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         /* Note: Using [] to match words is wrong, but using () doesn't seem to match 'function' on its own */
  83         addCallbackRegex(language,
  84                 "^[ \t]*[(public|protected|private|static|final)[ \t]*]*[ \t]*function[ \t]+&?[ \t]*([" ALPHA "_][" ALNUM "_]*)[[:space:]]*(\\(.*\\))",
  85                 NULL, function_cb);
  86         addTagRegex(language, "^[ \t]*(\\$|::\\$|\\$this->)([" ALPHA "_][" ALNUM "_]*)[ \t]*=",
  87                 "\\2", "v,variable,variables", NULL);
  88         addTagRegex(language, "^[ \t]*((var|public|protected|private|static)[ \t]+)+\\$([" ALPHA "_][" ALNUM "_]*)[ \t]*[=;]",
  89                 "\\3", "v,variable,variables", NULL);
  90
  91         /* function regex is covered by PHP regex */
  92         addTagRegex (language, "(^|[ \t])([A-Za-z0-9_]+)[ \t]*[=:][ \t]*function[ \t]*\\(",
  93                 "\\2", "j,jsfunction,javascript functions", NULL);
  94         addTagRegex (language, "(^|[ \t])([A-Za-z0-9_.]+)\\.([A-Za-z0-9_]+)[ \t]*=[ \t]*function[ \t]*\\(",
  95                 "\\2.\\3", "j,jsfunction,javascript functions", NULL);
  96         addTagRegex (language, "(^|[ \t])([A-Za-z0-9_.]+)\\.([A-Za-z0-9_]+)[ \t]*=[ \t]*function[ \t]*\\(",
  97                 "\\3", "j,jsfunction,javascript functions", NULL);
  98 }
  99
 100
 101 static void function_cb(const char *line, const regexMatch *matches, unsigned int count)
 102 {
 103         char *name, *arglist;
 104         char kind = 'f';
 105         static char *kindName = "function";
 106         tagEntryInfo e;
 107         const regexMatch *match_funcname = NULL;
 108         const regexMatch *match_arglist = NULL;
 109
 110         if (count > 2)
 111         {
 112                 match_funcname = &matches[count - 2];
 113                 match_arglist = &matches[count - 1];
 114         }
 115
 116         if (match_funcname != NULL)
 117         {
 118                 name = xMalloc(match_funcname->length + 1, char);
 119                 strncpy(name, line + match_funcname->start, match_funcname->length);
 120                 *(name+match_funcname->length) = '\x0';
 121                 arglist = xMalloc(match_arglist->length + 1, char);
 122                 strncpy(arglist, line + match_arglist->start, match_arglist->length);
 123                 *(arglist+match_arglist->length) = '\x0';
 124
 125                 initTagEntry (&e, name);
 126                 e.kind = kind;
 127                 e.kindName = kindName;
 128                 e.extensionFields.arglist = arglist;
 129                 makeTagEntry (&e);
 130
 131                 eFree(name);
 132                 eFree(arglist);
 133         }
 134 }
 135
 136 /* Create parser definition structure */
 137 extern parserDefinition* PhpParser (void)
 138 {
 139         static const char *const extensions [] = { "php", "php3", "phtml", NULL };
 140         parserDefinition* def = parserNew ("PHP");
 141         def->extensions = extensions;
 142         def->initialize = installPHPRegex;
 143         def->regex      = TRUE;
 144         return def;
 145 }
 146
 147 #if 0
 148
 149 static boolean isLetter(const int c)
 150 {
 151         return (boolean)(isalpha(c) || (c >= 127  &&  c <= 255));
 152 }
 153
 154 static boolean isVarChar1(const int c)
 155 {
 156         return (boolean)(isLetter (c)  ||  c == '_');
 157 }
 158
 159 static boolean isVarChar(const int c)
 160 {
 161         return (boolean)(isVarChar1 (c) || isdigit (c));
 162 }
 163
 164 static void findPhpTags (void)
 165 {
 166         vString *name = vStringNew ();
 167         const unsigned char *line;
 168
 169         while ((line = fileReadLine ()) != NULL)
 170         {
 171                 const unsigned char *cp = line;
 172                 const char* f;
 173
 174                 while (isspace (*cp))
 175                         cp++;
 176
 177                 if (*(const char*)cp == '$'  &&  isVarChar1 (*(const char*)(cp+1)))
 178                 {
 179                         cp += 1;
 180                         vStringClear (name);
 181                         while (isVarChar ((int) *cp))
 182                         {
 183                                 vStringPut (name, (int) *cp);
 184                                 ++cp;
 185                         }
 186                         while (isspace ((int) *cp))
 187                                 ++cp;
 188                         if (*(const char*) cp == '=')
 189                         {
 190                                 vStringTerminate (name);
 191                                 makeSimpleTag (name, PhpKinds, K_VARIABLE);
 192                                 vStringClear (name);
 193                         }
 194                 }
 195                 else if ((f = strstr ((const char*) cp, "function")) != NULL &&
 196                         (f == (const char*) cp || isspace ((int) f [-1])) &&
 197                         isspace ((int) f [8]))
 198                 {
 199                         cp = ((const unsigned char *) f) + 8;
 200
 201                         while (isspace ((int) *cp))
 202                                 ++cp;
 203
 204                         if (*cp == '&') /* skip reference character and following whitespace */
 205                         {
 206                                 cp++;
 207
 208                                 while (isspace ((int) *cp))
 209                                         ++cp;
 210                         }
 211
 212                         vStringClear (name);
 213                         while (isalnum ((int) *cp)  ||  *cp == '_')
 214                         {
 215                                 vStringPut (name, (int) *cp);
 216                                 ++cp;
 217                         }
 218                         vStringTerminate (name);
 219                         makeSimpleTag (name, PhpKinds, K_FUNCTION);
 220                         vStringClear (name);
 221                 }
 222                 else if (strncmp ((const char*) cp, "class", (size_t) 5) == 0 &&
 223                                  isspace ((int) cp [5]))
 224                 {
 225                         cp += 5;
 226
 227                         while (isspace ((int) *cp))
 228                                 ++cp;
 229                         vStringClear (name);
 230                         while (isalnum ((int) *cp)  ||  *cp == '_')
 231                         {
 232                                 vStringPut (name, (int) *cp);
 233                                 ++cp;
 234                         }
 235                         vStringTerminate (name);
 236                         makeSimpleTag (name, PhpKinds, K_CLASS);
 237                         vStringClear (name);
 238                 }
 239                 else if (strncmp ((const char*) cp, "define", (size_t) 6) == 0 &&
 240                                  ! isalnum ((int) cp [6]))
 241                 {
 242                         cp += 6;
 243
 244                         while (isspace ((int) *cp))
 245                                 ++cp;
 246                         if (*cp != '(')
 247                                 continue;
 248                         ++cp;
 249
 250                         while (isspace ((int) *cp))
 251                                 ++cp;
 252                         if ((*cp == '\'') || (*cp == '"'))
 253                                 ++cp;
 254                         else if (! ((*cp == '_')  || isalnum ((int) *cp)))
 255                                 continue;
 256
 257                         vStringClear (name);
 258                         while (isalnum ((int) *cp)  ||  *cp == '_')
 259                         {
 260                                 vStringPut (name, (int) *cp);
 261                                 ++cp;
 262                         }
 263                         vStringTerminate (name);
 264                         makeSimpleTag (name, PhpKinds, K_DEFINE);
 265                         vStringClear (name);
 266                 }
 267         }
 268         vStringDelete (name);
 269 }
 270
 271 extern parserDefinition* PhpParser (void)
 272 {
 273         static const char *const extensions [] = { "php", "php3", "phtml", NULL };
 274         parserDefinition* def = parserNew ("PHP");
 275         def->kinds      = PhpKinds;
 276         def->kindCount  = KIND_COUNT (PhpKinds);
 277         def->extensions = extensions;
 278         def->parser     = findPhpTags;
 279         return def;
 280 }
 281
 282 #endif
 283
 284 /* vi:set tabstop=4 shiftwidth=4: */