diff --git a/Makefile b/Makefile index 0183be9..f8e2e90 100644 --- a/Makefile +++ b/Makefile @@ -129,8 +129,7 @@ cover-cmd: .coverprofile-cmd showcover-cmd: .coverprofile-cmd go tool cover -html .coverprofile-cmd -.PHONY: cpu.out -cpu.out: +cpu.out: $(sources) go test -v -run TestMMLFile -cpuprofile cpu.out cpu: cpu.out diff --git a/chardefine.go b/chardefine.go index d31fee2..6a7df8c 100644 --- a/chardefine.go +++ b/chardefine.go @@ -8,6 +8,7 @@ import ( const ( charClassEscape = '\\' charClassBanned = "\\[]^-\b\f\n\r\t\v" + charSeqBanned = "\\\"\b\f\n\r\t\v" ) func newChar( @@ -44,11 +45,15 @@ func (p *charParser) isSingleChar() bool { return !p.not && len(p.chars) == 1 && len(p.ranges) == 0 } -func (p *charParser) format(_ *registry, f formatFlags) string { +func (p *charParser) format(_ *registry, _ formatOptions) string { if p.not && len(p.chars) == 0 && len(p.ranges) == 0 { return "." } + if p.isSingleChar() { + return "\"" + string(escape(charClassEscape, []rune(charSeqBanned), p.chars)) + "\"" + } + esc := func(c ...rune) []rune { return escape(charClassEscape, []rune(charClassBanned), c) } diff --git a/choicedefine.go b/choicedefine.go index feb633f..ddea381 100644 --- a/choicedefine.go +++ b/choicedefine.go @@ -134,24 +134,45 @@ func (d *choiceDefinition) builder() builder { return d.cbuilder } -func (d *choiceDefinition) format(r *registry, f formatFlags) string { +func (d *choiceDefinition) formatChoice(r *registry, o formatOptions) string { var chars []rune + sep := []rune{' ', '|', ' '} + if o.mode&formatPretty > 0 { + sep = []rune{'\n', '|', ' '} + } + for i := range d.options { if i > 0 { - chars = append(chars, []rune(" | ")...) + chars = append(chars, sep...) } optionDef := r.definition[d.options[i]] if optionDef.commitType()&userDefined != 0 { chars = append(chars, []rune(optionDef.nodeName())...) } else { - chars = append(chars, []rune(optionDef.format(r, f))...) + chars = append(chars, []rune(optionDef.format(r, o))...) } } return string(chars) } +func (d *choiceDefinition) format(r *registry, o formatOptions) string { + oo := o + oo.mode &^= formatPretty + f := d.formatChoice(r, oo) + if o.mode&formatPretty == 0 { + return string(f) + } + + if len(f) <= o.targetWidth { + return string(f) + } + + f = d.formatChoice(r, o) + return string(f) +} + func (p *choiceParser) generate(w io.Writer, done map[string]bool) error { if done[p.name] { return nil diff --git a/define.go b/define.go index 6bc8600..354e8b0 100644 --- a/define.go +++ b/define.go @@ -204,7 +204,6 @@ func addDefinition(s *Syntax, n *Node) error { func define(s *Syntax, syntaxTree *Node) error { syntaxTree = dropComments(syntaxTree) - for _, n := range syntaxTree.Nodes { if err := addDefinition(s, n); err != nil { return err diff --git a/format.go b/format.go new file mode 100644 index 0000000..74e4e7c --- /dev/null +++ b/format.go @@ -0,0 +1,638 @@ +package treerack + +import ( + "bytes" + "fmt" + "io" + "strings" + "unicode" +) + +const initialTargetWidth = 112 + +type commentFormat int + +const ( + commentFormatNone commentFormat = iota + standaloneComment + headerComment + suffixComment + inlineComment +) + +type formatItem struct { + commentFormat commentFormat + node *Node +} + +type formatGroup struct { + items []formatItem +} + +func topLevelCommentFormat(ast *Node, i int, n *Node) commentFormat { + if n.Name != "comment" { + return commentFormatNone + } + + if i > 0 && + ast.Nodes[i-1].Name != "comment" && + !strings.Contains(string(ast.tokens[ast.Nodes[i-1].To:n.From]), "\n") { + return suffixComment + } + + if len(ast.Nodes) == i+1 { + return standaloneComment + } + + next := ast.Nodes[i+1] + if next.Name == "comment" { + return standaloneComment + } + + var lines int + space := ast.tokens[n.To:next.From] + for _, s := range space { + if s == '\n' { + lines++ + } + } + + if lines >= 2 { + return standaloneComment + } + + return headerComment +} + +func groupASTByComments(ast *Node) []formatGroup { + var ( + groups []formatGroup + currentGroup formatGroup + ) + + for i, n := range ast.Nodes { + last := len(currentGroup.items) - 1 + cf := topLevelCommentFormat(ast, i, n) + item := formatItem{ + commentFormat: cf, + node: n, + } + + if cf == commentFormatNone { + if last >= 0 && currentGroup.items[last].commentFormat == standaloneComment { + groups = append(groups, currentGroup) + currentGroup.items = nil + } + + currentGroup.items = append(currentGroup.items, item) + continue + } + + if cf == suffixComment { + currentGroup.items = append(currentGroup.items, item) + continue + } + + if last >= 0 { + groups = append(groups, currentGroup) + } + + currentGroup.items = []formatItem{item} + } + + groups = append(groups, currentGroup) + return groups +} + +func trimComment(text string) string { + var inBlockComment, inLineComment bool + tr := []rune(text) + rr := make([]rune, 0, len(tr)) + for i := 0; i < len(tr); i++ { + r := tr[i] + if inBlockComment { + if r != '*' || len(tr) <= i+1 || tr[i+1] != '/' { + rr = append(rr, r) + continue + } + + rr = append(rr, '*', '/') + inBlockComment = false + if len(tr) > i+2 && !unicode.IsSpace(tr[i+2]) { + rr = append(rr, ' ') + } + + i++ + continue + } + + if inLineComment { + rr = append(rr, r) + inLineComment = r != '\n' + continue + } + + if r == '/' && len(tr) > i+1 && tr[i+1] == '*' { + rr = append(rr, '/', '*') + inBlockComment = true + i++ + continue + } + + if r == '/' && len(tr) > i+1 && tr[i+1] == '/' { + rr = append(rr, '/', '/') + inLineComment = true + if len(tr) > i+2 && tr[i+2] != ' ' { + rr = append(rr, ' ') + } + + i++ + continue + } + + if r == '\n' || len(rr) > 0 && !unicode.IsSpace(rr[len(rr)-1]) { + rr = append(rr, r) + } + } + + lines := strings.Split(string(rr), "\n") + for i := range lines { + lines[i] = strings.TrimRightFunc(lines[i], unicode.IsSpace) + } + + return strings.Join(lines, "\n") +} + +func formatComment(out io.Writer, n *Node) error { + text := n.Text() + text = trimComment(text) + _, err := fmt.Fprint(out, text) + return err +} + +func formatDefinitionName(item formatItem) string { + name := item.node.Nodes[0].Text() + flags := make([]string, 0, len(item.node.Nodes)-2) + for i := 1; i < len(item.node.Nodes)-1; i++ { + flags = append(flags, item.node.Nodes[i].Name) + } + + if len(flags) > 0 { + name += ":" + strings.Join(flags, ":") + } + + return name +} + +func formatItemNames(g formatGroup) ([]string, int) { + var maxWidth int + ordered := make([]string, 0, len(g.items)) + for _, item := range g.items { + if item.commentFormat != commentFormatNone { + ordered = append(ordered, "") + continue + } + + name := formatDefinitionName(item) + maxWidth = max(maxWidth, len(name)) + ordered = append(ordered, name) + } + + return ordered, maxWidth +} + +func formatAnyChar(out io.Writer) error { + _, err := fmt.Fprint(out, ".") + return err +} + +func formatCharClass(out io.Writer, n *Node) error { + _, err := fmt.Fprint(out, n.Text()) + return err +} + +func formatCharSequence(out io.Writer, n *Node) error { + _, err := fmt.Fprint(out, n.Text()) + return err +} + +func formatSymbol(out io.Writer, n *Node) error { + _, err := fmt.Fprint(out, n.Text()) + return err +} + +func decTargetWidth(w, by int) int { + if w <= 0 { + return w + } + + w -= by + if w < 0 { + w = 0 + } + + return w +} + +func formatSequenceItemNode(out io.Writer, targetWidth int, n *Node) error { + var ( + min, max int + err error + ) + + fprint := func(a ...any) { + if err != nil { + return + } + + _, err = fmt.Fprint(out, a...) + } + + if len(n.Nodes) == 2 { + if min, max, err = getQuantity(n.Nodes[1]); err != nil { + return err + } + } + + min, max = normalizeItemRange(min, max) + needsQuantifier := min != 1 || max != 1 + isChoice := n.Nodes[0].Name == "choice" + isChoiceOfMultiple := isChoice && len(n.Nodes[0].Nodes) > 1 + isSequence := n.Nodes[0].Name == "sequence" + isSequenceOfMultiple := isSequence && len(n.Nodes[0].Nodes) > 1 + needsGrouping := isChoiceOfMultiple || isSequenceOfMultiple + if needsGrouping { + var buf bytes.Buffer + targetWidth = decTargetWidth(targetWidth, 2) + if err := formatExpression(&buf, targetWidth, n.Nodes[0]); err != nil { + return err + } + + multiline := strings.Contains(buf.String(), "\n") + if multiline { + lines := strings.Split(buf.String(), "\n") + fprint("( ") + fprint(lines[0]) + for _, l := range lines[1:] { + fprint("\n ") + fprint(l) + } + + fprint("\n )") + } else { + fprint("(") + if _, err := io.Copy(out, &buf); err != nil { + return err + } + + fprint(")") + } + } else { + if err := formatExpression(out, targetWidth, n.Nodes[0]); err != nil { + return err + } + } + + if !needsQuantifier { + return nil + } + + if min == 0 && max == 1 { + fprint("?") + return err + } + + if min == 0 && max < 0 { + fprint("*") + return err + } + + if min == 1 && max < 0 { + fprint("+") + return err + } + + fprint("{") + if min == max { + fprint(min) + } else { + if min > 0 { + fprint(min) + } + + fprint(",") + if max >= 0 { + fprint(max) + } + } + + return err +} + +func formatSequenceItemNodes(out io.Writer, targetWidth int, n []*Node) error { + sep := " " + if targetWidth >= 0 { + sep = "\n " + } + + for i, ni := range n { + if i > 0 { + if _, err := fmt.Fprint(out, sep); err != nil { + return err + } + } + + if ni.Name == "comment" { + if err := formatComment(out, ni); err != nil { + return err + } + + continue + } + + if err := formatSequenceItemNode(out, targetWidth, ni); err != nil { + return err + } + } + + return nil +} + +func formatSequence(out io.Writer, targetWidth int, n []*Node) error { + var buf bytes.Buffer + if err := formatSequenceItemNodes(&buf, -1, n); err != nil { + return err + } + + if targetWidth >= 0 && buf.Len() > targetWidth { + (&buf).Reset() + if err := formatSequenceItemNodes(&buf, targetWidth, n); err != nil { + return err + } + } + + _, err := io.Copy(out, &buf) + return err +} + +func formatChoiceOptionNodes(out io.Writer, targetWidth int, n []*Node) error { + sep, commentSep := " | ", " " + if targetWidth >= 0 { + sep, commentSep = "\n| ", "\n" + } + + for i, ni := range n { + if ni.Name == "comment" { + if i > 0 { + if _, err := fmt.Fprint(out, commentSep); err != nil { + return err + } + } + + if err := formatComment(out, ni); err != nil { + return err + } + + continue + } + + if i > 0 { + if _, err := fmt.Fprint(out, sep); err != nil { + return err + } + } + + if err := formatExpression(out, targetWidth, ni); err != nil { + return err + } + } + + return nil +} + +func formatChoice(out io.Writer, targetWidth int, n []*Node) error { + var buf bytes.Buffer + if err := formatChoiceOptionNodes(&buf, -1, n); err != nil { + return err + } + + if targetWidth >= 0 && buf.Len() > targetWidth { + (&buf).Reset() + if err := formatChoiceOptionNodes(&buf, targetWidth, n); err != nil { + return err + } + } + + _, err := io.Copy(out, &buf) + return err +} + +func formatExpression(out io.Writer, targetWidth int, n *Node) error { + var err error + switch n.Name { + case "comment": + err = formatComment(out, n) + case "any-char": + err = formatAnyChar(out) + case "char-class": + err = formatCharClass(out, n) + case "char-sequence": + err = formatCharSequence(out, n) + case "symbol": + err = formatSymbol(out, n) + case "sequence": + err = formatSequence(out, targetWidth, n.Nodes) + case "choice": + err = formatChoice(out, targetWidth, n.Nodes) + } + + return err +} + +func formatDefinition(out io.Writer, targetWidth, namesWidth int, pad, name string, n *Node) error { + if _, err := fmt.Fprintf(out, "%s%s = ", name, pad[:namesWidth-len(name)]); err != nil { + return err + } + + var buf bytes.Buffer + targetWidth = decTargetWidth(targetWidth, namesWidth+3) + if err := formatExpression(&buf, targetWidth, n.Nodes[len(n.Nodes)-1]); err != nil { + return err + } + + var err error + fprint := func(a ...any) { + if err != nil { + return + } + + _, err = fmt.Fprint(out, a...) + } + + lines := strings.Split(buf.String(), "\n") + fprint(lines[0]) + for _, l := range lines[1:] { + fprint("\n ") + fprint(pad) + fprint(l) + } + + fprint(";") + return err +} + +func formatASTGroup(out io.Writer, g formatGroup) error { + if g.items[0].commentFormat == standaloneComment { + return formatComment(out, g.items[0].node) + } + + hasHeaderComment := g.items[0].commentFormat == headerComment + if hasHeaderComment { + if err := formatComment(out, g.items[0].node); err != nil { + return err + } + + g.items = g.items[1:] + } + + names, namesWidth := formatItemNames(g) + pad := strings.Join(make([]string, namesWidth+1), " ") + for i, item := range g.items { + name := names[i] + if item.commentFormat == suffixComment { + if _, err := fmt.Fprint(out, " "); err != nil { + return err + } + + formatComment(out, item.node) + continue + } + + if hasHeaderComment || i > 0 { + if _, err := fmt.Fprintln(out); err != nil { + return err + } + } + + if err := formatDefinition( + out, + initialTargetWidth, + namesWidth, + pad, + name, + item.node, + ); err != nil { + return err + } + } + + return nil +} + +func formatAST(out io.Writer, ast *Node) error { + // drop whitespace comments + // use line comments by default + // comment types: + // - standalone comment + // - header comment + // - suffix comment + // - inline comment + // + // standalone comment: + // - preceeded by definition or at least two empty lines and followed by at least two empty lines + // - separate it by two empty lines above and below + // + // header comment: + // - separated from the subsequent definition by zero or one empty lines + // - separate it by two empty lines above and one empty line below + // + // suffix comment: + // - starts on the same line as the definition it belongs to + // - append to the definition + // - if it consists of multiple lines, append a new line below + // + // inline comment: + // - it's inside a definition + // - if it's before the eq sign, discard name padding and use block comment + // - if it's in an expression, and falls on its own line, and fits on the previous line, put it there + // - if it's in an expression, and falls on its own line, use a line comment + // - if it's in an expression, and it's followed by non-comment on the same line, use block comment + // - if it consists of multiple lines, append a new line below the definition + + groups := groupASTByComments(ast) + for i, g := range groups { + if i > 0 { + if _, err := fmt.Fprint(out, "\n\n"); err != nil { + return err + } + } + + if err := formatASTGroup(out, g); err != nil { + return err + } + } + + return nil +} + +func formatDefinitions(out io.Writer, s *Syntax) error { + var o formatOptions + o.mode = formatPretty + o.targetWidth = initialTargetWidth + + var ( + namesWidth int + orderedDefs []string + ) + + defs := make(map[string]definition) + for _, def := range s.registry.definitions { + if def.commitType()&userDefined == 0 { + continue + } + + defName := def.nodeName() + ct := def.commitType() + ct &^= userDefined + if sq, ok := def.(*sequenceDefinition); ok && sq.isCharSequence(s.registry) { + ct &^= NoWhitespace + } + + if ct != None { + defName = fmt.Sprintf("%s:%v", defName, ct) + } + + orderedDefs = append(orderedDefs, defName) + defs[defName] = def + namesWidth = max(namesWidth, len([]rune(defName))) + } + + o.targetWidth = decTargetWidth(o.targetWidth, namesWidth+3) + pad := strings.Join(make([]string, namesWidth+1), " ") + for _, name := range orderedDefs { + def := defs[name] + f := def.format(s.registry, o) + lines := strings.Split(f, "\n") + if _, err := fmt.Fprintf( + out, + "%s%s = %s", + name, + pad[:namesWidth-len(name)], + lines[0], + ); err != nil { + return err + } + + for _, l := range lines[1:] { + if _, err := fmt.Fprintf(out, "\n%s %s", pad, l); err != nil { + return err + } + } + + if _, err := fmt.Fprint(out, ";\n"); err != nil { + return err + } + } + + return nil +} diff --git a/format_test.go b/format_test.go index d8925a8..430bf32 100644 --- a/format_test.go +++ b/format_test.go @@ -35,7 +35,7 @@ func testDefinitionFormatItem(t *testing.T, treerack *Syntax, f formatFlags, tes t.Fatal("failed to register definition") } - output := def.format(s.registry, f) + output := def.format(s.registry, formatOptions{mode: f}) if output != test.output { t.Error("invalid definition format") t.Log("got: ", output) @@ -63,23 +63,23 @@ func TestCharFormat(t *testing.T) { }, { title: "one char", definition: "[a]", - output: "[a]", + output: `"a"`, }, { title: "escaped char", definition: "[\\a]", - output: "[a]", + output: `"a"`, }, { title: "escaped control char", definition: "[\\^]", - output: "[\\^]", + output: `"^"`, }, { title: "escaped whitespace char", definition: "[\\n]", - output: "[\\n]", + output: `"\n"`, }, { title: "escaped verbatim whitespace char", definition: "[\n]", - output: "[\\n]", + output: `"\n"`, }, { title: "escaped range", definition: "[\\b-\\v]", @@ -123,7 +123,7 @@ func TestSequenceFormat(t *testing.T) { }, { title: "quantifiers, 0-or-more, single char", syntax: `def = "a"*`, - output: `[a]*`, + output: `"a"*`, }, { title: "quantifiers, 0-or-more", syntax: `def = "abc"*`, @@ -131,7 +131,7 @@ func TestSequenceFormat(t *testing.T) { }, { title: "quantifiers, 1-or-more, single char", syntax: `def = "a"+`, - output: `[a]+`, + output: `"a"+`, }, { title: "quantifiers, 1-or-more", syntax: `def = "abc"+`, @@ -139,7 +139,7 @@ func TestSequenceFormat(t *testing.T) { }, { title: "quantifiers, 0-or-one, single char", syntax: `def = "a"?`, - output: `[a]?`, + output: `"a"?`, }, { title: "quantifiers, 0-or-one", syntax: `def = "abc"?`, @@ -147,7 +147,7 @@ func TestSequenceFormat(t *testing.T) { }, { title: "quantifiers, exact number, single char", syntax: `def = "a"{3}`, - output: `[a]{3}`, + output: `"a"{3}`, }, { title: "quantifiers, exact number", syntax: `def = "abc"{3}`, @@ -155,7 +155,7 @@ func TestSequenceFormat(t *testing.T) { }, { title: "quantifiers, max, single char", syntax: `def = "a"{0, 3}`, - output: `[a]{,3}`, + output: `"a"{,3}`, }, { title: "quantifiers, max", syntax: `def = "abc"{0, 3}`, @@ -163,7 +163,7 @@ func TestSequenceFormat(t *testing.T) { }, { title: "quantifiers, min, single char", syntax: `def = "a"{3,}`, - output: `[a]{3,}`, + output: `"a"{3,}`, }, { title: "quantifiers, min", syntax: `def = "abc"{3,}`, @@ -171,7 +171,7 @@ func TestSequenceFormat(t *testing.T) { }, { title: "quantifiers, range, single char", syntax: `def = "a"{3, 9}`, - output: `[a]{3,9}`, + output: `"a"{3,9}`, }, { title: "quantifiers, range", syntax: `def = "abc"{3, 9}`, @@ -183,7 +183,7 @@ func TestSequenceFormat(t *testing.T) { }, { title: "choice in sequence, single char", syntax: `def = "a" ("b" | "c")`, - output: `[a] ([b] | [c])`, + output: `"a" ("b" | "c")`, }, { title: "choice in sequence", syntax: `def = "abc" ("def" | "ghi")`, @@ -191,7 +191,7 @@ func TestSequenceFormat(t *testing.T) { }, { title: "grouped quantifier, single char", syntax: `def = ("a" "b"){3}`, - output: `([a] [b]){3}`, + output: `("a" "b"){3}`, }, { title: "grouped quantifier", syntax: `def = ("abc" "def"){3}`, @@ -203,7 +203,7 @@ func TestChoiceFormat(t *testing.T) { testDefinitionFormat(t, formatNone, []formatDefinitionTestItem{{ title: "choice of char sequences, single char", syntax: `def = "a" | "b" | "c"`, - output: `[a] | [b] | [c]`, + output: `"a" | "b" | "c"`, }, { title: "choice of char sequences", syntax: `def = "abc" | "def" | "ghi"`, @@ -211,7 +211,7 @@ func TestChoiceFormat(t *testing.T) { }, { title: "choice of inline sequences, single char", syntax: `def = "a" "b" | "c" "d" | "e" "f"`, - output: `[a] [b] | [c] [d] | [e] [f]`, + output: `"a" "b" | "c" "d" | "e" "f"`, }, { title: "choice of inline sequences", syntax: `def = "abc" "def" | "ghi" "jkl" | "mno" "pqr"`, @@ -223,8 +223,92 @@ func TestChoiceFormat(t *testing.T) { }}) } -func TestMultiLine(t *testing.T) { -} +const testDoc = `/* + foo +*/ +// bar +// bar +// baz +/* foo + bar baz */ // foo bar baz +wschar:alias = " " | "\t" | "\n" | "\b" | "\f" | "\r" | "\v"; +wsc:ws = wschar | comment; +block-comment:alias:nows = "/*" ("*" [^/] | [^*])* "*/"; +line-comment:alias:nows = "//" [^\n]*; +comment-segment:alias:nows = line-comment | block-comment; +ws-no-nl:alias:nows = " " + | "\t" + | "\b" + /* this one */ /* is a */ + | "\f" + /* form feed */ /* for sure */ + | "\r" + | "\v"; +comment:nows = comment-segment + /* segment is not the best name */ /* but */ + (ws-no-nl* "\n"? ws-no-nl* /* fine */ comment-segment)*; +any-char = "."; // equivalent to [^] -func TestLineSplit(t *testing.T) { +// caution: newline is accepted +class-not = "^"; +class-char:nows = [^\\\[\]\^\-] | "\\" .; +char-range:nows = class-char "-" class-char; +char-class:nows = "[" class-not? (class-char | char-range)* "]"; + +// newline is accepted +sequence-char:nows = [^\\"] | "\\" .; +char-sequence:nows = "\"" sequence-char* "\""; +terminal:alias = any-char | char-class | char-sequence; +symbol:nows = [^\\ \n\t\b\f\r\v/.\[\]\"{}\^+*?|():=;]+; +group:alias = "(" expression ")"; +number:alias:nows = [0-9]+; +count = number; +count-quantifier = "{" count "}"; +range-from = number; +range-to = number; +range-quantifier = "{" range-from? "," range-to? "}"; +one-or-more = "+"; +zero-or-more = "*"; +zero-or-one = "?"; +quantity:alias = count-quantifier | range-quantifier | one-or-more | zero-or-more | zero-or-one; +item:nows = (terminal | symbol | group) quantity?; +sequence = item+; +option:alias = terminal | symbol | group | sequence; + +// DOC: how the order matters +choice = option ("|" option)+; + +// DOC: not having 'not' needs some tricks sometimes + +expression:alias = terminal | symbol | group | sequence | choice; +alias = "alias"; +ws = "ws"; +nows = "nows"; +kw = "kw"; +nokw = "nokw"; +failpass = "failpass"; +root = "root"; +flag:alias = alias | ws | nows | kw | nokw | failpass | root; +definition-name:alias:nows = symbol (":" flag)*; +definition = definition-name "=" expression; +definitions:alias = definition (";"+ definition)*; +syntax:root = ";"* definitions? ";"*;` + +func TestDocFormat(t *testing.T) { + in := bytes.NewBufferString(testDoc) + s := &Syntax{} + if err := s.ReadSyntax(in); err != nil { + t.Fatal(err) + } + + out := bytes.NewBuffer(nil) + if err := s.Format(out); err != nil { + t.Fatal(err) + } + + if out.String() != testDoc { + t.Log(testDoc) + t.Log(out.String()) + t.Fatal() + } } diff --git a/head.go b/head.go index 7695d32..f1d38fc 100644 --- a/head.go +++ b/head.go @@ -1,4 +1,4 @@ package treerack // generated with scripts/createhead.go -const headCode = "import (\n\t\"strconv\"\n\t\"errors\"\n\t\"io\"\n\t\"strings\"\n\t\"unicode\"\n\t\"fmt\"\n\t\"bufio\"\n)\n\ntype charParser struct {\n\tname\tstring\n\tid\tint\n\tnot\tbool\n\tchars\t[]rune\n\tranges\t[][]rune\n}\ntype charBuilder struct {\n\tname\tstring\n\tid\tint\n}\n\nfunc (p *charParser) nodeName() string {\n\treturn p.name\n}\nfunc (p *charParser) nodeID() int {\n\treturn p.id\n}\nfunc (p *charParser) commitType() commitType {\n\treturn alias\n}\nfunc matchChar(chars []rune, ranges [][]rune, not bool, char rune) bool {\n\tfor _, ci := range chars {\n\t\tif ci == char {\n\t\t\treturn !not\n\t\t}\n\t}\n\tfor _, ri := range ranges {\n\t\tif char >= ri[0] && char <= ri[1] {\n\t\t\treturn !not\n\t\t}\n\t}\n\treturn not\n}\nfunc (p *charParser) match(t rune) bool {\n\treturn matchChar(p.chars, p.ranges, p.not, t)\n}\nfunc (p *charParser) parse(c *context) {\n\tif tok, ok := c.token(); !ok || !p.match(tok) {\n\t\tif c.offset > c.failOffset {\n\t\t\tc.failOffset = c.offset\n\t\t\tc.failingParser = nil\n\t\t}\n\t\tc.fail(c.offset)\n\t\treturn\n\t}\n\tc.success(c.offset + 1)\n}\nfunc (b *charBuilder) nodeName() string {\n\treturn b.name\n}\nfunc (b *charBuilder) nodeID() int {\n\treturn b.id\n}\nfunc (b *charBuilder) build(c *context) ([]*node, bool) {\n\treturn nil, false\n}\n\ntype sequenceParser struct {\n\tname\t\tstring\n\tid\t\tint\n\tcommit\t\tcommitType\n\titems\t\t[]parser\n\tranges\t\t[][]int\n\tgeneralizations\t[]int\n\tallChars\tbool\n}\ntype sequenceBuilder struct {\n\tname\t\tstring\n\tid\t\tint\n\tcommit\t\tcommitType\n\titems\t\t[]builder\n\tranges\t\t[][]int\n\tgeneralizations\t[]int\n\tallChars\tbool\n}\n\nfunc (p *sequenceParser) nodeName() string {\n\treturn p.name\n}\nfunc (p *sequenceParser) nodeID() int {\n\treturn p.id\n}\nfunc (p *sequenceParser) commitType() commitType {\n\treturn p.commit\n}\nfunc (p *sequenceParser) parse(c *context) {\n\tif !p.allChars {\n\t\tif c.results.pending(c.offset, p.id) {\n\t\t\tc.fail(c.offset)\n\t\t\treturn\n\t\t}\n\t\tc.results.markPending(c.offset, p.id)\n\t}\n\tvar (\n\t\tcurrentCount\tint\n\t\tparsed\t\tbool\n\t)\n\titemIndex := 0\n\tfrom := c.offset\n\tto := c.offset\n\tfor itemIndex < len(p.items) {\n\t\tp.items[itemIndex].parse(c)\n\t\tif !c.matchLast {\n\t\t\tif currentCount >= p.ranges[itemIndex][0] {\n\t\t\t\titemIndex++\n\t\t\t\tcurrentCount = 0\n\t\t\t\tcontinue\n\t\t\t}\n\t\t\tc.offset = from\n\t\t\tif c.fromResults(p) {\n\t\t\t\tif to > c.failOffset {\n\t\t\t\t\tc.failOffset = -1\n\t\t\t\t\tc.failingParser = nil\n\t\t\t\t}\n\t\t\t\tif !p.allChars {\n\t\t\t\t\tc.results.unmarkPending(from, p.id)\n\t\t\t\t}\n\t\t\t\treturn\n\t\t\t}\n\t\t\tif c.failingParser == nil && p.commit&userDefined != 0 && p.commit&whitespace == 0 && p.commit&failPass == 0 {\n\t\t\t\tc.failingParser = p\n\t\t\t}\n\t\t\tc.fail(from)\n\t\t\tif !p.allChars {\n\t\t\t\tc.results.unmarkPending(from, p.id)\n\t\t\t}\n\t\t\treturn\n\t\t}\n\t\tparsed = c.offset > to\n\t\tif parsed {\n\t\t\tcurrentCount++\n\t\t}\n\t\tto = c.offset\n\t\tif !parsed || p.ranges[itemIndex][1] > 0 && currentCount == p.ranges[itemIndex][1] {\n\t\t\titemIndex++\n\t\t\tcurrentCount = 0\n\t\t}\n\t}\n\tif p.commit&noKeyword != 0 && c.isKeyword(from, to) {\n\t\tif c.failingParser == nil && p.commit&userDefined != 0 && p.commit&whitespace == 0 && p.commit&failPass == 0 {\n\t\t\tc.failingParser = p\n\t\t}\n\t\tc.fail(from)\n\t\tif !p.allChars {\n\t\t\tc.results.unmarkPending(from, p.id)\n\t\t}\n\t\treturn\n\t}\n\tfor _, g := range p.generalizations {\n\t\tif c.results.pending(from, g) {\n\t\t\tc.results.setMatch(from, g, to)\n\t\t}\n\t}\n\tif to > c.failOffset {\n\t\tc.failOffset = -1\n\t\tc.failingParser = nil\n\t}\n\tc.results.setMatch(from, p.id, to)\n\tc.success(to)\n\tif !p.allChars {\n\t\tc.results.unmarkPending(from, p.id)\n\t}\n}\nfunc (b *sequenceBuilder) nodeName() string {\n\treturn b.name\n}\nfunc (b *sequenceBuilder) nodeID() int {\n\treturn b.id\n}\nfunc (b *sequenceBuilder) build(c *context) ([]*node, bool) {\n\tto, ok := c.results.longestMatch(c.offset, b.id)\n\tif !ok {\n\t\treturn nil, false\n\t}\n\tfrom := c.offset\n\tparsed := to > from\n\tif b.allChars {\n\t\tc.offset = to\n\t\tif b.commit&alias != 0 {\n\t\t\treturn nil, true\n\t\t}\n\t\treturn []*node{{Name: b.name, From: from, To: to, tokens: c.tokens}}, true\n\t} else if parsed {\n\t\tc.results.dropMatchTo(c.offset, b.id, to)\n\t\tfor _, g := range b.generalizations {\n\t\t\tc.results.dropMatchTo(c.offset, g, to)\n\t\t}\n\t} else {\n\t\tif c.results.pending(c.offset, b.id) {\n\t\t\treturn nil, false\n\t\t}\n\t\tc.results.markPending(c.offset, b.id)\n\t\tfor _, g := range b.generalizations {\n\t\t\tc.results.markPending(c.offset, g)\n\t\t}\n\t}\n\tvar (\n\t\titemIndex\tint\n\t\tcurrentCount\tint\n\t\tnodes\t\t[]*node\n\t)\n\tfor itemIndex < len(b.items) {\n\t\titemFrom := c.offset\n\t\tn, ok := b.items[itemIndex].build(c)\n\t\tif !ok {\n\t\t\titemIndex++\n\t\t\tcurrentCount = 0\n\t\t\tcontinue\n\t\t}\n\t\tif c.offset > itemFrom {\n\t\t\tnodes = append(nodes, n...)\n\t\t\tcurrentCount++\n\t\t\tif b.ranges[itemIndex][1] > 0 && currentCount == b.ranges[itemIndex][1] {\n\t\t\t\titemIndex++\n\t\t\t\tcurrentCount = 0\n\t\t\t}\n\t\t\tcontinue\n\t\t}\n\t\tif currentCount < b.ranges[itemIndex][0] {\n\t\t\tfor i := 0; i < b.ranges[itemIndex][0]-currentCount; i++ {\n\t\t\t\tnodes = append(nodes, n...)\n\t\t\t}\n\t\t}\n\t\titemIndex++\n\t\tcurrentCount = 0\n\t}\n\tif !parsed {\n\t\tc.results.unmarkPending(from, b.id)\n\t\tfor _, g := range b.generalizations {\n\t\t\tc.results.unmarkPending(from, g)\n\t\t}\n\t}\n\tif b.commit&alias != 0 {\n\t\treturn nodes, true\n\t}\n\treturn []*node{{Name: b.name, From: from, To: to, Nodes: nodes, tokens: c.tokens}}, true\n}\n\ntype choiceParser struct {\n\tname\t\tstring\n\tid\t\tint\n\tcommit\t\tcommitType\n\toptions\t\t[]parser\n\tgeneralizations\t[]int\n}\ntype choiceBuilder struct {\n\tname\t\tstring\n\tid\t\tint\n\tcommit\t\tcommitType\n\toptions\t\t[]builder\n\tgeneralizations\t[]int\n}\n\nfunc (p *choiceParser) nodeName() string {\n\treturn p.name\n}\nfunc (p *choiceParser) nodeID() int {\n\treturn p.id\n}\nfunc (p *choiceParser) commitType() commitType {\n\treturn p.commit\n}\nfunc (p *choiceParser) parse(c *context) {\n\tif c.fromResults(p) {\n\t\treturn\n\t}\n\tif c.results.pending(c.offset, p.id) {\n\t\tc.fail(c.offset)\n\t\treturn\n\t}\n\tc.results.markPending(c.offset, p.id)\n\tvar (\n\t\tmatch\t\tbool\n\t\toptionIndex\tint\n\t\tfoundMatch\tbool\n\t\tfailingParser\tparser\n\t)\n\tfrom := c.offset\n\tto := c.offset\n\tinitialFailOffset := c.failOffset\n\tinitialFailingParser := c.failingParser\n\tfailOffset := initialFailOffset\n\tfor {\n\t\tfoundMatch = false\n\t\toptionIndex = 0\n\t\tfor optionIndex < len(p.options) {\n\t\t\tp.options[optionIndex].parse(c)\n\t\t\toptionIndex++\n\t\t\tif !c.matchLast {\n\t\t\t\tif c.failOffset > failOffset {\n\t\t\t\t\tfailOffset = c.failOffset\n\t\t\t\t\tfailingParser = c.failingParser\n\t\t\t\t}\n\t\t\t}\n\t\t\tif !c.matchLast || match && c.offset <= to {\n\t\t\t\tc.offset = from\n\t\t\t\tcontinue\n\t\t\t}\n\t\t\tmatch = true\n\t\t\tfoundMatch = true\n\t\t\tto = c.offset\n\t\t\tc.offset = from\n\t\t\tc.results.setMatch(from, p.id, to)\n\t\t}\n\t\tif !foundMatch {\n\t\t\tbreak\n\t\t}\n\t}\n\tif match {\n\t\tif p.commit&noKeyword != 0 && c.isKeyword(from, to) {\n\t\t\tif c.failingParser == nil && p.commit&userDefined != 0 && p.commit&whitespace == 0 && p.commit&failPass == 0 {\n\t\t\t\tc.failingParser = p\n\t\t\t}\n\t\t\tc.fail(from)\n\t\t\tc.results.unmarkPending(from, p.id)\n\t\t\treturn\n\t\t}\n\t\tif failOffset > to {\n\t\t\tc.failOffset = failOffset\n\t\t\tc.failingParser = failingParser\n\t\t} else if to > initialFailOffset {\n\t\t\tc.failOffset = -1\n\t\t\tc.failingParser = nil\n\t\t} else {\n\t\t\tc.failOffset = initialFailOffset\n\t\t\tc.failingParser = initialFailingParser\n\t\t}\n\t\tc.success(to)\n\t\tc.results.unmarkPending(from, p.id)\n\t\treturn\n\t}\n\tif failOffset > initialFailOffset {\n\t\tc.failOffset = failOffset\n\t\tc.failingParser = failingParser\n\t\tif c.failingParser == nil && p.commitType()&userDefined != 0 && p.commitType()&whitespace == 0 && p.commitType()&failPass == 0 {\n\t\t\tc.failingParser = p\n\t\t}\n\t}\n\tc.results.setNoMatch(from, p.id)\n\tc.fail(from)\n\tc.results.unmarkPending(from, p.id)\n}\nfunc (b *choiceBuilder) nodeName() string {\n\treturn b.name\n}\nfunc (b *choiceBuilder) nodeID() int {\n\treturn b.id\n}\nfunc (b *choiceBuilder) build(c *context) ([]*node, bool) {\n\tto, ok := c.results.longestMatch(c.offset, b.id)\n\tif !ok {\n\t\treturn nil, false\n\t}\n\tfrom := c.offset\n\tparsed := to > from\n\tif parsed {\n\t\tc.results.dropMatchTo(c.offset, b.id, to)\n\t\tfor _, g := range b.generalizations {\n\t\t\tc.results.dropMatchTo(c.offset, g, to)\n\t\t}\n\t} else {\n\t\tif c.results.pending(c.offset, b.id) {\n\t\t\treturn nil, false\n\t\t}\n\t\tc.results.markPending(c.offset, b.id)\n\t\tfor _, g := range b.generalizations {\n\t\t\tc.results.markPending(c.offset, g)\n\t\t}\n\t}\n\tvar option builder\n\tfor _, o := range b.options {\n\t\tif c.results.hasMatchTo(c.offset, o.nodeID(), to) {\n\t\t\toption = o\n\t\t\tbreak\n\t\t}\n\t}\n\tn, _ := option.build(c)\n\tif !parsed {\n\t\tc.results.unmarkPending(from, b.id)\n\t\tfor _, g := range b.generalizations {\n\t\t\tc.results.unmarkPending(from, g)\n\t\t}\n\t}\n\tif b.commit&alias != 0 {\n\t\treturn n, true\n\t}\n\treturn []*node{{Name: b.name, From: from, To: to, Nodes: n, tokens: c.tokens}}, true\n}\n\ntype idSet struct{ ids []uint }\n\nfunc divModBits(id int) (int, int) {\n\treturn id / strconv.IntSize, id % strconv.IntSize\n}\nfunc (s *idSet) set(id int) {\n\td, m := divModBits(id)\n\tif d >= len(s.ids) {\n\t\tif d < cap(s.ids) {\n\t\t\ts.ids = s.ids[:d+1]\n\t\t} else {\n\t\t\ts.ids = s.ids[:cap(s.ids)]\n\t\t\tfor i := cap(s.ids); i <= d; i++ {\n\t\t\t\ts.ids = append(s.ids, 0)\n\t\t\t}\n\t\t}\n\t}\n\ts.ids[d] |= 1 << uint(m)\n}\nfunc (s *idSet) unset(id int) {\n\td, m := divModBits(id)\n\tif d >= len(s.ids) {\n\t\treturn\n\t}\n\ts.ids[d] &^= 1 << uint(m)\n}\nfunc (s *idSet) has(id int) bool {\n\td, m := divModBits(id)\n\tif d >= len(s.ids) {\n\t\treturn false\n\t}\n\treturn s.ids[d]&(1< offset {\n\t\treturn ints\n\t}\n\tif cap(ints) > offset {\n\t\tints = ints[:offset+1]\n\t\treturn ints\n\t}\n\tints = ints[:cap(ints)]\n\tfor i := len(ints); i <= offset; i++ {\n\t\tints = append(ints, nil)\n\t}\n\treturn ints\n}\nfunc ensureOffsetIDs(ids []*idSet, offset int) []*idSet {\n\tif len(ids) > offset {\n\t\treturn ids\n\t}\n\tif cap(ids) > offset {\n\t\tids = ids[:offset+1]\n\t\treturn ids\n\t}\n\tids = ids[:cap(ids)]\n\tfor i := len(ids); i <= offset; i++ {\n\t\tids = append(ids, nil)\n\t}\n\treturn ids\n}\nfunc (r *results) setMatch(offset, id, to int) {\n\tr.match = ensureOffsetInts(r.match, offset)\n\tfor i := 0; i < len(r.match[offset]); i += 2 {\n\t\tif r.match[offset][i] != id || r.match[offset][i+1] != to {\n\t\t\tcontinue\n\t\t}\n\t\treturn\n\t}\n\tr.match[offset] = append(r.match[offset], id, to)\n}\nfunc (r *results) setNoMatch(offset, id int) {\n\tif len(r.match) > offset {\n\t\tfor i := 0; i < len(r.match[offset]); i += 2 {\n\t\t\tif r.match[offset][i] != id {\n\t\t\t\tcontinue\n\t\t\t}\n\t\t\treturn\n\t\t}\n\t}\n\tr.noMatch = ensureOffsetIDs(r.noMatch, offset)\n\tif r.noMatch[offset] == nil {\n\t\tr.noMatch[offset] = &idSet{}\n\t}\n\tr.noMatch[offset].set(id)\n}\nfunc (r *results) hasMatchTo(offset, id, to int) bool {\n\tif len(r.match) <= offset {\n\t\treturn false\n\t}\n\tfor i := 0; i < len(r.match[offset]); i += 2 {\n\t\tif r.match[offset][i] != id {\n\t\t\tcontinue\n\t\t}\n\t\tif r.match[offset][i+1] == to {\n\t\t\treturn true\n\t\t}\n\t}\n\treturn false\n}\nfunc (r *results) longestMatch(offset, id int) (int, bool) {\n\tif len(r.match) <= offset {\n\t\treturn 0, false\n\t}\n\tvar found bool\n\tto := -1\n\tfor i := 0; i < len(r.match[offset]); i += 2 {\n\t\tif r.match[offset][i] != id {\n\t\t\tcontinue\n\t\t}\n\t\tif r.match[offset][i+1] > to {\n\t\t\tto = r.match[offset][i+1]\n\t\t}\n\t\tfound = true\n\t}\n\treturn to, found\n}\nfunc (r *results) longestResult(offset, id int) (int, bool, bool) {\n\tif len(r.noMatch) > offset && r.noMatch[offset] != nil && r.noMatch[offset].has(id) {\n\t\treturn 0, false, true\n\t}\n\tto, ok := r.longestMatch(offset, id)\n\treturn to, ok, ok\n}\nfunc (r *results) dropMatchTo(offset, id, to int) {\n\tfor i := 0; i < len(r.match[offset]); i += 2 {\n\t\tif r.match[offset][i] != id {\n\t\t\tcontinue\n\t\t}\n\t\tif r.match[offset][i+1] == to {\n\t\t\tr.match[offset][i] = -1\n\t\t\treturn\n\t\t}\n\t}\n}\nfunc (r *results) resetPending() {\n\tr.isPending = nil\n}\nfunc (r *results) pending(offset, id int) bool {\n\tif len(r.isPending) <= id {\n\t\treturn false\n\t}\n\tfor i := range r.isPending[id] {\n\t\tif r.isPending[id][i] == offset {\n\t\t\treturn true\n\t\t}\n\t}\n\treturn false\n}\nfunc (r *results) markPending(offset, id int) {\n\tr.isPending = ensureOffsetInts(r.isPending, id)\n\tfor i := range r.isPending[id] {\n\t\tif r.isPending[id][i] == -1 {\n\t\t\tr.isPending[id][i] = offset\n\t\t\treturn\n\t\t}\n\t}\n\tr.isPending[id] = append(r.isPending[id], offset)\n}\nfunc (r *results) unmarkPending(offset, id int) {\n\tfor i := range r.isPending[id] {\n\t\tif r.isPending[id][i] == offset {\n\t\t\tr.isPending[id][i] = -1\n\t\t\tbreak\n\t\t}\n\t}\n}\n\ntype context struct {\n\treader\t\tio.RuneReader\n\tkeywords\t[]parser\n\toffset\t\tint\n\treadOffset\tint\n\tconsumed\tint\n\toffsetLimit\tint\n\tfailOffset\tint\n\tfailingParser\tparser\n\treadErr\t\terror\n\teof\t\tbool\n\tresults\t\t*results\n\ttokens\t\t[]rune\n\tmatchLast\tbool\n}\n\nfunc newContext(r io.RuneReader, keywords []parser) *context {\n\treturn &context{reader: r, keywords: keywords, results: &results{}, offsetLimit: -1, failOffset: -1}\n}\nfunc (c *context) read() bool {\n\tif c.eof || c.readErr != nil {\n\t\treturn false\n\t}\n\ttoken, n, err := c.reader.ReadRune()\n\tif err != nil {\n\t\tif errors.Is(err, io.EOF) {\n\t\t\tif n == 0 {\n\t\t\t\tc.eof = true\n\t\t\t\treturn false\n\t\t\t}\n\t\t} else {\n\t\t\tc.readErr = err\n\t\t\treturn false\n\t\t}\n\t}\n\tc.readOffset++\n\tif token == unicode.ReplacementChar {\n\t\tc.readErr = errInvalidUnicodeCharacter\n\t\treturn false\n\t}\n\tc.tokens = append(c.tokens, token)\n\treturn true\n}\nfunc (c *context) token() (rune, bool) {\n\tif c.offset == c.offsetLimit {\n\t\treturn 0, false\n\t}\n\tif c.offset == c.readOffset {\n\t\tif !c.read() {\n\t\t\treturn 0, false\n\t\t}\n\t}\n\treturn c.tokens[c.offset], true\n}\nfunc (c *context) fromResults(p parser) bool {\n\tto, m, ok := c.results.longestResult(c.offset, p.nodeID())\n\tif !ok {\n\t\treturn false\n\t}\n\tif m {\n\t\tc.success(to)\n\t} else {\n\t\tc.fail(c.offset)\n\t}\n\treturn true\n}\nfunc (c *context) isKeyword(from, to int) bool {\n\tol := c.offsetLimit\n\tc.offsetLimit = to\n\tdefer func() {\n\t\tc.offsetLimit = ol\n\t}()\n\tfor _, kw := range c.keywords {\n\t\tc.offset = from\n\t\tkw.parse(c)\n\t\tif c.matchLast && c.offset == to {\n\t\t\treturn true\n\t\t}\n\t}\n\treturn false\n}\nfunc (c *context) success(to int) {\n\tc.offset = to\n\tc.matchLast = true\n\tif to > c.consumed {\n\t\tc.consumed = to\n\t}\n}\nfunc (c *context) fail(offset int) {\n\tc.offset = offset\n\tc.matchLast = false\n}\nfunc findLine(tokens []rune, offset int) (line, column int) {\n\tif offset < 0 {\n\t\treturn 0, 0\n\t}\n\ttokens = tokens[:offset]\n\tfor i := range tokens {\n\t\tcolumn++\n\t\tif tokens[i] == '\\n' {\n\t\t\tcolumn = 0\n\t\t\tline++\n\t\t}\n\t}\n\treturn\n}\nfunc (c *context) parseError(p parser) error {\n\tdefinition := p.nodeName()\n\tflagIndex := strings.Index(definition, \":\")\n\tif flagIndex > 0 {\n\t\tdefinition = definition[:flagIndex]\n\t}\n\tif c.failingParser == nil {\n\t\tc.failOffset = c.consumed\n\t}\n\tline, col := findLine(c.tokens, c.failOffset)\n\treturn &parseError{Offset: c.failOffset, Line: line, Column: col, Definition: definition}\n}\nfunc (c *context) finalizeParse(root parser) error {\n\tfp := c.failingParser\n\tif fp == nil {\n\t\tfp = root\n\t}\n\tto, match, found := c.results.longestResult(0, root.nodeID())\n\tif !found || !match || found && match && to < c.readOffset {\n\t\treturn c.parseError(fp)\n\t}\n\tc.read()\n\tif c.eof {\n\t\treturn nil\n\t}\n\tif c.readErr != nil {\n\t\treturn c.readErr\n\t}\n\treturn c.parseError(root)\n}\n\ntype node struct {\n\tName\tstring\n\tNodes\t[]*node\n\tFrom\tint\n\tTo\tint\n\ttokens\t[]rune\n}\n\nfunc (n *node) Tokens() []rune {\n\treturn n.tokens\n}\nfunc (n *node) String() string {\n\treturn fmt.Sprintf(\"%s:%d:%d:%s\", n.Name, n.From, n.To, n.Text())\n}\nfunc (n *node) Text() string {\n\treturn string(n.Tokens()[n.From:n.To])\n}\n\ntype commitType int\n\nconst (\n\tnone\tcommitType\t= 0\n\talias\tcommitType\t= 1 << iota\n\twhitespace\n\tnoWhitespace\n\tkeyword\n\tnoKeyword\n\tfailPass\n\troot\n\tuserDefined\n)\n\ntype formatFlags int\n\nconst (\n\tformatNone\tformatFlags\t= 0\n\tformatPretty\tformatFlags\t= 1 << iota\n\tformatIncludeComments\n)\n\ntype parseError struct {\n\tInput\t\tstring\n\tOffset\t\tint\n\tLine\t\tint\n\tColumn\t\tint\n\tDefinition\tstring\n}\ntype parser interface {\n\tnodeName() string\n\tnodeID() int\n\tcommitType() commitType\n\tparse(*context)\n}\ntype builder interface {\n\tnodeName() string\n\tnodeID() int\n\tbuild(*context) ([]*node, bool)\n}\n\nvar errInvalidUnicodeCharacter = errors.New(\"invalid unicode character\")\n\nfunc (pe *parseError) Error() string {\n\treturn fmt.Sprintf(\"%s:%d:%d:parse failed, parsing: %s\", pe.Input, pe.Line+1, pe.Column+1, pe.Definition)\n}\nfunc parseInput(r io.Reader, p parser, b builder, kw []parser) (*node, error) {\n\tc := newContext(bufio.NewReader(r), kw)\n\tp.parse(c)\n\tif c.readErr != nil {\n\t\treturn nil, c.readErr\n\t}\n\tif err := c.finalizeParse(p); err != nil {\n\t\tif perr, ok := err.(*parseError); ok {\n\t\t\tperr.Input = \"\"\n\t\t}\n\t\treturn nil, err\n\t}\n\tc.offset = 0\n\tc.results.resetPending()\n\tn, _ := b.build(c)\n\treturn n[0], nil\n}\n" +const headCode = "import (\n\t\"strconv\"\n\t\"errors\"\n\t\"io\"\n\t\"strings\"\n\t\"unicode\"\n\t\"fmt\"\n\t\"bufio\"\n)\n\ntype charParser struct {\n\tname\tstring\n\tid\tint\n\tnot\tbool\n\tchars\t[]rune\n\tranges\t[][]rune\n}\ntype charBuilder struct {\n\tname\tstring\n\tid\tint\n}\n\nfunc (p *charParser) nodeName() string {\n\treturn p.name\n}\nfunc (p *charParser) nodeID() int {\n\treturn p.id\n}\nfunc (p *charParser) commitType() commitType {\n\treturn alias\n}\nfunc matchChar(chars []rune, ranges [][]rune, not bool, char rune) bool {\n\tfor _, ci := range chars {\n\t\tif ci == char {\n\t\t\treturn !not\n\t\t}\n\t}\n\tfor _, ri := range ranges {\n\t\tif char >= ri[0] && char <= ri[1] {\n\t\t\treturn !not\n\t\t}\n\t}\n\treturn not\n}\nfunc (p *charParser) match(t rune) bool {\n\treturn matchChar(p.chars, p.ranges, p.not, t)\n}\nfunc (p *charParser) parse(c *context) {\n\tif tok, ok := c.token(); !ok || !p.match(tok) {\n\t\tif c.offset > c.failOffset {\n\t\t\tc.failOffset = c.offset\n\t\t\tc.failingParser = nil\n\t\t}\n\t\tc.fail(c.offset)\n\t\treturn\n\t}\n\tc.success(c.offset + 1)\n}\nfunc (b *charBuilder) nodeName() string {\n\treturn b.name\n}\nfunc (b *charBuilder) nodeID() int {\n\treturn b.id\n}\nfunc (b *charBuilder) build(c *context) ([]*node, bool) {\n\treturn nil, false\n}\n\ntype sequenceParser struct {\n\tname\t\tstring\n\tid\t\tint\n\tcommit\t\tcommitType\n\titems\t\t[]parser\n\tranges\t\t[][]int\n\tgeneralizations\t[]int\n\tallChars\tbool\n}\ntype sequenceBuilder struct {\n\tname\t\tstring\n\tid\t\tint\n\tcommit\t\tcommitType\n\titems\t\t[]builder\n\tranges\t\t[][]int\n\tgeneralizations\t[]int\n\tallChars\tbool\n}\n\nfunc (p *sequenceParser) nodeName() string {\n\treturn p.name\n}\nfunc (p *sequenceParser) nodeID() int {\n\treturn p.id\n}\nfunc (p *sequenceParser) commitType() commitType {\n\treturn p.commit\n}\nfunc (p *sequenceParser) parse(c *context) {\n\tif !p.allChars {\n\t\tif c.results.pending(c.offset, p.id) {\n\t\t\tc.fail(c.offset)\n\t\t\treturn\n\t\t}\n\t\tc.results.markPending(c.offset, p.id)\n\t}\n\tvar (\n\t\tcurrentCount\tint\n\t\tparsed\t\tbool\n\t)\n\titemIndex := 0\n\tfrom := c.offset\n\tto := c.offset\n\tfor itemIndex < len(p.items) {\n\t\tp.items[itemIndex].parse(c)\n\t\tif !c.matchLast {\n\t\t\tif currentCount >= p.ranges[itemIndex][0] {\n\t\t\t\titemIndex++\n\t\t\t\tcurrentCount = 0\n\t\t\t\tcontinue\n\t\t\t}\n\t\t\tc.offset = from\n\t\t\tif c.fromResults(p) {\n\t\t\t\tif to > c.failOffset {\n\t\t\t\t\tc.failOffset = -1\n\t\t\t\t\tc.failingParser = nil\n\t\t\t\t}\n\t\t\t\tif !p.allChars {\n\t\t\t\t\tc.results.unmarkPending(from, p.id)\n\t\t\t\t}\n\t\t\t\treturn\n\t\t\t}\n\t\t\tif c.failingParser == nil && p.commit&userDefined != 0 && p.commit&whitespace == 0 && p.commit&failPass == 0 {\n\t\t\t\tc.failingParser = p\n\t\t\t}\n\t\t\tc.fail(from)\n\t\t\tif !p.allChars {\n\t\t\t\tc.results.unmarkPending(from, p.id)\n\t\t\t}\n\t\t\treturn\n\t\t}\n\t\tparsed = c.offset > to\n\t\tif parsed {\n\t\t\tcurrentCount++\n\t\t}\n\t\tto = c.offset\n\t\tif !parsed || p.ranges[itemIndex][1] > 0 && currentCount == p.ranges[itemIndex][1] {\n\t\t\titemIndex++\n\t\t\tcurrentCount = 0\n\t\t}\n\t}\n\tif p.commit&noKeyword != 0 && c.isKeyword(from, to) {\n\t\tif c.failingParser == nil && p.commit&userDefined != 0 && p.commit&whitespace == 0 && p.commit&failPass == 0 {\n\t\t\tc.failingParser = p\n\t\t}\n\t\tc.fail(from)\n\t\tif !p.allChars {\n\t\t\tc.results.unmarkPending(from, p.id)\n\t\t}\n\t\treturn\n\t}\n\tfor _, g := range p.generalizations {\n\t\tif c.results.pending(from, g) {\n\t\t\tc.results.setMatch(from, g, to)\n\t\t}\n\t}\n\tif to > c.failOffset {\n\t\tc.failOffset = -1\n\t\tc.failingParser = nil\n\t}\n\tc.results.setMatch(from, p.id, to)\n\tc.success(to)\n\tif !p.allChars {\n\t\tc.results.unmarkPending(from, p.id)\n\t}\n}\nfunc (b *sequenceBuilder) nodeName() string {\n\treturn b.name\n}\nfunc (b *sequenceBuilder) nodeID() int {\n\treturn b.id\n}\nfunc (b *sequenceBuilder) build(c *context) ([]*node, bool) {\n\tto, ok := c.results.longestMatch(c.offset, b.id)\n\tif !ok {\n\t\treturn nil, false\n\t}\n\tfrom := c.offset\n\tparsed := to > from\n\tif b.allChars {\n\t\tc.offset = to\n\t\tif b.commit&alias != 0 {\n\t\t\treturn nil, true\n\t\t}\n\t\treturn []*node{{Name: b.name, From: from, To: to, tokens: c.tokens}}, true\n\t} else if parsed {\n\t\tc.results.dropMatchTo(c.offset, b.id, to)\n\t\tfor _, g := range b.generalizations {\n\t\t\tc.results.dropMatchTo(c.offset, g, to)\n\t\t}\n\t} else {\n\t\tif c.results.pending(c.offset, b.id) {\n\t\t\treturn nil, false\n\t\t}\n\t\tc.results.markPending(c.offset, b.id)\n\t\tfor _, g := range b.generalizations {\n\t\t\tc.results.markPending(c.offset, g)\n\t\t}\n\t}\n\tvar (\n\t\titemIndex\tint\n\t\tcurrentCount\tint\n\t\tnodes\t\t[]*node\n\t)\n\tfor itemIndex < len(b.items) {\n\t\titemFrom := c.offset\n\t\tn, ok := b.items[itemIndex].build(c)\n\t\tif !ok {\n\t\t\titemIndex++\n\t\t\tcurrentCount = 0\n\t\t\tcontinue\n\t\t}\n\t\tif c.offset > itemFrom {\n\t\t\tnodes = append(nodes, n...)\n\t\t\tcurrentCount++\n\t\t\tif b.ranges[itemIndex][1] > 0 && currentCount == b.ranges[itemIndex][1] {\n\t\t\t\titemIndex++\n\t\t\t\tcurrentCount = 0\n\t\t\t}\n\t\t\tcontinue\n\t\t}\n\t\tif currentCount < b.ranges[itemIndex][0] {\n\t\t\tfor i := 0; i < b.ranges[itemIndex][0]-currentCount; i++ {\n\t\t\t\tnodes = append(nodes, n...)\n\t\t\t}\n\t\t}\n\t\titemIndex++\n\t\tcurrentCount = 0\n\t}\n\tif !parsed {\n\t\tc.results.unmarkPending(from, b.id)\n\t\tfor _, g := range b.generalizations {\n\t\t\tc.results.unmarkPending(from, g)\n\t\t}\n\t}\n\tif b.commit&alias != 0 {\n\t\treturn nodes, true\n\t}\n\treturn []*node{{Name: b.name, From: from, To: to, Nodes: nodes, tokens: c.tokens}}, true\n}\n\ntype choiceParser struct {\n\tname\t\tstring\n\tid\t\tint\n\tcommit\t\tcommitType\n\toptions\t\t[]parser\n\tgeneralizations\t[]int\n}\ntype choiceBuilder struct {\n\tname\t\tstring\n\tid\t\tint\n\tcommit\t\tcommitType\n\toptions\t\t[]builder\n\tgeneralizations\t[]int\n}\n\nfunc (p *choiceParser) nodeName() string {\n\treturn p.name\n}\nfunc (p *choiceParser) nodeID() int {\n\treturn p.id\n}\nfunc (p *choiceParser) commitType() commitType {\n\treturn p.commit\n}\nfunc (p *choiceParser) parse(c *context) {\n\tif c.fromResults(p) {\n\t\treturn\n\t}\n\tif c.results.pending(c.offset, p.id) {\n\t\tc.fail(c.offset)\n\t\treturn\n\t}\n\tc.results.markPending(c.offset, p.id)\n\tvar (\n\t\tmatch\t\tbool\n\t\toptionIndex\tint\n\t\tfoundMatch\tbool\n\t\tfailingParser\tparser\n\t)\n\tfrom := c.offset\n\tto := c.offset\n\tinitialFailOffset := c.failOffset\n\tinitialFailingParser := c.failingParser\n\tfailOffset := initialFailOffset\n\tfor {\n\t\tfoundMatch = false\n\t\toptionIndex = 0\n\t\tfor optionIndex < len(p.options) {\n\t\t\tp.options[optionIndex].parse(c)\n\t\t\toptionIndex++\n\t\t\tif !c.matchLast {\n\t\t\t\tif c.failOffset > failOffset {\n\t\t\t\t\tfailOffset = c.failOffset\n\t\t\t\t\tfailingParser = c.failingParser\n\t\t\t\t}\n\t\t\t}\n\t\t\tif !c.matchLast || match && c.offset <= to {\n\t\t\t\tc.offset = from\n\t\t\t\tcontinue\n\t\t\t}\n\t\t\tmatch = true\n\t\t\tfoundMatch = true\n\t\t\tto = c.offset\n\t\t\tc.offset = from\n\t\t\tc.results.setMatch(from, p.id, to)\n\t\t}\n\t\tif !foundMatch {\n\t\t\tbreak\n\t\t}\n\t}\n\tif match {\n\t\tif p.commit&noKeyword != 0 && c.isKeyword(from, to) {\n\t\t\tif c.failingParser == nil && p.commit&userDefined != 0 && p.commit&whitespace == 0 && p.commit&failPass == 0 {\n\t\t\t\tc.failingParser = p\n\t\t\t}\n\t\t\tc.fail(from)\n\t\t\tc.results.unmarkPending(from, p.id)\n\t\t\treturn\n\t\t}\n\t\tif failOffset > to {\n\t\t\tc.failOffset = failOffset\n\t\t\tc.failingParser = failingParser\n\t\t} else if to > initialFailOffset {\n\t\t\tc.failOffset = -1\n\t\t\tc.failingParser = nil\n\t\t} else {\n\t\t\tc.failOffset = initialFailOffset\n\t\t\tc.failingParser = initialFailingParser\n\t\t}\n\t\tc.success(to)\n\t\tc.results.unmarkPending(from, p.id)\n\t\treturn\n\t}\n\tif failOffset > initialFailOffset {\n\t\tc.failOffset = failOffset\n\t\tc.failingParser = failingParser\n\t\tif c.failingParser == nil && p.commitType()&userDefined != 0 && p.commitType()&whitespace == 0 && p.commitType()&failPass == 0 {\n\t\t\tc.failingParser = p\n\t\t}\n\t}\n\tc.results.setNoMatch(from, p.id)\n\tc.fail(from)\n\tc.results.unmarkPending(from, p.id)\n}\nfunc (b *choiceBuilder) nodeName() string {\n\treturn b.name\n}\nfunc (b *choiceBuilder) nodeID() int {\n\treturn b.id\n}\nfunc (b *choiceBuilder) build(c *context) ([]*node, bool) {\n\tto, ok := c.results.longestMatch(c.offset, b.id)\n\tif !ok {\n\t\treturn nil, false\n\t}\n\tfrom := c.offset\n\tparsed := to > from\n\tif parsed {\n\t\tc.results.dropMatchTo(c.offset, b.id, to)\n\t\tfor _, g := range b.generalizations {\n\t\t\tc.results.dropMatchTo(c.offset, g, to)\n\t\t}\n\t} else {\n\t\tif c.results.pending(c.offset, b.id) {\n\t\t\treturn nil, false\n\t\t}\n\t\tc.results.markPending(c.offset, b.id)\n\t\tfor _, g := range b.generalizations {\n\t\t\tc.results.markPending(c.offset, g)\n\t\t}\n\t}\n\tvar option builder\n\tfor _, o := range b.options {\n\t\tif c.results.hasMatchTo(c.offset, o.nodeID(), to) {\n\t\t\toption = o\n\t\t\tbreak\n\t\t}\n\t}\n\tn, _ := option.build(c)\n\tif !parsed {\n\t\tc.results.unmarkPending(from, b.id)\n\t\tfor _, g := range b.generalizations {\n\t\t\tc.results.unmarkPending(from, g)\n\t\t}\n\t}\n\tif b.commit&alias != 0 {\n\t\treturn n, true\n\t}\n\treturn []*node{{Name: b.name, From: from, To: to, Nodes: n, tokens: c.tokens}}, true\n}\n\ntype idSet struct{ ids []uint }\n\nfunc divModBits(id int) (int, int) {\n\treturn id / strconv.IntSize, id % strconv.IntSize\n}\nfunc (s *idSet) set(id int) {\n\td, m := divModBits(id)\n\tif d >= len(s.ids) {\n\t\tif d < cap(s.ids) {\n\t\t\ts.ids = s.ids[:d+1]\n\t\t} else {\n\t\t\ts.ids = s.ids[:cap(s.ids)]\n\t\t\tfor i := cap(s.ids); i <= d; i++ {\n\t\t\t\ts.ids = append(s.ids, 0)\n\t\t\t}\n\t\t}\n\t}\n\ts.ids[d] |= 1 << uint(m)\n}\nfunc (s *idSet) unset(id int) {\n\td, m := divModBits(id)\n\tif d >= len(s.ids) {\n\t\treturn\n\t}\n\ts.ids[d] &^= 1 << uint(m)\n}\nfunc (s *idSet) has(id int) bool {\n\td, m := divModBits(id)\n\tif d >= len(s.ids) {\n\t\treturn false\n\t}\n\treturn s.ids[d]&(1< offset {\n\t\treturn ints\n\t}\n\tif cap(ints) > offset {\n\t\tints = ints[:offset+1]\n\t\treturn ints\n\t}\n\tints = ints[:cap(ints)]\n\tfor i := len(ints); i <= offset; i++ {\n\t\tints = append(ints, nil)\n\t}\n\treturn ints\n}\nfunc ensureOffsetIDs(ids []*idSet, offset int) []*idSet {\n\tif len(ids) > offset {\n\t\treturn ids\n\t}\n\tif cap(ids) > offset {\n\t\tids = ids[:offset+1]\n\t\treturn ids\n\t}\n\tids = ids[:cap(ids)]\n\tfor i := len(ids); i <= offset; i++ {\n\t\tids = append(ids, nil)\n\t}\n\treturn ids\n}\nfunc (r *results) setMatch(offset, id, to int) {\n\tr.match = ensureOffsetInts(r.match, offset)\n\tfor i := 0; i < len(r.match[offset]); i += 2 {\n\t\tif r.match[offset][i] != id || r.match[offset][i+1] != to {\n\t\t\tcontinue\n\t\t}\n\t\treturn\n\t}\n\tr.match[offset] = append(r.match[offset], id, to)\n}\nfunc (r *results) setNoMatch(offset, id int) {\n\tif len(r.match) > offset {\n\t\tfor i := 0; i < len(r.match[offset]); i += 2 {\n\t\t\tif r.match[offset][i] != id {\n\t\t\t\tcontinue\n\t\t\t}\n\t\t\treturn\n\t\t}\n\t}\n\tr.noMatch = ensureOffsetIDs(r.noMatch, offset)\n\tif r.noMatch[offset] == nil {\n\t\tr.noMatch[offset] = &idSet{}\n\t}\n\tr.noMatch[offset].set(id)\n}\nfunc (r *results) hasMatchTo(offset, id, to int) bool {\n\tif len(r.match) <= offset {\n\t\treturn false\n\t}\n\tfor i := 0; i < len(r.match[offset]); i += 2 {\n\t\tif r.match[offset][i] != id {\n\t\t\tcontinue\n\t\t}\n\t\tif r.match[offset][i+1] == to {\n\t\t\treturn true\n\t\t}\n\t}\n\treturn false\n}\nfunc (r *results) longestMatch(offset, id int) (int, bool) {\n\tif len(r.match) <= offset {\n\t\treturn 0, false\n\t}\n\tvar found bool\n\tto := -1\n\tfor i := 0; i < len(r.match[offset]); i += 2 {\n\t\tif r.match[offset][i] != id {\n\t\t\tcontinue\n\t\t}\n\t\tif r.match[offset][i+1] > to {\n\t\t\tto = r.match[offset][i+1]\n\t\t}\n\t\tfound = true\n\t}\n\treturn to, found\n}\nfunc (r *results) longestResult(offset, id int) (int, bool, bool) {\n\tif len(r.noMatch) > offset && r.noMatch[offset] != nil && r.noMatch[offset].has(id) {\n\t\treturn 0, false, true\n\t}\n\tto, ok := r.longestMatch(offset, id)\n\treturn to, ok, ok\n}\nfunc (r *results) dropMatchTo(offset, id, to int) {\n\tfor i := 0; i < len(r.match[offset]); i += 2 {\n\t\tif r.match[offset][i] != id {\n\t\t\tcontinue\n\t\t}\n\t\tif r.match[offset][i+1] == to {\n\t\t\tr.match[offset][i] = -1\n\t\t\treturn\n\t\t}\n\t}\n}\nfunc (r *results) resetPending() {\n\tr.isPending = nil\n}\nfunc (r *results) pending(offset, id int) bool {\n\tif len(r.isPending) <= id {\n\t\treturn false\n\t}\n\tfor i := range r.isPending[id] {\n\t\tif r.isPending[id][i] == offset {\n\t\t\treturn true\n\t\t}\n\t}\n\treturn false\n}\nfunc (r *results) markPending(offset, id int) {\n\tr.isPending = ensureOffsetInts(r.isPending, id)\n\tfor i := range r.isPending[id] {\n\t\tif r.isPending[id][i] == -1 {\n\t\t\tr.isPending[id][i] = offset\n\t\t\treturn\n\t\t}\n\t}\n\tr.isPending[id] = append(r.isPending[id], offset)\n}\nfunc (r *results) unmarkPending(offset, id int) {\n\tfor i := range r.isPending[id] {\n\t\tif r.isPending[id][i] == offset {\n\t\t\tr.isPending[id][i] = -1\n\t\t\tbreak\n\t\t}\n\t}\n}\n\ntype context struct {\n\treader\t\tio.RuneReader\n\tkeywords\t[]parser\n\toffset\t\tint\n\treadOffset\tint\n\tconsumed\tint\n\toffsetLimit\tint\n\tfailOffset\tint\n\tfailingParser\tparser\n\treadErr\t\terror\n\teof\t\tbool\n\tresults\t\t*results\n\ttokens\t\t[]rune\n\tmatchLast\tbool\n}\n\nfunc newContext(r io.RuneReader, keywords []parser) *context {\n\treturn &context{reader: r, keywords: keywords, results: &results{}, offsetLimit: -1, failOffset: -1}\n}\nfunc (c *context) read() bool {\n\tif c.eof || c.readErr != nil {\n\t\treturn false\n\t}\n\ttoken, n, err := c.reader.ReadRune()\n\tif err != nil {\n\t\tif errors.Is(err, io.EOF) {\n\t\t\tif n == 0 {\n\t\t\t\tc.eof = true\n\t\t\t\treturn false\n\t\t\t}\n\t\t} else {\n\t\t\tc.readErr = err\n\t\t\treturn false\n\t\t}\n\t}\n\tc.readOffset++\n\tif token == unicode.ReplacementChar {\n\t\tc.readErr = errInvalidUnicodeCharacter\n\t\treturn false\n\t}\n\tc.tokens = append(c.tokens, token)\n\treturn true\n}\nfunc (c *context) token() (rune, bool) {\n\tif c.offset == c.offsetLimit {\n\t\treturn 0, false\n\t}\n\tif c.offset == c.readOffset {\n\t\tif !c.read() {\n\t\t\treturn 0, false\n\t\t}\n\t}\n\treturn c.tokens[c.offset], true\n}\nfunc (c *context) fromResults(p parser) bool {\n\tto, m, ok := c.results.longestResult(c.offset, p.nodeID())\n\tif !ok {\n\t\treturn false\n\t}\n\tif m {\n\t\tc.success(to)\n\t} else {\n\t\tc.fail(c.offset)\n\t}\n\treturn true\n}\nfunc (c *context) isKeyword(from, to int) bool {\n\tol := c.offsetLimit\n\tc.offsetLimit = to\n\tdefer func() {\n\t\tc.offsetLimit = ol\n\t}()\n\tfor _, kw := range c.keywords {\n\t\tc.offset = from\n\t\tkw.parse(c)\n\t\tif c.matchLast && c.offset == to {\n\t\t\treturn true\n\t\t}\n\t}\n\treturn false\n}\nfunc (c *context) success(to int) {\n\tc.offset = to\n\tc.matchLast = true\n\tif to > c.consumed {\n\t\tc.consumed = to\n\t}\n}\nfunc (c *context) fail(offset int) {\n\tc.offset = offset\n\tc.matchLast = false\n}\nfunc findLine(tokens []rune, offset int) (line, column int) {\n\tif offset < 0 {\n\t\treturn 0, 0\n\t}\n\ttokens = tokens[:offset]\n\tfor i := range tokens {\n\t\tcolumn++\n\t\tif tokens[i] == '\\n' {\n\t\t\tcolumn = 0\n\t\t\tline++\n\t\t}\n\t}\n\treturn\n}\nfunc (c *context) parseError(p parser) error {\n\tdefinition := p.nodeName()\n\tflagIndex := strings.Index(definition, \":\")\n\tif flagIndex > 0 {\n\t\tdefinition = definition[:flagIndex]\n\t}\n\tif c.failingParser == nil {\n\t\tc.failOffset = c.consumed\n\t}\n\tline, col := findLine(c.tokens, c.failOffset)\n\treturn &parseError{Offset: c.failOffset, Line: line, Column: col, Definition: definition}\n}\nfunc (c *context) finalizeParse(root parser) error {\n\tfp := c.failingParser\n\tif fp == nil {\n\t\tfp = root\n\t}\n\tto, match, found := c.results.longestResult(0, root.nodeID())\n\tif !found || !match || found && match && to < c.readOffset {\n\t\treturn c.parseError(fp)\n\t}\n\tc.read()\n\tif c.eof {\n\t\treturn nil\n\t}\n\tif c.readErr != nil {\n\t\treturn c.readErr\n\t}\n\treturn c.parseError(root)\n}\n\ntype node struct {\n\tName\tstring\n\tNodes\t[]*node\n\tFrom\tint\n\tTo\tint\n\ttokens\t[]rune\n}\n\nfunc (n *node) Tokens() []rune {\n\treturn n.tokens\n}\nfunc (n *node) String() string {\n\treturn fmt.Sprintf(\"%s:%d:%d:%s\", n.Name, n.From, n.To, n.Text())\n}\nfunc (n *node) Text() string {\n\treturn string(n.Tokens()[n.From:n.To])\n}\n\ntype commitType int\n\nconst (\n\tnone\tcommitType\t= 0\n\talias\tcommitType\t= 1 << iota\n\twhitespace\n\tnoWhitespace\n\tkeyword\n\tnoKeyword\n\tfailPass\n\troot\n\tuserDefined\n)\n\ntype formatFlags int\n\nconst (\n\tformatNone\tformatFlags\t= 0\n\tformatPretty\tformatFlags\t= 1 << iota\n\tformatIncludeComments\n)\n\ntype formatOptions struct {\n\tmode\t\tformatFlags\n\ttargetWidth\tint\n}\ntype parseError struct {\n\tInput\t\tstring\n\tOffset\t\tint\n\tLine\t\tint\n\tColumn\t\tint\n\tDefinition\tstring\n}\ntype parser interface {\n\tnodeName() string\n\tnodeID() int\n\tcommitType() commitType\n\tparse(*context)\n}\ntype builder interface {\n\tnodeName() string\n\tnodeID() int\n\tbuild(*context) ([]*node, bool)\n}\n\nvar errInvalidUnicodeCharacter = errors.New(\"invalid unicode character\")\n\nfunc (pe *parseError) Error() string {\n\treturn fmt.Sprintf(\"%s:%d:%d:parse failed, parsing: %s\", pe.Input, pe.Line+1, pe.Column+1, pe.Definition)\n}\nfunc parseInput(r io.Reader, p parser, b builder, kw []parser) (*node, error) {\n\tc := newContext(bufio.NewReader(r), kw)\n\tp.parse(c)\n\tif c.readErr != nil {\n\t\treturn nil, c.readErr\n\t}\n\tif err := c.finalizeParse(p); err != nil {\n\t\tif perr, ok := err.(*parseError); ok {\n\t\t\tperr.Input = \"\"\n\t\t}\n\t\treturn nil, err\n\t}\n\tc.offset = 0\n\tc.results.resetPending()\n\tn, _ := b.build(c)\n\treturn n[0], nil\n}\n" diff --git a/headexported.go b/headexported.go index 7e7e1c8..375a7af 100644 --- a/headexported.go +++ b/headexported.go @@ -1,4 +1,4 @@ package treerack // generated with scripts/createhead.go -const headCodeExported = "import (\n\t\"strconv\"\n\t\"errors\"\n\t\"io\"\n\t\"strings\"\n\t\"unicode\"\n\t\"fmt\"\n\t\"bufio\"\n)\n\ntype charParser struct {\n\tname\tstring\n\tid\tint\n\tnot\tbool\n\tchars\t[]rune\n\tranges\t[][]rune\n}\ntype charBuilder struct {\n\tname\tstring\n\tid\tint\n}\n\nfunc (p *charParser) nodeName() string {\n\treturn p.name\n}\nfunc (p *charParser) nodeID() int {\n\treturn p.id\n}\nfunc (p *charParser) commitType() CommitType {\n\treturn Alias\n}\nfunc matchChar(chars []rune, ranges [][]rune, not bool, char rune) bool {\n\tfor _, ci := range chars {\n\t\tif ci == char {\n\t\t\treturn !not\n\t\t}\n\t}\n\tfor _, ri := range ranges {\n\t\tif char >= ri[0] && char <= ri[1] {\n\t\t\treturn !not\n\t\t}\n\t}\n\treturn not\n}\nfunc (p *charParser) match(t rune) bool {\n\treturn matchChar(p.chars, p.ranges, p.not, t)\n}\nfunc (p *charParser) parse(c *context) {\n\tif tok, ok := c.token(); !ok || !p.match(tok) {\n\t\tif c.offset > c.failOffset {\n\t\t\tc.failOffset = c.offset\n\t\t\tc.failingParser = nil\n\t\t}\n\t\tc.fail(c.offset)\n\t\treturn\n\t}\n\tc.success(c.offset + 1)\n}\nfunc (b *charBuilder) nodeName() string {\n\treturn b.name\n}\nfunc (b *charBuilder) nodeID() int {\n\treturn b.id\n}\nfunc (b *charBuilder) build(c *context) ([]*Node, bool) {\n\treturn nil, false\n}\n\ntype sequenceParser struct {\n\tname\t\tstring\n\tid\t\tint\n\tcommit\t\tCommitType\n\titems\t\t[]parser\n\tranges\t\t[][]int\n\tgeneralizations\t[]int\n\tallChars\tbool\n}\ntype sequenceBuilder struct {\n\tname\t\tstring\n\tid\t\tint\n\tcommit\t\tCommitType\n\titems\t\t[]builder\n\tranges\t\t[][]int\n\tgeneralizations\t[]int\n\tallChars\tbool\n}\n\nfunc (p *sequenceParser) nodeName() string {\n\treturn p.name\n}\nfunc (p *sequenceParser) nodeID() int {\n\treturn p.id\n}\nfunc (p *sequenceParser) commitType() CommitType {\n\treturn p.commit\n}\nfunc (p *sequenceParser) parse(c *context) {\n\tif !p.allChars {\n\t\tif c.results.pending(c.offset, p.id) {\n\t\t\tc.fail(c.offset)\n\t\t\treturn\n\t\t}\n\t\tc.results.markPending(c.offset, p.id)\n\t}\n\tvar (\n\t\tcurrentCount\tint\n\t\tparsed\t\tbool\n\t)\n\titemIndex := 0\n\tfrom := c.offset\n\tto := c.offset\n\tfor itemIndex < len(p.items) {\n\t\tp.items[itemIndex].parse(c)\n\t\tif !c.matchLast {\n\t\t\tif currentCount >= p.ranges[itemIndex][0] {\n\t\t\t\titemIndex++\n\t\t\t\tcurrentCount = 0\n\t\t\t\tcontinue\n\t\t\t}\n\t\t\tc.offset = from\n\t\t\tif c.fromResults(p) {\n\t\t\t\tif to > c.failOffset {\n\t\t\t\t\tc.failOffset = -1\n\t\t\t\t\tc.failingParser = nil\n\t\t\t\t}\n\t\t\t\tif !p.allChars {\n\t\t\t\t\tc.results.unmarkPending(from, p.id)\n\t\t\t\t}\n\t\t\t\treturn\n\t\t\t}\n\t\t\tif c.failingParser == nil && p.commit&userDefined != 0 && p.commit&Whitespace == 0 && p.commit&FailPass == 0 {\n\t\t\t\tc.failingParser = p\n\t\t\t}\n\t\t\tc.fail(from)\n\t\t\tif !p.allChars {\n\t\t\t\tc.results.unmarkPending(from, p.id)\n\t\t\t}\n\t\t\treturn\n\t\t}\n\t\tparsed = c.offset > to\n\t\tif parsed {\n\t\t\tcurrentCount++\n\t\t}\n\t\tto = c.offset\n\t\tif !parsed || p.ranges[itemIndex][1] > 0 && currentCount == p.ranges[itemIndex][1] {\n\t\t\titemIndex++\n\t\t\tcurrentCount = 0\n\t\t}\n\t}\n\tif p.commit&NoKeyword != 0 && c.isKeyword(from, to) {\n\t\tif c.failingParser == nil && p.commit&userDefined != 0 && p.commit&Whitespace == 0 && p.commit&FailPass == 0 {\n\t\t\tc.failingParser = p\n\t\t}\n\t\tc.fail(from)\n\t\tif !p.allChars {\n\t\t\tc.results.unmarkPending(from, p.id)\n\t\t}\n\t\treturn\n\t}\n\tfor _, g := range p.generalizations {\n\t\tif c.results.pending(from, g) {\n\t\t\tc.results.setMatch(from, g, to)\n\t\t}\n\t}\n\tif to > c.failOffset {\n\t\tc.failOffset = -1\n\t\tc.failingParser = nil\n\t}\n\tc.results.setMatch(from, p.id, to)\n\tc.success(to)\n\tif !p.allChars {\n\t\tc.results.unmarkPending(from, p.id)\n\t}\n}\nfunc (b *sequenceBuilder) nodeName() string {\n\treturn b.name\n}\nfunc (b *sequenceBuilder) nodeID() int {\n\treturn b.id\n}\nfunc (b *sequenceBuilder) build(c *context) ([]*Node, bool) {\n\tto, ok := c.results.longestMatch(c.offset, b.id)\n\tif !ok {\n\t\treturn nil, false\n\t}\n\tfrom := c.offset\n\tparsed := to > from\n\tif b.allChars {\n\t\tc.offset = to\n\t\tif b.commit&Alias != 0 {\n\t\t\treturn nil, true\n\t\t}\n\t\treturn []*Node{{Name: b.name, From: from, To: to, tokens: c.tokens}}, true\n\t} else if parsed {\n\t\tc.results.dropMatchTo(c.offset, b.id, to)\n\t\tfor _, g := range b.generalizations {\n\t\t\tc.results.dropMatchTo(c.offset, g, to)\n\t\t}\n\t} else {\n\t\tif c.results.pending(c.offset, b.id) {\n\t\t\treturn nil, false\n\t\t}\n\t\tc.results.markPending(c.offset, b.id)\n\t\tfor _, g := range b.generalizations {\n\t\t\tc.results.markPending(c.offset, g)\n\t\t}\n\t}\n\tvar (\n\t\titemIndex\tint\n\t\tcurrentCount\tint\n\t\tnodes\t\t[]*Node\n\t)\n\tfor itemIndex < len(b.items) {\n\t\titemFrom := c.offset\n\t\tn, ok := b.items[itemIndex].build(c)\n\t\tif !ok {\n\t\t\titemIndex++\n\t\t\tcurrentCount = 0\n\t\t\tcontinue\n\t\t}\n\t\tif c.offset > itemFrom {\n\t\t\tnodes = append(nodes, n...)\n\t\t\tcurrentCount++\n\t\t\tif b.ranges[itemIndex][1] > 0 && currentCount == b.ranges[itemIndex][1] {\n\t\t\t\titemIndex++\n\t\t\t\tcurrentCount = 0\n\t\t\t}\n\t\t\tcontinue\n\t\t}\n\t\tif currentCount < b.ranges[itemIndex][0] {\n\t\t\tfor i := 0; i < b.ranges[itemIndex][0]-currentCount; i++ {\n\t\t\t\tnodes = append(nodes, n...)\n\t\t\t}\n\t\t}\n\t\titemIndex++\n\t\tcurrentCount = 0\n\t}\n\tif !parsed {\n\t\tc.results.unmarkPending(from, b.id)\n\t\tfor _, g := range b.generalizations {\n\t\t\tc.results.unmarkPending(from, g)\n\t\t}\n\t}\n\tif b.commit&Alias != 0 {\n\t\treturn nodes, true\n\t}\n\treturn []*Node{{Name: b.name, From: from, To: to, Nodes: nodes, tokens: c.tokens}}, true\n}\n\ntype choiceParser struct {\n\tname\t\tstring\n\tid\t\tint\n\tcommit\t\tCommitType\n\toptions\t\t[]parser\n\tgeneralizations\t[]int\n}\ntype choiceBuilder struct {\n\tname\t\tstring\n\tid\t\tint\n\tcommit\t\tCommitType\n\toptions\t\t[]builder\n\tgeneralizations\t[]int\n}\n\nfunc (p *choiceParser) nodeName() string {\n\treturn p.name\n}\nfunc (p *choiceParser) nodeID() int {\n\treturn p.id\n}\nfunc (p *choiceParser) commitType() CommitType {\n\treturn p.commit\n}\nfunc (p *choiceParser) parse(c *context) {\n\tif c.fromResults(p) {\n\t\treturn\n\t}\n\tif c.results.pending(c.offset, p.id) {\n\t\tc.fail(c.offset)\n\t\treturn\n\t}\n\tc.results.markPending(c.offset, p.id)\n\tvar (\n\t\tmatch\t\tbool\n\t\toptionIndex\tint\n\t\tfoundMatch\tbool\n\t\tfailingParser\tparser\n\t)\n\tfrom := c.offset\n\tto := c.offset\n\tinitialFailOffset := c.failOffset\n\tinitialFailingParser := c.failingParser\n\tfailOffset := initialFailOffset\n\tfor {\n\t\tfoundMatch = false\n\t\toptionIndex = 0\n\t\tfor optionIndex < len(p.options) {\n\t\t\tp.options[optionIndex].parse(c)\n\t\t\toptionIndex++\n\t\t\tif !c.matchLast {\n\t\t\t\tif c.failOffset > failOffset {\n\t\t\t\t\tfailOffset = c.failOffset\n\t\t\t\t\tfailingParser = c.failingParser\n\t\t\t\t}\n\t\t\t}\n\t\t\tif !c.matchLast || match && c.offset <= to {\n\t\t\t\tc.offset = from\n\t\t\t\tcontinue\n\t\t\t}\n\t\t\tmatch = true\n\t\t\tfoundMatch = true\n\t\t\tto = c.offset\n\t\t\tc.offset = from\n\t\t\tc.results.setMatch(from, p.id, to)\n\t\t}\n\t\tif !foundMatch {\n\t\t\tbreak\n\t\t}\n\t}\n\tif match {\n\t\tif p.commit&NoKeyword != 0 && c.isKeyword(from, to) {\n\t\t\tif c.failingParser == nil && p.commit&userDefined != 0 && p.commit&Whitespace == 0 && p.commit&FailPass == 0 {\n\t\t\t\tc.failingParser = p\n\t\t\t}\n\t\t\tc.fail(from)\n\t\t\tc.results.unmarkPending(from, p.id)\n\t\t\treturn\n\t\t}\n\t\tif failOffset > to {\n\t\t\tc.failOffset = failOffset\n\t\t\tc.failingParser = failingParser\n\t\t} else if to > initialFailOffset {\n\t\t\tc.failOffset = -1\n\t\t\tc.failingParser = nil\n\t\t} else {\n\t\t\tc.failOffset = initialFailOffset\n\t\t\tc.failingParser = initialFailingParser\n\t\t}\n\t\tc.success(to)\n\t\tc.results.unmarkPending(from, p.id)\n\t\treturn\n\t}\n\tif failOffset > initialFailOffset {\n\t\tc.failOffset = failOffset\n\t\tc.failingParser = failingParser\n\t\tif c.failingParser == nil && p.commitType()&userDefined != 0 && p.commitType()&Whitespace == 0 && p.commitType()&FailPass == 0 {\n\t\t\tc.failingParser = p\n\t\t}\n\t}\n\tc.results.setNoMatch(from, p.id)\n\tc.fail(from)\n\tc.results.unmarkPending(from, p.id)\n}\nfunc (b *choiceBuilder) nodeName() string {\n\treturn b.name\n}\nfunc (b *choiceBuilder) nodeID() int {\n\treturn b.id\n}\nfunc (b *choiceBuilder) build(c *context) ([]*Node, bool) {\n\tto, ok := c.results.longestMatch(c.offset, b.id)\n\tif !ok {\n\t\treturn nil, false\n\t}\n\tfrom := c.offset\n\tparsed := to > from\n\tif parsed {\n\t\tc.results.dropMatchTo(c.offset, b.id, to)\n\t\tfor _, g := range b.generalizations {\n\t\t\tc.results.dropMatchTo(c.offset, g, to)\n\t\t}\n\t} else {\n\t\tif c.results.pending(c.offset, b.id) {\n\t\t\treturn nil, false\n\t\t}\n\t\tc.results.markPending(c.offset, b.id)\n\t\tfor _, g := range b.generalizations {\n\t\t\tc.results.markPending(c.offset, g)\n\t\t}\n\t}\n\tvar option builder\n\tfor _, o := range b.options {\n\t\tif c.results.hasMatchTo(c.offset, o.nodeID(), to) {\n\t\t\toption = o\n\t\t\tbreak\n\t\t}\n\t}\n\tn, _ := option.build(c)\n\tif !parsed {\n\t\tc.results.unmarkPending(from, b.id)\n\t\tfor _, g := range b.generalizations {\n\t\t\tc.results.unmarkPending(from, g)\n\t\t}\n\t}\n\tif b.commit&Alias != 0 {\n\t\treturn n, true\n\t}\n\treturn []*Node{{Name: b.name, From: from, To: to, Nodes: n, tokens: c.tokens}}, true\n}\n\ntype idSet struct{ ids []uint }\n\nfunc divModBits(id int) (int, int) {\n\treturn id / strconv.IntSize, id % strconv.IntSize\n}\nfunc (s *idSet) set(id int) {\n\td, m := divModBits(id)\n\tif d >= len(s.ids) {\n\t\tif d < cap(s.ids) {\n\t\t\ts.ids = s.ids[:d+1]\n\t\t} else {\n\t\t\ts.ids = s.ids[:cap(s.ids)]\n\t\t\tfor i := cap(s.ids); i <= d; i++ {\n\t\t\t\ts.ids = append(s.ids, 0)\n\t\t\t}\n\t\t}\n\t}\n\ts.ids[d] |= 1 << uint(m)\n}\nfunc (s *idSet) unset(id int) {\n\td, m := divModBits(id)\n\tif d >= len(s.ids) {\n\t\treturn\n\t}\n\ts.ids[d] &^= 1 << uint(m)\n}\nfunc (s *idSet) has(id int) bool {\n\td, m := divModBits(id)\n\tif d >= len(s.ids) {\n\t\treturn false\n\t}\n\treturn s.ids[d]&(1< offset {\n\t\treturn ints\n\t}\n\tif cap(ints) > offset {\n\t\tints = ints[:offset+1]\n\t\treturn ints\n\t}\n\tints = ints[:cap(ints)]\n\tfor i := len(ints); i <= offset; i++ {\n\t\tints = append(ints, nil)\n\t}\n\treturn ints\n}\nfunc ensureOffsetIDs(ids []*idSet, offset int) []*idSet {\n\tif len(ids) > offset {\n\t\treturn ids\n\t}\n\tif cap(ids) > offset {\n\t\tids = ids[:offset+1]\n\t\treturn ids\n\t}\n\tids = ids[:cap(ids)]\n\tfor i := len(ids); i <= offset; i++ {\n\t\tids = append(ids, nil)\n\t}\n\treturn ids\n}\nfunc (r *results) setMatch(offset, id, to int) {\n\tr.match = ensureOffsetInts(r.match, offset)\n\tfor i := 0; i < len(r.match[offset]); i += 2 {\n\t\tif r.match[offset][i] != id || r.match[offset][i+1] != to {\n\t\t\tcontinue\n\t\t}\n\t\treturn\n\t}\n\tr.match[offset] = append(r.match[offset], id, to)\n}\nfunc (r *results) setNoMatch(offset, id int) {\n\tif len(r.match) > offset {\n\t\tfor i := 0; i < len(r.match[offset]); i += 2 {\n\t\t\tif r.match[offset][i] != id {\n\t\t\t\tcontinue\n\t\t\t}\n\t\t\treturn\n\t\t}\n\t}\n\tr.noMatch = ensureOffsetIDs(r.noMatch, offset)\n\tif r.noMatch[offset] == nil {\n\t\tr.noMatch[offset] = &idSet{}\n\t}\n\tr.noMatch[offset].set(id)\n}\nfunc (r *results) hasMatchTo(offset, id, to int) bool {\n\tif len(r.match) <= offset {\n\t\treturn false\n\t}\n\tfor i := 0; i < len(r.match[offset]); i += 2 {\n\t\tif r.match[offset][i] != id {\n\t\t\tcontinue\n\t\t}\n\t\tif r.match[offset][i+1] == to {\n\t\t\treturn true\n\t\t}\n\t}\n\treturn false\n}\nfunc (r *results) longestMatch(offset, id int) (int, bool) {\n\tif len(r.match) <= offset {\n\t\treturn 0, false\n\t}\n\tvar found bool\n\tto := -1\n\tfor i := 0; i < len(r.match[offset]); i += 2 {\n\t\tif r.match[offset][i] != id {\n\t\t\tcontinue\n\t\t}\n\t\tif r.match[offset][i+1] > to {\n\t\t\tto = r.match[offset][i+1]\n\t\t}\n\t\tfound = true\n\t}\n\treturn to, found\n}\nfunc (r *results) longestResult(offset, id int) (int, bool, bool) {\n\tif len(r.noMatch) > offset && r.noMatch[offset] != nil && r.noMatch[offset].has(id) {\n\t\treturn 0, false, true\n\t}\n\tto, ok := r.longestMatch(offset, id)\n\treturn to, ok, ok\n}\nfunc (r *results) dropMatchTo(offset, id, to int) {\n\tfor i := 0; i < len(r.match[offset]); i += 2 {\n\t\tif r.match[offset][i] != id {\n\t\t\tcontinue\n\t\t}\n\t\tif r.match[offset][i+1] == to {\n\t\t\tr.match[offset][i] = -1\n\t\t\treturn\n\t\t}\n\t}\n}\nfunc (r *results) resetPending() {\n\tr.isPending = nil\n}\nfunc (r *results) pending(offset, id int) bool {\n\tif len(r.isPending) <= id {\n\t\treturn false\n\t}\n\tfor i := range r.isPending[id] {\n\t\tif r.isPending[id][i] == offset {\n\t\t\treturn true\n\t\t}\n\t}\n\treturn false\n}\nfunc (r *results) markPending(offset, id int) {\n\tr.isPending = ensureOffsetInts(r.isPending, id)\n\tfor i := range r.isPending[id] {\n\t\tif r.isPending[id][i] == -1 {\n\t\t\tr.isPending[id][i] = offset\n\t\t\treturn\n\t\t}\n\t}\n\tr.isPending[id] = append(r.isPending[id], offset)\n}\nfunc (r *results) unmarkPending(offset, id int) {\n\tfor i := range r.isPending[id] {\n\t\tif r.isPending[id][i] == offset {\n\t\t\tr.isPending[id][i] = -1\n\t\t\tbreak\n\t\t}\n\t}\n}\n\ntype context struct {\n\treader\t\tio.RuneReader\n\tkeywords\t[]parser\n\toffset\t\tint\n\treadOffset\tint\n\tconsumed\tint\n\toffsetLimit\tint\n\tfailOffset\tint\n\tfailingParser\tparser\n\treadErr\t\terror\n\teof\t\tbool\n\tresults\t\t*results\n\ttokens\t\t[]rune\n\tmatchLast\tbool\n}\n\nfunc newContext(r io.RuneReader, keywords []parser) *context {\n\treturn &context{reader: r, keywords: keywords, results: &results{}, offsetLimit: -1, failOffset: -1}\n}\nfunc (c *context) read() bool {\n\tif c.eof || c.readErr != nil {\n\t\treturn false\n\t}\n\ttoken, n, err := c.reader.ReadRune()\n\tif err != nil {\n\t\tif errors.Is(err, io.EOF) {\n\t\t\tif n == 0 {\n\t\t\t\tc.eof = true\n\t\t\t\treturn false\n\t\t\t}\n\t\t} else {\n\t\t\tc.readErr = err\n\t\t\treturn false\n\t\t}\n\t}\n\tc.readOffset++\n\tif token == unicode.ReplacementChar {\n\t\tc.readErr = ErrInvalidUnicodeCharacter\n\t\treturn false\n\t}\n\tc.tokens = append(c.tokens, token)\n\treturn true\n}\nfunc (c *context) token() (rune, bool) {\n\tif c.offset == c.offsetLimit {\n\t\treturn 0, false\n\t}\n\tif c.offset == c.readOffset {\n\t\tif !c.read() {\n\t\t\treturn 0, false\n\t\t}\n\t}\n\treturn c.tokens[c.offset], true\n}\nfunc (c *context) fromResults(p parser) bool {\n\tto, m, ok := c.results.longestResult(c.offset, p.nodeID())\n\tif !ok {\n\t\treturn false\n\t}\n\tif m {\n\t\tc.success(to)\n\t} else {\n\t\tc.fail(c.offset)\n\t}\n\treturn true\n}\nfunc (c *context) isKeyword(from, to int) bool {\n\tol := c.offsetLimit\n\tc.offsetLimit = to\n\tdefer func() {\n\t\tc.offsetLimit = ol\n\t}()\n\tfor _, kw := range c.keywords {\n\t\tc.offset = from\n\t\tkw.parse(c)\n\t\tif c.matchLast && c.offset == to {\n\t\t\treturn true\n\t\t}\n\t}\n\treturn false\n}\nfunc (c *context) success(to int) {\n\tc.offset = to\n\tc.matchLast = true\n\tif to > c.consumed {\n\t\tc.consumed = to\n\t}\n}\nfunc (c *context) fail(offset int) {\n\tc.offset = offset\n\tc.matchLast = false\n}\nfunc findLine(tokens []rune, offset int) (line, column int) {\n\tif offset < 0 {\n\t\treturn 0, 0\n\t}\n\ttokens = tokens[:offset]\n\tfor i := range tokens {\n\t\tcolumn++\n\t\tif tokens[i] == '\\n' {\n\t\t\tcolumn = 0\n\t\t\tline++\n\t\t}\n\t}\n\treturn\n}\nfunc (c *context) parseError(p parser) error {\n\tdefinition := p.nodeName()\n\tflagIndex := strings.Index(definition, \":\")\n\tif flagIndex > 0 {\n\t\tdefinition = definition[:flagIndex]\n\t}\n\tif c.failingParser == nil {\n\t\tc.failOffset = c.consumed\n\t}\n\tline, col := findLine(c.tokens, c.failOffset)\n\treturn &ParseError{Offset: c.failOffset, Line: line, Column: col, Definition: definition}\n}\nfunc (c *context) finalizeParse(root parser) error {\n\tfp := c.failingParser\n\tif fp == nil {\n\t\tfp = root\n\t}\n\tto, match, found := c.results.longestResult(0, root.nodeID())\n\tif !found || !match || found && match && to < c.readOffset {\n\t\treturn c.parseError(fp)\n\t}\n\tc.read()\n\tif c.eof {\n\t\treturn nil\n\t}\n\tif c.readErr != nil {\n\t\treturn c.readErr\n\t}\n\treturn c.parseError(root)\n}\n\ntype Node struct {\n\tName\tstring\n\tNodes\t[]*Node\n\tFrom\tint\n\tTo\tint\n\ttokens\t[]rune\n}\n\nfunc (n *Node) Tokens() []rune {\n\treturn n.tokens\n}\nfunc (n *Node) String() string {\n\treturn fmt.Sprintf(\"%s:%d:%d:%s\", n.Name, n.From, n.To, n.Text())\n}\nfunc (n *Node) Text() string {\n\treturn string(n.Tokens()[n.From:n.To])\n}\n\ntype CommitType int\n\nconst (\n\tNone\tCommitType\t= 0\n\tAlias\tCommitType\t= 1 << iota\n\tWhitespace\n\tNoWhitespace\n\tKeyword\n\tNoKeyword\n\tFailPass\n\tRoot\n\tuserDefined\n)\n\ntype formatFlags int\n\nconst (\n\tformatNone\tformatFlags\t= 0\n\tformatPretty\tformatFlags\t= 1 << iota\n\tformatIncludeComments\n)\n\ntype ParseError struct {\n\tInput\t\tstring\n\tOffset\t\tint\n\tLine\t\tint\n\tColumn\t\tint\n\tDefinition\tstring\n}\ntype parser interface {\n\tnodeName() string\n\tnodeID() int\n\tcommitType() CommitType\n\tparse(*context)\n}\ntype builder interface {\n\tnodeName() string\n\tnodeID() int\n\tbuild(*context) ([]*Node, bool)\n}\n\nvar ErrInvalidUnicodeCharacter = errors.New(\"invalid unicode character\")\n\nfunc (pe *ParseError) Error() string {\n\treturn fmt.Sprintf(\"%s:%d:%d:parse failed, parsing: %s\", pe.Input, pe.Line+1, pe.Column+1, pe.Definition)\n}\nfunc parseInput(r io.Reader, p parser, b builder, kw []parser) (*Node, error) {\n\tc := newContext(bufio.NewReader(r), kw)\n\tp.parse(c)\n\tif c.readErr != nil {\n\t\treturn nil, c.readErr\n\t}\n\tif err := c.finalizeParse(p); err != nil {\n\t\tif perr, ok := err.(*ParseError); ok {\n\t\t\tperr.Input = \"\"\n\t\t}\n\t\treturn nil, err\n\t}\n\tc.offset = 0\n\tc.results.resetPending()\n\tn, _ := b.build(c)\n\treturn n[0], nil\n}\n" +const headCodeExported = "import (\n\t\"strconv\"\n\t\"errors\"\n\t\"io\"\n\t\"strings\"\n\t\"unicode\"\n\t\"fmt\"\n\t\"bufio\"\n)\n\ntype charParser struct {\n\tname\tstring\n\tid\tint\n\tnot\tbool\n\tchars\t[]rune\n\tranges\t[][]rune\n}\ntype charBuilder struct {\n\tname\tstring\n\tid\tint\n}\n\nfunc (p *charParser) nodeName() string {\n\treturn p.name\n}\nfunc (p *charParser) nodeID() int {\n\treturn p.id\n}\nfunc (p *charParser) commitType() CommitType {\n\treturn Alias\n}\nfunc matchChar(chars []rune, ranges [][]rune, not bool, char rune) bool {\n\tfor _, ci := range chars {\n\t\tif ci == char {\n\t\t\treturn !not\n\t\t}\n\t}\n\tfor _, ri := range ranges {\n\t\tif char >= ri[0] && char <= ri[1] {\n\t\t\treturn !not\n\t\t}\n\t}\n\treturn not\n}\nfunc (p *charParser) match(t rune) bool {\n\treturn matchChar(p.chars, p.ranges, p.not, t)\n}\nfunc (p *charParser) parse(c *context) {\n\tif tok, ok := c.token(); !ok || !p.match(tok) {\n\t\tif c.offset > c.failOffset {\n\t\t\tc.failOffset = c.offset\n\t\t\tc.failingParser = nil\n\t\t}\n\t\tc.fail(c.offset)\n\t\treturn\n\t}\n\tc.success(c.offset + 1)\n}\nfunc (b *charBuilder) nodeName() string {\n\treturn b.name\n}\nfunc (b *charBuilder) nodeID() int {\n\treturn b.id\n}\nfunc (b *charBuilder) build(c *context) ([]*Node, bool) {\n\treturn nil, false\n}\n\ntype sequenceParser struct {\n\tname\t\tstring\n\tid\t\tint\n\tcommit\t\tCommitType\n\titems\t\t[]parser\n\tranges\t\t[][]int\n\tgeneralizations\t[]int\n\tallChars\tbool\n}\ntype sequenceBuilder struct {\n\tname\t\tstring\n\tid\t\tint\n\tcommit\t\tCommitType\n\titems\t\t[]builder\n\tranges\t\t[][]int\n\tgeneralizations\t[]int\n\tallChars\tbool\n}\n\nfunc (p *sequenceParser) nodeName() string {\n\treturn p.name\n}\nfunc (p *sequenceParser) nodeID() int {\n\treturn p.id\n}\nfunc (p *sequenceParser) commitType() CommitType {\n\treturn p.commit\n}\nfunc (p *sequenceParser) parse(c *context) {\n\tif !p.allChars {\n\t\tif c.results.pending(c.offset, p.id) {\n\t\t\tc.fail(c.offset)\n\t\t\treturn\n\t\t}\n\t\tc.results.markPending(c.offset, p.id)\n\t}\n\tvar (\n\t\tcurrentCount\tint\n\t\tparsed\t\tbool\n\t)\n\titemIndex := 0\n\tfrom := c.offset\n\tto := c.offset\n\tfor itemIndex < len(p.items) {\n\t\tp.items[itemIndex].parse(c)\n\t\tif !c.matchLast {\n\t\t\tif currentCount >= p.ranges[itemIndex][0] {\n\t\t\t\titemIndex++\n\t\t\t\tcurrentCount = 0\n\t\t\t\tcontinue\n\t\t\t}\n\t\t\tc.offset = from\n\t\t\tif c.fromResults(p) {\n\t\t\t\tif to > c.failOffset {\n\t\t\t\t\tc.failOffset = -1\n\t\t\t\t\tc.failingParser = nil\n\t\t\t\t}\n\t\t\t\tif !p.allChars {\n\t\t\t\t\tc.results.unmarkPending(from, p.id)\n\t\t\t\t}\n\t\t\t\treturn\n\t\t\t}\n\t\t\tif c.failingParser == nil && p.commit&userDefined != 0 && p.commit&Whitespace == 0 && p.commit&FailPass == 0 {\n\t\t\t\tc.failingParser = p\n\t\t\t}\n\t\t\tc.fail(from)\n\t\t\tif !p.allChars {\n\t\t\t\tc.results.unmarkPending(from, p.id)\n\t\t\t}\n\t\t\treturn\n\t\t}\n\t\tparsed = c.offset > to\n\t\tif parsed {\n\t\t\tcurrentCount++\n\t\t}\n\t\tto = c.offset\n\t\tif !parsed || p.ranges[itemIndex][1] > 0 && currentCount == p.ranges[itemIndex][1] {\n\t\t\titemIndex++\n\t\t\tcurrentCount = 0\n\t\t}\n\t}\n\tif p.commit&NoKeyword != 0 && c.isKeyword(from, to) {\n\t\tif c.failingParser == nil && p.commit&userDefined != 0 && p.commit&Whitespace == 0 && p.commit&FailPass == 0 {\n\t\t\tc.failingParser = p\n\t\t}\n\t\tc.fail(from)\n\t\tif !p.allChars {\n\t\t\tc.results.unmarkPending(from, p.id)\n\t\t}\n\t\treturn\n\t}\n\tfor _, g := range p.generalizations {\n\t\tif c.results.pending(from, g) {\n\t\t\tc.results.setMatch(from, g, to)\n\t\t}\n\t}\n\tif to > c.failOffset {\n\t\tc.failOffset = -1\n\t\tc.failingParser = nil\n\t}\n\tc.results.setMatch(from, p.id, to)\n\tc.success(to)\n\tif !p.allChars {\n\t\tc.results.unmarkPending(from, p.id)\n\t}\n}\nfunc (b *sequenceBuilder) nodeName() string {\n\treturn b.name\n}\nfunc (b *sequenceBuilder) nodeID() int {\n\treturn b.id\n}\nfunc (b *sequenceBuilder) build(c *context) ([]*Node, bool) {\n\tto, ok := c.results.longestMatch(c.offset, b.id)\n\tif !ok {\n\t\treturn nil, false\n\t}\n\tfrom := c.offset\n\tparsed := to > from\n\tif b.allChars {\n\t\tc.offset = to\n\t\tif b.commit&Alias != 0 {\n\t\t\treturn nil, true\n\t\t}\n\t\treturn []*Node{{Name: b.name, From: from, To: to, tokens: c.tokens}}, true\n\t} else if parsed {\n\t\tc.results.dropMatchTo(c.offset, b.id, to)\n\t\tfor _, g := range b.generalizations {\n\t\t\tc.results.dropMatchTo(c.offset, g, to)\n\t\t}\n\t} else {\n\t\tif c.results.pending(c.offset, b.id) {\n\t\t\treturn nil, false\n\t\t}\n\t\tc.results.markPending(c.offset, b.id)\n\t\tfor _, g := range b.generalizations {\n\t\t\tc.results.markPending(c.offset, g)\n\t\t}\n\t}\n\tvar (\n\t\titemIndex\tint\n\t\tcurrentCount\tint\n\t\tnodes\t\t[]*Node\n\t)\n\tfor itemIndex < len(b.items) {\n\t\titemFrom := c.offset\n\t\tn, ok := b.items[itemIndex].build(c)\n\t\tif !ok {\n\t\t\titemIndex++\n\t\t\tcurrentCount = 0\n\t\t\tcontinue\n\t\t}\n\t\tif c.offset > itemFrom {\n\t\t\tnodes = append(nodes, n...)\n\t\t\tcurrentCount++\n\t\t\tif b.ranges[itemIndex][1] > 0 && currentCount == b.ranges[itemIndex][1] {\n\t\t\t\titemIndex++\n\t\t\t\tcurrentCount = 0\n\t\t\t}\n\t\t\tcontinue\n\t\t}\n\t\tif currentCount < b.ranges[itemIndex][0] {\n\t\t\tfor i := 0; i < b.ranges[itemIndex][0]-currentCount; i++ {\n\t\t\t\tnodes = append(nodes, n...)\n\t\t\t}\n\t\t}\n\t\titemIndex++\n\t\tcurrentCount = 0\n\t}\n\tif !parsed {\n\t\tc.results.unmarkPending(from, b.id)\n\t\tfor _, g := range b.generalizations {\n\t\t\tc.results.unmarkPending(from, g)\n\t\t}\n\t}\n\tif b.commit&Alias != 0 {\n\t\treturn nodes, true\n\t}\n\treturn []*Node{{Name: b.name, From: from, To: to, Nodes: nodes, tokens: c.tokens}}, true\n}\n\ntype choiceParser struct {\n\tname\t\tstring\n\tid\t\tint\n\tcommit\t\tCommitType\n\toptions\t\t[]parser\n\tgeneralizations\t[]int\n}\ntype choiceBuilder struct {\n\tname\t\tstring\n\tid\t\tint\n\tcommit\t\tCommitType\n\toptions\t\t[]builder\n\tgeneralizations\t[]int\n}\n\nfunc (p *choiceParser) nodeName() string {\n\treturn p.name\n}\nfunc (p *choiceParser) nodeID() int {\n\treturn p.id\n}\nfunc (p *choiceParser) commitType() CommitType {\n\treturn p.commit\n}\nfunc (p *choiceParser) parse(c *context) {\n\tif c.fromResults(p) {\n\t\treturn\n\t}\n\tif c.results.pending(c.offset, p.id) {\n\t\tc.fail(c.offset)\n\t\treturn\n\t}\n\tc.results.markPending(c.offset, p.id)\n\tvar (\n\t\tmatch\t\tbool\n\t\toptionIndex\tint\n\t\tfoundMatch\tbool\n\t\tfailingParser\tparser\n\t)\n\tfrom := c.offset\n\tto := c.offset\n\tinitialFailOffset := c.failOffset\n\tinitialFailingParser := c.failingParser\n\tfailOffset := initialFailOffset\n\tfor {\n\t\tfoundMatch = false\n\t\toptionIndex = 0\n\t\tfor optionIndex < len(p.options) {\n\t\t\tp.options[optionIndex].parse(c)\n\t\t\toptionIndex++\n\t\t\tif !c.matchLast {\n\t\t\t\tif c.failOffset > failOffset {\n\t\t\t\t\tfailOffset = c.failOffset\n\t\t\t\t\tfailingParser = c.failingParser\n\t\t\t\t}\n\t\t\t}\n\t\t\tif !c.matchLast || match && c.offset <= to {\n\t\t\t\tc.offset = from\n\t\t\t\tcontinue\n\t\t\t}\n\t\t\tmatch = true\n\t\t\tfoundMatch = true\n\t\t\tto = c.offset\n\t\t\tc.offset = from\n\t\t\tc.results.setMatch(from, p.id, to)\n\t\t}\n\t\tif !foundMatch {\n\t\t\tbreak\n\t\t}\n\t}\n\tif match {\n\t\tif p.commit&NoKeyword != 0 && c.isKeyword(from, to) {\n\t\t\tif c.failingParser == nil && p.commit&userDefined != 0 && p.commit&Whitespace == 0 && p.commit&FailPass == 0 {\n\t\t\t\tc.failingParser = p\n\t\t\t}\n\t\t\tc.fail(from)\n\t\t\tc.results.unmarkPending(from, p.id)\n\t\t\treturn\n\t\t}\n\t\tif failOffset > to {\n\t\t\tc.failOffset = failOffset\n\t\t\tc.failingParser = failingParser\n\t\t} else if to > initialFailOffset {\n\t\t\tc.failOffset = -1\n\t\t\tc.failingParser = nil\n\t\t} else {\n\t\t\tc.failOffset = initialFailOffset\n\t\t\tc.failingParser = initialFailingParser\n\t\t}\n\t\tc.success(to)\n\t\tc.results.unmarkPending(from, p.id)\n\t\treturn\n\t}\n\tif failOffset > initialFailOffset {\n\t\tc.failOffset = failOffset\n\t\tc.failingParser = failingParser\n\t\tif c.failingParser == nil && p.commitType()&userDefined != 0 && p.commitType()&Whitespace == 0 && p.commitType()&FailPass == 0 {\n\t\t\tc.failingParser = p\n\t\t}\n\t}\n\tc.results.setNoMatch(from, p.id)\n\tc.fail(from)\n\tc.results.unmarkPending(from, p.id)\n}\nfunc (b *choiceBuilder) nodeName() string {\n\treturn b.name\n}\nfunc (b *choiceBuilder) nodeID() int {\n\treturn b.id\n}\nfunc (b *choiceBuilder) build(c *context) ([]*Node, bool) {\n\tto, ok := c.results.longestMatch(c.offset, b.id)\n\tif !ok {\n\t\treturn nil, false\n\t}\n\tfrom := c.offset\n\tparsed := to > from\n\tif parsed {\n\t\tc.results.dropMatchTo(c.offset, b.id, to)\n\t\tfor _, g := range b.generalizations {\n\t\t\tc.results.dropMatchTo(c.offset, g, to)\n\t\t}\n\t} else {\n\t\tif c.results.pending(c.offset, b.id) {\n\t\t\treturn nil, false\n\t\t}\n\t\tc.results.markPending(c.offset, b.id)\n\t\tfor _, g := range b.generalizations {\n\t\t\tc.results.markPending(c.offset, g)\n\t\t}\n\t}\n\tvar option builder\n\tfor _, o := range b.options {\n\t\tif c.results.hasMatchTo(c.offset, o.nodeID(), to) {\n\t\t\toption = o\n\t\t\tbreak\n\t\t}\n\t}\n\tn, _ := option.build(c)\n\tif !parsed {\n\t\tc.results.unmarkPending(from, b.id)\n\t\tfor _, g := range b.generalizations {\n\t\t\tc.results.unmarkPending(from, g)\n\t\t}\n\t}\n\tif b.commit&Alias != 0 {\n\t\treturn n, true\n\t}\n\treturn []*Node{{Name: b.name, From: from, To: to, Nodes: n, tokens: c.tokens}}, true\n}\n\ntype idSet struct{ ids []uint }\n\nfunc divModBits(id int) (int, int) {\n\treturn id / strconv.IntSize, id % strconv.IntSize\n}\nfunc (s *idSet) set(id int) {\n\td, m := divModBits(id)\n\tif d >= len(s.ids) {\n\t\tif d < cap(s.ids) {\n\t\t\ts.ids = s.ids[:d+1]\n\t\t} else {\n\t\t\ts.ids = s.ids[:cap(s.ids)]\n\t\t\tfor i := cap(s.ids); i <= d; i++ {\n\t\t\t\ts.ids = append(s.ids, 0)\n\t\t\t}\n\t\t}\n\t}\n\ts.ids[d] |= 1 << uint(m)\n}\nfunc (s *idSet) unset(id int) {\n\td, m := divModBits(id)\n\tif d >= len(s.ids) {\n\t\treturn\n\t}\n\ts.ids[d] &^= 1 << uint(m)\n}\nfunc (s *idSet) has(id int) bool {\n\td, m := divModBits(id)\n\tif d >= len(s.ids) {\n\t\treturn false\n\t}\n\treturn s.ids[d]&(1< offset {\n\t\treturn ints\n\t}\n\tif cap(ints) > offset {\n\t\tints = ints[:offset+1]\n\t\treturn ints\n\t}\n\tints = ints[:cap(ints)]\n\tfor i := len(ints); i <= offset; i++ {\n\t\tints = append(ints, nil)\n\t}\n\treturn ints\n}\nfunc ensureOffsetIDs(ids []*idSet, offset int) []*idSet {\n\tif len(ids) > offset {\n\t\treturn ids\n\t}\n\tif cap(ids) > offset {\n\t\tids = ids[:offset+1]\n\t\treturn ids\n\t}\n\tids = ids[:cap(ids)]\n\tfor i := len(ids); i <= offset; i++ {\n\t\tids = append(ids, nil)\n\t}\n\treturn ids\n}\nfunc (r *results) setMatch(offset, id, to int) {\n\tr.match = ensureOffsetInts(r.match, offset)\n\tfor i := 0; i < len(r.match[offset]); i += 2 {\n\t\tif r.match[offset][i] != id || r.match[offset][i+1] != to {\n\t\t\tcontinue\n\t\t}\n\t\treturn\n\t}\n\tr.match[offset] = append(r.match[offset], id, to)\n}\nfunc (r *results) setNoMatch(offset, id int) {\n\tif len(r.match) > offset {\n\t\tfor i := 0; i < len(r.match[offset]); i += 2 {\n\t\t\tif r.match[offset][i] != id {\n\t\t\t\tcontinue\n\t\t\t}\n\t\t\treturn\n\t\t}\n\t}\n\tr.noMatch = ensureOffsetIDs(r.noMatch, offset)\n\tif r.noMatch[offset] == nil {\n\t\tr.noMatch[offset] = &idSet{}\n\t}\n\tr.noMatch[offset].set(id)\n}\nfunc (r *results) hasMatchTo(offset, id, to int) bool {\n\tif len(r.match) <= offset {\n\t\treturn false\n\t}\n\tfor i := 0; i < len(r.match[offset]); i += 2 {\n\t\tif r.match[offset][i] != id {\n\t\t\tcontinue\n\t\t}\n\t\tif r.match[offset][i+1] == to {\n\t\t\treturn true\n\t\t}\n\t}\n\treturn false\n}\nfunc (r *results) longestMatch(offset, id int) (int, bool) {\n\tif len(r.match) <= offset {\n\t\treturn 0, false\n\t}\n\tvar found bool\n\tto := -1\n\tfor i := 0; i < len(r.match[offset]); i += 2 {\n\t\tif r.match[offset][i] != id {\n\t\t\tcontinue\n\t\t}\n\t\tif r.match[offset][i+1] > to {\n\t\t\tto = r.match[offset][i+1]\n\t\t}\n\t\tfound = true\n\t}\n\treturn to, found\n}\nfunc (r *results) longestResult(offset, id int) (int, bool, bool) {\n\tif len(r.noMatch) > offset && r.noMatch[offset] != nil && r.noMatch[offset].has(id) {\n\t\treturn 0, false, true\n\t}\n\tto, ok := r.longestMatch(offset, id)\n\treturn to, ok, ok\n}\nfunc (r *results) dropMatchTo(offset, id, to int) {\n\tfor i := 0; i < len(r.match[offset]); i += 2 {\n\t\tif r.match[offset][i] != id {\n\t\t\tcontinue\n\t\t}\n\t\tif r.match[offset][i+1] == to {\n\t\t\tr.match[offset][i] = -1\n\t\t\treturn\n\t\t}\n\t}\n}\nfunc (r *results) resetPending() {\n\tr.isPending = nil\n}\nfunc (r *results) pending(offset, id int) bool {\n\tif len(r.isPending) <= id {\n\t\treturn false\n\t}\n\tfor i := range r.isPending[id] {\n\t\tif r.isPending[id][i] == offset {\n\t\t\treturn true\n\t\t}\n\t}\n\treturn false\n}\nfunc (r *results) markPending(offset, id int) {\n\tr.isPending = ensureOffsetInts(r.isPending, id)\n\tfor i := range r.isPending[id] {\n\t\tif r.isPending[id][i] == -1 {\n\t\t\tr.isPending[id][i] = offset\n\t\t\treturn\n\t\t}\n\t}\n\tr.isPending[id] = append(r.isPending[id], offset)\n}\nfunc (r *results) unmarkPending(offset, id int) {\n\tfor i := range r.isPending[id] {\n\t\tif r.isPending[id][i] == offset {\n\t\t\tr.isPending[id][i] = -1\n\t\t\tbreak\n\t\t}\n\t}\n}\n\ntype context struct {\n\treader\t\tio.RuneReader\n\tkeywords\t[]parser\n\toffset\t\tint\n\treadOffset\tint\n\tconsumed\tint\n\toffsetLimit\tint\n\tfailOffset\tint\n\tfailingParser\tparser\n\treadErr\t\terror\n\teof\t\tbool\n\tresults\t\t*results\n\ttokens\t\t[]rune\n\tmatchLast\tbool\n}\n\nfunc newContext(r io.RuneReader, keywords []parser) *context {\n\treturn &context{reader: r, keywords: keywords, results: &results{}, offsetLimit: -1, failOffset: -1}\n}\nfunc (c *context) read() bool {\n\tif c.eof || c.readErr != nil {\n\t\treturn false\n\t}\n\ttoken, n, err := c.reader.ReadRune()\n\tif err != nil {\n\t\tif errors.Is(err, io.EOF) {\n\t\t\tif n == 0 {\n\t\t\t\tc.eof = true\n\t\t\t\treturn false\n\t\t\t}\n\t\t} else {\n\t\t\tc.readErr = err\n\t\t\treturn false\n\t\t}\n\t}\n\tc.readOffset++\n\tif token == unicode.ReplacementChar {\n\t\tc.readErr = ErrInvalidUnicodeCharacter\n\t\treturn false\n\t}\n\tc.tokens = append(c.tokens, token)\n\treturn true\n}\nfunc (c *context) token() (rune, bool) {\n\tif c.offset == c.offsetLimit {\n\t\treturn 0, false\n\t}\n\tif c.offset == c.readOffset {\n\t\tif !c.read() {\n\t\t\treturn 0, false\n\t\t}\n\t}\n\treturn c.tokens[c.offset], true\n}\nfunc (c *context) fromResults(p parser) bool {\n\tto, m, ok := c.results.longestResult(c.offset, p.nodeID())\n\tif !ok {\n\t\treturn false\n\t}\n\tif m {\n\t\tc.success(to)\n\t} else {\n\t\tc.fail(c.offset)\n\t}\n\treturn true\n}\nfunc (c *context) isKeyword(from, to int) bool {\n\tol := c.offsetLimit\n\tc.offsetLimit = to\n\tdefer func() {\n\t\tc.offsetLimit = ol\n\t}()\n\tfor _, kw := range c.keywords {\n\t\tc.offset = from\n\t\tkw.parse(c)\n\t\tif c.matchLast && c.offset == to {\n\t\t\treturn true\n\t\t}\n\t}\n\treturn false\n}\nfunc (c *context) success(to int) {\n\tc.offset = to\n\tc.matchLast = true\n\tif to > c.consumed {\n\t\tc.consumed = to\n\t}\n}\nfunc (c *context) fail(offset int) {\n\tc.offset = offset\n\tc.matchLast = false\n}\nfunc findLine(tokens []rune, offset int) (line, column int) {\n\tif offset < 0 {\n\t\treturn 0, 0\n\t}\n\ttokens = tokens[:offset]\n\tfor i := range tokens {\n\t\tcolumn++\n\t\tif tokens[i] == '\\n' {\n\t\t\tcolumn = 0\n\t\t\tline++\n\t\t}\n\t}\n\treturn\n}\nfunc (c *context) parseError(p parser) error {\n\tdefinition := p.nodeName()\n\tflagIndex := strings.Index(definition, \":\")\n\tif flagIndex > 0 {\n\t\tdefinition = definition[:flagIndex]\n\t}\n\tif c.failingParser == nil {\n\t\tc.failOffset = c.consumed\n\t}\n\tline, col := findLine(c.tokens, c.failOffset)\n\treturn &ParseError{Offset: c.failOffset, Line: line, Column: col, Definition: definition}\n}\nfunc (c *context) finalizeParse(root parser) error {\n\tfp := c.failingParser\n\tif fp == nil {\n\t\tfp = root\n\t}\n\tto, match, found := c.results.longestResult(0, root.nodeID())\n\tif !found || !match || found && match && to < c.readOffset {\n\t\treturn c.parseError(fp)\n\t}\n\tc.read()\n\tif c.eof {\n\t\treturn nil\n\t}\n\tif c.readErr != nil {\n\t\treturn c.readErr\n\t}\n\treturn c.parseError(root)\n}\n\ntype Node struct {\n\tName\tstring\n\tNodes\t[]*Node\n\tFrom\tint\n\tTo\tint\n\ttokens\t[]rune\n}\n\nfunc (n *Node) Tokens() []rune {\n\treturn n.tokens\n}\nfunc (n *Node) String() string {\n\treturn fmt.Sprintf(\"%s:%d:%d:%s\", n.Name, n.From, n.To, n.Text())\n}\nfunc (n *Node) Text() string {\n\treturn string(n.Tokens()[n.From:n.To])\n}\n\ntype CommitType int\n\nconst (\n\tNone\tCommitType\t= 0\n\tAlias\tCommitType\t= 1 << iota\n\tWhitespace\n\tNoWhitespace\n\tKeyword\n\tNoKeyword\n\tFailPass\n\tRoot\n\tuserDefined\n)\n\ntype formatFlags int\n\nconst (\n\tformatNone\tformatFlags\t= 0\n\tformatPretty\tformatFlags\t= 1 << iota\n\tformatIncludeComments\n)\n\ntype formatOptions struct {\n\tmode\t\tformatFlags\n\ttargetWidth\tint\n}\ntype ParseError struct {\n\tInput\t\tstring\n\tOffset\t\tint\n\tLine\t\tint\n\tColumn\t\tint\n\tDefinition\tstring\n}\ntype parser interface {\n\tnodeName() string\n\tnodeID() int\n\tcommitType() CommitType\n\tparse(*context)\n}\ntype builder interface {\n\tnodeName() string\n\tnodeID() int\n\tbuild(*context) ([]*Node, bool)\n}\n\nvar ErrInvalidUnicodeCharacter = errors.New(\"invalid unicode character\")\n\nfunc (pe *ParseError) Error() string {\n\treturn fmt.Sprintf(\"%s:%d:%d:parse failed, parsing: %s\", pe.Input, pe.Line+1, pe.Column+1, pe.Definition)\n}\nfunc parseInput(r io.Reader, p parser, b builder, kw []parser) (*Node, error) {\n\tc := newContext(bufio.NewReader(r), kw)\n\tp.parse(c)\n\tif c.readErr != nil {\n\t\treturn nil, c.readErr\n\t}\n\tif err := c.finalizeParse(p); err != nil {\n\t\tif perr, ok := err.(*ParseError); ok {\n\t\t\tperr.Input = \"\"\n\t\t}\n\t\treturn nil, err\n\t}\n\tc.offset = 0\n\tc.results.resetPending()\n\tn, _ := b.build(c)\n\treturn n[0], nil\n}\n" diff --git a/internal/self/self.go b/internal/self/self.go index ddda63d..c207325 100644 --- a/internal/self/self.go +++ b/internal/self/self.go @@ -771,6 +771,10 @@ const ( formatIncludeComments ) +type formatOptions struct { + mode formatFlags + targetWidth int +} type ParseError struct { Input string Offset int diff --git a/sequencedefine.go b/sequencedefine.go index 93fe0a3..99a126d 100644 --- a/sequencedefine.go +++ b/sequencedefine.go @@ -4,6 +4,7 @@ import ( "fmt" "io" "strconv" + "strings" ) type sequenceDefinition struct { @@ -43,26 +44,26 @@ func (d *sequenceDefinition) setID(id int) { d.id = id } func (d *sequenceDefinition) commitType() CommitType { return d.commit } func (d *sequenceDefinition) setCommitType(ct CommitType) { d.commit = ct } -func normalizeItemRange(item SequenceItem) SequenceItem { - if item.Min == 0 && item.Max == 0 { - item.Min, item.Max = 1, 1 - return item +func normalizeItemRange(min, max int) (int, int) { + if min == 0 && max == 0 { + min, max = 1, 1 + return min, max } - if item.Min <= 0 { - item.Min = 0 + if min <= 0 { + min = 0 } - if item.Max <= 0 { - item.Max = -1 + if max <= 0 { + max = -1 } - return item + return min, max } func (d *sequenceDefinition) initRanges() { for i, item := range d.items { - item = normalizeItemRange(item) + item.Min, item.Max = normalizeItemRange(item.Min, item.Max) d.items[i] = item d.ranges = append(d.ranges, []int{item.Min, item.Max}) } @@ -184,8 +185,8 @@ func (d *sequenceDefinition) builder() builder { func (d *sequenceDefinition) isCharSequence(r *registry) bool { for i := range d.originalItems { - item := normalizeItemRange(d.originalItems[i]) - if item.Min != 1 || item.Max != 1 { + min, max := normalizeItemRange(d.originalItems[i].Min, d.originalItems[i].Max) + if min != 1 || max != 1 { return false } @@ -199,98 +200,140 @@ func (d *sequenceDefinition) isCharSequence(r *registry) bool { return true } -func (d *sequenceDefinition) format(r *registry, f formatFlags) string { - if d.isCharSequence(r) { - if len(d.originalItems) == 1 { - itemDef := r.definition[d.originalItems[0].Name] - c, _ := itemDef.(*charParser) - return c.format(r, f) - } - - var chars []rune - for i := range d.originalItems { - itemDef := r.definition[d.originalItems[i].Name] - c, _ := itemDef.(*charParser) - chars = append(chars, c.chars[0]) - } - - chars = escape(charClassEscape, []rune(charClassBanned), chars) - return string(append([]rune{'"'}, append(chars, '"')...)) +func (d *sequenceDefinition) formatCharSequence(r *registry, o formatOptions) string { + if len(d.originalItems) == 1 { + itemDef := r.definition[d.originalItems[0].Name] + c, _ := itemDef.(*charParser) + return c.format(r, o) } var chars []rune + for i := range d.originalItems { + itemDef := r.definition[d.originalItems[i].Name] + c, _ := itemDef.(*charParser) + chars = append(chars, c.chars[0]) + } + + chars = escape(charClassEscape, []rune(charSeqBanned), chars) + return string(append([]rune{'"'}, append(chars, '"')...)) +} + +func formatSequenceItem(r *registry, item definition, needsGrouping bool, o formatOptions) []rune { + oi := o + if needsGrouping { + oi.targetWidth -= 2 + } + + itemString := item.format(r, oi) + if !needsGrouping { + return []rune(itemString) + } + + var chars []rune + multiLine := strings.Contains(itemString, "\n") + if !multiLine { + chars = append(chars, '(') + chars = append(chars, []rune(itemString)...) + chars = append(chars, ')') + return chars + } + + lines := strings.Split(itemString, "\n") + chars = append(chars, '(', ' ') + chars = append(chars, []rune(lines[0])...) + for _, l := range lines[1:] { + chars = append(chars, '\n', ' ', ' ') + chars = append(chars, []rune(l)...) + } + + chars = append(chars, '\n', ' ', ' ', ')') + return chars +} + +func (d *sequenceDefinition) formatSequence(r *registry, o formatOptions) []rune { + var chars []rune + sep := []rune{' '} + if o.mode&formatPretty > 0 { + sep = []rune{'\n', ' ', ' '} + } + for i := range d.originalItems { if len(chars) > 0 { - chars = append(chars, ' ') + chars = append(chars, sep...) } - item := normalizeItemRange(d.originalItems[i]) - needsQuantifier := item.Min != 1 || item.Max != 1 - - itemDef := r.definition[item.Name] + min, max := normalizeItemRange(d.originalItems[i].Min, d.originalItems[i].Max) + needsQuantifier := min != 1 || max != 1 + itemDef := r.definition[d.originalItems[i].Name] isSymbol := itemDef.commitType()&userDefined != 0 - ch, isChoice := itemDef.(*choiceDefinition) isChoiceOfMultiple := isChoice && len(ch.options) > 1 - seq, isSequence := itemDef.(*sequenceDefinition) isSequenceOfMultiple := isSequence && len(seq.originalItems) > 1 && !seq.isCharSequence(r) - needsGrouping := isChoiceOfMultiple || isSequenceOfMultiple - if isSymbol { chars = append(chars, []rune(itemDef.nodeName())...) } else { - if needsGrouping { - chars = append(chars, '(') - } - - chars = append(chars, []rune(itemDef.format(r, f))...) - - if needsGrouping { - chars = append(chars, ')') - } + chars = append(chars, formatSequenceItem(r, itemDef, needsGrouping, o)...) } if !needsQuantifier { continue } - if item.Min == 0 && item.Max == 1 { + if min == 0 && max == 1 { chars = append(chars, '?') continue } - if item.Min == 0 && item.Max < 0 { + if min == 0 && max < 0 { chars = append(chars, '*') continue } - if item.Min == 1 && item.Max < 0 { + if min == 1 && max < 0 { chars = append(chars, '+') continue } chars = append(chars, '{') - - if item.Min == item.Max { - chars = append(chars, []rune(strconv.Itoa(item.Min))...) + if min == max { + chars = append(chars, []rune(strconv.Itoa(min))...) } else { - if item.Min > 0 { - chars = append(chars, []rune(strconv.Itoa(item.Min))...) + if min > 0 { + chars = append(chars, []rune(strconv.Itoa(min))...) } chars = append(chars, ',') - - if item.Max >= 0 { - chars = append(chars, []rune(strconv.Itoa(item.Max))...) + if max >= 0 { + chars = append(chars, []rune(strconv.Itoa(max))...) } } chars = append(chars, '}') } - return string(chars) + return chars +} + +func (d *sequenceDefinition) format(r *registry, o formatOptions) string { + if d.isCharSequence(r) { + return d.formatCharSequence(r, o) + } + + oo := o + oo.mode &^= formatPretty + f := d.formatSequence(r, oo) + if o.mode&formatPretty == 0 { + return string(f) + } + + if len(f) <= o.targetWidth { + return string(f) + } + + f = d.formatSequence(r, o) + return string(f) } func (p *sequenceParser) generate(w io.Writer, done map[string]bool) error { diff --git a/syntax.go b/syntax.go index 3a1efa2..f401bd7 100644 --- a/syntax.go +++ b/syntax.go @@ -20,6 +20,7 @@ import ( "errors" "fmt" "io" + "strings" ) // SequenceItem represents a single element within a sequence definition, referencing another parser by name. @@ -54,6 +55,7 @@ type SequenceItem struct { // // 3. Execution: Use Parse() to process input or Generate() to create Go source code. type Syntax struct { + ast *Node registry *registry initialized bool errInitFailed error @@ -91,7 +93,7 @@ type definition interface { addGeneralization(int) parser() parser builder() builder - format(*registry, formatFlags) string + format(*registry, formatOptions) string } var ( @@ -110,26 +112,44 @@ var ( ) func (ct CommitType) String() string { - switch ct { - case None: + if ct == None || ct == userDefined { return "none" - case Alias: - return "alias" - case Whitespace: - return "whitespace" - case NoWhitespace: - return "no-whitespace" - case Keyword: - return "keyword" - case NoKeyword: - return "no-keyword" - case FailPass: - return "fail-pass" - case Root: - return "root" - default: + } + + if ct&^(Alias|Whitespace|NoWhitespace|Keyword|NoKeyword|FailPass|Root|userDefined) != 0 { return "unknown" } + + var cts []string + if ct&Alias > 0 { + cts = append(cts, "alias") + } + + if ct&Whitespace > 0 { + cts = append(cts, "ws") + } + + if ct&NoWhitespace > 0 { + cts = append(cts, "nows") + } + + if ct&Keyword > 0 { + cts = append(cts, "kw") + } + + if ct&NoKeyword > 0 { + cts = append(cts, "nokw") + } + + if ct&FailPass > 0 { + cts = append(cts, "failpass") + } + + if ct&Root > 0 { + cts = append(cts, "root") + } + + return strings.Join(cts, ":") } func duplicateDefinition(name string) error { @@ -143,6 +163,10 @@ func parserNotFound(name string) error { var symbolChars = []rune("\\ \n\t\b\f\r\v/.[]\"{}^+*?|():=;") func isValidSymbol(n string) bool { + if len(n) == 0 { + return false + } + runes := []rune(n) for _, r := range runes { if !matchChar(symbolChars, nil, true, r) { @@ -357,6 +381,7 @@ func (s *Syntax) ReadSyntax(r io.Reader) error { } n := mapSelfNode(sn) + s.ast = n return define(s, n) } @@ -523,3 +548,12 @@ func (s *Syntax) Parse(r io.Reader) (*Node, error) { return parseInput(r, s.root.parser(), s.root.builder(), s.keywordParsers()) } + +// Format prints the loaded syntax definition to the output in a formatted way. +func (s *Syntax) Format(out io.Writer) error { + if s.ast == nil { + return formatDefinitions(out, s) + } + + return formatAST(out, s.ast) +} diff --git a/syntax.treerack b/syntax.treerack index 44d1551..6e25955 100644 --- a/syntax.treerack +++ b/syntax.treerack @@ -1,11 +1,19 @@ +/* + foo +*/ +// bar + // bar + //baz + /* foo + bar baz */// foo bar baz wschar:alias = " " | "\t" | "\n" | "\b" | "\f" | "\r" | "\v"; wsc:ws = wschar | comment; block-comment:alias:nows = "/*" ("*" [^/] | [^*])* "*/"; line-comment:alias:nows = "//" [^\n]*; comment-segment:alias:nows = line-comment | block-comment; -ws-no-nl:alias:nows = " " | "\t" | "\b" | "\f" | "\r" | "\v"; -comment:nows = comment-segment (ws-no-nl* "\n"? ws-no-nl* comment-segment)*; +ws-no-nl:alias:nows = " " | "\t" | "\b" | /* this one */ /* is a */ "\f" /* form feed */ /* for sure */ | "\r" | "\v"; +comment:nows = comment-segment /* segment is not the best name */ /* but */ (ws-no-nl* "\n"? ws-no-nl* /* fine */ comment-segment)*; any-char = "."; // equivalent to [^] diff --git a/syntaxhead.go b/syntaxhead.go index 10e4890..39bb351 100644 --- a/syntaxhead.go +++ b/syntaxhead.go @@ -53,6 +53,11 @@ const ( formatIncludeComments ) +type formatOptions struct { + mode formatFlags + targetWidth int +} + // ParseError reports a failure to match the input text against the defined syntax. type ParseError struct {