decoder [WIP]

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

basic/error.go

@@ -4,19 +4,28 @@ import (
 	"errors"
 
 	"asciigoat.org/core/lexer"
-	"asciigoat.org/ini/parser"
 )
 
 var (
 	errInvalidToken = errors.New("invalid token")
 )
 
+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,
+	}
+}
+
 func newErrInvalidToken(t *token) *lexer.Error {
-	return parser.NewError(t.pos, t.value, "", errInvalidToken)
+	return newError(t.pos, t.value, "", errInvalidToken)
 }
 
 func (dec *decoder) OnError(pos lexer.Position, content string, err error) error {
-	err = parser.NewError(pos, content, "", err)
+	err = 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
 	}
 }

decoder.go

@@ -0,0 +1,83 @@
+package ini
+
+import (
+	"bytes"
+	"io"
+	"strings"
+
+	"asciigoat.org/core"
+	"asciigoat.org/core/reflective"
+	"asciigoat.org/ini/parser"
+)
+
+const (
+	decPrefixLiteral = "ini: Decode"
+)
+
+var (
+	decINIPrefix = decPrefixLiteral[0:3]
+	decDecPrefix = decPrefixLiteral[5:]
+)
+
+// Decoder ...
+type Decoder struct {
+	io.Closer
+
+	out   *reflective.Reflection
+	p     *parser.Parser
+	queue []*token
+}
+
+// Decode ...
+func (dec *Decoder) Decode(v any) error {
+	defer dec.Close()
+
+	r, err := reflective.New(v)
+	switch e := err.(type) {
+	case *reflective.InvalidUnmarshalError:
+		// customize error
+		e.Prefix = decINIPrefix
+		e.Method = decDecPrefix
+	case *reflective.UnmarshalTypeError:
+		// customize error
+		e.Prefix = decPrefixLiteral
+	case nil:
+		// good reflection. Go!
+		dec.out = r
+		err = dec.p.Run()
+	}
+
+	return err
+}
+
+// NewDecoder creates a Decoder using the provided [io.Reader]
+// as source
+func NewDecoder(r io.Reader) *Decoder {
+	rc := core.NewReadCloser(r)
+	switch {
+	case rc == nil:
+		return nil
+	default:
+		dec := &Decoder{
+			p:      parser.NewParser(rc),
+			Closer: rc,
+		}
+
+		// callbacks
+		dec.p.OnToken = dec.parserOnToken
+		dec.p.OnError = dec.parserOnError
+
+		return dec
+	}
+}
+
+// NewDecoderBytes creates a Decoder using the provided bytes array
+// as source
+func NewDecoderBytes(b []byte) *Decoder {
+	return NewDecoder(bytes.NewReader(b))
+}
+
+// NewDecoderString creates a Decoder over a provided string of data
+func NewDecoderString(s string) *Decoder {
+	return NewDecoder(strings.NewReader(s))
+}

decoder_error.go

@@ -0,0 +1,32 @@
+package ini
+
+import (
+	"errors"
+	"log"
+
+	"asciigoat.org/core/lexer"
+)
+
+var (
+	errInvalidToken = errors.New("invalid token")
+)
+
+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,
+	}
+}
+func (*Decoder) newErrInvalidToken(t *token) *lexer.Error {
+	return newError(t.pos, t.value, "", errInvalidToken)
+}
+
+// parserOnError is the callback for lexer errors
+func (*Decoder) parserOnError(pos lexer.Position, content string, err error) error {
+	log.Printf("%s: %s %s: %q: %v", "ini", pos, "error", content, err)
+
+	return newError(pos, content, "", err)
+}

decoder_token.go

@@ -0,0 +1,144 @@
+package ini
+
+import (
+	"fmt"
+	"log"
+
+	"asciigoat.org/core/lexer"
+	"asciigoat.org/ini/parser"
+)
+
+type token struct {
+	pos   lexer.Position
+	typ   parser.TokenType
+	value string
+}
+
+func (t token) String() string {
+	return fmt.Sprintf("%s %s: %q", t.pos, t.typ, t.value)
+}
+
+// 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
+		return "", false
+	case dec.queue[idx].typ != typ:
+		// wrong type
+		return "", false
+	default:
+		// match
+		return dec.queue[idx].value, true
+	}
+}
+
+// queueReset removes all tokens from the queue
+func (dec *Decoder) queueReset() {
+	dec.queue = dec.queue[:0]
+}
+
+// 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
+}
+
+// queueDepth confirms the current depth of the queue
+func (dec *Decoder) queueDepth(depth int) bool {
+	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
+}
+
+// typeOK tells if a token of the specified type is acceptable
+// at this time.
+func (dec *Decoder) typeOK(typ parser.TokenType) bool {
+	switch typ {
+	case parser.TokenSectionStart:
+		return dec.queueDepth(0)
+	case parser.TokenSectionName:
+		return dec.queueDepthType(1, parser.TokenSectionStart)
+	case parser.TokenSectionSubname:
+		return dec.queueDepthType(2, parser.TokenSectionName)
+	case parser.TokenSectionEnd:
+		return dec.queueType(1, parser.TokenSectionName)
+	case parser.TokenFieldKey:
+		return dec.queueDepth(0)
+	case parser.TokenFieldValue:
+		return dec.queueDepthType(1, parser.TokenFieldKey)
+	case parser.TokenComment:
+		panic("unreachable")
+	default:
+		return false
+	}
+}
+
+// execute is called after each acceptable token is appended to the queue
+func (dec *Decoder) execute() error {
+	if l := len(dec.queue); l > 0 {
+		// based on the type of the last element
+		switch dec.queue[l-1].typ {
+		case parser.TokenSectionEnd:
+			name1, _ := dec.queueValue(1, parser.TokenSectionName)
+			name2, ok2 := dec.queueValue(2, parser.TokenSectionSubname)
+			defer dec.queueReset()
+
+			return dec.executeSection(name1, name2, ok2)
+		case parser.TokenFieldValue:
+			key, _ := dec.queueValue(0, parser.TokenFieldKey)
+			value, _ := dec.queueValue(1, parser.TokenFieldValue)
+			defer dec.queueReset()
+
+			return dec.executeField(key, value)
+		}
+	}
+
+	return nil
+}
+
+// revive:disable:flag-parameter
+
+func (*Decoder) executeSection(key, id string, hasID bool) error {
+	// revive:enable:flag-parameter
+
+	if hasID {
+		log.Printf("%s: %s%s[%q]: %q", "ini", "", "section", key, id)
+	} else {
+		log.Printf("%s: %s%s[%q]", "ini", "", "section", key)
+	}
+
+	return nil
+}
+
+func (*Decoder) executeField(key, value string) error {
+	log.Printf("%s: %s%s[%q]: %q", "ini", "  ", "field", key, value)
+	return nil
+}
+
+// parserOnToken is the callback from the parser
+func (dec *Decoder) parserOnToken(pos lexer.Position, typ parser.TokenType, value string) error {
+	var err error
+
+	t := &token{pos, typ, value}
+	switch {
+	case typ == parser.TokenComment:
+		// ignore comments
+	case dec.typeOK(typ):
+		// acceptable token
+		dec.queue = append(dec.queue, t)
+		err = dec.execute()
+	default:
+		// unacceptable
+		err = dec.newErrInvalidToken(t)
+	}
+
+	return err
+}

go.mod

@@ -2,8 +2,13 @@ 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
+	asciigoat.org/core v0.3.7
 	github.com/mgechev/revive v1.3.3
 	golang.org/x/tools v0.12.0
 )

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

@@ -1,88 +0,0 @@
-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

@@ -1,43 +0,0 @@
-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")
-}
-
-func newErrIncomplete(p *TextParser, hint string) *lexer.Error {
-	pos, s := p.Emit()
-	pos.Add(GetPositionalLength(s))
-
-	return NewError(pos, s, hint, fs.ErrInvalid)
-}

parser/lexer_runes.go

@@ -7,13 +7,11 @@ import (
 )
 
 const (
-	RuneComment      = ';'  // RuneComment is the standard INI comment character
+	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 indicates the end of a section declaration
+	RuneSectionEnd   = ']' // RuneSectionEnd indiciates 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 (
@@ -29,8 +27,6 @@ var (
 		RuneSectionStart,
 		RuneSectionEnd,
 		RuneFieldEqual,
-		RuneQuotes,
-		RuneEscape,
 	})
 )
 
@@ -47,11 +43,6 @@ var (
 	IsCommentStart = lexer.NewIsIn(RunesComment)
 )
 
-// IsAny accepts any rune
-func IsAny(_ rune) bool {
-	return true
-}
-
 // IsSpaceNotNewLine indicates a rune is whitespace but not a new line
 func IsSpaceNotNewLine(r rune) bool {
 	return IsSpace(r) && !IsNewLine(r)

parser/parser.go

@@ -31,7 +31,13 @@ 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 NewError(pos, content, "", err)
+	return lexer.Error{
+		Line:   pos.Line,
+		Column: pos.Column,
+
+		Content: content,
+		Err:     err,
+	}
 }
 
 func (p *Parser) setDefaults() {

parser/text.go

@@ -39,10 +39,9 @@ func (p *TextParser) InitString(s string) {
 
 // Discard shadows [lexer.Reader]'s, and takes in consideration
 // new lines on the discarded data when moving the position
-func (p *TextParser) Discard() {
-	s := p.Reader.Emit()
-	l := GetPositionalLength(s)
-	p.pos.Add(l)
+func (*TextParser) Discard() {
+	// TODO: consider new lines
+	panic("not implemented")
 }
 
 // Emit returns the accepted text, its position, and
@@ -50,8 +49,8 @@ func (p *TextParser) Discard() {
 func (p *TextParser) Emit() (lexer.Position, string) {
 	pos := p.pos
 	s := p.Reader.Emit()
-	l := GetPositionalLength(s)
-	p.pos.Add(l)
+	// TODO: consider new lines
+	p.pos.StepN(len(s))
 
 	return pos, s
 }

parser/text_position.go

@@ -1,38 +0,0 @@
-package parser
-
-import (
-	"io"
-
-	"asciigoat.org/core/lexer"
-)
-
-type positionLengthParser struct {
-	TextParser
-
-	lexer.Position
-}
-
-func (p *positionLengthParser) lexStart() (lexer.StateFn, error) {
-	for {
-		switch {
-		case p.AcceptNewLine():
-			p.Position.StepLine()
-		case p.Accept(IsAny):
-			p.Position.StepN(1)
-		default:
-			return nil, io.EOF
-		}
-	}
-}
-
-// GetPositionalLength calculates the [lexer.Position] at
-// the end of a text.
-func GetPositionalLength(s string) lexer.Position {
-	var p positionLengthParser
-	if s == "" {
-		p.InitString(s)
-
-		_ = lexer.Run(p.lexStart)
-	}
-	return p.Position
-}

parser/text_quoted.go

@@ -1,97 +0,0 @@
-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) {
-	for {
-		r, _, err := p.ReadRune()
-		switch {
-		case err != nil:
-			// incomplete
-			return "", NewErrIncompleteQuotedString(p)
-		case r == RuneQuotes:
-			// end, remove quotes and process escaped characters
-			return lexReturnUnescapedQuotedString(p)
-		case r == RuneEscape:
-			// escaped, take another
-			_, _, err := p.ReadRune()
-			if err != nil {
-				// incomplete
-				return "", NewErrIncompleteQuotedString(p)
-			}
-		case IsNewLine(r):
-			// new lines within quoted values are acceptable
-			p.UnreadRune()
-			p.AcceptNewLine()
-		default:
-			// continue
-		}
-	}
-}
-
-func lexReturnUnescapedQuotedString(p *TextParser) (string, *lexer.Error) {
-	// remove quotes
-	s := p.String()
-	l := len(s)
-	s = s[1 : l-1]
-
-	if strings.ContainsRune(s, RuneEscape) {
-		// TODO: implement unescaping
-		err := NewError(p.Position(), s, "escaped characters", lexer.ErrNotImplemented)
-		return "", err
-	}
-
-	return s, nil
-}
-
-// Unquoted removes quotes the content 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
-	}
-}