helm/metadata/extractor/meta_lex.l

   1  /******************************************************************/
   2  /*  Copyright (C) 2000, HELM Team                                 */
   3  /*                                                                */
   4  /* This file is part of HELM, an Hypertextual, Electronic         */
   5  /* Library of Mathematics, developed at the Computer Science      */
   6  /* Department, University of Bologna, Italy.                      */
   7  /*                                                                */
   8  /* HELM is free software; you can redistribute it and/or          */
   9  /* modify it under the terms of the GNU General Public License    */
  10  /* as published by the Free Software Foundation; either version   */
  11  /* 2 of the License, or (at your option) any later version.       */
  12  /*                                                                */
  13  /* HELM is distributed in the hope that it will be useful,        */
  14  /* but WITHOUT ANY WARRANTY; without even the implied warranty of */
  15  /* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the   */
  16  /* GNU General Public License for more details.                   */
  17  /*                                                                */
  18  /* You should have received a copy of the GNU General Public      */
  19  /* License along with HELM; if not, write to the Free Software    */
  20  /* Foundation, Inc., 59 Temple Place - Suite 330, Boston,         */
  21  /* MA  02111-1307, USA.                                           */
  22  /*                                                                */
  23  /* For details, see the HELM World-Wide-Web page,                 */
  24  /* http://cs.unibo.it/helm/.                                      */
  25  /******************************************************************/
  26
  27  /***************************************************************/
  28  /*                        META_LEXAN                           */
  29  /*                 Automatic Metadata Extractor                */
  30  /*           First draft 11/12/2001, by Andrea Asperti         */
  31  /*      more bugs added by domenico lordi on mon 12/17/2001    */
  32  /***************************************************************/
  33
  34  /***************************************************************/
  35  /* 1. Inclusion of header files.                               */
  36  /***************************************************************/
  37
  38 %{
  39 #include                <string.h>
  40 #include                <stdlib.h>
  41 #include                <sys/stat.h>
  42 #include                "sthandler.h"
  43 %}
  44
  45  /***************************************************************/
  46  /* 2. Constants and Variables Definitions                      */
  47  /***************************************************************/
  48
  49 %{
  50 #define                 NOWHERE   0
  51 #define                 CONST     1
  52 #define                 MUTIND    2
  53 #define                 MUTCONSTRUCT  3
  54 #define                 SORT      4
  55
  56 #define                 INBODY    0
  57 #define                 MAINHYP   1
  58 #define                 INHYP     2
  59 #define                 INCONCL   3
  60 #define                 MAINCONCL 4
  61 #define                 INTYPE    5
  62 #define                 NOTFOUND  6
  63
  64 #define                 HERE      0
  65 #define                 AFTER     1
  66
  67
  68 int                     where = NOWHERE;
  69 int                     found = NOTFOUND;
  70 int                     position = INBODY;
  71 int                     first_child = HERE;
  72 int                     skip = 0;     // boolean to skip the insertion of a URI
  73 int                     no_open_source = 0;
  74 int                     spine_depth = 0;
  75 int                     depth = 0;
  76 int                     tmp_n;
  77 char                    sep = '"';
  78 char                    *xpointer = "#xpointer(1/";
  79 char                    *uri;
  80 char                    *tmp;
  81
  82 void search(char *uri, int first_child, int position, int depth);
  83 %}
  84
  85  /***************************************************************/
  86  /* 3. Regular definitions.                                     */
  87  /***************************************************************/
  88
  89 uri                     [^"]+
  90 digits                  [0-9]+
  91 value                   [^"]+
  92
  93  /***************************************************************/
  94  /* 4. Rules.                                                   */
  95  /***************************************************************/
  96
  97
  98 %%
  99
 100 "<Variable"[^>]*">"(" "|\n)*"<body" {
 101                     position = INBODY; // Variables have both a body and a type
 102                    }
 103
 104 "</body>"(" "|\n)*"<type" {
 105                     position = INTYPE; // Variables have both a body and a type
 106                     first_child = HERE;
 107                     no_open_source = 0;
 108                     spine_depth = 0;
 109                     depth = 0;
 110                    }
 111
 112 "<decl"            |
 113 "<def"             {
 114                     if (position == INTYPE)
 115                        position = MAINHYP;
 116                     else if (position == MAINHYP)
 117                         { position = INHYP;
 118                           no_open_source++;};
 119                    }
 120
 121 "</decl>"          |
 122 "</def>"           {
 123                     if (position == INHYP)
 124                      {
 125                       no_open_source--;
 126                       if (no_open_source == 0)
 127                         {
 128                          position = MAINHYP;
 129                          depth++;
 130                          first_child = HERE;
 131                         }
 132                      }
 133                     else if (position == MAINHYP)
 134                       {
 135                        position = INTYPE;
 136                        spine_depth++;
 137                        depth = 0;
 138                        first_child = HERE;
 139                       }
 140                     /* bug? first_child = HERE; */
 141                    }
 142
 143
 144 .|\n               {
 145                    }
 146
 147 "<LAMBDA"          |
 148 "<MUTCASE"         |
 149 "<FIX"             |
 150 "<COFIX"           {
 151                           first_child = AFTER;
 152                    }
 153
 154 "<REL"             {
 155                     if (((position == INTYPE) | (position == MAINHYP)) &&
 156                        (first_child == HERE))
 157                      {
 158                        if (position == INTYPE) /* REL on the spine */
 159                          {
 160                            position = INCONCL;
 161                            search("Rel",first_child,position,spine_depth);
 162                          }
 163                        else search("Rel",first_child,position,depth);
 164                        first_child = AFTER;
 165                      }
 166                    }
 167
 168 "<SORT"(" "|\n)+"value=\""{value}   {
 169                     if (((position == INTYPE) | (position == MAINHYP)) &&
 170                        (first_child == HERE))
 171                      {
 172                        tmp=(char *)malloc((sizeof('a')*200));
 173                        strcpy(tmp,yytext);
 174                        strsep(&tmp,&sep);
 175                        if (position == INTYPE) /* SORT on the spine */
 176                          {
 177                            position = INCONCL;
 178                            search(tmp,first_child,position,spine_depth);
 179                          }
 180                        else search(tmp,first_child,position,depth);
 181                        first_child = AFTER;
 182                      }
 183                    }
 184
 185 "<VAR"             {
 186                      skip = 1;
 187                      first_child = AFTER;
 188                    }
 189
 190 "<CONST"           {
 191                      if (position == INTYPE) /* CONST on the spine */
 192                         position = INCONCL;
 193                      where = CONST;
 194                    }
 195
 196 "<MUTIND"          {
 197                      if (position == INTYPE) /* MUTIND on the spine */
 198                         position = INCONCL;
 199                      where = MUTIND;
 200                    }
 201
 202 "<MUTCONSTRUCT"    {
 203                      if (position == INTYPE) /* MUTCONSTRUCT on the spine */
 204                         position = INCONCL;
 205                      where = MUTCONSTRUCT;
 206                    }
 207
 208 "uri=\""{uri}      {
 209                          if (!skip) {
 210                             uri=(char *)malloc((sizeof('a')*200));
 211                             strcpy(uri,yytext);
 212                             strsep(&uri,&sep);
 213                             if (where == CONST)
 214                               {
 215                                 if (position == INCONCL)
 216                                   search(uri,first_child,position,spine_depth);
 217                                 else search(uri,first_child,position,depth);
 218                                 where = NOWHERE;
 219                                 first_child = AFTER;
 220                                 free(uri);
 221                               };
 222                          } else skip = 0;
 223                    }
 224
 225 "noType=\""{digits} {
 226                          if ((where == MUTIND) || (where == MUTCONSTRUCT))
 227                           { strsep(&yytext,&sep);
 228                             tmp=(char *)malloc((sizeof(sep)*(strlen(yytext)+1)));
 229                             strcpy(tmp,yytext);
 230                             tmp_n = atoi(tmp)+1;
 231                             sprintf(tmp,"%d",tmp_n);
 232                             strcat(uri,"#xpointer(1/");
 233                             strcat(uri,tmp);
 234                           };
 235                          if (where == MUTIND)
 236                              {
 237                                strcat(uri,")");
 238                                if (position == INCONCL)
 239                                   search(uri,first_child,position,spine_depth);
 240                                else search(uri,first_child,position,depth);
 241                                free(uri);
 242                                free(tmp);
 243                                where = NOWHERE;
 244                                first_child = AFTER;};
 245                    }
 246
 247 "noConstr=\""{digits} {
 248                          if (where == MUTCONSTRUCT)
 249                           { strsep(&yytext,&sep);
 250                             tmp=(char *)malloc((sizeof(sep)*(strlen(yytext)+1)));
 251                             strcpy(tmp,yytext);
 252                             strcat(uri,"/");
 253                             strcat(uri,tmp);
 254                             strcat(uri,")");
 255                             if (position == INCONCL)
 256                               search(uri,first_child,position,spine_depth);
 257                             else search(uri,first_child,position,depth);
 258                             free(uri);
 259                             free(tmp);
 260                             where = NOWHERE;
 261                             first_child = AFTER;};
 262                    }
 263
 264
 265
 266 %%
 267
 268  /***************************************************************/
 269  /* 6. Auxiliary functions.                                     */
 270  /***************************************************************/
 271
 272 int main(int argc, char *argv[])
 273 {
 274     struct stat buf;
 275
 276     /* FILE *debug; */
 277
 278     if (argc != 4)
 279     {
 280         fprintf(stderr, "Usage: meta <object_uri> <body_file> <type_file>\n");
 281         exit(1);
 282     }
 283
 284
 285     /* initialize the symbol table */
 286     init_symbol_table();
 287
 288     // We process the body
 289     if (!stat(argv[2],&buf))
 290     {
 291         yyin = fopen(argv[2], "r");
 292         position = INBODY;
 293         yylex();
 294         fclose(yyin);
 295      }
 296
 297     // We process the type
 298     yyin = fopen(argv[3], "r");
 299     position = INTYPE;
 300     first_child = HERE;
 301     no_open_source = 0;
 302     spine_depth = 0;
 303     depth = 0;
 304     yylex();
 305     fclose(yyin);
 306     print_all(argv[1]);
 307
 308     return 0;
 309 }
 310
 311
 312 void search(uri,first_child,position,depth)
 313 char               *uri;
 314 int                first_child;
 315 int                position;
 316 {
 317                    if (position == MAINHYP)
 318                       {
 319                        if (first_child == HERE)
 320                            found = search_bucket(uri,MAINHYP,depth);
 321                        else
 322                            found = search_bucket(uri,INHYP,0);
 323                       }
 324                    else if (position == INCONCL)
 325                       {
 326                        if (first_child == HERE)
 327                            found = search_bucket(uri,MAINCONCL,depth);
 328                        else
 329                            found = search_bucket(uri,INCONCL,0);
 330                       }
 331
 332                    else
 333                       found = search_bucket(uri,position,depth);
 334                    /*
 335                    if (found == NOTFOUND)
 336                          fprintf(stderr,"here = %d, pos = %d, uri = %s\n", first_child,position, uri); */
 337 }
 338 /*
 339                       (first_child == HERE)
 340                       {
 341                        if (position == MAINHYP)
 342                           found = search_bucket(uri,MAINHYP,depth);
 343                        else if (position == INCONCL)
 344                           found = search_bucket(uri,MAINCONCL,0);
 345                        else if (position == INHYP)
 346                           found = search_bucket(uri,INHYP,0);
 347                           if (found == NOTFOUND)
 348                           printf( "pos = %d, uri = %s\n", MAINCONCL, uri);
 349                        }
 350                    else if ((position == MAINHYP) && (first_child == AFTER))
 351                         found = search_bucket(uri,INHYP,0);
 352                    else found = search_bucket(uri,position,0);
 353                    if (found == NOTFOUND)
 354                          printf( "pos = %d, uri = %s\n", position, uri);
 355                    } */
 356
 357 int yywrap() {
 358                return 1;
 359              }
 360
 361
 362
 363