helm/metadata/create_V7_mowgli/METADATA/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
  83  /***************************************************************/
  84  /* 3. Regular definitions.                                     */
  85  /***************************************************************/
  86
  87 uri                     [^"]+
  88 digits                  [0-9]+
  89 value                   [^"]+
  90
  91  /***************************************************************/
  92  /* 4. Rules.                                                   */
  93  /***************************************************************/
  94
  95
  96 %%
  97
  98 "<Variable"[^>]*">"(" "|\n)*"<body" {
  99                     position = INBODY; // Variables have both a body and a type
 100                    }
 101
 102 "</body>"(" "|\n)*"<type" {
 103                     position = INTYPE; // Variables have both a body and a type
 104                    }
 105
 106 "<decl"            |
 107 "<def"             {
 108                     if (position == INTYPE)
 109                        position = MAINHYP;
 110                     else if (position == MAINHYP)
 111                         { position = INHYP;
 112                           no_open_source++;};
 113                    }
 114
 115 "</decl>"          |
 116 "</def"            {
 117                     if (position == INHYP)
 118                      {
 119                       no_open_source--;
 120                       if (no_open_source == 0)
 121                         {
 122                          position = MAINHYP;
 123                          depth++;
 124                          first_child = HERE;
 125                         }
 126                      }
 127                     else if (position == MAINHYP)
 128                       {
 129                        position = INTYPE;
 130                        spine_depth++;
 131                        depth = 0;
 132                       }
 133                     first_child = HERE;
 134                    }
 135
 136
 137 .|\n               {
 138                    }
 139
 140 "<LAMBDA"          |
 141 "<MUTCASE"         |
 142 "<FIX"             |
 143 "<COFIX"           {
 144                           first_child = AFTER;
 145                    }
 146
 147 "<REL"             {
 148                     if (((position == INTYPE) | (position == MAINHYP)) &&
 149                        (first_child == HERE))
 150                      {
 151                        if (position == INTYPE) /* REL on the spine */
 152                          {
 153                            position = INCONCL;
 154                            search("Rel",first_child,position,spine_depth);
 155                          }
 156                        else search("Rel",first_child,position,depth);
 157                        first_child = AFTER;
 158                      }
 159                    }
 160
 161 "<SORT"(" "|\n)+"value=\""{value}   {
 162                     if ((position == INTYPE) | (position == MAINHYP))
 163                      {
 164                        tmp=(char *)malloc((sizeof('a')*200));
 165                        strcpy(tmp,yytext);
 166                        strsep(&tmp,&sep);
 167                        if (position == INTYPE) /* SORT on the spine */
 168                          {
 169                            position = INCONCL;
 170                            search(tmp,first_child,position,spine_depth);
 171                          }
 172                        else search(tmp,first_child,position,depth);
 173                        first_child = AFTER;
 174                      }
 175                    }
 176
 177 "<VAR"             {
 178                      skip = 1;
 179                      first_child = AFTER;
 180                    }
 181
 182 "<CONST"           {
 183                      if (position == INTYPE) /* CONST on the spine */
 184                         position = INCONCL;
 185                      where = CONST;
 186                    }
 187
 188 "<MUTIND"          {
 189                      if (position == INTYPE) /* MUTIND on the spine */
 190                         position = INCONCL;
 191                      where = MUTIND;
 192                    }
 193
 194 "<MUTCONSTRUCT"    {
 195                      if (position == INTYPE) /* MUTCONSTRUCT on the spine */
 196                         position = INCONCL;
 197                      where = MUTCONSTRUCT;
 198                    }
 199
 200 "uri=\""{uri}      {
 201                          if (!skip) {
 202                             uri=(char *)malloc((sizeof('a')*200));
 203                             strcpy(uri,yytext);
 204                             strsep(&uri,&sep);
 205                             if (where == CONST)
 206                               {
 207                                 if (position == INCONCL)
 208                                   search(uri,first_child,position,spine_depth);
 209                                 else search(uri,first_child,position,depth);
 210                                 where = NOWHERE;
 211                                 first_child = AFTER;
 212                                 free(uri);
 213                               };
 214                          } else skip = 0;
 215                    }
 216
 217 "noType=\""{digits} {
 218                          if ((where == MUTIND) || (where == MUTCONSTRUCT))
 219                           { strsep(&yytext,&sep);
 220                             tmp=(char *)malloc((sizeof(sep)*(strlen(yytext)+1)));
 221                             strcpy(tmp,yytext);
 222                             tmp_n = atoi(tmp)+1;
 223                             sprintf(tmp,"%d",tmp_n);
 224                             strcat(uri,"#xpointer(1/");
 225                             strcat(uri,tmp);
 226                           };
 227                          if (where == MUTIND)
 228                              {
 229                                strcat(uri,")");
 230                                if (position == INCONCL)
 231                                   search(uri,first_child,position,spine_depth);
 232                                else search(uri,first_child,position,depth);
 233                                free(uri);
 234                                free(tmp);
 235                                where = NOWHERE;
 236                                first_child = AFTER;};
 237                    }
 238
 239 "noConstr=\""{digits} {
 240                          if (where == MUTCONSTRUCT)
 241                           { strsep(&yytext,&sep);
 242                             tmp=(char *)malloc((sizeof(sep)*(strlen(yytext)+1)));
 243                             strcpy(tmp,yytext);
 244                             strcat(uri,"/");
 245                             strcat(uri,tmp);
 246                             strcat(uri,")");
 247                             if (position == INCONCL)
 248                               search(uri,first_child,position,spine_depth);
 249                             else search(uri,first_child,position,depth);
 250                             free(uri);
 251                             free(tmp);
 252                             where = NOWHERE;
 253                             first_child = AFTER;};
 254                    }
 255
 256
 257
 258 %%
 259
 260  /***************************************************************/
 261  /* 6. Auxiliary functions.                                     */
 262  /***************************************************************/
 263
 264 main(int argc, char *argv[])
 265 {
 266                    struct stat buf;
 267                    FILE *outrel, *outsort;
 268
 269                    init_symbol_table();
 270                    if (!(outrel = fopen("forward_rel.rdf","a")))
 271                      {
 272                       fprintf(stderr, "error in openinf file forward_rel.rdf\n");
 273                       exit(-1);
 274                      }
 275                    if (!(outsort = fopen("forward_sort.rdf","a")))
 276                      {
 277                       fprintf(stderr, "error in openinf file forward_rel.rdf\n");
 278                       exit(-1);
 279                      }
 280                    // We process the body
 281                    if (!stat("tmp/body.xml",&buf)) {
 282                       yyin = fopen("tmp/body.xml", "r");
 283                       position = INBODY;
 284                       yylex();
 285                       fclose(yyin);
 286                    }
 287
 288                    // We process the type
 289                    yyin = fopen("tmp/type.xml", "r");
 290                    position = INTYPE;
 291                    first_child = HERE;
 292                    no_open_source = 0;
 293                    spine_depth = 0;
 294                    depth = 0;
 295                    yylex();
 296
 297                    printf("<?xml version=\"1.0\" encoding=\"ISO-8859-1\"?>\n\n");
 298 printf("<!DOCTYPE rdf:RDF [
 299         <!ENTITY rdfns 'http://www.w3.org/1999/02/22-rdf-syntax-ns#'>
 300         <!ENTITY hthns 'http://www.cs.unibo.it/helm/schemas/schema-helmth#'>
 301         <!ENTITY hns 'http://www.cs.unibo.it/helm/schemas/schema-helm#'>
 302
 303    ]>\n");
 304
 305                    printf("<rdf:RDF xml:lang=\"en\" xmlns:rdf=\"&rdfns;\" xmlns:h=\"&hns;\" xmlns:hth=\"&hthns;\">\n");
 306                    printf("\t<h:Object rdf:about=\"");
 307                    printf("%s",argv[1]);
 308                    printf("\">\n");
 309                    print_all(argv[1],outrel,outsort);
 310                    printf("\t</h:Object>\n");
 311                    printf("</rdf:RDF>\n");
 312                    fclose(yyin);
 313                    }
 314
 315 search(uri,first_child,position,depth)
 316 char               *uri;
 317 int                first_child;
 318 int                position;
 319 {
 320                    if (position == MAINHYP)
 321                       {
 322                        if (first_child == HERE)
 323                            found = search_bucket(uri,MAINHYP,depth);
 324                        else
 325                            found = search_bucket(uri,INHYP,0);
 326                       }
 327                    else if (position == INCONCL)
 328                       {
 329                        if (first_child == HERE)
 330                            found = search_bucket(uri,MAINCONCL,depth);
 331                        else
 332                            found = search_bucket(uri,INCONCL,0);
 333                       }
 334
 335                    else
 336                       found = search_bucket(uri,position,depth);
 337                    /* if (found == NOTFOUND)
 338                          fprintf(stderr,"pos = %d, uri = %s\n", position, uri); */
 339 }
 340 /*
 341                       (first_child == HERE)
 342                       {
 343                        if (position == MAINHYP)
 344                           found = search_bucket(uri,MAINHYP,depth);
 345                        else if (position == INCONCL)
 346                           found = search_bucket(uri,MAINCONCL,0);
 347                        else if (position == INHYP)
 348                           found = search_bucket(uri,INHYP,0);
 349                           if (found == NOTFOUND)
 350                           printf( "pos = %d, uri = %s\n", MAINCONCL, uri);
 351                        }
 352                    else if ((position == MAINHYP) && (first_child == AFTER))
 353                         found = search_bucket(uri,INHYP,0);
 354                    else found = search_bucket(uri,position,0);
 355                    if (found == NOTFOUND)
 356                          printf( "pos = %d, uri = %s\n", position, uri);
 357                    } */
 358
 359 int yywrap() {
 360                return 1;
 361              }
 362