epan/dtd_preparse.l

   1 %option noyywrap
   2 %option nounput
   3 %option prefix="Dtd_PreParse_"
   4 %option never-interactive
   5 %option caseless
   6 %option outfile="dtd_preparse.c"
   7
   8 %{
   9         /*
  10          * dtd_preparser.l
  11          *
  12          * an XML dissector for wireshark
  13          *
  14          * DTD Preparser -  import a dtd file into a GString
  15          *                                      including files, removing comments
  16          *                  and resolving %entities;
  17          *
  18          * Copyright 2004, Luis E. Garcia Ontanon <luis.ontanon@gmail.com>
  19          *
  20          * $Id$
  21          *
  22          * Wireshark - Network traffic analyzer
  23          * By Gerald Combs <gerald@wireshark.org>
  24          * Copyright 1998 Gerald Combs
  25          *
  26          * This program is free software; you can redistribute it and/or
  27          * modify it under the terms of the GNU General Public License
  28          * as published by the Free Software Foundation; either version 2
  29          * of the License, or (at your option) any later version.
  30          *
  31          * This program is distributed in the hope that it will be useful,
  32          * but WITHOUT ANY WARRANTY; without even the implied warranty of
  33          * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  34          * GNU General Public License for more details.
  35          *
  36          * You should have received a copy of the GNU General Public License
  37          * along with this program; if not, write to the Free Software
  38          * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
  39          */
  40
  41 #include <glib.h>
  42 #include <string.h>
  43 #include <errno.h>
  44 #include <stdio.h>
  45 #include "dtd.h"
  46
  47 #define ECHO g_string_append(current,yytext);
  48
  49 static GString* current;
  50 static GString* output;
  51 static GHashTable* entities;
  52 static gchar* entity_name;
  53 static GString* error;
  54
  55 static const gchar* dtd_dirname;
  56 static const gchar* filename;
  57 static guint linenum;
  58
  59 static gchar* replace_entity(gchar* s);
  60 static const gchar* location(void);
  61
  62 %}
  63 xmlpi_start "<?"
  64 xmlpi_stop  "?>"
  65 xmlpi_chars .
  66
  67 comment_start "<!--"
  68 comment_stop "-->"
  69 special_start "<!"
  70 special_stop ">"
  71
  72 entity_start     "<!"[[:blank:]\n]*entity[[:blank:]\n]*"%"
  73 system     SYSTEM
  74 filename   [^"]+
  75
  76
  77 name [A-Za-z][-:A-Za-z0-9_\.]*
  78
  79 quote "\""
  80 percent [%]
  81 escaped_quote "\\\""
  82 non_quote [^"%]+
  83
  84 avoid_editor_bug ["]
  85
  86 entity        [%&][A-Za-z][-A-Za-z0-9_]*;
  87
  88 whitespace [[blank:]]+
  89 newline    \n
  90 %START OUTSIDE IN_COMMENT IN_ENTITY NAMED_ENTITY IN_QUOTE ENTITY_DONE XMLPI
  91 %%
  92
  93
  94 {entity}                                                if (current) g_string_sprintfa(current,"%s\n%s\n",replace_entity(yytext),location());
  95
  96 {whitespace}                                    if (current) g_string_append(current," ");
  97
  98 <OUTSIDE>{xmlpi_start}                  { g_string_append(current,yytext); BEGIN XMLPI; }
  99 <XMLPI>{xmlpi_chars}                    { g_string_append(current,yytext); }
 100 <XMLPI>{newline}                                { g_string_append(current,yytext); }
 101 <XMLPI>{xmlpi_stop}                             { g_string_append(current,yytext); BEGIN OUTSIDE; }
 102
 103 <OUTSIDE>{comment_start}                { current = NULL; BEGIN IN_COMMENT; }
 104 <IN_COMMENT>[^-]?                               |
 105 <IN_COMMENT>[-]                                 ;
 106 <IN_COMMENT>{comment_stop}              { current = output; BEGIN OUTSIDE; }
 107
 108 {newline}                                               {
 109         linenum++;
 110         if (current) g_string_sprintfa(current,"%s\n",location());
 111 }
 112
 113
 114 <OUTSIDE>{entity_start}                 { BEGIN IN_ENTITY; }
 115 <IN_ENTITY>{name}                               { entity_name = g_strdup_printf("%%%s;",yytext); BEGIN NAMED_ENTITY; }
 116 <NAMED_ENTITY>{quote}                   { current = g_string_new(location()); BEGIN IN_QUOTE; }
 117 <IN_QUOTE>{quote}                               { g_hash_table_insert(entities,entity_name,current);  BEGIN ENTITY_DONE; }
 118 <IN_QUOTE>{percent}                             |
 119 <IN_QUOTE>{non_quote}                   |
 120 <IN_QUOTE>{escaped_quote}               g_string_append(current,yytext);
 121 <NAMED_ENTITY>{system}                  {
 122     g_string_sprintfa(error,"at %s:%u: file inclusion is not supported!", filename, linenum);
 123     yyterminate();
 124 }
 125 <ENTITY_DONE>{special_stop}             { current = output; g_string_append(current,"\n"); BEGIN OUTSIDE; }
 126
 127 %%
 128
 129 static gchar* replace_entity(gchar* entity) {
 130         GString* replacement;
 131
 132         *entity = '%';
 133
 134         replacement = g_hash_table_lookup(entities,entity);
 135
 136         if (replacement) {
 137                 return replacement->str;
 138         } else {
 139                 g_string_sprintfa(error,"dtd_preparse: in file '%s': entity %s does not exists\n", filename, entity);
 140                 return "";
 141         }
 142
 143 }
 144
 145 static const gchar* location(void) {
 146         static gchar* loc = NULL;
 147
 148         if (loc) g_free(loc);
 149
 150         loc = g_strdup_printf("<? wireshark:location %s:%u ?>", filename, linenum);
 151
 152         return loc;
 153 }
 154
 155 static gboolean free_gstring_hash_items(gpointer k,gpointer v,gpointer p _U_) {
 156         g_free(k);
 157         g_string_free(v,TRUE);
 158         return TRUE;
 159 }
 160
 161 extern GString* dtd_preparse(const gchar* dname,const  gchar* fname, GString* err) {
 162         gchar* fullname = g_strdup_printf("%s%c%s",dname,G_DIR_SEPARATOR,fname);
 163
 164         dtd_dirname = dname;
 165         filename = fname;
 166         linenum = 1;
 167
 168         yyin = fopen(fullname,"r");
 169
 170         if (!yyin) {
 171                 if (err)
 172                         g_string_sprintfa(err, "Could not open file: '%s', error: %s",fullname,strerror(errno));
 173
 174                 return NULL;
 175         }
 176
 177         error = err;
 178
 179         entities = g_hash_table_new(g_str_hash,g_str_equal);
 180         current = output = g_string_new(location());
 181
 182         BEGIN OUTSIDE;
 183
 184         yylex();
 185
 186         fclose(yyin);
 187
 188         yyrestart(NULL);
 189
 190         g_hash_table_foreach_remove(entities,free_gstring_hash_items,NULL);
 191         g_hash_table_destroy(entities);
 192
 193     g_free(fullname);
 194
 195         return output;
 196 }