tagmanager/parse.c

   1 /*
   2 *
   3 *   Copyright (c) 1996-2001, Darren Hiebert
   4 *
   5 *   This source code is released for free distribution under the terms of the
   6 *   GNU General Public License.
   7 *
   8 *   This module contains functions for managing source languages and
   9 *   dispatching files to the appropriate language parser.
  10 */
  11
  12 /*
  13 *   INCLUDE FILES
  14 */
  15 #include "general.h"    /* must always come first */
  16
  17 #include <string.h>
  18 #include <mio/mio.h>
  19
  20
  21 #include "entry.h"
  22 #include "main.h"
  23 #define OPTION_WRITE
  24 #include "options.h"
  25 #include "parsers.h"
  26 #include "read.h"
  27 #include "vstring.h"
  28
  29 /*
  30 *   DATA DEFINITIONS
  31 */
  32 static parserDefinitionFunc* BuiltInParsers[] = { PARSER_LIST };
  33 parserDefinition** LanguageTable = NULL;
  34 static unsigned int LanguageCount = 0;
  35 tagEntryFunction TagEntryFunction = NULL;
  36 tagEntrySetArglistFunction TagEntrySetArglistFunction = NULL;
  37
  38 /*
  39 *   FUNCTION DEFINITIONS
  40 */
  41
  42 extern void makeSimpleTag (const vString* const name,
  43                            kindOption* const kinds, const int kind)
  44 {
  45     if (name != NULL  &&  vStringLength (name) > 0)
  46     {
  47         tagEntryInfo e;
  48         initTagEntry (&e, vStringValue (name));
  49
  50         e.kindName = kinds [kind].name;
  51         e.kind     = kinds [kind].letter;
  52
  53         makeTagEntry (&e);
  54     }
  55 }
  56
  57
  58 extern void makeSimpleScopedTag (const vString* const name,
  59                                  kindOption* const kinds, const int kind,
  60                                  const char* scope, const char *scope2,
  61                                  const char *laccess)
  62 {
  63     if (name != NULL  &&  vStringLength (name) > 0)
  64     {
  65         tagEntryInfo e;
  66         initTagEntry (&e, vStringValue (name));
  67
  68         e.kindName = kinds [kind].name;
  69         e.kind     = kinds [kind].letter;
  70         e.extensionFields.scope[0] = scope;
  71         e.extensionFields.scope[1] = scope2;
  72         e.extensionFields.access = laccess;
  73
  74         makeTagEntry (&e);
  75     }
  76 }
  77
  78 /*
  79 *   parserDescription mapping management
  80 */
  81
  82 extern parserDefinition* parserNew (const char* name)
  83 {
  84     parserDefinition* result = xCalloc (1, parserDefinition);
  85     result->name = eStrdup (name);
  86     return result;
  87 }
  88
  89 extern const char *getLanguageName (const langType language)
  90 {
  91     /*Assert (0 <= language  &&  language < (int) LanguageCount);*/
  92     if (language < 0) return NULL;
  93         return LanguageTable [language]->name;
  94 }
  95
  96 extern langType getNamedLanguage (const char *const name)
  97 {
  98     langType result = LANG_IGNORE;
  99     unsigned int i;
 100     Assert (name != NULL);
 101         for (i = 0  ;  i < LanguageCount  &&  result == LANG_IGNORE  ;  ++i)
 102     {
 103         if (LanguageTable [i]->name != NULL)
 104             if (stricmp (name, LanguageTable [i]->name) == 0)
 105                 result = i;
 106     }
 107     return result;
 108 }
 109
 110 static langType getExtensionLanguage (const char *const extension)
 111 {
 112     langType result = LANG_IGNORE;
 113     unsigned int i;
 114     for (i = 0  ;  i < LanguageCount  &&  result == LANG_IGNORE  ;  ++i)
 115     {
 116         stringList* const exts = LanguageTable [i]->currentExtensions;
 117         if (exts != NULL  &&  stringListExtensionMatched (exts, extension))
 118             result = i;
 119     }
 120     return result;
 121 }
 122
 123 static langType getPatternLanguage (const char *const fileName)
 124 {
 125     langType result = LANG_IGNORE;
 126     const char* base = baseFilename (fileName);
 127     unsigned int i;
 128     for (i = 0  ;  i < LanguageCount  &&  result == LANG_IGNORE  ;  ++i)
 129     {
 130         stringList* const ptrns = LanguageTable [i]->currentPatterns;
 131         if (ptrns != NULL  &&  stringListFileMatched (ptrns, base))
 132             result = i;
 133     }
 134     return result;
 135 }
 136
 137 #ifdef SYS_INTERPRETER
 138
 139 /*  The name of the language interpreter, either directly or as the argument
 140  *  to "env".
 141  */
 142 static vString* determineInterpreter (const char* const cmd)
 143 {
 144     vString* const interpreter = vStringNew ();
 145     const char* p = cmd;
 146     do
 147     {
 148         vStringClear (interpreter);
 149         for ( ;  isspace (*p)  ;  ++p)
 150             ; /* no-op */
 151         for ( ;  *p != '\0'  &&  ! isspace (*p)  ;  ++p)
 152             vStringPut (interpreter, (int) *p);
 153         vStringTerminate (interpreter);
 154     } while (strcmp (vStringValue (interpreter), "env") == 0);
 155     return interpreter;
 156 }
 157
 158 static langType getInterpreterLanguage (const char *const fileName)
 159 {
 160     langType result = LANG_IGNORE;
 161     FILE* const fp = g_fopen (fileName, "r");
 162     if (fp != NULL)
 163     {
 164         vString* const vLine = vStringNew ();
 165         const char* const line = readLine (vLine, fp);
 166         if (line != NULL  &&  line [0] == '#'  &&  line [1] == '!')
 167         {
 168             const char* const lastSlash = strrchr (line, '/');
 169             const char *const cmd = lastSlash != NULL ? lastSlash+1 : line+2;
 170             vString* const interpreter = determineInterpreter (cmd);
 171             result = getExtensionLanguage (vStringValue (interpreter));
 172             vStringDelete (interpreter);
 173         }
 174         vStringDelete (vLine);
 175         fclose (fp);
 176     }
 177     return result;
 178 }
 179
 180 #endif
 181
 182 extern langType getFileLanguage (const char *const fileName)
 183 {
 184     langType language = Option.language;
 185     if (language == LANG_AUTO)
 186     {
 187         language = getExtensionLanguage (fileExtension (fileName));
 188         if (language == LANG_IGNORE)
 189             language = getPatternLanguage (fileName);
 190 #ifdef SYS_INTERPRETER
 191         if (language == LANG_IGNORE  &&  isExecutable (fileName))
 192             language = getInterpreterLanguage (fileName);
 193 #endif
 194     }
 195     return language;
 196 }
 197
 198 extern void printLanguageMap (const langType language)
 199 {
 200     boolean first = TRUE;
 201     unsigned int i;
 202     stringList* map = LanguageTable [language]->currentPatterns;
 203     Assert (0 <= language  &&  language < (int) LanguageCount);
 204     for (i = 0  ;  map != NULL  &&  i < stringListCount (map)  ;  ++i)
 205     {
 206         printf ("%s(%s)", (first ? "" : " "),
 207                 vStringValue (stringListItem (map, i)));
 208         first = FALSE;
 209     }
 210     map = LanguageTable [language]->currentExtensions;
 211     for (i = 0  ;  map != NULL  &&  i < stringListCount (map)  ;  ++i)
 212     {
 213         printf ("%s.%s", (first ? "" : " "),
 214                 vStringValue (stringListItem (map, i)));
 215         first = FALSE;
 216     }
 217 }
 218
 219 extern void installLanguageMapDefault (const langType language)
 220 {
 221     Assert (language >= 0);
 222     if (LanguageTable [language]->currentPatterns != NULL)
 223         stringListDelete (LanguageTable [language]->currentPatterns);
 224     if (LanguageTable [language]->currentExtensions != NULL)
 225         stringListDelete (LanguageTable [language]->currentExtensions);
 226
 227     if (LanguageTable [language]->patterns == NULL)
 228         LanguageTable [language]->currentPatterns = stringListNew ();
 229     else
 230     {
 231         LanguageTable [language]->currentPatterns =
 232             stringListNewFromArgv (LanguageTable [language]->patterns);
 233     }
 234     if (LanguageTable [language]->extensions == NULL)
 235         LanguageTable [language]->currentExtensions = stringListNew ();
 236     else
 237     {
 238         LanguageTable [language]->currentExtensions =
 239             stringListNewFromArgv (LanguageTable [language]->extensions);
 240     }
 241 }
 242
 243 extern void installLanguageMapDefaults (void)
 244 {
 245     unsigned int i;
 246     for (i = 0  ;  i < LanguageCount  ;  ++i)
 247     {
 248         installLanguageMapDefault (i);
 249     }
 250 }
 251
 252 extern void clearLanguageMap (const langType language)
 253 {
 254     Assert (0 <= language  &&  language < (int) LanguageCount);
 255     stringListClear (LanguageTable [language]->currentPatterns);
 256     stringListClear (LanguageTable [language]->currentExtensions);
 257 }
 258
 259 extern void addLanguagePatternMap (const langType language, const char* ptrn)
 260 {
 261     vString* const str = vStringNewInit (ptrn);
 262     Assert (0 <= language  &&  language < (int) LanguageCount);
 263     if (LanguageTable [language]->currentPatterns == NULL)
 264         LanguageTable [language]->currentPatterns = stringListNew ();
 265     stringListAdd (LanguageTable [language]->currentPatterns, str);
 266 }
 267
 268 extern void addLanguageExtensionMap (const langType language,
 269                                      const char* extension)
 270 {
 271     vString* const str = vStringNewInit (extension);
 272     Assert (0 <= language  &&  language < (int) LanguageCount);
 273     stringListAdd (LanguageTable [language]->currentExtensions, str);
 274 }
 275
 276 extern void enableLanguages (const boolean state)
 277 {
 278     unsigned int i;
 279     for (i = 0  ;  i < LanguageCount  ;  ++i)
 280         LanguageTable [i]->enabled = state;
 281 }
 282
 283 extern void enableLanguage (const langType language, const boolean state)
 284 {
 285     Assert (0 <= language  &&  language < (int) LanguageCount);
 286     LanguageTable [language]->enabled = state;
 287 }
 288
 289 static void initializeParsers (void)
 290 {
 291     unsigned int i;
 292     for (i = 0  ;  i < LanguageCount  ;  ++i)
 293         if (LanguageTable [i]->initialize != NULL)
 294             (LanguageTable [i]->initialize) ((langType) i);
 295 }
 296
 297 extern void initializeParsing (void)
 298 {
 299     unsigned int builtInCount;
 300     unsigned int i;
 301
 302     builtInCount = sizeof (BuiltInParsers) / sizeof (BuiltInParsers [0]);
 303     LanguageTable = xMalloc (builtInCount, parserDefinition*);
 304
 305     for (i = 0  ;  i < builtInCount  ;  ++i)
 306     {
 307         parserDefinition* const def = (*BuiltInParsers [i]) ();
 308         if (def != NULL)
 309         {
 310             boolean accepted = FALSE;
 311             if (def->name == NULL  ||  def->name[0] == '\0')
 312                 error (FATAL, "parser definition must contain name\n");
 313             else if (def->regex)
 314             {
 315 #ifdef HAVE_REGEX
 316                 def->parser = findRegexTags;
 317                 accepted = TRUE;
 318 #endif
 319             }
 320             else if ((def->parser == NULL)  ==  (def->parser2 == NULL))
 321                 error (FATAL,
 322         "%s parser definition must define one and only one parsing routine\n",
 323                        def->name);
 324             else
 325                 accepted = TRUE;
 326             if (accepted)
 327             {
 328                 def->id = LanguageCount++;
 329                 LanguageTable [def->id] = def;
 330             }
 331         }
 332     }
 333     enableLanguages (TRUE);
 334     initializeParsers ();
 335 }
 336
 337 extern void freeParserResources (void)
 338 {
 339     unsigned int i;
 340     for (i = 0  ;  i < LanguageCount  ;  ++i)
 341     {
 342         freeList (&LanguageTable [i]->currentPatterns);
 343         freeList (&LanguageTable [i]->currentExtensions);
 344         eFree (LanguageTable [i]->name);
 345         LanguageTable [i]->name = NULL;
 346         eFree (LanguageTable [i]);
 347     }
 348     eFree (LanguageTable);
 349     LanguageTable = NULL;
 350     LanguageCount = 0;
 351 }
 352
 353 /*
 354 *   Option parsing
 355 */
 356
 357 extern void processLanguageDefineOption (const char *const option,
 358                                          const char *const __unused__ parameter)
 359 {
 360 #ifdef HAVE_REGEX
 361     if (parameter [0] == '\0')
 362         error (WARNING, "No language specified for \"%s\" option", option);
 363     else if (getNamedLanguage (parameter) != LANG_IGNORE)
 364         error (WARNING, "Language \"%s\" already defined", parameter);
 365     else
 366     {
 367         unsigned int i = LanguageCount++;
 368         parserDefinition* const def = parserNew (parameter);
 369         def->parser            = findRegexTags;
 370         def->currentPatterns   = stringListNew ();
 371         def->currentExtensions = stringListNew ();
 372         def->regex             = TRUE;
 373         def->enabled           = TRUE;
 374         def->id                = i;
 375         LanguageTable = xRealloc (LanguageTable, i + 1, parserDefinition*);
 376         LanguageTable [i] = def;
 377     }
 378 #else
 379     error (WARNING, "regex support not available; required for --%s option",
 380            option);
 381 #endif
 382 }
 383
 384 static kindOption *langKindOption (const langType language, const int flag)
 385 {
 386     unsigned int i;
 387     kindOption* result = NULL;
 388     const parserDefinition* lang;
 389     Assert (0 <= language  &&  language < (int) LanguageCount);
 390     lang = LanguageTable [language];
 391     for (i=0  ;  i < lang->kindCount  &&  result == NULL  ;  ++i)
 392         if (lang->kinds [i].letter == flag)
 393             result = &lang->kinds [i];
 394     return result;
 395 }
 396
 397 extern void processLegacyKindOption (const char *const parameter)
 398 {
 399     const langType lang = getNamedLanguage ("c");
 400     boolean clear = FALSE;
 401     const char* p = parameter;
 402     boolean mode = TRUE;
 403     int c;
 404
 405     error (WARNING, "-i option is deprecated; use --c-types option instead");
 406     if (*p == '=')
 407     {
 408         clear = TRUE;
 409         ++p;
 410     }
 411     if (clear  &&  *p != '+'  &&  *p != '-')
 412     {
 413         unsigned int i;
 414         for (i = 0  ;  i < LanguageTable [lang]->kindCount  ;  ++i)
 415             LanguageTable [lang]->kinds [i].enabled = FALSE;
 416         Option.include.fileNames= FALSE;
 417         Option.include.fileScope= FALSE;
 418     }
 419     while ((c = *p++) != '\0') switch (c)
 420     {
 421         case '+': mode = TRUE;  break;
 422         case '-': mode = FALSE; break;
 423
 424         case 'F': Option.include.fileNames = mode; break;
 425         case 'S': Option.include.fileScope = mode; break;
 426
 427         default:
 428         {
 429             kindOption* const opt = langKindOption (lang, c);
 430             if (opt != NULL)
 431                 opt->enabled = mode;
 432             else
 433                 error (WARNING, "Unsupported parameter '%c' for -i option", c);
 434         } break;
 435     }
 436 }
 437
 438 static void disableLanguageKinds (const langType language)
 439 {
 440     if (LanguageTable [language]->regex)
 441 #ifdef HAVE_REGEX
 442         disableRegexKinds (language);
 443 #else
 444         ;
 445 #endif
 446     else
 447     {
 448         unsigned int i;
 449         for (i = 0  ;  i < LanguageTable [language]->kindCount  ;  ++i)
 450             LanguageTable [language]->kinds [i].enabled = FALSE;
 451     }
 452 }
 453
 454 static boolean enableLanguageKind (const langType language,
 455                                    const int kind, const boolean mode)
 456 {
 457     boolean result = FALSE;
 458     if (LanguageTable [language]->regex)
 459 #ifdef HAVE_REGEX
 460         result = enableRegexKind (language, kind, mode);
 461 #else
 462         ;
 463 #endif
 464     else
 465     {
 466         kindOption* const opt = langKindOption (language, kind);
 467         if (opt != NULL)
 468         {
 469             opt->enabled = mode;
 470             result = TRUE;
 471         }
 472     }
 473     return result;
 474 }
 475
 476 static void processLangKindOption (const langType language,
 477                                    const char *const option,
 478                                    const char *const parameter)
 479 {
 480     const char *p = parameter;
 481     boolean mode = TRUE;
 482     int c;
 483
 484     Assert (0 <= language  &&  language < (int) LanguageCount);
 485     if (*p != '+'  &&  *p != '-')
 486         disableLanguageKinds (language);
 487     while ((c = *p++) != '\0') switch (c)
 488     {
 489         case '+': mode = TRUE;  break;
 490         case '-': mode = FALSE; break;
 491
 492         default:
 493         {
 494             if (! enableLanguageKind (language, c, mode))
 495                 error (WARNING, "Unsupported parameter '%c' for --%s option",
 496                     c, option);
 497         } break;
 498     }
 499 }
 500
 501 extern boolean processKindOption (const char *const option,
 502                                   const char *const parameter)
 503 {
 504     boolean handled = FALSE;
 505     const char* const dash = strchr (option, '-');
 506     if (dash != NULL  &&
 507         (strcmp (dash + 1, "types") == 0  ||  strcmp (dash + 1, "kinds") == 0))
 508     {
 509         langType language;
 510         vString* langName = vStringNew ();
 511         vStringNCopyS (langName, option, dash - option);
 512         language = getNamedLanguage (vStringValue (langName));
 513         if (language == LANG_IGNORE)
 514             error (WARNING, "Unknown language specified in \"%s\" option", option);
 515         else
 516             processLangKindOption (language, option, parameter);
 517         vStringDelete (langName);
 518         handled = TRUE;
 519     }
 520     return handled;
 521 }
 522
 523 static void printLangugageKindOption (const kindOption* const kind)
 524 {
 525     printf ("          %c  %s%s\n", kind->letter,
 526         kind->description != NULL ? kind->description :
 527             (kind->name != NULL ? kind->name : ""),
 528         kind->enabled ? "" : " [off]");
 529 }
 530
 531 static void printLangugageKindOptions (const langType language)
 532 {
 533     const parserDefinition* lang;
 534     Assert (0 <= language  &&  language < (int) LanguageCount);
 535     lang = LanguageTable [language];
 536     if (lang->kinds != NULL  ||  lang->regex)
 537     {
 538         unsigned int i;
 539         char* const name = newLowerString (lang->name);
 540         printf ("  --%s-types=[+|-]kinds\n", name);
 541         eFree (name);
 542         if (lang->kinds != NULL)
 543             for (i = 0  ;  i < lang->kindCount  ;  ++i)
 544                 printLangugageKindOption (lang->kinds + i);
 545 #ifdef HAVE_REGEX
 546         /*printRegexKindOptions (language);*/ /* unused */
 547 #endif
 548     }
 549 }
 550
 551 extern void printKindOptions (void)
 552 {
 553     unsigned int i;
 554
 555     printf (
 556  "\n  The following options are used to specify which language-specific tag\n\
 557   types (or kinds) should be included in the tag file. \"Kinds\" is a group of\n\
 558   one-letter flags designating kinds of tags to either include or exclude from\n\
 559   the output. Each letter or group of letters may be preceded by either '+' to\n\
 560   add it to those already included, or '-' to exclude it from the output. In\n\
 561   the absence of any preceding '+' or '-' sign, only those kinds listed in\n\
 562   \"kinds\" will be included in the output. Below each option is a list of the\n\
 563   flags accepted. All kinds are enabled by default unless otherwise noted.\n\n");
 564
 565     for (i = 0  ;  i < LanguageCount  ;  ++i)
 566         printLangugageKindOptions (i);
 567 }
 568
 569 /*
 570 *   File parsing
 571 */
 572
 573 static void makeFileTag (const char *const fileName)
 574 {
 575     if (Option.include.fileNames)
 576     {
 577         tagEntryInfo tag;
 578         initTagEntry (&tag, baseFilename (fileName));
 579
 580         tag.isFileEntry     = TRUE;
 581         tag.lineNumberEntry = TRUE;
 582         tag.lineNumber      = 1;
 583         tag.kindName        = "file";
 584         tag.kind            = 'F';
 585
 586         makeTagEntry (&tag);
 587     }
 588 }
 589
 590 static boolean createTagsForFile (const char *const fileName,
 591                                   const langType language,
 592                                   const unsigned int passCount)
 593 {
 594     boolean retried = FALSE;
 595
 596     if (fileOpen (fileName, language))
 597     {
 598
 599         makeFileTag (fileName);
 600
 601         if (LanguageTable [language]->parser != NULL)
 602             LanguageTable [language]->parser ();
 603         else if (LanguageTable [language]->parser2 != NULL)
 604             retried = LanguageTable [language]->parser2 (passCount);
 605
 606
 607         fileClose ();
 608     }
 609
 610     return retried;
 611 }
 612
 613 static boolean createTagsWithFallback (const char *const fileName,
 614                                        const langType language)
 615 {
 616     const unsigned long numTags = TagFile.numTags.added;
 617     MIOPos tagFilePosition;
 618     unsigned int passCount = 0;
 619     boolean tagFileResized = FALSE;
 620
 621     mio_getpos (TagFile.mio, &tagFilePosition);
 622     while (createTagsForFile (fileName, language, ++passCount))
 623     {
 624         /*  Restore prior state of tag file.
 625          */
 626         mio_setpos (TagFile.mio, &tagFilePosition);
 627         TagFile.numTags.added = numTags;
 628         tagFileResized = TRUE;
 629     }
 630     return tagFileResized;
 631 }
 632
 633 extern boolean parseFile (const char *const fileName)
 634 {
 635     boolean tagFileResized = FALSE;
 636     langType language = Option.language;
 637     if (Option.language == LANG_AUTO)
 638         language = getFileLanguage (fileName);
 639     Assert (language != LANG_AUTO);
 640         if (Option.filter)
 641             openTagFile ();
 642
 643         tagFileResized = createTagsWithFallback (fileName, language);
 644
 645         addTotals (1, 0L, 0L);
 646
 647     return tagFileResized;
 648 }
 649
 650 /* vi:set tabstop=8 shiftwidth=4 nowrap: */