metadata/extractor/meta_lex_ind.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 int                     inductive_type = 0;
  78 int                     constructor = 0;
  79 int                     deep_type = 0;
  80 char                    sep = '"';
  81 char                    *xpointer = "#xpointer(1/";
  82 char                    *uri;
  83 char                    *tmp;
  84 char                    *source_uri;
  85 char                    *source_uri_prefix;
  86 int                     waiting_for_name = 0;
  87
  88 void search(char *uri, int first_child, int position, int depth);
  89 %}
  90
  91  /***************************************************************/
  92  /* 3. Regular definitions.                                     */
  93  /***************************************************************/
  94
  95 uri                     [^"]+
  96 digits                  [0-9]+
  97 value                   [^"]+
  98 id                      [a-zA-Z]([-_'a-zA-Z0-9])*
  99
 100  /***************************************************************/
 101  /* 4. Rules.                                                   */
 102  /***************************************************************/
 103
 104
 105 %%
 106
 107 "<InductiveType"   {
 108                      /* fprintf(stderr,"uno"); */
 109                      init_symbol_table();
 110                      no_open_source = 0;
 111                      depth = 0;
 112                      spine_depth = 0;
 113                      /* fprintf(stderr,"due"); */
 114                      inductive_type++;
 115                      constructor=0;
 116                      position = INTYPE;
 117                      first_child = HERE;
 118                      waiting_for_name = 1;
 119
 120                      tmp = (char *)malloc(sizeof('a')*128);
 121                      strcpy(source_uri,source_uri_prefix);
 122                      sprintf(tmp,"#xpointer(1/%d)", inductive_type);
 123                      strcat(source_uri,tmp);
 124                      /* fprintf(stderr,"cinque"); */
 125                      free(tmp);
 126                    }
 127
 128 "</arity>"         {
 129                      print_all(source_uri);
 130                      /* print_file(); */
 131                    }
 132
 133 "<Constructor"     { init_symbol_table();
 134                      no_open_source = 0;
 135                      depth = 0;
 136                      spine_depth = 0;
 137                      constructor++;
 138                      position = INTYPE;
 139                      first_child = HERE;
 140                      waiting_for_name = 1;
 141
 142                      tmp = (char *)malloc(sizeof('a')*128);
 143                      strcpy(source_uri,source_uri_prefix);
 144                      sprintf(tmp,"#xpointer(1/%d/%d)",inductive_type,constructor);
 145                      strcat(source_uri,tmp);
 146                      free(tmp);
 147                    }
 148
 149 "</Constructor>"   { print_all(source_uri);
 150                      /* print_file(); */
 151                    }
 152
 153 "<decl"            |
 154 "<def"            {
 155                     if (position == INTYPE)
 156                        position = MAINHYP;
 157                     else if (position == MAINHYP)
 158                         { position = INHYP;
 159                           no_open_source = 1;}
 160                     else if (position == INHYP) no_open_source++;
 161                    }
 162
 163 "</decl>"          |
 164 "</def>"           {
 165                     if (position == INHYP)
 166                      {
 167                       no_open_source--;
 168                       if (no_open_source == 0)
 169                         {
 170                          position = MAINHYP;
 171                          depth++;
 172                          first_child = HERE;
 173                         }
 174                      }
 175                     else if (position == MAINHYP)
 176                       {
 177                        position = INTYPE;
 178                        spine_depth++;
 179                        depth = 0;
 180                        first_child = HERE;
 181                       }
 182                    }
 183
 184
 185 .|\n               {
 186                    }
 187
 188 "<LAMBDA"          |
 189 "<MUTCASE"         |
 190 "<FIX"             |
 191 "<COFIX"           {
 192                           first_child = AFTER;
 193                    }
 194
 195 "<REL"             {
 196                     if (((position == INTYPE) | (position == MAINHYP)) &&
 197                        (first_child == HERE))
 198                      {
 199                        if (position == INTYPE) /* REL on the spine */
 200                          {
 201                            position = INCONCL;
 202                            search("Rel",first_child,position,spine_depth);
 203                          }
 204                        else search("Rel",first_child,position,depth);
 205                        first_child = AFTER;
 206                      }
 207                    }
 208
 209 "<SORT"(" "|\n)+"value=\""{value}   {
 210                     if (((position == INTYPE) | (position == MAINHYP)) &&
 211                        (first_child == HERE))
 212                      {
 213                        tmp=(char *)malloc((sizeof('a')*200));
 214                        strcpy(tmp,yytext);
 215                        strsep(&tmp,&sep);
 216                        if (position == INTYPE) /* SORT on the spine */
 217                          {
 218                            position = INCONCL;
 219                            search(tmp,first_child,position,spine_depth);
 220                          }
 221                        else search(tmp,first_child,position,depth);
 222                        first_child = AFTER;
 223                      }
 224                    }
 225
 226 "<VAR"             {
 227                      skip = 1;
 228                      first_child = AFTER;
 229                    }
 230
 231 "<CONST"           {
 232                      if (position == INTYPE) /* CONST on the spine */
 233                         position = INCONCL;
 234                      where = CONST;
 235                    }
 236
 237 "<MUTIND"          {
 238                      if (position == INTYPE) /* MUTIND on the spine */
 239                         position = INCONCL;
 240                      where = MUTIND;
 241                    }
 242
 243 "<MUTCONSTRUCT"    {
 244                      if (position == INTYPE) /* MUTCONSTRUCT on the spine */
 245                         position = INCONCL;
 246                      where = MUTCONSTRUCT;
 247                    }
 248
 249 "uri=\""{uri}      {
 250                          if (!skip) {
 251                             uri=(char *)malloc((sizeof('a')*200));
 252                             strcpy(uri,yytext);
 253                             strsep(&uri,&sep);
 254                             if (where == CONST)
 255                               {
 256                                 if (position == INCONCL)
 257                                   search(uri,first_child,position,spine_depth);
 258                                 else search(uri,first_child,position,depth);
 259                                 where = NOWHERE;
 260                                 first_child = AFTER;
 261                                 free(uri);
 262                               };
 263                          } else skip = 0;
 264                    }
 265
 266 "noType=\""{digits} {
 267                          if ((where == MUTIND) || (where == MUTCONSTRUCT))
 268                           { strsep(&yytext,&sep);
 269                             tmp=(char *)malloc((sizeof(sep)*(strlen(yytext)+1)));
 270                             strcpy(tmp,yytext);
 271                             tmp_n = atoi(tmp)+1;
 272                             sprintf(tmp,"%d",tmp_n);
 273                             strcat(uri,"#xpointer(1/");
 274                             strcat(uri,tmp);
 275                           };
 276                          if (where == MUTIND)
 277                              {
 278                                strcat(uri,")");
 279                                if (position == INCONCL)
 280                                   search(uri,first_child,position,spine_depth);
 281                                else search(uri,first_child,position,depth);
 282                                free(uri);
 283                                free(tmp);
 284                                where = NOWHERE;
 285                                first_child = AFTER;};
 286                    }
 287
 288 "noConstr=\""{digits} {
 289                          if (where == MUTCONSTRUCT)
 290                           { strsep(&yytext,&sep);
 291                             tmp=(char *)malloc((sizeof(sep)*(strlen(yytext)+1)));
 292                             strcpy(tmp,yytext);
 293                             strcat(uri,"/");
 294                             strcat(uri,tmp);
 295                             strcat(uri,")");
 296                             if (position == INCONCL)
 297                               search(uri,first_child,position,spine_depth);
 298                             else search(uri,first_child,position,depth);
 299                             free(uri);
 300                             free(tmp);
 301                             where = NOWHERE;
 302                             first_child = AFTER;};
 303                    }
 304
 305 "name=\""{id} {
 306                  if (waiting_for_name == 1) {
 307                     waiting_for_name = 0;
 308                     strsep(&yytext,&sep);
 309                     tmp=(char *)malloc((sizeof(sep)*(strlen(yytext)+1)));
 310                     strcpy(tmp,yytext);
 311                     print_name(tmp,source_uri);
 312                     free(tmp);
 313                  }
 314               }
 315
 316
 317 %%
 318
 319  /***************************************************************/
 320  /* 6. Auxiliary functions.                                     */
 321  /***************************************************************/
 322
 323 int main(int argc, char *argv[])
 324 {
 325     /* FILE *debug; */
 326
 327     if (argc != 3)
 328     {
 329         fprintf(stderr, "Usage: meta_ind <object_uri> <inductive_type_file>\n");
 330         exit(1);
 331     }
 332
 333     source_uri = malloc((sizeof('a')*2000));
 334     source_uri_prefix=argv[1];
 335     /* fprintf(stderr,"qua"); */
 336     yyin = fopen(argv[2], "r");
 337     yylex();
 338
 339     return 0;
 340 }
 341
 342 void search(uri,first_child,position,depth)
 343 char               *uri;
 344 int                first_child;
 345 int                position;
 346 {
 347                    if (position == MAINHYP)
 348                       {
 349                        if (first_child == HERE)
 350                            found = search_bucket(uri,MAINHYP,depth);
 351                        else
 352                            found = search_bucket(uri,INHYP,0);
 353                       }
 354                    else if (position == INCONCL)
 355                       {
 356                        if (first_child == HERE)
 357                            found = search_bucket(uri,MAINCONCL,depth);
 358                        else
 359                            found = search_bucket(uri,INCONCL,0);
 360                       }
 361
 362                    else
 363                       found = search_bucket(uri,position,depth);
 364                    /* if (found == NOTFOUND)
 365                          printf( "pos = %d, uri = %s\n", position, uri);  */
 366 }
 367
 368
 369 int yywrap() {
 370                return 1;
 371              }
 372
 373
 374
 375