git.kianting.info Git - clo/blob - src/parser.ts

   1 /**
   2  * parser.ts - parser and js generator of clo.
   3  */
   4 import * as p from 'typescript-parsec';
   5 import { Token } from 'typescript-parsec';
   6
   7 /**
   8  *
   9  * # REPRESENTATION
  10  */
  11
  12 /**
  13  * convert a `tkTree` AST to S-expr string
  14  * @param t the `tkTree`
  15  * @returns S-expr String
  16  *
  17 export function tkTreeToSExp(t: tkTree): string{
  18     var str = "";
  19
  20     if (Array.isArray(t)){
  21         let strArray = t.map((x)=>tkTreeToSExp(x));
  22         str = "(" + strArray.join("◎") + ")";
  23     }else{
  24         if (t=== undefined){
  25             str = "%undefined"
  26         }else{
  27             str = t;
  28         }
  29     }
  30
  31     return str;
  32 }*/
  33
  34 type tkTree = string | tkTree[];
  35
  36 enum TokenKind {
  37     Seperator, // ---
  38     Semicolon, // ;
  39     Number,
  40     Op,
  41     ExprMark, // @
  42     ExcapeAt, // \@
  43     Paren,
  44     SpaceNL, // \s\t\n\r
  45     Id,
  46     Str,
  47     Comment, // /* ooo */
  48 }
  49
  50 /**
  51  * Parsing
  52  */
  53 const lexer = p.buildLexer([
  54     [true, /^\d+(\.\d+)?/g, TokenKind.Number],
  55     [true, /^[\\][\\]/g, TokenKind.Op],
  56     [true, /^\\\@/g, TokenKind.ExcapeAt],
  57     [true, /^\/\*([^/]|\/[^*])*\*\//g, TokenKind.Comment],
  58     [true, /^\;/g, TokenKind.Semicolon],
  59     [true, /^[-][-][-]/g, TokenKind.Seperator],
  60     [true, /^[\+\-\*\/\&\|\!\^\<\>\~\=\?]+/g, TokenKind.Op],
  61     [true, /^\@/g, TokenKind.ExprMark],
  62     [true, /^[()\[\]{}]/g, TokenKind.Paren],
  63     [true, /^[\"]([^\"]|[\\].)*[\"]/g, TokenKind.Str],
  64     [true, /^[\']([^\']|[\\].)*[\']/g, TokenKind.Str],
  65     [true, /^[()\[\]{}]/g, TokenKind.Paren],
  66     [true, /^[^\/\\\@\s\n\t\r;]+/g, TokenKind.Id],
  67     [true, /^(\s|\n|\r|\t)+/g, TokenKind.SpaceNL],
  68
  69 ]);
  70
  71 /**
  72  *
  73  * # TEST
  74  */
  75
  76
  77
  78 const PROG = p.rule<TokenKind, tkTree>();
  79 const SEGMENT = p.rule<TokenKind, tkTree>();
  80 const IMPORT = p.rule<TokenKind, tkTree>();
  81 const IMPORTS = p.rule<TokenKind, tkTree>();
  82 const SEMICOLON = p.rule<TokenKind, tkTree>();
  83 const NOT_AT_TEXT = p.rule<TokenKind, tkTree>();
  84 const CONTENT = p.rule<TokenKind, tkTree>();
  85
  86
  87 function applySegment(input: [Token<TokenKind>, Token<TokenKind>[],
  88         Token<TokenKind>]): tkTree[]{
  89     let unpackedInnerExprs = input[1].map((x)=>{return x.text});
  90     return ["%exprs", unpackedInnerExprs];
  91 }
  92
  93 function applySemiColon(value: Token<TokenKind.Semicolon>): tkTree{
  94     return value.text;
  95 }
  96
  97 function applyParts(first: tkTree,
  98                     second: [Token<TokenKind>, tkTree]):tkTree {
  99     return ["%clo", first , second[1]];
 100 }
 101
 102
 103 function applyComment(value: Token<TokenKind.Comment>): tkTree[]{
 104     return [value.text];
 105 }
 106
 107
 108 function applyImport(input: [Token<TokenKind>,Token<TokenKind>[], tkTree]) : tkTree{
 109     let importTail = input[1].map(x=>x.text);
 110     return ["import"].concat(importTail);
 111 };
 112
 113
 114 /*
 115 function applyImportComment(input: [Token<TokenKind>,Token<TokenKind>[],
 116     tkTree, Token<TokenKind.Comment>]) : tkTree{
 117     let importTail = input[1].map(x=>x.text);
 118     let comment = [input[3].text];
 119     return ["import"].concat(importTail).concat(comment);
 120 };*/
 121
 122 function applyImports(input : [tkTree, tkTree[]]): tkTree{
 123     let resultBody = [input[0]].concat(input[1]);
 124     let resultWrapper = ["%import", resultBody];
 125     return resultWrapper;
 126 };
 127
 128
 129
 130
 131 function applyNotAtText(value : Token<TokenKind>): tkTree{
 132     if (value.text == "\\\@"){
 133         return '@';
 134     }
 135     else{return value.text;}
 136 };
 137
 138 function applyText (input : tkTree): tkTree[]{
 139     return ["%text", input];
 140 };
 141
 142 function applyContent(input : tkTree[]): tkTree[]{
 143     return ["%content", input];
 144 };
 145
 146 function applySpaceNL(value : Token<TokenKind.SpaceNL>): tkTree{
 147     return value.text;
 148 }
 149
 150 /**
 151  * IMPORTEE:  Number, Op, Paren, Id, Str, Comment,
 152  */
 153 let IMPORTEE = p.alt(p.tok(TokenKind.Number),
 154     p.tok(TokenKind.Op),
 155     p.tok(TokenKind.Paren),
 156     p.tok(TokenKind.Id),
 157     p.tok(TokenKind.Str),
 158     p.tok(TokenKind.SpaceNL),
 159     p.tok(TokenKind.Comment));
 160
 161 let NOT_AT = p.alt(p.tok(TokenKind.Seperator),
 162     p.tok(TokenKind.Semicolon),
 163     p.tok(TokenKind.Number),
 164     p.tok(TokenKind.ExcapeAt),
 165     p.tok(TokenKind.Op),
 166     p.tok(TokenKind.Paren),
 167     p.tok(TokenKind.SpaceNL),
 168     p.tok(TokenKind.Id),
 169     p.tok(TokenKind.Str),
 170     p.tok(TokenKind.Comment),
 171     );
 172
 173 /**
 174  * PROG : IMPORTS '---' CONTENT;
 175  */
 176 PROG.setPattern(
 177     p.lrec_sc(IMPORTS, p.seq(p.str('---'), CONTENT), applyParts)
 178
 179 )
 180
 181 /**
 182  * NOT_AT_TEXT : NOT_AT
 183  */
 184 NOT_AT_TEXT.setPattern(
 185     p.apply(NOT_AT, applyNotAtText)
 186 );
 187
 188 IMPORTS.setPattern(
 189     p.apply( p.seq(IMPORT, p.rep(IMPORT)), applyImports)
 190 );
 191
 192 /**
 193  * IMPORT :
 194  * 'import' IMPORTEE* SEMICOLON |
 195  * COMMENT |
 196  */
 197 IMPORT.setPattern(
 198     p.alt(
 199         p.apply(p.seq(p.str('import'), p.rep_sc(IMPORTEE), SEMICOLON),
 200             applyImport),
 201         p.apply(p.tok(TokenKind.Comment), applyComment),
 202         p.apply(p.tok(TokenKind.SpaceNL), applySpaceNL)
 203
 204     )
 205 );
 206
 207 /**
 208  * SEMICOLON : ';';
 209  */
 210 SEMICOLON.setPattern(
 211     p.apply(p.tok(TokenKind.Semicolon), applySemiColon)
 212 );
 213
 214
 215
 216 /**
 217  * SEGMENT : '@' NOT_AT* '@' |
 218  * (NOT_AT_TEXT | EXCAPE_AT)*
 219  */
 220 SEGMENT.setPattern(
 221     p.alt(
 222         p.apply(p.rep_sc(NOT_AT_TEXT), applyText),
 223         p.apply(p.seq(p.str('@'), p.rep(NOT_AT), p.str('@')), applySegment),
 224     )
 225 );
 226
 227 /**
 228  * CONTENT : SEGMENT*
 229  */
 230 CONTENT.setPattern(
 231     p.apply(p.rep(SEGMENT), applyContent)
 232 );
 233
 234
 235
 236 /**
 237  * the head part of the output JS code : before import
 238  */
 239 let outputHead = `
 240 /* clo, a typesetting engine, generated JS file*/
 241 /* CLO:  beginning of head*/
 242 import * as clo from "clo";
 243
 244 cl = clo.initClo();
 245 /* CLO:  end of head*/\n`
 246
 247 /**
 248  * the middle part of the output JS code : between import part and content part
 249  */
 250 let outputMiddle =`
 251 /* CLO:  beginning of middle part*/
 252 cl.mainText = /* CLO: end of middle part*/
 253 `
 254 let outputEnd =`
 255 /* CLO: beginning of end part*/
 256 cl.generatePdf();
 257 /*CLO : end of end part*/
 258 `
 259
 260 /**
 261  * Convert `tree` (ASTTree; `tkTree`) to JS Code.
 262  */
 263 export function treeToJS(tree : tkTree): string{
 264     let head = tree[0];
 265     if (head == "%clo"){
 266         let totalResult = outputHead + treeToJS(tree[1]) +
 267             outputMiddle + treeToJS(tree[2]) + outputEnd;
 268         return totalResult;
 269     }
 270     if (head == "%import"){
 271         let imports = tree[1];
 272         if (Array.isArray(imports)){
 273             let importsText = imports.map(
 274                 (x)=>{
 275                     if (Array.isArray(x)){
 276                         return x.join('') + ';';
 277                     }
 278                     else{
 279                         return x;
 280                     }
 281                 });
 282             let importTextCombined = importsText.join('');
 283             return importTextCombined;
 284         }
 285         else{
 286             return imports;
 287         }
 288     }
 289     if (head == "%content"){
 290         let tail = tree[1];
 291         if (Array.isArray(tail)){
 292             if (tail.length == 1){
 293                 return treeToJS(tail);
 294             }
 295             let tailStrings = tail.map((x)=>treeToJS(x));
 296             return "(" + tailStrings.join(').concat(') + ");";
 297         }else{
 298             return tail;
 299         }
 300     }
 301     if (head == "%text"){
 302         let textContents = tree[1];
 303         if (Array.isArray(textContents)){
 304             let decoratedArray = textContents
 305                                 .flatMap(x=>String(x))
 306                                 .map(x=>x.replace("\`","\\\`"));
 307
 308             return "[`" + decoratedArray.join("\`, \`") + "`]";
 309         }else{
 310             let decorated = textContents.replace("\`","\\\`");
 311
 312             return "[`" + decorated + "`]";
 313         }
 314     }
 315
 316     if (head == "%exprs"){
 317         let content = tree[1];
 318         if (Array.isArray(content)){
 319             let flattenContent = content.flat();
 320             return flattenContent.join('');
 321         }
 322         else{
 323             return content;
 324         }
 325
 326     }
 327     else{
 328         if (Array.isArray(tree)){
 329             return tree.join('');
 330         }else{
 331             return tree;
 332         }
 333     }
 334 }
 335
 336
 337 /**
 338  * `inputText` to `tkTree` (ASTTree)
 339  */
 340 export function inputTextToTree(inputText : string){
 341 return p.expectSingleResult(
 342     p.expectEOF(PROG.parse(lexer.parse(inputText))));
 343 }