tagmanager/php.c

   1 /*
   2 *   $Id$
   3 *
   4 *   Copyright (c) 2000, Jesus Castagnetto <jmcastagnetto@zkey.com>
   5 *
   6 *   This source code is released for free distribution under the terms of the
   7 *   GNU General Public License.
   8 *
   9 *   This module contains functions for generating tags for the PHP web page
  10 *   scripting language. Only recognizes functions and classes, not methods or
  11 *   variables.
  12 *
  13 *   Parsing PHP defines by Pavel Hlousek <pavel.hlousek@seznam.cz>, Apr 2003.
  14 */
  15
  16 /*
  17 *   INCLUDE FILES
  18 */
  19 #include "general.h"  /* must always come first */
  20
  21 #include <string.h>
  22 #include "main.h"
  23 #include "parse.h"
  24 #include "read.h"
  25 #include "vstring.h"
  26
  27 /*
  28 *   DATA DEFINITIONS
  29 */
  30 typedef enum {
  31         K_CLASS, K_DEFINE, K_FUNCTION, K_VARIABLE
  32 } phpKind;
  33
  34 #if 0
  35 static kindOption PhpKinds [] = {
  36         { TRUE, 'c', "class",    "classes" },
  37         { TRUE, 'd', "define",   "constant definitions" },
  38         { TRUE, 'f', "function", "functions" },
  39         { TRUE, 'v', "variable", "variables" }
  40 };
  41 #endif
  42
  43 /*
  44 *   FUNCTION DEFINITIONS
  45 */
  46
  47 /* JavaScript patterns are duplicated in jscript.c */
  48
  49 /*
  50  * Cygwin doesn't support non-ASCII characters in character classes.
  51  * This isn't a good solution to the underlying problem, because we're still
  52  * making assumptions about the character encoding.
  53  * Really, these regular expressions need to concentrate on what marks the
  54  * end of an identifier, and we need something like iconv to take into
  55  * account the user's locale (or an override on the command-line.)
  56  */
  57 /*
  58 #ifdef __CYGWIN__
  59 #define ALPHA "[:alpha:]"
  60 #define ALNUM "[:alnum:]"
  61 #else
  62 #define ALPHA "A-Za-z\x7f-\xff"
  63 #define ALNUM "0-9A-Za-z\x7f-\xff"
  64 #endif
  65 */
  66 /* "A-Za-z\x7f-\xff" fails on other locales than "C" and so skip it */
  67 #define ALPHA "[:alpha:]"
  68 #define ALNUM "[:alnum:]"
  69
  70 static void function_cb(const char *line, const regexMatch *matches, unsigned int count);
  71
  72 static void installPHPRegex (const langType language)
  73 {
  74         addTagRegex(language, "^[ \t]*((final|abstract)[ \t]+)*class[ \t]+([" ALPHA "_][" ALNUM "_]*)",
  75                 "\\3", "c,class,classes", NULL);
  76         addTagRegex(language, "^[ \t]*interface[ \t]+([" ALPHA "_][" ALNUM "_]*)",
  77                 "\\1", "i,interface,interfaces", NULL);
  78         addTagRegex(language, "^[ \t]*define[ \t]*\\([ \t]*['\"]?([" ALPHA "_][" ALNUM "_]*)",
  79                 "\\1", "m,macro,macros", NULL);
  80         addTagRegex(language, "^[ \t]*const[ \t]*([" ALPHA "_][" ALNUM "_]*)[ \t]*[=;]",
  81                 "\\1", "m,macro,macros", NULL);
  82         addCallbackRegex(language,
  83         "^[ \t]*[(public|protected|private|static)[ \t]*]*[ \t]*function[ \t]+&?[ \t]*([" ALPHA "_][" ALNUM "_]*)[[:space:]]*(\\(.*\\))",
  84                         NULL, function_cb);
  85         addTagRegex(language, "^[ \t]*(\\$|::\\$|\\$this->)([" ALPHA "_][" ALNUM "_]*)[ \t]*=",
  86                 "\\2", "v,variable,variables", NULL);
  87         addTagRegex(language, "^[ \t]*((var|public|protected|private|static)[ \t]+)+\\$([" ALPHA "_][" ALNUM "_]*)[ \t]*[=;]",
  88                 "\\3", "v,variable,variables", NULL);
  89
  90         /* function regex is covered by PHP regex */
  91         addTagRegex (language, "(^|[ \t])([A-Za-z0-9_]+)[ \t]*[=:][ \t]*function[ \t]*\\(",
  92                 "\\2", "j,jsfunction,javascript functions", NULL);
  93         addTagRegex (language, "(^|[ \t])([A-Za-z0-9_.]+)\\.([A-Za-z0-9_]+)[ \t]*=[ \t]*function[ \t]*\\(",
  94                 "\\2.\\3", "j,jsfunction,javascript functions", NULL);
  95         addTagRegex (language, "(^|[ \t])([A-Za-z0-9_.]+)\\.([A-Za-z0-9_]+)[ \t]*=[ \t]*function[ \t]*\\(",
  96                 "\\3", "j,jsfunction,javascript functions", NULL);
  97 }
  98
  99
 100 static void function_cb(const char *line, const regexMatch *matches, unsigned int count)
 101 {
 102         char *name, *arglist;
 103         char kind = 'f';
 104         static char *kindName = "function";
 105         tagEntryInfo e;
 106         const regexMatch *match_funcname = NULL;
 107         const regexMatch *match_arglist = NULL;
 108
 109         if (count > 2)
 110         {
 111                 match_funcname = &matches[count - 2];
 112                 match_arglist = &matches[count - 1];
 113         }
 114
 115         if (match_funcname != NULL)
 116         {
 117                 name = xMalloc(match_funcname->length + 1, char);
 118                 strncpy(name, line + match_funcname->start, match_funcname->length);
 119                 *(name+match_funcname->length) = '\x0';
 120                 arglist = xMalloc(match_arglist->length + 1, char);
 121                 strncpy(arglist, line + match_arglist->start, match_arglist->length);
 122                 *(arglist+match_arglist->length) = '\x0';
 123
 124                 initTagEntry (&e, name);
 125                 e.kind = kind;
 126                 e.kindName = kindName;
 127                 e.extensionFields.arglist = arglist;
 128                 makeTagEntry (&e);
 129         }
 130 }
 131
 132 /* Create parser definition structure */
 133 extern parserDefinition* PhpParser (void)
 134 {
 135         static const char *const extensions [] = { "php", "php3", "phtml", NULL };
 136         parserDefinition* def = parserNew ("PHP");
 137         def->extensions = extensions;
 138         def->initialize = installPHPRegex;
 139         def->regex      = TRUE;
 140         return def;
 141 }
 142
 143 #if 0
 144
 145 static boolean isLetter(const int c)
 146 {
 147         return (boolean)(isalpha(c) || (c >= 127  &&  c <= 255));
 148 }
 149
 150 static boolean isVarChar1(const int c)
 151 {
 152         return (boolean)(isLetter (c)  ||  c == '_');
 153 }
 154
 155 static boolean isVarChar(const int c)
 156 {
 157         return (boolean)(isVarChar1 (c) || isdigit (c));
 158 }
 159
 160 static void findPhpTags (void)
 161 {
 162         vString *name = vStringNew ();
 163         const unsigned char *line;
 164
 165         while ((line = fileReadLine ()) != NULL)
 166         {
 167                 const unsigned char *cp = line;
 168                 const char* f;
 169
 170                 while (isspace (*cp))
 171                         cp++;
 172
 173                 if (*(const char*)cp == '$'  &&  isVarChar1 (*(const char*)(cp+1)))
 174                 {
 175                         cp += 1;
 176                         vStringClear (name);
 177                         while (isVarChar ((int) *cp))
 178                         {
 179                                 vStringPut (name, (int) *cp);
 180                                 ++cp;
 181                         }
 182                         while (isspace ((int) *cp))
 183                                 ++cp;
 184                         if (*(const char*) cp == '=')
 185                         {
 186                                 vStringTerminate (name);
 187                                 makeSimpleTag (name, PhpKinds, K_VARIABLE);
 188                                 vStringClear (name);
 189                         }
 190                 }
 191                 else if ((f = strstr ((const char*) cp, "function")) != NULL &&
 192                         (f == (const char*) cp || isspace ((int) f [-1])) &&
 193                         isspace ((int) f [8]))
 194                 {
 195                         cp = ((const unsigned char *) f) + 8;
 196
 197                         while (isspace ((int) *cp))
 198                                 ++cp;
 199
 200                         if (*cp == '&') /* skip reference character and following whitespace */
 201                         {
 202                                 cp++;
 203
 204                                 while (isspace ((int) *cp))
 205                                         ++cp;
 206                         }
 207
 208                         vStringClear (name);
 209                         while (isalnum ((int) *cp)  ||  *cp == '_')
 210                         {
 211                                 vStringPut (name, (int) *cp);
 212                                 ++cp;
 213                         }
 214                         vStringTerminate (name);
 215                         makeSimpleTag (name, PhpKinds, K_FUNCTION);
 216                         vStringClear (name);
 217                 }
 218                 else if (strncmp ((const char*) cp, "class", (size_t) 5) == 0 &&
 219                                  isspace ((int) cp [5]))
 220                 {
 221                         cp += 5;
 222
 223                         while (isspace ((int) *cp))
 224                                 ++cp;
 225                         vStringClear (name);
 226                         while (isalnum ((int) *cp)  ||  *cp == '_')
 227                         {
 228                                 vStringPut (name, (int) *cp);
 229                                 ++cp;
 230                         }
 231                         vStringTerminate (name);
 232                         makeSimpleTag (name, PhpKinds, K_CLASS);
 233                         vStringClear (name);
 234                 }
 235                 else if (strncmp ((const char*) cp, "define", (size_t) 6) == 0 &&
 236                                  ! isalnum ((int) cp [6]))
 237                 {
 238                         cp += 6;
 239
 240                         while (isspace ((int) *cp))
 241                                 ++cp;
 242                         if (*cp != '(')
 243                                 continue;
 244                         ++cp;
 245
 246                         while (isspace ((int) *cp))
 247                                 ++cp;
 248                         if ((*cp == '\'') || (*cp == '"'))
 249                                 ++cp;
 250                         else if (! ((*cp == '_')  || isalnum ((int) *cp)))
 251                                 continue;
 252
 253                         vStringClear (name);
 254                         while (isalnum ((int) *cp)  ||  *cp == '_')
 255                         {
 256                                 vStringPut (name, (int) *cp);
 257                                 ++cp;
 258                         }
 259                         vStringTerminate (name);
 260                         makeSimpleTag (name, PhpKinds, K_DEFINE);
 261                         vStringClear (name);
 262                 }
 263         }
 264         vStringDelete (name);
 265 }
 266
 267 extern parserDefinition* PhpParser (void)
 268 {
 269         static const char *const extensions [] = { "php", "php3", "phtml", NULL };
 270         parserDefinition* def = parserNew ("PHP");
 271         def->kinds      = PhpKinds;
 272         def->kindCount  = KIND_COUNT (PhpKinds);
 273         def->extensions = extensions;
 274         def->parser     = findPhpTags;
 275         return def;
 276 }
 277
 278 #endif
 279
 280 /* vi:set tabstop=4 shiftwidth=4: */