NT2chCtrl45/html/HtmlElement_2chparser.cs

   1 using System;\r
   2 using System.Collections.Generic;\r
   3 using System.Text;\r
   4 using NT2chObject;\r
   5 \r
   6 namespace NT2chCtrl.html\r
   7 {\r
   8     public partial class HtmlElement\r
   9     {\r
  10         private bool parse2chRes(string source, IAmbiguousSearch search)\r
  11         {\r
  12             //int idx1 = 0;\r
  13             int idx2 = 0;\r
  14             int idx3, idx4;\r
  15             int length = source.Length;\r
  16             int textStart = 0;\r
  17             //StringElement sElem = null;\r
  18             HtmlElement hElem = null;\r
  19 \r
  20             //StringBuilder sb = new StringBuilder(source.Length);\r
  21 \r
  22             for (int idx1 = 0; idx1 < length; idx1++)\r
  23             {\r
  24                 idx2 = source.IndexOf('<', idx1);\r
  25                 if (idx2 < 0)\r
  26                     break;\r
  27                 if (idx2 < length - 1)\r
  28                 {\r
  29 \r
  30                     switch (source[idx2 + 1])\r
  31                     {\r
  32                         case 'a':\r
  33                         case 'A':\r
  34                             idx3 = source.IndexOf(" href=", idx2 + 2, StringComparison.OrdinalIgnoreCase);\r
  35                             if (idx3 == idx2 + 2)\r
  36                             {\r
  37                                 idx4 = source.IndexOf(">", idx3 + 6, StringComparison.OrdinalIgnoreCase);\r
  38                                 if (idx4 > 0)\r
  39                                 {\r
  40                                     //sElem = new StringElement(this, source.Substring(textStart, idx2 - textStart));\r
  41                                     parse2chRes2(this, source, textStart, idx2, search);\r
  42                                     textStart = idx4 + 1;\r
  43                                     idx1 = textStart - 1;\r
  44                                 }\r
  45                             }\r
  46                             break;\r
  47                         case 'b':\r
  48                         case 'B':\r
  49                             idx3 = source.IndexOf("r", idx2 + 2, StringComparison.OrdinalIgnoreCase);\r
  50                             if (idx3 == idx2 + 2)\r
  51                             {\r
  52                                 idx4 = source.IndexOf(">", idx3 + 1, StringComparison.OrdinalIgnoreCase);\r
  53                                 if (idx4 > 0)\r
  54                                 {\r
  55                                     //sElem = new StringElement(this, source.Substring(textStart, idx2 - textStart));\r
  56                                     parse2chRes2(this, source, textStart, idx2, search);\r
  57                                     hElem = new HtmlElement(this, "br");\r
  58                                     hElem.setClosed(true);\r
  59                                     textStart = idx4 + 1;\r
  60                                     idx1 = textStart - 1;\r
  61                                 }\r
  62                             }\r
  63                             break;\r
  64                         case '/':\r
  65                             idx3 = source.IndexOf("a>", idx2 + 2, StringComparison.OrdinalIgnoreCase);\r
  66                             if (idx3 == idx2 + 2)\r
  67                             {\r
  68                                     //sElem = new StringElement(this, source.Substring(textStart, idx2 - textStart));\r
  69                                     parse2chRes2(this, source, textStart, idx2, search);\r
  70                                     textStart = idx3 + 2;\r
  71                                     idx1 = textStart - 1;\r
  72                              }\r
  73                             break;\r
  74                         default:\r
  75                             break;\r
  76                     }\r
  77                 }\r
  78                 else\r
  79                 {\r
  80                     break;\r
  81                 }\r
  82 \r
  83             } //while (textStart < length);\r
  84 \r
  85             if (textStart < length)\r
  86             {\r
  87                 //sElem = new StringElement(this, source.Substring(textStart, length - textStart));\r
  88                 parse2chRes2(this, source, textStart, length, search);\r
  89             }\r
  90             return true;\r
  91         }\r
  92 \r
  93         private static int findNonDisplayAscii(string source, int start, int end)\r
  94         {\r
  95             int length = source.Length;\r
  96             length = Math.Min(length, end);\r
  97             for (int i = start; i < length; i++)\r
  98             {\r
  99                 char c = source[i];\r
 100                 if(c <= ' ' || c >= 0x7f){\r
 101                     return i;\r
 102                 }\r
 103             }\r
 104             return length;\r
 105         }\r
 106 \r
 107         private bool parseEmphasisText(HtmlElement pElem, string source, int start, int end, IAmbiguousSearch search)\r
 108         {\r
 109             HtmlElement hElem;\r
 110             HtmlAttribute hAttr;\r
 111             StringElement sElem;\r
 112 \r
 113             int textLength = end - start;\r
 114 \r
 115             string text = source.Substring(start, textLength);\r
 116 \r
 117             if (search == null)\r
 118             {\r
 119                 sElem = new StringElement(pElem, text);\r
 120                 return true;\r
 121             }\r
 122             int idx = 0;\r
 123             int findIdx, endIdx;\r
 124             do{\r
 125                 if (!search.match(text, idx, out findIdx, out endIdx))\r
 126                     break;\r
 127                 if (idx < findIdx)\r
 128                 {\r
 129                     sElem = new StringElement(pElem,\r
 130                                     text.Substring(idx, findIdx - idx));\r
 131                 }\r
 132                 hElem = new HtmlElement(pElem, "span");\r
 133                 hAttr = new HtmlAttribute("emphasis");\r
 134                 hElem.addAttribute(hAttr);\r
 135                 sElem = new StringElement(hElem,\r
 136                                 text.Substring(findIdx, endIdx - findIdx));\r
 137                 hElem.setClosed(true);\r
 138                 idx = endIdx;\r
 139             }while(idx < textLength);\r
 140 \r
 141             if (idx < textLength)\r
 142             {\r
 143                 sElem = new StringElement(pElem,\r
 144                                 text.Substring(idx, textLength - idx));\r
 145             }\r
 146             return true;\r
 147         }\r
 148 \r
 149         private bool parse2chRes2(HtmlElement pElem, string source, int start, int end, IAmbiguousSearch search)\r
 150         {\r
 151             bool bRet = false;\r
 152             int textStart = start;\r
 153             int textEnd = 0;\r
 154             int state = 0;\r
 155             int idx1, idx2;\r
 156             StringElement sElem;\r
 157             HtmlElement hElem;\r
 158             HtmlAttribute hAttr;\r
 159             string sLink;\r
 160             string prefix;\r
 161 \r
 162             for (int i = start; i < end; i++)\r
 163             {\r
 164                 switch (source[i])\r
 165                 {\r
 166                     case '>':\r
 167                     case '＞':\r
 168                         if (state == 0)\r
 169                         {\r
 170                             textEnd = i;\r
 171                             state = 1;\r
 172                         }\r
 173                         else if (state == 1)\r
 174                         {\r
 175                             textEnd = i - 1;\r
 176                         }\r
 177                         break;\r
 178                     case '0':\r
 179                     case '1':\r
 180                     case '2':\r
 181                     case '3':\r
 182                     case '4':\r
 183                     case '5':\r
 184                     case '6':\r
 185                     case '7':\r
 186                     case '8':\r
 187                     case '9':\r
 188                         if (state == 1 || state == 2 || state == 3)\r
 189                             state = 2;\r
 190                         break;\r
 191                     case '-':\r
 192                     case ',':\r
 193                         if (state == 2)\r
 194                             state = 3;\r
 195                         else\r
 196                             state = 0;\r
 197                         break;\r
 198                     case 'h':\r
 199                     case 'H':\r
 200                     case 't':\r
 201                     case 'T':\r
 202                         if (state == 2 || state == 3)\r
 203                         {\r
 204 \r
 205                             //sElem = new StringElement(this,\r
 206                             //    source.Substring(textStart, textEnd - textStart));\r
 207                             parseEmphasisText(this, source, textStart, textEnd, search);\r
 208                             hElem = new HtmlElement(this, "span");\r
 209                             hAttr = new HtmlAttribute("res-link");\r
 210                             hElem.addAttribute(hAttr);\r
 211                             hElem.setClosed(true);\r
 212                             if (state == 3)\r
 213                                 textStart = i - 2;\r
 214                             else if (state == 2)\r
 215                                 textStart = i - 1;\r
 216                             sElem = new StringElement(hElem, source.Substring(textEnd, textStart - textEnd));\r
 217                         }\r
 218                         idx1 = source.IndexOf("ttp://", i, StringComparison.OrdinalIgnoreCase);\r
 219                         if (idx1 == i || idx1 == (i + 1))\r
 220                         {\r
 221                             if (idx1 == i)\r
 222                                 prefix = "h";\r
 223                             else\r
 224                                 prefix = string.Empty;\r
 225                             idx2 = findNonDisplayAscii(source, idx1 + 6, end);\r
 226                             sLink = source.Substring(i, idx2 - i);\r
 227                             if (chkGraphicLink(sLink))\r
 228                             {\r
 229                                 hElem = new HtmlElement(this, "img");\r
 230                                 hAttr = new HtmlAttribute("src", sLink);\r
 231                                 hElem.addAttribute(hAttr);\r
 232                                 hElem.setClosed(true);\r
 233                             }\r
 234                             hElem = new HtmlElement(this, "a");\r
 235                             hAttr = new HtmlAttribute("href", prefix+sLink);\r
 236                             hElem.addAttribute(hAttr);\r
 237                             hElem.setClosed(true);\r
 238                             sElem = new StringElement(hElem, sLink);\r
 239                             textStart = idx2;\r
 240                             i = idx2 - 1;\r
 241                         }\r
 242                         else\r
 243                         {\r
 244                             idx1 = source.IndexOf("ttps://", i, StringComparison.OrdinalIgnoreCase);\r
 245                             if (idx1 == i || idx1 == (i + 1))\r
 246                             {\r
 247                                 if (idx1 == i)\r
 248                                     prefix = "h";\r
 249                                 else\r
 250                                     prefix = string.Empty;\r
 251                                 idx2 = findNonDisplayAscii(source, idx1 + 7, end);\r
 252                                 sLink = source.Substring(i, idx2 - i);\r
 253                                 if (chkGraphicLink(sLink))\r
 254                                 {\r
 255                                     hElem = new HtmlElement(this, "img");\r
 256                                     hAttr = new HtmlAttribute("src", sLink);\r
 257                                     hElem.addAttribute(hAttr);\r
 258                                     hElem.setClosed(true);\r
 259                                 }\r
 260                                 hElem = new HtmlElement(this, "a");\r
 261                                 hAttr = new HtmlAttribute("href", prefix+sLink);\r
 262                                 hElem.addAttribute(hAttr);\r
 263                                 hElem.setClosed(true);\r
 264                                 sElem = new StringElement(hElem, sLink);\r
 265                                 textStart = idx2;\r
 266                                 i = idx2 - 1;\r
 267                             }\r
 268                             else\r
 269                             {\r
 270                             }\r
 271                         }\r
 272                         state = 0;\r
 273                         break;\r
 274                     case 's':\r
 275                         if (state == 2 || state == 3)\r
 276                         {\r
 277 \r
 278                             //sElem = new StringElement(this,\r
 279                             //    source.Substring(textStart, textEnd - textStart));\r
 280                             parseEmphasisText(this, source, textStart, textEnd, search);\r
 281                             hElem = new HtmlElement(this, "span");\r
 282                             hAttr = new HtmlAttribute("res-link");\r
 283                             hElem.addAttribute(hAttr);\r
 284                             hElem.setClosed(true);\r
 285                             if (state == 3)\r
 286                                 textStart = i - 2;\r
 287                             else if (state == 2)\r
 288                                 textStart = i - 1;\r
 289                             sElem = new StringElement(hElem, source.Substring(textEnd, textStart - textEnd));\r
 290                             state = 0;\r
 291                         }\r
 292                         idx1 = source.IndexOf("sssp://", i, StringComparison.OrdinalIgnoreCase);\r
 293                         if (idx1 == i || idx1 == (i + 1))\r
 294                         {\r
 295                             idx2 = findNonDisplayAscii(source, idx1 + 7, end);\r
 296                             sLink = "http" + source.Substring(i+4, idx2 - i - 4);\r
 297                             if (chkGraphicLink(sLink))\r
 298                             {\r
 299                                 hElem = new HtmlElement(this, "img");\r
 300                                 hAttr = new HtmlAttribute("src", sLink);\r
 301                                 hElem.addAttribute(hAttr);\r
 302                                 hAttr = new HtmlAttribute("sssp");\r
 303                                 hElem.addAttribute(hAttr);\r
 304                                 hElem.setClosed(true);\r
 305                                 textStart = idx2;\r
 306                                 i = idx2 - 1;\r
 307                             }\r
 308                         }\r
 309                         break;\r
 310                     default:\r
 311                         if (state == 2 || state == 3)\r
 312                         {\r
 313                             //sElem = new StringElement(this,\r
 314                             //    source.Substring(textStart, textEnd - textStart));\r
 315                             parseEmphasisText(this, source, textStart, textEnd, search);\r
 316                             hElem = new HtmlElement(this, "span");\r
 317                             hAttr = new HtmlAttribute("res-link");\r
 318                             hElem.addAttribute(hAttr);\r
 319                             hElem.setClosed(true);\r
 320                             if (state == 3)\r
 321                                 textStart = i - 2;\r
 322                             else if (state == 2)\r
 323                                 textStart = i - 1;\r
 324                             sElem = new StringElement(hElem, source.Substring(textEnd, textStart - textEnd));\r
 325                             state = 0;\r
 326                         }\r
 327                         break;\r
 328                 }\r
 329             }\r
 330             if (textStart < end)\r
 331             {\r
 332                 if (state == 2 || state == 3)\r
 333                 {\r
 334                     //sElem = new StringElement(this,\r
 335                     //    source.Substring(textStart, textEnd - textStart));\r
 336                     parseEmphasisText(this, source, textStart, textEnd, search);\r
 337                     hElem = new HtmlElement(this, "span");\r
 338                     hAttr = new HtmlAttribute("res-link");\r
 339                     hElem.addAttribute(hAttr);\r
 340                     hElem.setClosed(true);\r
 341                     if (state == 3)\r
 342                         textStart = end - 1;\r
 343                     else if (state == 2)\r
 344                         textStart = end;\r
 345                     sElem = new StringElement(hElem, source.Substring(textEnd, textStart - textEnd));\r
 346                     if (state == 3)\r
 347                     {\r
 348                         sElem = new StringElement(this, source.Substring(end -1, 1));\r
 349                     }\r
 350                 }\r
 351                 else\r
 352                 {\r
 353                     //sElem = new StringElement(this, source.Substring(textStart, end - textStart));\r
 354                     parseEmphasisText(this, source, textStart, end, search);\r
 355                 }\r
 356             }\r
 357             return bRet;\r
 358         }\r
 359 \r
 360         bool chkGraphicLink(string src)\r
 361         {\r
 362             //if (!chkShowThumbnail())\r
 363             //    return false;\r
 364 \r
 365             if (src == null || src.Length < 4)\r
 366                 return false;\r
 367             int idx = src.LastIndexOf('.');\r
 368             if (idx < 0)\r
 369                 return false;\r
 370 \r
 371             string suffix = src.Substring(idx + 1).ToLower();\r
 372             switch (suffix)\r
 373             {\r
 374                 case "png":\r
 375                 case "jpg":\r
 376                 case "jpeg":\r
 377                 case "bmp":\r
 378                 case "gif":\r
 379                     return true;\r
 380                 default:\r
 381                     return false;\r
 382             }\r
 383         }\r
 384 \r
 385         private bool parse2chRes_old(string source)\r
 386         {\r
 387 \r
 388             if (source == null || source.Length == 0)\r
 389                 return false;\r
 390 \r
 391             int textStart = 0;\r
 392             int state = 0;\r
 393             string tagName, attrName=null, attrVal;\r
 394             StringElement sElem;\r
 395             HtmlAttribute hAttr;\r
 396             HtmlElement hElem = null;\r
 397             bool dquote = false;\r
 398             bool squote = false;\r
 399 \r
 400 \r
 401             int length = source.Length;\r
 402             for (int i = 0; i < length; i++)\r
 403             {\r
 404                 switch (HtmlParser.getCharToken(source[i]))\r
 405                 {\r
 406                     case  HtmlParser.CHAR_TOKEN.LT:\r
 407                         if (state == 0 && textStart < i)\r
 408                         {\r
 409                             sElem = new StringElement(\r
 410                                 this, source.Substring(textStart, i - textStart));\r
 411                         }\r
 412                         state = 1;\r
 413                         textStart = length;\r
 414                         break;\r
 415                     case HtmlParser.CHAR_TOKEN.SLASH:\r
 416                         if (state == 0)\r
 417                             break;\r
 418                         else if (state == 6 && (dquote || squote))\r
 419                         {\r
 420                             break;\r
 421                         }\r
 422                         else if (state == 1)\r
 423                         {\r
 424                             state = 8;\r
 425                         }\r
 426                         else\r
 427                         {                            \r
 428                             state = 7;\r
 429                         }\r
 430                         break;\r
 431                     case HtmlParser.CHAR_TOKEN.GT:\r
 432                         if (state == 0){\r
 433                             break;\r
 434                         }if (state == 1 || state == 8){\r
 435                             return false;\r
 436                         }\r
 437                         else if (state == 2)\r
 438                         {\r
 439                             tagName = source.Substring(textStart, i - textStart);\r
 440                             hElem = new HtmlElement(this, tagName);\r
 441                         }\r
 442                         else if (state == 7)\r
 443                         {\r
 444                             hElem.setClosed(true);\r
 445                         }\r
 446                         else if (state == 9)\r
 447                         {\r
 448                             tagName = source.Substring(textStart, i - textStart);\r
 449                             int count = mChildren.Count;\r
 450                             for (int j = count - 1; j >= 0; j--)\r
 451                             {\r
 452                                 HtmlElement child = mChildren[j];\r
 453                                 if (child.Closed())\r
 454                                     continue;\r
 455 \r
 456                                 if (tagName.Equals(child.getTagName()))\r
 457                                 {\r
 458                                     child.setClosed(true);\r
 459                                     if (j < (count-1))\r
 460                                     {\r
 461                                         child.mChildren.AddRange(\r
 462                                             mChildren.GetRange(j+1, count - 1 - j));\r
 463                                         mChildren.RemoveRange(j+1, count - 1 - j);\r
 464                                     }\r
 465                                     break;\r
 466                                 }\r
 467                             }\r
 468                         }\r
 469                         hElem = null;\r
 470                         state = 0;\r
 471                         textStart = i + 1;\r
 472                         break;\r
 473                     case  HtmlParser.CHAR_TOKEN.WHITESPACE:\r
 474                     case  HtmlParser.CHAR_TOKEN.NL:\r
 475                         if (state == 1)\r
 476                             return false;\r
 477                         else if (state == 2)\r
 478                         {\r
 479                             tagName = source.Substring(textStart, i - textStart);\r
 480                             hElem = new HtmlElement(this, tagName);\r
 481                             state = 3;\r
 482                         }\r
 483                         else if (state == 4)\r
 484                         {\r
 485                             attrName = source.Substring(textStart, i - textStart);\r
 486                             state = 5;\r
 487                         }\r
 488                         else if (state == 5)\r
 489                         {\r
 490                             hAttr = new HtmlAttribute(attrName);\r
 491                             hElem.addAttribute(hAttr);\r
 492                             state = 3;\r
 493                             dquote = squote = false;\r
 494                         }\r
 495                         else if (state == 6)\r
 496                         {\r
 497                             attrVal = source.Substring(textStart, i - textStart);\r
 498                             hAttr = new HtmlAttribute(attrName, attrVal);\r
 499                             hElem.addAttribute(hAttr);\r
 500                             state = 3;\r
 501                             dquote = squote = false;\r
 502                         }\r
 503                         break;\r
 504                     case HtmlParser.CHAR_TOKEN.EQUAL:\r
 505                         if (state != 0 && state != 4 && state != 5)\r
 506                             return false;\r
 507                         if (state == 0)\r
 508                             break;\r
 509                         if (state == 4)\r
 510                         {\r
 511                             attrName = source.Substring(textStart, i - textStart);\r
 512                         }\r
 513                         textStart = i + 1;\r
 514                         state = 6;\r
 515                         break;\r
 516                     case HtmlParser.CHAR_TOKEN.DQUOTE:\r
 517                         if (state == 0)\r
 518                             break;\r
 519                         if(state != 6)\r
 520                             return false;\r
 521                         if (squote)\r
 522                             break;\r
 523                         if (dquote)\r
 524                         {\r
 525                             attrVal = source.Substring(textStart, i - textStart);\r
 526                             hAttr = new HtmlAttribute(attrName, attrVal);\r
 527                             hElem.addAttribute(hAttr);\r
 528                             state = 3;\r
 529                             dquote = false;\r
 530                         }\r
 531                         else\r
 532                         {\r
 533                             dquote = true;\r
 534                             textStart = i + 1;\r
 535                         }\r
 536                         break;\r
 537                     case HtmlParser.CHAR_TOKEN.SQUOTE:\r
 538                         if (state == 0)\r
 539                             break;\r
 540                         if(state != 6)\r
 541                             return false;\r
 542                         if (dquote)\r
 543                             break;\r
 544                         if (squote)\r
 545                         {\r
 546                             attrVal = source.Substring(textStart, i - textStart);\r
 547                             hAttr = new HtmlAttribute(attrName, attrVal);\r
 548                             hElem.addAttribute(hAttr);\r
 549                             state = 3;\r
 550                             squote = false;\r
 551                         }\r
 552                         else\r
 553                         {\r
 554                             squote = true;\r
 555                             textStart = i + 1;\r
 556                         }\r
 557                         break;\r
 558                     case HtmlParser.CHAR_TOKEN.ALPHA:\r
 559                         if(state == 0)\r
 560                             break;\r
 561                         else if (state == 1)\r
 562                         {\r
 563                             state = 2;\r
 564                             textStart = i;\r
 565                         }\r
 566                         else if (state == 3)\r
 567                         {\r
 568                             textStart = i;\r
 569                             state = 4;\r
 570                         }\r
 571                         else if (state == 8)\r
 572                         {\r
 573                             textStart = i;\r
 574                             state = 9;\r
 575                         }\r
 576                         else if(state == 7)\r
 577                         {\r
 578                             return false;\r
 579                         }\r
 580                         break;\r
 581                     default:\r
 582                         if(state == 7)\r
 583                         {\r
 584                             return false;\r
 585                         }\r
 586                         break;\r
 587                 }\r
 588             }\r
 589             if (state == 0 && textStart < length)\r
 590             {\r
 591                 sElem = new StringElement(\r
 592                     this, source.Substring(textStart, length - textStart));\r
 593             }\r
 594             return false;\r
 595         }\r
 596     }\r
 597 }\r