|
| 1 | +using System; |
| 2 | +using System.Collections.Generic; |
| 3 | +using System.IO; |
| 4 | +using System.Linq; |
| 5 | +using System.Xml.Linq; |
| 6 | + |
| 7 | +using Irony.Ast; |
| 8 | +using Irony.Parsing; |
| 9 | + |
| 10 | +namespace Java.Interop.Tools.JavaSource { |
| 11 | + |
| 12 | + using static IronyExtensions; |
| 13 | + |
| 14 | + public partial class SourceJavadocToXmldocGrammar { |
| 15 | + |
| 16 | + public class HtmlBnfTerms { |
| 17 | + internal HtmlBnfTerms () |
| 18 | + { |
| 19 | + } |
| 20 | + |
| 21 | + internal void CreateRules (SourceJavadocToXmldocGrammar grammar) |
| 22 | + { |
| 23 | + AllHtmlTerms.Rule = InlineDeclaration |
| 24 | + | PBlockDeclaration |
| 25 | + ; |
| 26 | + |
| 27 | + InlineDeclaration.Rule = ParsedCharacterData |
| 28 | + | FontStyleDeclaration |
| 29 | + /* |
| 30 | + | PhraseDeclaration |
| 31 | + | SpecialDeclaration |
| 32 | + | FormCtrlDeclaration |
| 33 | + */ |
| 34 | + | grammar.InlineTagsTerms.AllInlineTerms |
| 35 | + | UnknownHtmlElementStart |
| 36 | + ; |
| 37 | + InlineDeclarations.MakeStarRule (grammar, InlineDeclaration); |
| 38 | + |
| 39 | + var fontstyle_tt = new NonTerminal ("<tt>") { |
| 40 | + Rule = CreateStartElement ("tt", grammar) + InlineDeclarations + CreateEndElement ("tt", grammar), |
| 41 | + AstConfig = new AstNodeConfig { |
| 42 | + NodeCreator = (context, parseNode) => { |
| 43 | + parseNode.AstNode = new XElement ("c", |
| 44 | + parseNode.ChildNodes |
| 45 | + .Select (c => c.AstNode ?? "")); |
| 46 | + }, |
| 47 | + }, |
| 48 | + }; |
| 49 | + |
| 50 | + var fontstyle_i = new NonTerminal ("<i>", ConcatChildNodes) { |
| 51 | + Rule = CreateStartElement ("i", grammar) + InlineDeclarations + CreateEndElement ("i", grammar), |
| 52 | + }; |
| 53 | + |
| 54 | + FontStyleDeclaration.Rule = fontstyle_tt | fontstyle_i; |
| 55 | + |
| 56 | + PBlockDeclaration.Rule = |
| 57 | + CreateStartElement ("p", grammar) + InlineDeclarations + CreateEndElement ("p", grammar, optional:true) |
| 58 | + ; |
| 59 | + PBlockDeclaration.AstConfig.NodeCreator = (context, parseNode) => { |
| 60 | + parseNode.AstNode = new XElement ("para", |
| 61 | + parseNode.ChildNodes |
| 62 | + .Select (c => c.AstNode ?? "")); |
| 63 | + }; |
| 64 | + } |
| 65 | + |
| 66 | + public readonly NonTerminal AllHtmlTerms = new NonTerminal (nameof (AllHtmlTerms), ConcatChildNodes); |
| 67 | + |
| 68 | + |
| 69 | + // https://www.w3.org/TR/html401/struct/global.html#h-7.5.3 |
| 70 | +// public readonly Terminal ParsedCharacterData = new RegexBasedTerminal (nameof (ParsedCharacterData), "[^<{@}]*") { |
| 71 | +// public readonly Terminal ParsedCharacterData = new WikiTextTerminal (nameof (ParsedCharacterData)) {* |
| 72 | + public readonly Terminal ParsedCharacterData = new CharacterDataTerminal ("#PCDATA") { |
| 73 | +// Priority = TerminalPriority.Low, |
| 74 | + AstConfig = new AstNodeConfig { |
| 75 | + NodeCreator = (context, parseNode) => parseNode.AstNode = parseNode.Token.Value.ToString (), |
| 76 | + }, |
| 77 | + }; |
| 78 | + |
| 79 | + // https://www.w3.org/TR/html4/sgml/dtd.html#inline |
| 80 | + public readonly NonTerminal InlineDeclaration = new NonTerminal (nameof (InlineDeclaration), ConcatChildNodes); |
| 81 | + public readonly NonTerminal InlineDeclarations = new NonTerminal (nameof (InlineDeclarations), ConcatChildNodes); |
| 82 | + // https://www.w3.org/TR/html4/sgml/dtd.html#fontstyle |
| 83 | + public readonly NonTerminal FontStyleDeclaration = new NonTerminal (nameof (FontStyleDeclaration), ConcatChildNodes); |
| 84 | + // https://www.w3.org/TR/html4/sgml/dtd.html#phrase |
| 85 | + public readonly NonTerminal PhraseDeclaration = new NonTerminal (nameof (PhraseDeclaration), ConcatChildNodes); |
| 86 | + // https://www.w3.org/TR/html4/sgml/dtd.html#special |
| 87 | + public readonly NonTerminal SpecialDeclaration = new NonTerminal (nameof (SpecialDeclaration), ConcatChildNodes); |
| 88 | + // https://www.w3.org/TR/html4/sgml/dtd.html#formctrl |
| 89 | + public readonly NonTerminal FormCtrlDeclaration = new NonTerminal (nameof (FormCtrlDeclaration), ConcatChildNodes); |
| 90 | + // https://www.w3.org/TR/html4/sgml/dtd.html#block |
| 91 | + public readonly NonTerminal BlockDeclaration = new NonTerminal (nameof (BlockDeclaration), ConcatChildNodes); |
| 92 | + public readonly NonTerminal PBlockDeclaration = new NonTerminal (nameof (PBlockDeclaration), ConcatChildNodes); |
| 93 | + |
| 94 | + public readonly Terminal UnknownHtmlElementStart = new UnknownHtmlElementStartTerminal (nameof (UnknownHtmlElementStart)) { |
| 95 | + AstConfig = new AstNodeConfig { |
| 96 | + NodeCreator = (context, parseNode) => parseNode.AstNode = parseNode.Token.Value.ToString (), |
| 97 | + }, |
| 98 | + }; |
| 99 | + |
| 100 | + static NonTerminal CreateStartElement (string startElement, Grammar grammar) |
| 101 | + { |
| 102 | + var start = new NonTerminal ("<" + startElement + ">", nodeCreator: (context, parseNode) => parseNode.AstNode = "") { |
| 103 | + Rule = grammar.ToTerm ("<" + startElement + ">") | "<" + startElement.ToUpperInvariant () + ">", |
| 104 | + }; |
| 105 | + return start; |
| 106 | + } |
| 107 | + |
| 108 | + static NonTerminal CreateEndElement (string endElement, Grammar grammar, bool optional = false) |
| 109 | + { |
| 110 | + var end = new NonTerminal (endElement, nodeCreator: (context, parseNode) => parseNode.AstNode = "") { |
| 111 | + Rule = grammar.ToTerm ("</" + endElement + ">") | "</" + endElement.ToUpperInvariant () + ">", |
| 112 | + }; |
| 113 | + if (optional) { |
| 114 | + end.Rule |= grammar.Empty; |
| 115 | + } |
| 116 | + return end; |
| 117 | + } |
| 118 | + } |
| 119 | + } |
| 120 | + |
| 121 | + // Based in part on WikiTextTerminal |
| 122 | + class CharacterDataTerminal : Terminal { |
| 123 | + |
| 124 | + private char[] _stopChars; |
| 125 | + |
| 126 | + public CharacterDataTerminal (string name) |
| 127 | + : base (name) |
| 128 | + { |
| 129 | + base.Priority = TerminalPriority.Low; |
| 130 | + } |
| 131 | + |
| 132 | + public override void Init (GrammarData grammarData) |
| 133 | + { |
| 134 | + base.Init (grammarData); |
| 135 | + var stopCharSet = new Irony.CharHashSet (); |
| 136 | + foreach(var term in grammarData.Terminals) { |
| 137 | + var firsts = term.GetFirsts (); |
| 138 | + if (firsts == null) |
| 139 | + continue; |
| 140 | + foreach (var first in firsts) { |
| 141 | + if (string.IsNullOrEmpty (first)) |
| 142 | + continue; |
| 143 | + stopCharSet.Add (first [0]); |
| 144 | + } |
| 145 | + } |
| 146 | + _stopChars = stopCharSet.ToArray(); |
| 147 | + } |
| 148 | + |
| 149 | + public override Token TryMatch (ParsingContext context, ISourceStream source) |
| 150 | + { |
| 151 | + var stopIndex = source.Text.IndexOfAny (_stopChars, source.Location.Position); |
| 152 | + if (stopIndex == source.Location.Position) |
| 153 | + return null; |
| 154 | + if (stopIndex < 0) |
| 155 | + stopIndex = source.Text.Length; |
| 156 | + source.PreviewPosition = stopIndex; |
| 157 | + |
| 158 | + // preserve leading whitespace, if present. |
| 159 | + int start = source.Location.Position; |
| 160 | + while (start > 0 && char.IsWhiteSpace (source.Text, start-1)) { |
| 161 | + start--; |
| 162 | + } |
| 163 | + var content = source.Text.Substring (start, stopIndex - start); |
| 164 | + |
| 165 | + return source.CreateToken (this.OutputTerminal, content); |
| 166 | + } |
| 167 | + } |
| 168 | + |
| 169 | + class UnknownHtmlElementStartTerminal : Terminal { |
| 170 | + |
| 171 | + public UnknownHtmlElementStartTerminal (string name) |
| 172 | + : base (name) |
| 173 | + { |
| 174 | + base.Priority = TerminalPriority.Low-1; |
| 175 | + } |
| 176 | + |
| 177 | + public override void Init (GrammarData grammarData) |
| 178 | + { |
| 179 | + base.Init (grammarData); |
| 180 | + } |
| 181 | + |
| 182 | + public override Token TryMatch (ParsingContext context, ISourceStream source) |
| 183 | + { |
| 184 | + if (source.Text [source.Location.Position] != '<') |
| 185 | + return null; |
| 186 | + source.PreviewPosition += 1; |
| 187 | + return source.CreateToken (this.OutputTerminal, "<"); |
| 188 | + } |
| 189 | + } |
| 190 | +} |
0 commit comments