refactor init phase - definitions

char.go

@@ -32,7 +32,7 @@ func (p *charParser) setCommitType(ct CommitType) {}
 func (p *charParser) validate(*registry) error    { return nil }
 func (p *charParser) init(*registry)              {}
 
-func (p *charParser) setIncludedBy(r *registry, includedBy int) {
+func (p *charParser) setIncludedBy(includedBy int) {
 	if intsContain(p.includedBy, includedBy) {
 		return
 	}
@@ -40,18 +40,8 @@ func (p *charParser) setIncludedBy(r *registry, includedBy int) {
 	p.includedBy = append(p.includedBy, includedBy)
 }
 
-func (p *charParser) parser(r *registry) parser {
+func (p *charParser) parser() parser   { return p }
-	if _, ok := r.parser(p.name); ok {
+func (p *charParser) builder() builder { return p }
-		return p
-	}
-
-	r.setParser(p)
-	return p
-}
-
-func (p *charParser) builder() builder {
-	return p
-}
 
 func matchChars(chars []rune, ranges [][]rune, not bool, char rune) bool {
 	for _, ci := range chars {

choice.go

@@ -5,8 +5,10 @@ type choiceDefinition struct {
 	id          int
 	commit      CommitType
 	elements    []string
+	elementDefs []definition
 	includedBy  []int
 	cbuilder    *choiceBuilder
+	cparser     *choiceParser
 	validated   bool
 	initialized bool
 }
@@ -48,7 +50,6 @@ func (d *choiceDefinition) validate(r *registry) error {
 	}
 
 	d.validated = true
-
 	for i := range d.elements {
 		e, ok := r.definitions[d.elements[i]]
 		if !ok {
@@ -63,101 +64,79 @@ func (d *choiceDefinition) validate(r *registry) error {
 	return nil
 }
 
+func (d *choiceDefinition) ensureBuilder() {
+	if d.cbuilder != nil {
+		return
+	}
+
+	d.cbuilder = &choiceBuilder{
+		name:       d.name,
+		id:         d.id,
+		commit:     d.commit,
+		includedBy: &idSet{},
+	}
+}
+
+func (d *choiceDefinition) initElements(r *registry) {
+	for _, e := range d.elements {
+		def := r.definitions[e]
+		d.elementDefs = append(d.elementDefs, def)
+		def.init(r)
+		d.cbuilder.elements = append(d.cbuilder.elements, def.builder())
+		def.setIncludedBy(d.id)
+	}
+}
+
 func (d *choiceDefinition) init(r *registry) {
 	if d.initialized {
 		return
 	}
 
 	d.initialized = true
-
+	d.ensureBuilder()
-	if d.cbuilder == nil {
+	d.initElements(r)
-		d.cbuilder = &choiceBuilder{
-			name:       d.name,
-			id:         d.id,
-			commit:     d.commit,
-			includedBy: &idSet{},
-		}
-	}
-
-	for _, e := range d.elements {
-		def := r.definitions[e]
-		d.cbuilder.elements = append(d.cbuilder.elements, def.builder())
-		def.init(r)
-		def.setIncludedBy(r, d.id)
-	}
 }
 
-func (d *choiceDefinition) setIncludedBy(r *registry, includedBy int) {
+func (d *choiceDefinition) setIncludedBy(includedBy int) {
 	if intsContain(d.includedBy, includedBy) {
 		return
 	}
 
 	d.includedBy = append(d.includedBy, includedBy)
-
+	d.ensureBuilder()
-	if d.cbuilder == nil {
-		d.cbuilder = &choiceBuilder{
-			name:       d.name,
-			id:         d.id,
-			commit:     d.commit,
-			includedBy: &idSet{},
-		}
-	}
-
 	d.cbuilder.includedBy.set(includedBy)
-
+	for _, e := range d.elementDefs {
-	for _, e := range d.elements {
+		e.setIncludedBy(includedBy)
-		r.definitions[e].setIncludedBy(r, includedBy)
 	}
 }
 
-// TODO:
+func (d *choiceDefinition) createParser() {
-// - it may be possible to initialize the parsers non-recursively
+	d.cparser = &choiceParser{
-// - maybe the whole definition, parser and builder can be united
-
-func (d *choiceDefinition) parser(r *registry) parser {
-	p, ok := r.parser(d.name)
-	if ok {
-		return p
-	}
-
-	cp := &choiceParser{
 		name:       d.name,
 		id:         d.id,
 		commit:     d.commit,
 		includedBy: d.includedBy,
 	}
-
-	r.setParser(cp)
-
-	var elements []parser
-	for _, e := range d.elements {
-		element, ok := r.parser(e)
-		if ok {
-			elements = append(elements, element)
-			continue
-		}
-
-		element = r.definitions[e].parser(r)
-		elements = append(elements, element)
-	}
-
-	cp.elements = elements
-	return cp
 }
 
-func (d *choiceDefinition) builder() builder {
+func (d *choiceDefinition) createElementParsers() {
-	if d.cbuilder == nil {
+	for _, def := range d.elementDefs {
-		d.cbuilder = &choiceBuilder{
+		element := def.parser()
-			name:       d.name,
+		d.cparser.elements = append(d.cparser.elements, element)
-			id:         d.id,
-			commit:     d.commit,
-			includedBy: &idSet{},
 	}
-	}
-
-	return d.cbuilder
 }
 
+func (d *choiceDefinition) parser() parser {
+	if d.cparser != nil {
+		return d.cparser
+	}
+
+	d.createParser()
+	d.createElementParsers()
+	return d.cparser
+}
+
+func (d *choiceDefinition) builder() builder { return d.cbuilder }
 func (p *choiceParser) nodeName() string     { return p.name }
 func (p *choiceParser) nodeID() int          { return p.id }
 

notes.txt

@@ -22,6 +22,8 @@ code generation go:
 - char matches can be generated into switches
 code generation js
 documentation flag
+support custom tokenization
+streaming
 
 [problems]
 can the root be an alias? check the commit mechanism

parse.go

@@ -1,74 +0,0 @@
-package treerack
-
-import "fmt"
-
-type definition interface {
-	nodeName() string
-	setNodeName(string)
-	nodeID() int
-	commitType() CommitType
-	setCommitType(CommitType)
-	setID(int)
-	validate(*registry) error
-	init(*registry)
-	setIncludedBy(*registry, int)
-	parser(*registry) parser
-	builder() builder
-}
-
-type parser interface {
-	nodeName() string
-	nodeID() int
-	parse(*context)
-}
-
-type builder interface {
-	nodeName() string
-	nodeID() int
-	build(*context) ([]*Node, bool)
-}
-
-func parserNotFound(name string) error {
-	return fmt.Errorf("parser not found: %s", name)
-}
-
-func cannotIncludeParsers(name string) error {
-	return fmt.Errorf("parser: %s cannot include other parsers", name)
-}
-
-func intsContain(is []int, i int) bool {
-	for _, ii := range is {
-		if ii == i {
-			return true
-		}
-	}
-
-	return false
-}
-
-func parse(p parser, c *context) error {
-	p.parse(c)
-	if c.readErr != nil {
-		return c.readErr
-	}
-
-	if !c.match {
-		return ErrInvalidInput
-	}
-
-	if err := c.finalize(p); err != nil {
-		return err
-	}
-
-	return nil
-}
-
-func build(b builder, c *context) *Node {
-	c.offset = 0
-	n, ok := b.build(c)
-	if !ok || len(n) != 1 {
-		panic("damaged parse result")
-	}
-
-	return n[0]
-}

registry.go

@@ -28,11 +28,6 @@ func (r *registry) definition(name string) (definition, bool) {
 	return d, ok
 }
 
-func (r *registry) parser(name string) (parser, bool) {
-	p, ok := r.parsers[name]
-	return p, ok
-}
-
 func (r *registry) setDefinition(d definition) error {
 	if _, ok := r.definitions[d.nodeName()]; ok {
 		return duplicateDefinition(d.nodeName())
@@ -48,10 +43,6 @@ func (r *registry) setDefinition(d definition) error {
 	return nil
 }
 
-func (r *registry) setParser(p parser) {
-	r.parsers[p.nodeName()] = p
-}
-
 func (r *registry) getDefinitions() []definition {
 	var defs []definition
 	for _, def := range r.definitions {

sequence.go

@@ -5,9 +5,11 @@ type sequenceDefinition struct {
 	id          int
 	commit      CommitType
 	items       []SequenceItem
+	itemDefs    []definition
 	includedBy  []int
 	ranges      [][]int
 	sbuilder    *sequenceBuilder
+	sparser     *sequenceParser
 	allChars    bool
 	validated   bool
 	initialized bool
@@ -54,7 +56,6 @@ func (d *sequenceDefinition) validate(r *registry) error {
 	}
 
 	d.validated = true
-
 	for i := range d.items {
 		ii, ok := r.definition(d.items[i].Name)
 		if !ok {
@@ -70,34 +71,23 @@ func (d *sequenceDefinition) validate(r *registry) error {
 }
 
 func (d *sequenceDefinition) includeItems() bool {
-	return len(d.items) == 1 && d.items[0].Min == 1 && d.items[0].Max == 1
+	return len(d.items) == 1 && d.items[0].Max == 1
 }
 
-func (d *sequenceDefinition) init(r *registry) {
+func (d *sequenceDefinition) ensureBuilder() {
-	if d.initialized {
+	if d.sbuilder != nil {
 		return
 	}
 
-	d.initialized = true
-
-	for i := range d.items {
-		if d.items[i].Min == 0 && d.items[i].Max == 0 {
-			d.items[i].Min, d.items[i].Max = 1, 1
-		} else if d.items[i].Max == 0 {
-			d.items[i].Max = -1
-		}
-	}
-
-	if d.sbuilder == nil {
 	d.sbuilder = &sequenceBuilder{
 		name:       d.name,
 		id:         d.id,
 		commit:     d.commit,
 		includedBy: &idSet{},
 	}
-	}
+}
 
-	allChars := true
+func (d *sequenceDefinition) initRanges() {
 	for _, item := range d.items {
 		if item.Min == 0 && item.Max == 0 {
 			item.Min, item.Max = 1, 1
@@ -106,101 +96,84 @@ func (d *sequenceDefinition) init(r *registry) {
 		}
 
 		d.ranges = append(d.ranges, []int{item.Min, item.Max})
+	}
+}
 
+func (d *sequenceDefinition) initItems(r *registry) {
+	allChars := true
+	for _, item := range d.items {
 		def := r.definitions[item.Name]
+		d.itemDefs = append(d.itemDefs, def)
+		def.init(r)
 		d.sbuilder.items = append(d.sbuilder.items, def.builder())
-
 		if allChars {
 			if _, isChar := def.(*charParser); !isChar {
 				allChars = false
 			}
 		}
-
-		def.init(r)
 	}
 
-	d.sbuilder.ranges = d.ranges
 	d.sbuilder.allChars = allChars
 	d.allChars = allChars
+}
 
-	if !d.includeItems() {
+func (d *sequenceDefinition) init(r *registry) {
+	if d.initialized {
 		return
 	}
 
-	r.definitions[d.items[0].Name].setIncludedBy(r, d.id)
+	d.initialized = true
+	d.initRanges()
+	d.ensureBuilder()
+	d.sbuilder.ranges = d.ranges
+	d.initItems(r)
+	if d.includeItems() {
+		d.itemDefs[0].setIncludedBy(d.id)
+	}
 }
 
-func (d *sequenceDefinition) setIncludedBy(r *registry, includedBy int) {
+func (d *sequenceDefinition) setIncludedBy(includedBy int) {
 	if intsContain(d.includedBy, includedBy) {
 		return
 	}
 
 	d.includedBy = append(d.includedBy, includedBy)
-
+	d.ensureBuilder()
-	if d.sbuilder == nil {
-		d.sbuilder = &sequenceBuilder{
-			name:       d.name,
-			id:         d.id,
-			commit:     d.commit,
-			includedBy: &idSet{},
-		}
-	}
-
 	d.sbuilder.includedBy.set(includedBy)
-
+	if d.includeItems() {
-	if !d.includeItems() {
+		d.itemDefs[0].setIncludedBy(includedBy)
-		return
 	}
-
-	r.definitions[d.items[0].Name].setIncludedBy(r, includedBy)
 }
 
-func (d *sequenceDefinition) parser(r *registry) parser {
+func (d *sequenceDefinition) createParser() {
-	p, ok := r.parser(d.name)
+	d.sparser = &sequenceParser{
-	if ok {
-		return p
-	}
-
-	sp := &sequenceParser{
 		name:       d.name,
 		id:         d.id,
 		commit:     d.commit,
 		includedBy: d.includedBy,
 		allChars:   d.allChars,
+		ranges:     d.ranges,
 	}
-
-	r.setParser(sp)
-
-	var items []parser
-	for _, item := range d.items {
-		pi, ok := r.parser(item.Name)
-		if ok {
-			items = append(items, pi)
-			continue
-		}
-
-		pi = r.definitions[item.Name].parser(r)
-		items = append(items, pi)
-	}
-
-	sp.items = items
-	sp.ranges = d.ranges
-	return sp
 }
 
-func (d *sequenceDefinition) builder() builder {
+func (d *sequenceDefinition) createItemParsers() {
-	if d.sbuilder == nil {
+	for _, item := range d.itemDefs {
-		d.sbuilder = &sequenceBuilder{
+		pi := item.parser()
-			name:       d.name,
+		d.sparser.items = append(d.sparser.items, pi)
-			id:         d.id,
-			commit:     d.commit,
-			includedBy: &idSet{},
 	}
-	}
-
-	return d.sbuilder
 }
 
+func (d *sequenceDefinition) parser() parser {
+	if d.sparser != nil {
+		return d.sparser
+	}
+
+	d.createParser()
+	d.createItemParsers()
+	return d.sparser
+}
+
+func (d *sequenceDefinition) builder() builder { return d.sbuilder }
 func (p *sequenceParser) nodeName() string     { return p.name }
 func (p *sequenceParser) nodeID() int          { return p.id }
 

syntax.go

@@ -33,6 +33,32 @@ type Syntax struct {
 	builder      builder
 }
 
+type definition interface {
+	nodeName() string
+	setNodeName(string)
+	nodeID() int
+	commitType() CommitType
+	setCommitType(CommitType)
+	setID(int)
+	validate(*registry) error
+	init(*registry)
+	setIncludedBy(int)
+	parser() parser
+	builder() builder
+}
+
+type parser interface {
+	nodeName() string
+	nodeID() int
+	parse(*context)
+}
+
+type builder interface {
+	nodeName() string
+	nodeID() int
+	build(*context) ([]*Node, bool)
+}
+
 var (
 	ErrSyntaxInitialized       = errors.New("syntax initialized")
 	ErrInitFailed              = errors.New("init failed")
@@ -49,6 +75,14 @@ var (
 	ErrInvalidSymbolName       = errors.New("invalid symbol name")
 )
 
+func duplicateDefinition(name string) error {
+	return fmt.Errorf("duplicate definition: %s", name)
+}
+
+func parserNotFound(name string) error {
+	return fmt.Errorf("parser not found: %s", name)
+}
+
 const symbolChars = "^\\\\ \\n\\t\\b\\f\\r\\v/.\\[\\]\\\"{}\\^+*?|():=;"
 
 func parseSymbolChars(c []rune) []rune {
@@ -62,10 +96,6 @@ func parseSymbolChars(c []rune) []rune {
 
 var symbolCharRunes = parseSymbolChars([]rune(symbolChars))
 
-func duplicateDefinition(name string) error {
-	return fmt.Errorf("duplicate definition: %s", name)
-}
-
 func isValidSymbol(n string) bool {
 	runes := []rune(n)
 	for _, r := range runes {
@@ -78,6 +108,16 @@ func isValidSymbol(n string) bool {
 
 }
 
+func intsContain(is []int, i int) bool {
+	for _, ii := range is {
+		if ii == i {
+			return true
+		}
+	}
+
+	return false
+}
+
 func (s *Syntax) applyRoot(d definition) error {
 	explicitRoot := d.commitType()&Root != 0
 	if explicitRoot && s.explicitRoot {
@@ -247,7 +287,7 @@ func (s *Syntax) Init() error {
 	}
 
 	s.root.init(s.registry)
-	s.parser = s.root.parser(s.registry)
+	s.parser = s.root.parser()
 	s.builder = s.root.builder()
 
 	s.initialized = true
@@ -262,17 +302,30 @@ func (s *Syntax) Generate(w io.Writer) error {
 	return ErrNotImplemented
 }
 
-// TODO: optimize top sequences to save memory, or just support streaming, or combine the two
-
 func (s *Syntax) Parse(r io.Reader) (*Node, error) {
 	if err := s.Init(); err != nil {
 		return nil, err
 	}
 
 	c := newContext(bufio.NewReader(r))
-	if err := parse(s.parser, c); err != nil {
+	s.parser.parse(c)
+	if c.readErr != nil {
+		return nil, c.readErr
+	}
+
+	if !c.match {
+		return nil, ErrInvalidInput
+	}
+
+	if err := c.finalize(s.parser); err != nil {
 		return nil, err
 	}
 
-	return build(s.builder, c), nil
+	c.offset = 0
+	n, ok := s.builder.build(c)
+	if !ok || len(n) != 1 {
+		panic("damaged parse result")
+	}
+
+	return n[0], nil
 }