parser: Unescaped [WIP]

Signed-off-by: Alejandro Mery <amery@jpi.io>

.vscode/settings.json

@@ -0,0 +1,7 @@
+{
+    "cSpell.words": [
+        "asciigoat",
+        "Subname",
+        "unescapes"
+    ]
+}

basic/error.go

@@ -4,6 +4,7 @@ import (
 	"errors"
 
 	"asciigoat.org/core/lexer"
+	"asciigoat.org/ini/parser"
 )
 
 var (
@@ -11,23 +12,11 @@ var (
 )
 
 func newErrInvalidToken(t *token) *lexer.Error {
-	err := &lexer.Error{
-		Line:    t.pos.Line,
-		Column:  t.pos.Column,
-		Content: t.value,
-		Err:     errInvalidToken,
-	}
-	return err
+	return parser.NewError(t.pos, t.value, "", errInvalidToken)
 }
 
 func (dec *decoder) OnError(pos lexer.Position, content string, err error) error {
-	err = &lexer.Error{
-		Line:    pos.Line,
-		Column:  pos.Column,
-		Content: content,
-		Err:     err,
-	}
-
+	err = parser.NewError(pos, content, "", err)
 	dec.executeFinal()
 	return err
 }

basic/token.go

@@ -32,21 +32,21 @@ func (dec *decoder) executeFinal() {
 func (dec *decoder) execute(typ parser.TokenType) {
 	switch typ {
 	case parser.TokenSectionEnd:
-		name1, ok1 := dec.getValue(1, parser.TokenSectionName)
+		name1, ok1 := dec.queueValue(1, parser.TokenSectionName)
 
 		if ok1 {
-			name2, ok2 := dec.getValue(2, parser.TokenSectionSubname)
+			name2, ok2 := dec.queueValue(2, parser.TokenSectionSubname)
 
 			dec.addSection(name1, name2, ok2)
 		}
 
-		dec.reset()
+		dec.queueReset()
 	case parser.TokenFieldValue:
-		key, _ := dec.getValue(0, parser.TokenFieldKey)
-		value, _ := dec.getValue(1, parser.TokenFieldValue)
+		key, _ := dec.queueValue(0, parser.TokenFieldKey)
+		value, _ := dec.queueValue(1, parser.TokenFieldValue)
 
 		dec.addField(key, value)
-		dec.reset()
+		dec.queueReset()
 	}
 }
 
@@ -82,7 +82,8 @@ func (dec *decoder) addField(key, value string) {
 	}
 }
 
-func (dec *decoder) getValue(idx int, typ parser.TokenType) (string, bool) {
+// queueValue extracts the value of element on the queue if the type matches.
+func (dec *decoder) queueValue(idx int, typ parser.TokenType) (string, bool) {
 	switch {
 	case idx < 0 || idx >= len(dec.queue):
 		// out of range
@@ -95,40 +96,48 @@ func (dec *decoder) getValue(idx int, typ parser.TokenType) (string, bool) {
 	}
 }
 
-func (dec *decoder) reset() {
+// queueReset removes all tokens from the queue
+func (dec *decoder) queueReset() {
 	dec.queue = dec.queue[:0]
 }
 
-func (dec *decoder) depth(depth int) bool {
+// queueDepth confirms the current depth of the queue
+func (dec *decoder) queueDepth(depth int) bool {
 	return len(dec.queue) == depth
 }
 
-func (dec *decoder) depthAfter(depth int, typ parser.TokenType) bool {
-	_, ok := dec.getValue(depth-1, typ)
-	if ok {
-		return len(dec.queue) == depth
+// queueDepthType confirms the current depth of the queue and the type of the last
+// element.
+func (dec *decoder) queueDepthType(depth int, typ parser.TokenType) bool {
+	if dec.queueDepth(depth) {
+		return dec.queueType(depth-1, typ)
 	}
 	return false
 }
 
+// queueType tells if the specified element on the queue is of the required type.
+func (dec *decoder) queueType(idx int, typ parser.TokenType) bool {
+	_, ok := dec.queueValue(idx, typ)
+	return ok
+}
+
 func (dec *decoder) typeOK(typ parser.TokenType) bool {
 	switch typ {
 	case parser.TokenSectionStart, parser.TokenFieldKey:
 		// first token only
-		return dec.depth(0)
+		return dec.queueDepth(0)
 	case parser.TokenSectionName:
 		// right after TokenSectionStart
-		return dec.depthAfter(1, parser.TokenSectionStart)
+		return dec.queueDepthType(1, parser.TokenSectionStart)
 	case parser.TokenSectionSubname:
 		// right after TokenSectionName
-		return dec.depthAfter(2, parser.TokenSectionName)
+		return dec.queueDepthType(2, parser.TokenSectionName)
 	case parser.TokenSectionEnd:
 		// only on a section with name
-		_, ok := dec.getValue(1, parser.TokenSectionName)
-		return ok
+		return dec.queueType(1, parser.TokenSectionName)
 	case parser.TokenFieldValue:
 		// right after a TokenFieldKey
-		return dec.depthAfter(1, parser.TokenFieldKey)
+		return dec.queueDepthType(1, parser.TokenFieldKey)
 	default:
 		// never
 		return false
@@ -149,6 +158,8 @@ func (dec *decoder) OnToken(pos lexer.Position, typ parser.TokenType, value stri
 		return nil
 	default:
 		// unacceptable
-		return newErrInvalidToken(t)
+		err := newErrInvalidToken(t)
+		dec.executeFinal()
+		return err
 	}
 }

go.mod

@@ -2,6 +2,11 @@ module asciigoat.org/ini
 
 go 1.19
 
+replace (
+	asciigoat.org/core => ../core
+	darvaza.org/core => ../../darvaza.org/core
+)
+
 require (
 	asciigoat.org/core v0.3.9
 	github.com/mgechev/revive v1.3.3

go.sum

@@ -1,5 +1,3 @@
-asciigoat.org/core v0.3.9 h1:hgDDz4ecm3ZvehX++m8A/IzAt+B5oDPiRtxatzfUHPQ=
-asciigoat.org/core v0.3.9/go.mod h1:CAaHwyw8MpAq4a1MYtN2dxJrsK+hmIdW50OndaQZYPI=
 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 v0.1.0 h1:2JHa3hbYf5D9dsgseMKAmc/MZ109otzgNFk5s87H9Pc=

parser/comma_array.go

@@ -0,0 +1,88 @@
+package parser
+
+import (
+	"strings"
+
+	"asciigoat.org/core/lexer"
+)
+
+type commaArrayParser struct {
+	TextParser
+
+	out []string
+}
+
+func (p *commaArrayParser) lexStart() (lexer.StateFn, error) {
+	for {
+		r, _, err := p.ReadRune()
+		switch {
+		case err != nil:
+			// EOF
+			return nil, err
+		case r == RuneQuotes:
+			// Quoted Value
+			return p.lexQuotedString, nil
+		case IsNewLine(r):
+			// new lines are acceptable when parsing a string for
+			// comma delimited arrays. but make sure we discard it
+			// complete
+			p.UnreadRune()
+			p.AcceptNewLine()
+			p.Discard()
+		case lexer.IsSpace(r):
+			// discard whitespace outside quotes
+			p.Discard()
+		default:
+			p.UnreadRune()
+			return p.lexWord, nil
+		}
+	}
+}
+
+func (p *commaArrayParser) lexWord() (lexer.StateFn, error) {
+	for {
+		r, _, err := p.ReadRune()
+		switch {
+		case err != nil:
+			// done. store what we got and move on
+			_, s := p.Emit()
+			p.out = append(p.out, s)
+			return nil, err
+		case r == ',':
+			// done
+			_, s := p.Emit()
+			// remove comma, trim and append to output
+			s = strings.TrimRightFunc(s[:len(s)-1], IsSpace)
+			p.out = append(p.out, s)
+			return p.lexStart, nil
+		}
+	}
+}
+
+func (p *commaArrayParser) lexQuotedString() (lexer.StateFn, error) {
+	s, err := lexQuotedString(&p.TextParser)
+	if err != nil {
+		return nil, err
+	}
+
+	p.Discard()
+	p.out = append(p.out, s)
+	return p.lexStart, nil
+}
+
+func (p *commaArrayParser) Run() ([]string, error) {
+	err := lexer.Run(p.lexStart)
+
+	return p.out, err
+}
+
+// SplitCommaArray splits comma separated strings, removing whitespace
+// and respecting quoted literals.
+func SplitCommaArray(s string) ([]string, error) {
+	if s != "" {
+		var p commaArrayParser
+		p.InitString(s)
+		return p.Run()
+	}
+	return nil, nil
+}

parser/error.go

@@ -0,0 +1,62 @@
+package parser
+
+import (
+	"io/fs"
+
+	"asciigoat.org/core/lexer"
+)
+
+// NewError creates a lexer.Error using a lexer.Position
+func NewError(pos lexer.Position, content, hint string, err error) *lexer.Error {
+	return &lexer.Error{
+		Line:    pos.Line,
+		Column:  pos.Column,
+		Content: content,
+		Hint:    hint,
+		Err:     err,
+	}
+}
+
+// ErrPlusPosition returns a copy of the given [lexer.Error]
+// offsetting the Line/Column information.
+func ErrPlusPosition(pos lexer.Position, e *lexer.Error) *lexer.Error {
+	pos.Add(lexer.Position{
+		Line:   e.Line,
+		Column: e.Column,
+	})
+
+	return NewError(pos, e.Content, e.Hint, e.Err)
+}
+
+// NewErrIncompleteQuotedString returns a [lexer.Error]
+// indicating the quoted string being parsed wasn't correctly
+// terminated
+func NewErrIncompleteQuotedString(p *TextParser) *lexer.Error {
+	return newErrIncomplete(p, "incomplete quoted string")
+}
+
+// NewErrIncompleteEscaped returns a [lexer.Error]
+// indicating the text being parsed wasn't correctly
+// terminated
+func NewErrIncompleteEscaped(p *TextParser) *lexer.Error {
+	return newErrIncomplete(p, "incomplete escaped string")
+}
+
+func newErrIncomplete(p *TextParser, hint string) *lexer.Error {
+	pos, s := p.Emit()
+	pos.Add(GetPositionalLength(s))
+
+	return NewError(pos, s, hint, fs.ErrInvalid)
+}
+
+// NewErrInvalidEscapeSequence returns a [lexer.Error] indicating
+// the specified sequence, at the end of the accepted buffer,
+// is invalid
+func NewErrInvalidEscapeSequence(p *TextParser, seq string) *lexer.Error {
+	pos, s := p.Position(), p.String()
+
+	s = s[:len(s)-len(seq)]
+	pos.Add(GetPositionalLength(s))
+
+	return NewError(pos, seq, "invalid escape character", fs.ErrInvalid)
+}

parser/lexer_runes.go

@@ -7,11 +7,13 @@ import (
 )
 
 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
+	RuneComment      = ';'  // RuneComment is the standard INI comment character
+	RuneCommentExtra = '#'  // RuneCommentExtra is UNIX shell's comment character
+	RuneSectionStart = '['  // RuneSectionStart indicates the start of a section declaration
+	RuneSectionEnd   = ']'  // RuneSectionEnd indicates the end of a section declaration
+	RuneFieldEqual   = '='  // RuneFieldEqual separates field keys from their values
+	RuneQuotes       = '"'  // RuneQuotes indicates the start and end of a quoted value
+	RuneEscape       = '\\' // RuneEscape indicates the next rune is escaped
 )
 
 var (
@@ -27,6 +29,8 @@ var (
 		RuneSectionStart,
 		RuneSectionEnd,
 		RuneFieldEqual,
+		RuneQuotes,
+		RuneEscape,
 	})
 )
 

parser/parser.go

@@ -31,13 +31,7 @@ func defaultOnToken(pos lexer.Position, typ TokenType, value string) error {
 func defaultOnError(pos lexer.Position, content string, err error) error {
 	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,
-	}
+	return NewError(pos, content, "", err)
 }
 
 func (p *Parser) setDefaults() {

parser/text_quoted.go

@@ -0,0 +1,135 @@
+package parser
+
+import (
+	"strings"
+
+	"asciigoat.org/core/lexer"
+)
+
+// AcceptQuotedString consumes a quoted string from the source
+// and returns it unquoted and unescaped
+func (p *TextParser) AcceptQuotedString() (string, bool, error) {
+	r, _, err := p.ReadRune()
+	switch {
+	case err != nil:
+		// nothing here
+		return "", false, err
+	case r != RuneQuotes:
+		// not for us
+		p.UnreadRune()
+		return "", false, nil
+	default:
+		// let's roll
+		s, err := lexQuotedString(p)
+		switch {
+		case err != nil:
+			// bad quoted string
+			return "", false, err
+		default:
+			// success
+			return s, true, nil
+		}
+	}
+}
+
+func lexQuotedString(p *TextParser) (string, *lexer.Error) {
+	s, ok, err := lexQuotedStringNoEscape(p)
+	switch {
+	case err != nil:
+		return "", err
+	case ok:
+		return s, nil
+	default:
+		// escape character detected
+		return lexQuotedStringEscaped(p)
+	}
+}
+
+func lexQuotedStringNoEscape(p *TextParser) (string, bool, *lexer.Error) {
+	for {
+		r, _, err := p.ReadRune()
+		switch {
+		case err != nil:
+			// incomplete
+			return "", false, NewErrIncompleteQuotedString(p)
+		case r == RuneQuotes:
+			// end, just remove the quotes
+			s := p.String()
+			l := len(s)
+			return s[1 : l-1], true, nil
+		case r == RuneEscape:
+			// things just got complicated...
+			p.UnreadRune()
+			return "", false, nil
+		case IsNewLine(r):
+			// new lines within quoted values are acceptable
+			p.UnreadRune()
+			p.AcceptNewLine()
+		default:
+			// continue
+		}
+	}
+}
+
+// Unquoted removes quotes the content and unescapes the content
+func lexQuotedStringEscaped(p *TextParser) (string, *lexer.Error) {
+	var result strings.Builder
+
+	// append what was accepted before the escape character
+	_, _ = result.WriteString(p.String()[1:])
+
+	for {
+		r, _, err := p.ReadRune()
+		switch {
+		case err != nil:
+			// incomplete quoted
+			return "", NewErrIncompleteQuotedString(p)
+		case r == RuneQuotes:
+			// end
+			return result.String(), nil
+		case r == RuneEscape:
+			// escaped
+			r2, _, err := p.ReadRune()
+			switch {
+			case err != nil:
+				// incomplete escaped
+				return "", NewErrIncompleteEscaped(p)
+			case IsNewLine(r2):
+				// escaped new line, skip
+				p.UnreadRune()
+				p.AcceptNewLine()
+			default:
+				// TODO: check valid escape character and
+				// append to result
+				s := string([]rune{r, r2})
+				err := NewErrInvalidEscapeSequence(p, s)
+				return "", err
+			}
+		default:
+			// normal, append to result
+			_, _ = result.WriteRune(r)
+		}
+	}
+}
+
+// Unquoted removes quotes and unescapes the content
+func Unquoted(s string) (string, error) {
+	var p TextParser
+	if s == "" {
+		return "", nil
+	}
+
+	p.InitString(s)
+	unquoted, ok, err := p.AcceptQuotedString()
+	switch {
+	case err != nil:
+		// bad string
+		return "", err
+	case ok:
+		// success
+		return unquoted, nil
+	default:
+		// not quoted
+		return s, nil
+	}
+}