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 gchar* dirname;
  60 gchar* filename;
  61 guint linenum;
  62
  63 static gchar* replace_entity(gchar* s);
  64 static const gchar* location(void);
  65 static gchar* load_entity_file(gchar* filename);
  66                 /* [:blank:]+file[:blank:]*=[:blank:]*["] */
  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 GET_FNAME_OPEN_QUOTE GET_FNAME GET_FNAME_CLOSE_QUOTE 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}                  { BEGIN GET_FNAME_OPEN_QUOTE; }
 128 <GET_FNAME_OPEN_QUOTE>{quote}   { BEGIN GET_FNAME; }
 129 <GET_FNAME>{filename}               {  g_hash_table_insert(entities,entity_name,load_entity_file(yytext));  BEGIN GET_FNAME_CLOSE_QUOTE; }
 130 <GET_FNAME_CLOSE_QUOTE>{quote}  { BEGIN ENTITY_DONE; }
 131 <ENTITY_DONE>{special_stop}             { current = output; g_string_append(current,"\n"); BEGIN OUTSIDE; }
 132
 133 %%
 134
 135 static gchar* load_entity_file(gchar* fname) {
 136         gchar* fullname = g_strdup_printf("%s%s",dirname,fname);
 137         gchar* save_filename = filename;
 138         guint save_linenum = linenum;
 139         FILE* fp = fopen(fullname,"r");
 140         GString* filetext;
 141         gchar* retstr;
 142         gchar c;
 143
 144         g_free(fullname);
 145
 146         if (!fp) {
 147                 g_string_sprintfa(error,"at %s:%u: could not load file %s: %s", filename, linenum, fname, strerror(errno));
 148                 return "";
 149         }
 150
 151         filename = fname;
 152         linenum = 1;
 153
 154         filetext = g_string_new(location());
 155
 156         while(( c = fgetc(fp) )) {
 157                 g_string_append_c(filetext,c);
 158                 if(c == '\n') {
 159                         g_string_append(filetext,location());
 160                         linenum++;
 161                 }
 162         }
 163
 164         retstr = filetext->str;
 165         g_string_free(filetext,FALSE);
 166
 167         if ( ferror(fp) ) {
 168                 g_string_sprintfa(error,"at %s:%u: problem reading file %s: %s", filename, linenum, fname, strerror(errno));
 169         }
 170
 171         filename = save_filename;
 172         save_linenum = linenum;
 173
 174         return retstr;
 175 }
 176
 177 static gchar* replace_entity(gchar* entity) {
 178         GString* replacement;
 179
 180         *entity = '%';
 181
 182         replacement = g_hash_table_lookup(entities,entity);
 183
 184         if (replacement) {
 185                 return replacement->str;
 186         } else {
 187                 g_string_sprintfa(error,"dtd_preparse: in file '%s': %s does not exists\n", filename, entity);
 188                 return "";
 189         }
 190
 191 }
 192
 193 static const gchar* location(void) {
 194         static GString* loc = NULL;
 195         guint i = include_stack_ptr + 1;
 196
 197         if (loc) {
 198                 g_string_truncate(loc,0);
 199         } else {
 200                 loc = g_string_new("");
 201         }
 202
 203         g_string_sprintfa(loc,"<? ethereal:location ");
 204
 205         while (i--) {
 206                         g_string_sprintfa(loc, "%s:%u from",
 207                                                           filename,
 208                                                           linenum);
 209         }
 210
 211         g_string_truncate(loc,(loc->len) - 4);
 212
 213         g_string_sprintfa(loc,"?>");
 214
 215         return loc->str;
 216 }
 217
 218 static gboolean free_gstring_hash_items(gpointer k,gpointer v,gpointer p _U_) {
 219         g_free(k);
 220         g_string_free(v,TRUE);
 221         return TRUE;
 222 }
 223
 224 extern GString* dtd_preparse(gchar* dname, gchar* fname, GString* err) {
 225         gchar* fullname = g_strdup_printf("%s%s",dname,fname);
 226
 227         dirname = dname;
 228         filename = fname;
 229
 230         yyin = fopen(fullname,"r");
 231
 232         g_free(fullname);
 233
 234         if (!yyin) {
 235                 if (err)
 236                         g_string_sprintfa(err, "Could not open file: '%s', error: %s",filename,strerror(errno));
 237
 238                 return NULL;
 239         }
 240
 241         filename = filename;
 242         linenum = 1;
 243
 244         error = err;
 245
 246         entities = g_hash_table_new(g_str_hash,g_str_equal);
 247         current = output = g_string_new(location());
 248
 249         BEGIN OUTSIDE;
 250
 251         yylex();
 252
 253         yyrestart(NULL);
 254
 255         g_hash_table_foreach_remove(entities,free_gstring_hash_items,NULL);
 256         g_hash_table_destroy(entities);
 257
 258         return output;
 259 }