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 export 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 function applyPartsWithoutImport(parsed: [Token<TokenKind>, tkTree]):tkTree {
 103 return ["%clo", "" , parsed[1]];
 104 }
 105
 106
 107 function applyComment(value: Token<TokenKind.Comment>): tkTree[]{
 108     return [value.text];
 109 }
 110
 111
 112 function applyImport(input: [Token<TokenKind>,Token<TokenKind>[], tkTree]) : tkTree{
 113     let importTail = input[1].map(x=>x.text);
 114     return ["import"].concat(importTail);
 115 };
 116
 117
 118 /*
 119 function applyImportComment(input: [Token<TokenKind>,Token<TokenKind>[],
 120     tkTree, Token<TokenKind.Comment>]) : tkTree{
 121     let importTail = input[1].map(x=>x.text);
 122     let comment = [input[3].text];
 123     return ["import"].concat(importTail).concat(comment);
 124 };*/
 125
 126 function applyImports(input : [tkTree, tkTree[]]): tkTree{
 127     let resultBody = [input[0]].concat(input[1]);
 128     let resultWrapper = ["%import", resultBody];
 129     return resultWrapper;
 130 };
 131
 132
 133
 134
 135 function applyNotAtText(value : Token<TokenKind>): tkTree{
 136     if (value.text == "\\\@"){
 137         return '@';
 138     }
 139     else{return value.text;}
 140 };
 141
 142 function applyText (input : tkTree): tkTree[]{
 143     return ["%text", input];
 144 };
 145
 146 function applyContent(input : tkTree[]): tkTree[]{
 147     return ["%content", input];
 148 };
 149
 150 function applySpaceNL(value : Token<TokenKind.SpaceNL>): tkTree{
 151     return value.text;
 152 }
 153
 154 /**
 155  * IMPORTEE:  Number, Op, Paren, Id, Str, Comment,
 156  */
 157 let IMPORTEE = p.alt(p.tok(TokenKind.Number),
 158     p.tok(TokenKind.Op),
 159     p.tok(TokenKind.Paren),
 160     p.tok(TokenKind.Id),
 161     p.tok(TokenKind.Str),
 162     p.tok(TokenKind.SpaceNL),
 163     p.tok(TokenKind.Comment));
 164
 165 let NOT_AT = p.alt(p.tok(TokenKind.Seperator),
 166     p.tok(TokenKind.Semicolon),
 167     p.tok(TokenKind.Number),
 168     p.tok(TokenKind.ExcapeAt),
 169     p.tok(TokenKind.Op),
 170     p.tok(TokenKind.Paren),
 171     p.tok(TokenKind.SpaceNL),
 172     p.tok(TokenKind.Id),
 173     p.tok(TokenKind.Str),
 174     p.tok(TokenKind.Comment),
 175     );
 176
 177 /**
 178  * PROG : IMPORTS '---' CONTENT | '---' CONTNENT
 179  */
 180 PROG.setPattern(
 181     p.alt(
 182         p.lrec_sc(IMPORTS, p.seq(p.str('---'), CONTENT), applyParts),
 183         p.apply(p.seq(p.str('---'), CONTENT), applyPartsWithoutImport))
 184
 185 )
 186
 187 /**
 188  * NOT_AT_TEXT : NOT_AT
 189  */
 190 NOT_AT_TEXT.setPattern(
 191     p.apply(NOT_AT, applyNotAtText)
 192 );
 193
 194 IMPORTS.setPattern(
 195     p.apply( p.seq(IMPORT, p.rep(IMPORT)), applyImports)
 196 );
 197
 198 /**
 199  * IMPORT :
 200  * 'import' IMPORTEE* SEMICOLON |
 201  * COMMENT |
 202  */
 203 IMPORT.setPattern(
 204     p.alt(
 205         p.apply(p.seq(p.str('import'), p.rep_sc(IMPORTEE), SEMICOLON),
 206             applyImport),
 207         p.apply(p.tok(TokenKind.Comment), applyComment),
 208         p.apply(p.tok(TokenKind.SpaceNL), applySpaceNL)
 209
 210     )
 211 );
 212
 213 /**
 214  * SEMICOLON : ';';
 215  */
 216 SEMICOLON.setPattern(
 217     p.apply(p.tok(TokenKind.Semicolon), applySemiColon)
 218 );
 219
 220
 221
 222 /**
 223  * SEGMENT : '@' NOT_AT* '@' |
 224  * (NOT_AT_TEXT | EXCAPE_AT)*
 225  */
 226 SEGMENT.setPattern(
 227     p.alt(
 228         p.apply(p.rep_sc(NOT_AT_TEXT), applyText),
 229         p.apply(p.seq(p.str('@'), p.rep(NOT_AT), p.str('@')), applySegment),
 230     )
 231 );
 232
 233 /**
 234  * CONTENT : SEGMENT*
 235  */
 236 CONTENT.setPattern(
 237     p.apply(p.rep(SEGMENT), applyContent)
 238 );
 239
 240
 241
 242 /**
 243  * the head part of the output JS code : before import
 244  */
 245 let outputHead = `
 246 /* clo, a typesetting engine, generated JS file*/
 247 /* CLO:  beginning of head*/
 248
 249 let cloLib = require("./src/libclo/index.js");
 250 let clo = new cloLib.Clo();
 251
 252 /* CLO:  end of head*/\n`
 253
 254 /**
 255  * the middle part of the output JS code : between import part and content part
 256  */
 257 let outputMiddle =`
 258 /* CLO:  beginning of middle part*/
 259 clo.mainStream = /* CLO: end of middle part*/
 260 `
 261 let outputEnd =`
 262 /* CLO: beginning of end part*/
 263 clo.generatePdf();
 264 /*CLO : end of end part*/
 265 `
 266
 267 /**
 268  * Convert `tree` (ASTTree; `tkTree`) to JS Code.
 269  */
 270 export function treeToJS(tree : tkTree): string{
 271
 272     let head = tree[0];
 273     if (head == "%clo"){
 274         let totalResult = outputHead + treeToJS(tree[1]) +
 275             outputMiddle + treeToJS(tree[2]) + outputEnd;
 276         return totalResult;
 277     }
 278     if (head == "%import"){
 279         let imports = tree[1];
 280         if (Array.isArray(imports)){
 281             let importsText = imports.map(
 282                 (x)=>{
 283                     if (Array.isArray(x)){
 284                         return x.join('') + ';';
 285                     }
 286                     else{
 287                         return x;
 288                     }
 289                 });
 290             let importTextCombined = importsText.join('');
 291             return importTextCombined;
 292         }
 293         else{
 294             return imports;
 295         }
 296     }
 297     if (head == "%content"){
 298         let tail = tree[1];
 299         if (Array.isArray(tail)){
 300             if (tail.length == 1){
 301                 return tail.map((x)=>treeToJS(x)).join("').concat('")+ ";";
 302             }
 303             let tailStrings = tail.map((x)=>treeToJS(x));
 304             return "(" + tailStrings.join(').concat(') + ");";
 305         }else{
 306             return tail;
 307         }
 308     }
 309     if (head == "%text"){
 310         let textContents = tree[1];
 311         if (Array.isArray(textContents)){
 312             let decoratedArray = textContents
 313                                 .flatMap(x=>String(x))
 314                                 .map(x=>x.replace("\`","\\\`"));
 315
 316             return "[`" + decoratedArray.join("\`, \`") + "`]";
 317         }else{
 318             let decorated = textContents.replace("\`","\\\`");
 319
 320             return "[`" + decorated + "`]";
 321         }
 322     }
 323
 324     if (head == "%exprs"){
 325         let content = tree[1];
 326         if (Array.isArray(content)){
 327             let flattenContent = content.flat();
 328             return flattenContent.join('');
 329         }
 330         else{
 331             return content;
 332         }
 333
 334     }
 335     else{
 336         if (Array.isArray(tree)){
 337             return tree.join('');
 338         }else{
 339             return tree;
 340         }
 341     }
 342 }
 343
 344
 345 /**
 346  * `inputText` to `tkTree` (ASTTree)
 347  */
 348 export function inputTextToTree(inputText : string){
 349 return p.expectSingleResult(
 350     p.expectEOF(PROG.parse(lexer.parse(inputText))));
 351 }