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 ethereal
  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          * Ethereal - Network traffic analyzer
  23          * By Gerald Combs <gerald@ethereal.com>
  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 MAX_INCLUDE_DEPTH 10
  48 YY_BUFFER_STATE include_stack[MAX_INCLUDE_DEPTH];
  49 int include_stack_ptr = 0;
  50
  51 #define ECHO g_string_append(current,yytext);
  52
  53 GString* current;
  54 GString* output;
  55 GHashTable* entities;
  56 gchar* entity_name;
  57 GString* error;
  58
  59 const gchar* dirname;
  60 const gchar* filename;
  61 guint linenum;
  62
  63 GString* textstr;
  64
  65 static gchar* replace_entity(gchar* s);
  66 static const gchar* location(void);
  67
  68 %}
  69 xmlpi_start "<?"
  70 xmlpi_stop  "?>"
  71 xmlpi_chars .
  72
  73 comment_start "<!--"
  74 comment_stop "-->"
  75 special_start "<!"
  76 special_stop ">"
  77
  78 entity_start     "<!"[[:blank:]\n]*entity[[:blank:]\n]*"%"
  79 system     SYSTEM
  80 filename   [^"]+
  81
  82
  83 name [A-Za-z][-:A-Za-z0-9_]*
  84
  85 quote "\""
  86 percent [%]
  87 escaped_quote "\\\""
  88 non_quote [^"%]+
  89
  90 avoid_editor_bug ["]
  91
  92 entity        [%&][A-Za-z][-A-Za-z0-9_]*;
  93
  94 whitespace [[blank:]]+
  95 newline    \n
  96 %START OUTSIDE IN_COMMENT IN_ENTITY NAMED_ENTITY IN_QUOTE ENTITY_DONE XMLPI
  97 %%
  98
  99
 100 {entity}                                                if (current) g_string_sprintfa(current,"%s\n%s\n",replace_entity(yytext),location());
 101
 102 {whitespace}                                    if (current) g_string_append(current," ");
 103
 104 <OUTSIDE>{xmlpi_start}                  { g_string_append(current,yytext); BEGIN XMLPI; }
 105 <XMLPI>{xmlpi_chars}                    { g_string_append(current,yytext); }
 106 <XMLPI>{newline}                                { g_string_append(current,yytext); }
 107 <XMLPI>{xmlpi_stop}                             { g_string_append(current,yytext); BEGIN OUTSIDE; }
 108
 109 <OUTSIDE>{comment_start}                { current = NULL; BEGIN IN_COMMENT; }
 110 <IN_COMMENT>[^-]?                               |
 111 <IN_COMMENT>[-]                                 ;
 112 <IN_COMMENT>{comment_stop}              { current = output; BEGIN OUTSIDE; }
 113
 114 {newline}                                               {
 115         linenum++;
 116         if (current) g_string_sprintfa(current,"%s\n",location());
 117 }
 118
 119
 120 <OUTSIDE>{entity_start}                 { BEGIN IN_ENTITY; }
 121 <IN_ENTITY>{name}                               { entity_name = g_strdup_printf("%%%s;",yytext); BEGIN NAMED_ENTITY; }
 122 <NAMED_ENTITY>{quote}                   { current = g_string_new(location()); BEGIN IN_QUOTE; }
 123 <IN_QUOTE>{quote}                               { g_hash_table_insert(entities,entity_name,current);  BEGIN ENTITY_DONE; }
 124 <IN_QUOTE>{percent}                             |
 125 <IN_QUOTE>{non_quote}                   |
 126 <IN_QUOTE>{escaped_quote}               g_string_append(current,yytext);
 127 <NAMED_ENTITY>{system}                  {
 128     g_string_sprintfa(error,"at %s:%u: file inclusion is not supported!", filename, linenum);
 129     yyterminate();
 130 }
 131 <ENTITY_DONE>{special_stop}             { current = output; g_string_append(current,"\n"); BEGIN OUTSIDE; }
 132
 133 %%
 134
 135 static gchar* replace_entity(gchar* entity) {
 136         GString* replacement;
 137
 138         *entity = '%';
 139
 140         replacement = g_hash_table_lookup(entities,entity);
 141
 142         if (replacement) {
 143                 return replacement->str;
 144         } else {
 145                 g_string_sprintfa(error,"dtd_preparse: in file '%s': entity %s does not exists\n", filename, entity);
 146                 return "";
 147         }
 148
 149 }
 150
 151 static const gchar* location(void) {
 152         static GString* loc = NULL;
 153         guint i = include_stack_ptr + 1;
 154
 155         if (loc) {
 156                 g_string_truncate(loc,0);
 157         } else {
 158                 loc = g_string_new("");
 159         }
 160
 161         g_string_sprintfa(loc,"<? ethereal:location ");
 162
 163         while (i--) {
 164                         g_string_sprintfa(loc, "%s:%u from",
 165                                                           filename,
 166                                                           linenum);
 167         }
 168
 169         g_string_truncate(loc,(loc->len) - 4);
 170
 171         g_string_sprintfa(loc,"?>");
 172
 173         return loc->str;
 174 }
 175
 176 static gboolean free_gstring_hash_items(gpointer k,gpointer v,gpointer p _U_) {
 177         g_free(k);
 178         g_string_free(v,TRUE);
 179         return TRUE;
 180 }
 181
 182 extern GString* dtd_preparse(const gchar* dname,const  gchar* fname, GString* err) {
 183         gchar* fullname = g_strdup_printf("%s%c%s",dname,G_DIR_SEPARATOR,fname);
 184
 185         dirname = dname;
 186         filename = fname;
 187
 188         yyin = fopen(fullname,"r");
 189
 190         g_free(fullname);
 191
 192         if (!yyin) {
 193                 if (err)
 194                         g_string_sprintfa(err, "Could not open file: '%s', error: %s",filename,strerror(errno));
 195
 196                 return NULL;
 197         }
 198
 199         filename = filename;
 200         linenum = 1;
 201
 202         error = err;
 203
 204         entities = g_hash_table_new(g_str_hash,g_str_equal);
 205         current = output = g_string_new(location());
 206
 207         BEGIN OUTSIDE;
 208
 209         yylex();
 210
 211         yyrestart(NULL);
 212
 213         g_hash_table_foreach_remove(entities,free_gstring_hash_items,NULL);
 214         g_hash_table_destroy(entities);
 215
 216         return output;
 217 }