php.c

   1 /*
   2 *   $Id: php.c 734 2009-08-20 23:33:54Z jafl $
   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
  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 #ifdef __CYGWIN__
  58 #define ALPHA "[:alpha:]"
  59 #define ALNUM "[:alnum:]"
  60 #else
  61 #define ALPHA "A-Za-z\x7f-\xff"
  62 #define ALNUM "0-9A-Za-z\x7f-\xff"
  63 #endif
  64
  65 static void installPHPRegex (const langType language)
  66 {
  67         addTagRegex(language, "^[ \t]*((final|abstract)[ \t]+)*class[ \t]+([" ALPHA "_][" ALNUM "_]*)",
  68                 "\\3", "c,class,classes", NULL);
  69         addTagRegex(language, "^[ \t]*interface[ \t]+([" ALPHA "_][" ALNUM "_]*)",
  70                 "\\1", "i,interface,interfaces", NULL);
  71         addTagRegex(language, "^[ \t]*define[ \t]*\\([ \t]*['\"]?([" ALPHA "_][" ALNUM "_]*)",
  72                 "\\1", "d,define,constant definitions", NULL);
  73         addTagRegex(language, "^[ \t]*((static|public|protected|private)[ \t]+)*function[ \t]+&?[ \t]*([" ALPHA "_][" ALNUM "_]*)",
  74                 "\\3", "f,function,functions", NULL);
  75         addTagRegex(language, "^[ \t]*(\\$|::\\$|\\$this->)([" ALPHA "_][" ALNUM "_]*)[ \t]*=",
  76                 "\\2", "v,variable,variables", NULL);
  77         addTagRegex(language, "^[ \t]*((var|public|protected|private|static)[ \t]+)+\\$([" ALPHA "_][" ALNUM "_]*)[ \t]*[=;]",
  78                 "\\3", "v,variable,variables", NULL);
  79
  80         /* function regex is covered by PHP regex */
  81         addTagRegex (language, "(^|[ \t])([A-Za-z0-9_]+)[ \t]*[=:][ \t]*function[ \t]*\\(",
  82                 "\\2", "j,jsfunction,javascript functions", NULL);
  83         addTagRegex (language, "(^|[ \t])([A-Za-z0-9_.]+)\\.([A-Za-z0-9_]+)[ \t]*=[ \t]*function[ \t]*\\(",
  84                 "\\2.\\3", "j,jsfunction,javascript functions", NULL);
  85         addTagRegex (language, "(^|[ \t])([A-Za-z0-9_.]+)\\.([A-Za-z0-9_]+)[ \t]*=[ \t]*function[ \t]*\\(",
  86                 "\\3", "j,jsfunction,javascript functions", NULL);
  87 }
  88
  89 /* Create parser definition structure */
  90 extern parserDefinition* PhpParser (void)
  91 {
  92         static const char *const extensions [] = { "php", "php3", "phtml", NULL };
  93         parserDefinition* def = parserNew ("PHP");
  94         def->extensions = extensions;
  95         def->initialize = installPHPRegex;
  96         def->regex      = TRUE;
  97         return def;
  98 }
  99
 100 #if 0
 101
 102 static boolean isLetter(const int c)
 103 {
 104         return (boolean)(isalpha(c) || (c >= 127  &&  c <= 255));
 105 }
 106
 107 static boolean isVarChar1(const int c)
 108 {
 109         return (boolean)(isLetter (c)  ||  c == '_');
 110 }
 111
 112 static boolean isVarChar(const int c)
 113 {
 114         return (boolean)(isVarChar1 (c) || isdigit (c));
 115 }
 116
 117 static void findPhpTags (void)
 118 {
 119         vString *name = vStringNew ();
 120         const unsigned char *line;
 121
 122         while ((line = fileReadLine ()) != NULL)
 123         {
 124                 const unsigned char *cp = line;
 125                 const char* f;
 126
 127                 while (isspace (*cp))
 128                         cp++;
 129
 130                 if (*(const char*)cp == '$'  &&  isVarChar1 (*(const char*)(cp+1)))
 131                 {
 132                         cp += 1;
 133                         vStringClear (name);
 134                         while (isVarChar ((int) *cp))
 135                         {
 136                                 vStringPut (name, (int) *cp);
 137                                 ++cp;
 138                         }
 139                         while (isspace ((int) *cp))
 140                                 ++cp;
 141                         if (*(const char*) cp == '=')
 142                         {
 143                                 vStringTerminate (name);
 144                                 makeSimpleTag (name, PhpKinds, K_VARIABLE);
 145                                 vStringClear (name);
 146                         }
 147                 }
 148                 else if ((f = strstr ((const char*) cp, "function")) != NULL &&
 149                         (f == (const char*) cp || isspace ((int) f [-1])) &&
 150                         isspace ((int) f [8]))
 151                 {
 152                         cp = ((const unsigned char *) f) + 8;
 153
 154                         while (isspace ((int) *cp))
 155                                 ++cp;
 156
 157                         if (*cp == '&') /* skip reference character and following whitespace */
 158                         {
 159                                 cp++;
 160
 161                                 while (isspace ((int) *cp))
 162                                         ++cp;
 163                         }
 164
 165                         vStringClear (name);
 166                         while (isalnum ((int) *cp)  ||  *cp == '_')
 167                         {
 168                                 vStringPut (name, (int) *cp);
 169                                 ++cp;
 170                         }
 171                         vStringTerminate (name);
 172                         makeSimpleTag (name, PhpKinds, K_FUNCTION);
 173                         vStringClear (name);
 174                 }
 175                 else if (strncmp ((const char*) cp, "class", (size_t) 5) == 0 &&
 176                                  isspace ((int) cp [5]))
 177                 {
 178                         cp += 5;
 179
 180                         while (isspace ((int) *cp))
 181                                 ++cp;
 182                         vStringClear (name);
 183                         while (isalnum ((int) *cp)  ||  *cp == '_')
 184                         {
 185                                 vStringPut (name, (int) *cp);
 186                                 ++cp;
 187                         }
 188                         vStringTerminate (name);
 189                         makeSimpleTag (name, PhpKinds, K_CLASS);
 190                         vStringClear (name);
 191                 }
 192                 else if (strncmp ((const char*) cp, "define", (size_t) 6) == 0 &&
 193                                  ! isalnum ((int) cp [6]))
 194                 {
 195                         cp += 6;
 196
 197                         while (isspace ((int) *cp))
 198                                 ++cp;
 199                         if (*cp != '(')
 200                                 continue;
 201                         ++cp;
 202
 203                         while (isspace ((int) *cp))
 204                                 ++cp;
 205                         if ((*cp == '\'') || (*cp == '"'))
 206                                 ++cp;
 207                         else if (! ((*cp == '_')  || isalnum ((int) *cp)))
 208                                 continue;
 209
 210                         vStringClear (name);
 211                         while (isalnum ((int) *cp)  ||  *cp == '_')
 212                         {
 213                                 vStringPut (name, (int) *cp);
 214                                 ++cp;
 215                         }
 216                         vStringTerminate (name);
 217                         makeSimpleTag (name, PhpKinds, K_DEFINE);
 218                         vStringClear (name);
 219                 }
 220         }
 221         vStringDelete (name);
 222 }
 223
 224 extern parserDefinition* PhpParser (void)
 225 {
 226         static const char *const extensions [] = { "php", "php3", "phtml", NULL };
 227         parserDefinition* def = parserNew ("PHP");
 228         def->kinds      = PhpKinds;
 229         def->kindCount  = KIND_COUNT (PhpKinds);
 230         def->extensions = extensions;
 231         def->parser     = findPhpTags;
 232         return def;
 233 }
 234
 235 #endif
 236
 237 /* vi:set tabstop=4 shiftwidth=4: */