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                <postgresql/libpq-fe.h>
  43 #include                "sthandler.h"
  44 %}
  45
  46  /***************************************************************/
  47  /* 2. Constants and Variables Definitions                      */
  48  /***************************************************************/
  49
  50 %{
  51 #define                 NOWHERE   0
  52 #define                 CONST     1
  53 #define                 MUTIND    2
  54 #define                 MUTCONSTRUCT  3
  55 #define                 SORT      4
  56
  57 #define                 INBODY    0
  58 #define                 MAINHYP   1
  59 #define                 INHYP     2
  60 #define                 INCONCL   3
  61 #define                 MAINCONCL 4
  62 #define                 INTYPE    5
  63 #define                 NOTFOUND  6
  64
  65 #define                 HERE      0
  66 #define                 AFTER     1
  67
  68
  69 int                     where = NOWHERE;
  70 int                     found = NOTFOUND;
  71 int                     position = INBODY;
  72 int                     first_child = HERE;
  73 int                     skip = 0;     // boolean to skip the insertion of a URI
  74 int                     no_open_source = 0;
  75 int                     spine_depth = 0;
  76 int                     depth = 0;
  77 int                     tmp_n;
  78 char                    sep = '"';
  79 char                    *xpointer = "#xpointer(1/";
  80 char                    *uri;
  81 char                    *tmp;
  82 %}
  83
  84  /***************************************************************/
  85  /* 3. Regular definitions.                                     */
  86  /***************************************************************/
  87
  88 uri                     [^"]+
  89 digits                  [0-9]+
  90 value                   [^"]+
  91
  92  /***************************************************************/
  93  /* 4. Rules.                                                   */
  94  /***************************************************************/
  95
  96
  97 %%
  98
  99 "<Variable"[^>]*">"(" "|\n)*"<body" {
 100                     position = INBODY; // Variables have both a body and a type
 101                    }
 102
 103 "</body>"(" "|\n)*"<type" {
 104                     position = INTYPE; // Variables have both a body and a type
 105                     first_child = HERE;
 106                     no_open_source = 0;
 107                     spine_depth = 0;
 108                     depth = 0;
 109                    }
 110
 111 "<decl"            |
 112 "<def"             {
 113                     if (position == INTYPE)
 114                        position = MAINHYP;
 115                     else if (position == MAINHYP)
 116                         { position = INHYP;
 117                           no_open_source++;};
 118                    }
 119
 120 "</decl>"          |
 121 "</def>"           {
 122                     if (position == INHYP)
 123                      {
 124                       no_open_source--;
 125                       if (no_open_source == 0)
 126                         {
 127                          position = MAINHYP;
 128                          depth++;
 129                          first_child = HERE;
 130                         }
 131                      }
 132                     else if (position == MAINHYP)
 133                       {
 134                        position = INTYPE;
 135                        spine_depth++;
 136                        depth = 0;
 137                        first_child = HERE;
 138                       }
 139                     /* bug? first_child = HERE; */
 140                    }
 141
 142
 143 .|\n               {
 144                    }
 145
 146 "<LAMBDA"          |
 147 "<MUTCASE"         |
 148 "<FIX"             |
 149 "<COFIX"           {
 150                           first_child = AFTER;
 151                    }
 152
 153 "<REL"             {
 154                     if (((position == INTYPE) | (position == MAINHYP)) &&
 155                        (first_child == HERE))
 156                      {
 157                        if (position == INTYPE) /* REL on the spine */
 158                          {
 159                            position = INCONCL;
 160                            search("Rel",first_child,position,spine_depth);
 161                          }
 162                        else search("Rel",first_child,position,depth);
 163                        first_child = AFTER;
 164                      }
 165                    }
 166
 167 "<SORT"(" "|\n)+"value=\""{value}   {
 168                     if (((position == INTYPE) | (position == MAINHYP)) &&
 169                        (first_child == HERE))
 170                      {
 171                        tmp=(char *)malloc((sizeof('a')*200));
 172                        strcpy(tmp,yytext);
 173                        strsep(&tmp,&sep);
 174                        if (position == INTYPE) /* SORT on the spine */
 175                          {
 176                            position = INCONCL;
 177                            search(tmp,first_child,position,spine_depth);
 178                          }
 179                        else search(tmp,first_child,position,depth);
 180                        first_child = AFTER;
 181                      }
 182                    }
 183
 184 "<VAR"             {
 185                      skip = 1;
 186                      first_child = AFTER;
 187                    }
 188
 189 "<CONST"           {
 190                      if (position == INTYPE) /* CONST on the spine */
 191                         position = INCONCL;
 192                      where = CONST;
 193                    }
 194
 195 "<MUTIND"          {
 196                      if (position == INTYPE) /* MUTIND on the spine */
 197                         position = INCONCL;
 198                      where = MUTIND;
 199                    }
 200
 201 "<MUTCONSTRUCT"    {
 202                      if (position == INTYPE) /* MUTCONSTRUCT on the spine */
 203                         position = INCONCL;
 204                      where = MUTCONSTRUCT;
 205                    }
 206
 207 "uri=\""{uri}      {
 208                          if (!skip) {
 209                             uri=(char *)malloc((sizeof('a')*200));
 210                             strcpy(uri,yytext);
 211                             strsep(&uri,&sep);
 212                             if (where == CONST)
 213                               {
 214                                 if (position == INCONCL)
 215                                   search(uri,first_child,position,spine_depth);
 216                                 else search(uri,first_child,position,depth);
 217                                 where = NOWHERE;
 218                                 first_child = AFTER;
 219                                 free(uri);
 220                               };
 221                          } else skip = 0;
 222                    }
 223
 224 "noType=\""{digits} {
 225                          if ((where == MUTIND) || (where == MUTCONSTRUCT))
 226                           { strsep(&yytext,&sep);
 227                             tmp=(char *)malloc((sizeof(sep)*(strlen(yytext)+1)));
 228                             strcpy(tmp,yytext);
 229                             tmp_n = atoi(tmp)+1;
 230                             sprintf(tmp,"%d",tmp_n);
 231                             strcat(uri,"#xpointer(1/");
 232                             strcat(uri,tmp);
 233                           };
 234                          if (where == MUTIND)
 235                              {
 236                                strcat(uri,")");
 237                                if (position == INCONCL)
 238                                   search(uri,first_child,position,spine_depth);
 239                                else search(uri,first_child,position,depth);
 240                                free(uri);
 241                                free(tmp);
 242                                where = NOWHERE;
 243                                first_child = AFTER;};
 244                    }
 245
 246 "noConstr=\""{digits} {
 247                          if (where == MUTCONSTRUCT)
 248                           { strsep(&yytext,&sep);
 249                             tmp=(char *)malloc((sizeof(sep)*(strlen(yytext)+1)));
 250                             strcpy(tmp,yytext);
 251                             strcat(uri,"/");
 252                             strcat(uri,tmp);
 253                             strcat(uri,")");
 254                             if (position == INCONCL)
 255                               search(uri,first_child,position,spine_depth);
 256                             else search(uri,first_child,position,depth);
 257                             free(uri);
 258                             free(tmp);
 259                             where = NOWHERE;
 260                             first_child = AFTER;};
 261                    }
 262
 263
 264
 265 %%
 266
 267  /***************************************************************/
 268  /* 6. Auxiliary functions.                                     */
 269  /***************************************************************/
 270
 271 main(int argc, char *argv[])
 272 {
 273     struct stat buf;
 274
 275     char       *pghost,
 276                *pgport,
 277                *pgoptions,
 278                *pgtty;
 279     char       *dbName;
 280
 281     /* FILE *debug; */
 282
 283     PGconn     *conn;
 284     PGresult   *res;
 285
 286     /*
 287      * begin, by setting the parameters for a backend connection if the
 288      * parameters are null, then the system will try to use reasonable
 289      * defaults by looking up environment variables or, failing that,
 290      * using hardwired constants
 291      */
 292
 293     /* make a connection to the database */
 294     conn = PQconnectdb("user=helm dbname=mowgli");
 295
 296     /*
 297      * check to see that the backend connection was successfully made
 298      */
 299     if (PQstatus(conn) == CONNECTION_BAD)
 300     {
 301         fprintf(stderr, "Connection to database '%s' failed.\n", dbName);
 302         fprintf(stderr, "%s", PQerrorMessage(conn));
 303         exit_nicely(conn);
 304     }
 305
 306     /* debug = fopen("/tmp/trace.out","w"); */
 307     /* PQtrace(conn, debug);  */
 308
 309     /* initialize the symbol table */
 310     init_symbol_table();
 311
 312     // We process the body
 313     if (!stat("tmp/body.xml",&buf))
 314     {
 315         yyin = fopen("tmp/body.xml", "r");
 316         position = INBODY;
 317         yylex();
 318         fclose(yyin);
 319      }
 320
 321     // We process the type
 322     yyin = fopen("tmp/type.xml", "r");
 323     position = INTYPE;
 324     first_child = HERE;
 325     no_open_source = 0;
 326     spine_depth = 0;
 327     depth = 0;
 328     yylex();
 329     fclose(yyin);
 330     print_all(argv[1],conn);
 331 }
 332
 333
 334 search(uri,first_child,position,depth)
 335 char               *uri;
 336 int                first_child;
 337 int                position;
 338 {
 339                    if (position == MAINHYP)
 340                       {
 341                        if (first_child == HERE)
 342                            found = search_bucket(uri,MAINHYP,depth);
 343                        else
 344                            found = search_bucket(uri,INHYP,0);
 345                       }
 346                    else if (position == INCONCL)
 347                       {
 348                        if (first_child == HERE)
 349                            found = search_bucket(uri,MAINCONCL,depth);
 350                        else
 351                            found = search_bucket(uri,INCONCL,0);
 352                       }
 353
 354                    else
 355                       found = search_bucket(uri,position,depth);
 356                    /*
 357                    if (found == NOTFOUND)
 358                          fprintf(stderr,"here = %d, pos = %d, uri = %s\n", first_child,position, uri); */
 359 }
 360 /*
 361                       (first_child == HERE)
 362                       {
 363                        if (position == MAINHYP)
 364                           found = search_bucket(uri,MAINHYP,depth);
 365                        else if (position == INCONCL)
 366                           found = search_bucket(uri,MAINCONCL,0);
 367                        else if (position == INHYP)
 368                           found = search_bucket(uri,INHYP,0);
 369                           if (found == NOTFOUND)
 370                           printf( "pos = %d, uri = %s\n", MAINCONCL, uri);
 371                        }
 372                    else if ((position == MAINHYP) && (first_child == AFTER))
 373                         found = search_bucket(uri,INHYP,0);
 374                    else found = search_bucket(uri,position,0);
 375                    if (found == NOTFOUND)
 376                          printf( "pos = %d, uri = %s\n", position, uri);
 377                    } */
 378
 379 int yywrap() {
 380                return 1;
 381              }
 382
 383
 384
 385