); err != nil {
+		return nil, err
+	}
+
+	// ignore whitespace between key and the '=' sign
+	if p.src.AcceptAll(IsSpaceNotNewLine) {
+		p.stepString()
+	}
+
+	r, _, err := p.src.ReadRune()
+	switch {
+	case err != nil:
+		return p.emitError("", err)
+	case r != RuneFieldEqual:
+		return p.emitInvalidRune(r)
+	}
+
+	// ignore whitespace between the '=' and the value
+	if p.src.AcceptAll(IsSpaceNotNewLine) {
+		p.stepString()
+	}
+
+	p.src.AcceptAll(IsNotNewLine)
+	if err := p.emitString(TokenFieldValue); err != nil {
+		return nil, err
+	}
+
+	return p.lexStart, err
+}

parser/lexer_error.go

@@ -1,15 +0,0 @@
-package parser
-
-import "asciigoat.org/core/lexer"
-
-func (p *Parser) emitError(content string, err error) (lexer.StateFn, error) {
-	err2 := p.OnError(p.pos, content, err)
-	switch {
-	case err2 != nil:
-		// return wrapped error
-		return nil, err2
-	default:
-		// return original error
-		return nil, err
-	}
-}

parser/lexer_queue.go

@@ -1,44 +0,0 @@
-package parser
-
-import "log"
-
-func (p *Parser) push(tok Token) {
-	n := len(p.queue)
-	p.queue = append(p.queue, tok)
-
-	log.Printf("queue[%v]: %s", n, tok)
-}
-
-func (p *Parser) pushString(typ TokenType) {
-	s := p.src.Emit()
-
-	el := Token{
-		Type:     typ,
-		Value:    s,
-		Position: p.pos,
-	}
-
-	p.pos.StepN(len(s))
-
-	p.push(el)
-}
-
-// stepLine discards the data and moves the position
-// to the next line
-func (p *Parser) stepLine() {
-	p.src.Discard()
-	p.pos.StepLine()
-}
-
-// stepRune discards the data and moves the position
-// on rune forward on the same line
-func (p *Parser) stepRune() {
-	p.src.Discard()
-	p.pos.Step()
-}
-
-func (p *Parser) stepString() string {
-	s := p.src.Emit()
-	p.pos.StepN(len(s))
-	return s
-}

parser/lexer_runes.go

@@ -1,19 +1,67 @@
 package parser
 
-import "asciigoat.org/core/lexer"
+import (
+	"strings"
+
+	"asciigoat.org/core/lexer"
+)
+
+const (
+	RuneComment      = ';' // RuneComment is the standard dosini comment character
+	RuneCommentExtra = '#' // RuneCommentExtra is UNIX shell's comment character
+	RuneSectionStart = '[' // RuneSectionStart indicates the start of a section declaration
+	RuneSectionEnd   = ']' // RuneSectionEnd indiciates the end of a section declaration
+	RuneFieldEqual   = '=' // RuneFieldEqual separates field keys from their values
+)
 
 var (
-	// IsNewLine tells if a rune represents a line break or the start of one
-	IsNewLine = lexer.NewIsIn("\n\r")
-	// IsSpace tells if a rune is considered whitespace by unicode
-	IsSpace = lexer.IsSpace
-	// IsNotNewLine tells if a rune is anything other than line breaks
+	// RunesComment is a string containing all runes acceptable to start comments
+	RunesComment = string([]rune{
+		RuneComment,
+		RuneCommentExtra,
+	})
+	// RunesSpecial is a string containing all the runes with special meaning
+	RunesSpecial = string([]rune{
+		RuneComment,
+		RuneCommentExtra,
+		RuneSectionStart,
+		RuneSectionEnd,
+		RuneFieldEqual,
+	})
+)
+
+var (
+	// IsNewLine tells if the rune indicates a line break or the start of one
+	IsNewLine = lexer.NewIsIn("\r\n")
+	// IsNotNewLine tells if the rune is not a line break nor the start of one
 	IsNotNewLine = lexer.NewIsNot(IsNewLine)
-	// IsNotSpace tells if a rune is anything other than whitespace
+	// IsSpace tells if the rune is considered whitespace by Unicode
+	IsSpace = lexer.IsSpace
+	// IsNotSpace tells if the rune is not considered whitespace by Unicode
 	IsNotSpace = lexer.NewIsNot(IsSpace)
+	// IsCommentStart ...
+	IsCommentStart = lexer.NewIsIn(RunesComment)
 )
 
 // IsSpaceNotNewLine indicates a rune is whitespace but not a new line
 func IsSpaceNotNewLine(r rune) bool {
 	return IsSpace(r) && !IsNewLine(r)
 }
+
+// IsSectionStart indicates the rune starts the section declaration
+func IsSectionStart(r rune) bool { return r == RuneSectionStart }
+
+// IsSectionEnd indicates the rune ends the section declaration
+func IsSectionEnd(r rune) bool { return r == RuneSectionEnd }
+
+// IsName indicates a rune is acceptable for section or field names
+func IsName(r rune) bool {
+	switch {
+	case IsSpace(r):
+		return false
+	case strings.ContainsRune(RunesSpecial, r):
+		return false
+	default:
+		return true
+	}
+}

parser/parser.go

@@ -3,6 +3,7 @@ package parser
 
 import (
 	"io"
+	"log"
 
 	"asciigoat.org/core/lexer"
 )
@@ -10,61 +11,91 @@ import (
 // Parser parses a dosini-style document
 type Parser struct {
 	src *lexer.Reader
+	pos lexer.Position
 
-	pos   lexer.Position
-	queue []Token
+	// OnToken is called for each identified token. if it returns an error
+	// parsing is interrupted.
+	OnToken func(pos lexer.Position, typ TokenType, value string) error
 
-	// OnSection is called after a [section] is parsed.
-	// Returning an error will abort the process.
-	OnSection func(pos lexer.Position, name, subname string, hasSubname bool) error
-
-	// OnField is called after a `key = value` entry is parsed
-	// Returning an error will abort the process.
-	OnField func(pos lexer.Position, key, value string) error
-
-	// OnComment is called after a comment is parsed
-	// Returning an error will abort the process.
-	OnComment func(pos lexer.Position, comment string) error
-
-	// OnError is called after each parsing error, which you are allowed to
-	// override.
-	// OnError is called for EOF as well, but this error isn't returned as such by
-	// Parser.Run(). The caller will receive (nil, nil) instead indicating the
-	// processes terminated correctly.
+	// OnError is called in case of a parsing error, and it's allowed
+	// to replace the error returned by [Parser.Run].
+	// OnError is called for io.EOF, but [Parser.Run] will consider it
+	// normal termination.
 	OnError func(pos lexer.Position, content string, err error) error
 }
 
-func defaultOnSection(_ lexer.Position, _, _ string, _ bool) error { return nil }
-func defaultOnField(_ lexer.Position, _, _ string) error           { return nil }
-func defaultOnComment(_ lexer.Position, _ string) error            { return nil }
+func defaultOnToken(pos lexer.Position, typ TokenType, value string) error {
+	log.Printf("%s:%v:%v: %q", typ, pos.Line, pos.Column, value)
+	return nil
+}
 
 func defaultOnError(pos lexer.Position, content string, err error) error {
-	return &lexer.Error{
-		Line:    pos.Line,
-		Column:  pos.Column,
+	log.Printf("%s:%v:%v: %q: %s", "error", pos.Line, pos.Column, content, err)
+
+	return lexer.Error{
+		Line:   pos.Line,
+		Column: pos.Column,
+
 		Content: content,
 		Err:     err,
 	}
 }
 
 func (p *Parser) setDefaults() {
-	if p.OnSection == nil {
-		p.OnSection = defaultOnSection
+	if p.OnToken == nil {
+		p.OnToken = defaultOnToken
 	}
-
-	if p.OnField == nil {
-		p.OnField = defaultOnField
-	}
-
-	if p.OnComment == nil {
-		p.OnComment = defaultOnComment
-	}
-
 	if p.OnError == nil {
 		p.OnError = defaultOnError
 	}
 }
 
+func (p *Parser) emitString(typ TokenType) error {
+	s := p.src.Emit()
+	err := p.OnToken(p.pos, typ, s)
+	p.pos.StepN(len(s))
+
+	return err
+}
+
+func (p *Parser) emitError(content string, err error) (lexer.StateFn, error) {
+	err2 := p.OnError(p.pos, content, err)
+	switch {
+	case err2 != nil:
+		// return wrapped error
+		return nil, err2
+	default:
+		// return original error
+		return nil, err
+	}
+}
+
+func (p *Parser) emitInvalidRune(r rune) (lexer.StateFn, error) {
+	return p.emitError(string([]rune{r}), lexer.ErrUnacceptableRune)
+}
+
+// stepLine discards the data and moves the position
+// to the next line.
+func (p *Parser) stepLine() {
+	p.src.Discard()
+	p.pos.StepLine()
+}
+
+// stepRune discards the data and moves the position
+// one rune forward on the same line.
+func (p *Parser) stepRune() {
+	p.src.Discard()
+	p.pos.Step()
+}
+
+// stepString discards the data and moves the position
+// forward on the same line the length of the discarded
+// content.
+func (p *Parser) stepString() {
+	s := p.src.Emit()
+	p.pos.StepN(len(s))
+}
+
 // NewParser creates a dosini-style parser using
 // an [io.Reader] as source
 func NewParser(r io.Reader) *Parser {

parser/token.go

@@ -1,11 +1,6 @@
 package parser
 
 //go:generate go run golang.org/x/tools/cmd/stringer -type=TokenType
-import (
-	"fmt"
-
-	"asciigoat.org/core/lexer"
-)
 
 // A TokenType is a type of Token
 type TokenType uint
@@ -13,6 +8,12 @@ type TokenType uint
 const (
 	// TokenUnknown represents a Token that hasn't been identified
 	TokenUnknown TokenType = iota
+	// TokenSectionStart indicates the opening marker of a section declaration.
+	// The left squared bracket.
+	TokenSectionStart
+	// TokenSectionEnd indicates the closing marker of a section declaration.
+	// The right squared bracket.
+	TokenSectionEnd
 	// TokenSectionName represents the section name between the squared brackets
 	TokenSectionName
 	// TokenSectionSubname represents a secondary name in the section represented
@@ -21,21 +22,10 @@ const (
 	// [section_name "section_subname"]
 	TokenSectionSubname
 	// TokenComment represents a comment, including the initial ';' or '#' until
-	// the end of the line
+	// the end of the line.
 	TokenComment
 	// TokenFieldKey represents a field name in a `key = value` entry
 	TokenFieldKey
 	// TokenFieldValue represents a field value in a `key = value` entry
 	TokenFieldValue
 )
-
-// A Token is an element from the document
-type Token struct {
-	Type     TokenType
-	Position lexer.Position
-	Value    string
-}
-
-func (t Token) String() string {
-	return fmt.Sprintf("%s:%v:%v: %q", t.Type, t.Position.Line, t.Position.Column, t.Value)
-}

parser/tokentype_string.go

@@ -9,16 +9,18 @@ func _() {
 	// Re-run the stringer command to generate them again.
 	var x [1]struct{}
 	_ = x[TokenUnknown-0]
-	_ = x[TokenSectionName-1]
-	_ = x[TokenSectionSubname-2]
-	_ = x[TokenComment-3]
-	_ = x[TokenFieldKey-4]
-	_ = x[TokenFieldValue-5]
+	_ = x[TokenSectionStart-1]
+	_ = x[TokenSectionEnd-2]
+	_ = x[TokenSectionName-3]
+	_ = x[TokenSectionSubname-4]
+	_ = x[TokenComment-5]
+	_ = x[TokenFieldKey-6]
+	_ = x[TokenFieldValue-7]
 }
 
-const _TokenType_name = "TokenUnknownTokenSectionNameTokenSectionSubnameTokenCommentTokenFieldKeyTokenFieldValue"
+const _TokenType_name = "TokenUnknownTokenSectionStartTokenSectionEndTokenSectionNameTokenSectionSubnameTokenCommentTokenFieldKeyTokenFieldValue"
 
-var _TokenType_index = [...]uint8{0, 12, 28, 47, 59, 72, 87}
+var _TokenType_index = [...]uint8{0, 12, 29, 44, 60, 79, 91, 104, 119}
 
 func (i TokenType) String() string {
 	if i >= TokenType(len(_TokenType_index)-1) {