YAZ: cclqfile.c Source File

00001 /* This file is part of the YAZ toolkit.
00002  * Copyright (C) 1995-2008 Index Data
00003  * See the file LICENSE for details.
00004  */
00010 #include <stdio.h>
00011 #include <stdlib.h>
00012 #include <string.h>
00013 
00014 #include <yaz/tokenizer.h>
00015 #include <yaz/ccl.h>
00016 #include <yaz/log.h>
00017 
00018 #define MAX_QUAL 128
00019 
00020 int ccl_qual_field2(CCL_bibset bibset, const char *cp, const char *qual_name,
00021                     const char **addinfo)
00022 {
00023     yaz_tok_cfg_t yt = yaz_tok_cfg_create();
00024 
00025     int type_ar[MAX_QUAL];
00026     int value_ar[MAX_QUAL];
00027     char *svalue_ar[MAX_QUAL];
00028     char *attsets[MAX_QUAL];
00029     int pair_no = 0;
00030     char *type_str = 0;
00031     int t;
00032     yaz_tok_parse_t tp;
00033 
00034     yaz_tok_cfg_single_tokens(yt, ",=");
00035 
00036     tp = yaz_tok_parse_buf(yt, cp);
00037 
00038     yaz_tok_cfg_destroy(yt);
00039     *addinfo = 0;
00040     
00041     t = yaz_tok_move(tp);
00042     while (t == YAZ_TOK_STRING)
00043     {
00044         /* we don't know what lead is yet */
00045         char *lead_str = xstrdup(yaz_tok_parse_string(tp));
00046         const char *value_str = 0;
00047         int type = 0, value = 0; /* indicates attribute value UNSET  */
00048 
00049         t = yaz_tok_move(tp);
00050         if (t == ',')
00051         {
00052             /* full attribute spec: set, type = value */
00053             /* lead is attribute set */
00054             attsets[pair_no] = lead_str;
00055             t = yaz_tok_move(tp);
00056             if (t != YAZ_TOK_STRING)
00057             {
00058                 *addinfo = "token expected";
00059                 goto out;
00060             }
00061             xfree(type_str);
00062             type_str = xstrdup(yaz_tok_parse_string(tp));
00063             if (yaz_tok_move(tp) != '=')
00064             {
00065                 *addinfo = "= expected";
00066                 goto out;
00067             }
00068         }
00069         else if (t == '=')
00070         {
00071             /* lead is attribute type */
00072             /* attribute set omitted: type = value */
00073             attsets[pair_no] = 0;
00074             xfree(type_str);
00075             type_str = lead_str;
00076         }
00077         else
00078         {
00079             /* lead is first of a list of qualifier aliaeses */
00080             /* qualifier alias: q1 q2 ... */
00081             char *qlist[10];
00082             int i = 0;
00083 
00084             qlist[i++] = lead_str;
00085 
00086             while ((t=yaz_tok_move(tp)) == YAZ_TOK_STRING)
00087             {
00088                 if (i < sizeof(qlist)/sizeof(*qlist)-1)
00089                     qlist[i++] = xstrdup(yaz_tok_parse_string(tp));
00090             }
00091             qlist[i] = 0;
00092             yaz_tok_parse_destroy(tp);
00093             ccl_qual_add_combi (bibset, qual_name, (const char **) qlist);
00094             for (i = 0; qlist[i]; i++)
00095                 xfree(qlist[i]);
00096             return 0;
00097         }
00098         while (1) /* comma separated attribute value list */
00099         {
00100             t = yaz_tok_move(tp);
00101             /* must have a value now */
00102             if (t != YAZ_TOK_STRING)
00103             {
00104                 *addinfo = "value token expected";
00105                 goto out;
00106             }
00107             value_str = yaz_tok_parse_string(tp);
00108             
00109             if (sscanf(type_str, "%d", &type) == 1)
00110                 ;
00111             else if (strlen(type_str) != 1)
00112             {
00113                 *addinfo = "bad attribute type";
00114                 goto out;
00115             }
00116             else
00117             {
00118                 switch (*type_str)
00119                 {
00120                 case 'u':
00121                 case 'U':
00122                     type = CCL_BIB1_USE;
00123                     break;
00124                 case 'r':
00125                 case 'R':
00126                     type = CCL_BIB1_REL;
00127                     if (!ccl_stricmp (value_str, "o"))
00128                         value = CCL_BIB1_REL_ORDER;
00129                     else if (!ccl_stricmp (value_str, "r"))
00130                         value = CCL_BIB1_REL_PORDER;
00131                     break;                
00132                 case 'p':
00133                 case 'P':
00134                     type = CCL_BIB1_POS;
00135                     break;
00136                 case 's':
00137                 case 'S':
00138                     type = CCL_BIB1_STR;
00139                     if (!ccl_stricmp (value_str, "pw"))
00140                         value = CCL_BIB1_STR_WP;
00141                     if (!ccl_stricmp (value_str, "al"))
00142                         value = CCL_BIB1_STR_AND_LIST;
00143                     if (!ccl_stricmp (value_str, "ol"))
00144                         value = CCL_BIB1_STR_OR_LIST;
00145                     break;                
00146                 case 't':
00147                 case 'T':
00148                     type = CCL_BIB1_TRU;
00149                     if (!ccl_stricmp (value_str, "l"))
00150                         value = CCL_BIB1_TRU_CAN_LEFT;
00151                     else if (!ccl_stricmp (value_str, "r"))
00152                         value = CCL_BIB1_TRU_CAN_RIGHT;
00153                     else if (!ccl_stricmp (value_str, "b"))
00154                         value = CCL_BIB1_TRU_CAN_BOTH;
00155                     else if (!ccl_stricmp (value_str, "n"))
00156                         value = CCL_BIB1_TRU_CAN_NONE;
00157                     break;                
00158                 case 'c':
00159                 case 'C':
00160                     type = CCL_BIB1_COM;
00161                     break;
00162                 }
00163             }
00164             if (type == 0)
00165             {
00166                 /* type was not set in switch above */
00167                 *addinfo = "bad attribute type";
00168                 goto out;
00169             }
00170             type_ar[pair_no] = type;
00171             if (value)
00172             {
00173                 value_ar[pair_no] = value;
00174                 svalue_ar[pair_no] = 0;
00175             }
00176             else if (*value_str >= '0' && *value_str <= '9')
00177             {
00178                 value_ar[pair_no] = atoi (value_str);
00179                 svalue_ar[pair_no] = 0;
00180             }
00181             else
00182             {
00183                 value_ar[pair_no] = 0;
00184                 svalue_ar[pair_no] = xstrdup(value_str);
00185             }
00186             pair_no++;
00187             if (pair_no == MAX_QUAL)
00188             {
00189                 *addinfo = "too many attribute values";
00190                 goto out;
00191             }
00192             t = yaz_tok_move(tp);
00193             if (t != ',')
00194                 break;
00195             attsets[pair_no] = attsets[pair_no-1];
00196         }
00197     }
00198  out:
00199     xfree(type_str);
00200     type_str = 0;
00201 
00202     yaz_tok_parse_destroy(tp);
00203 
00204     if (*addinfo)
00205     {
00206         int i;
00207         for (i = 0; i<pair_no; i++)
00208         {
00209             xfree(attsets[i]);
00210             xfree(svalue_ar[i]);
00211         }
00212         return -1;
00213     }
00214     ccl_qual_add_set(bibset, qual_name, pair_no, type_ar, value_ar, svalue_ar,
00215                      attsets);
00216     return 0;
00217 }
00218 
00219 void ccl_qual_field(CCL_bibset bibset, const char *cp, const char *qual_name)
00220 {
00221     const char *addinfo;
00222     ccl_qual_field2(bibset, cp, qual_name, &addinfo);
00223     if (addinfo)
00224         yaz_log(YLOG_WARN, "ccl_qual_field2 fail: %s", addinfo);
00225 }
00226 
00227 void ccl_qual_fitem (CCL_bibset bibset, const char *cp, const char *qual_name)
00228 {
00229     if (*qual_name == '@')
00230         ccl_qual_add_special(bibset, qual_name+1, cp);
00231     else
00232         ccl_qual_field(bibset, cp, qual_name);
00233 }
00234 
00235 void ccl_qual_buf(CCL_bibset bibset, const char *buf)
00236 {
00237     const char *cp1 = buf;
00238     char line[256];
00239     while (1)
00240     {
00241         const char *cp2 = cp1;
00242         int len;
00243         while (*cp2 && !strchr("\r\n", *cp2))
00244             cp2++;
00245         len = cp2 - cp1;
00246         if (len > 0)
00247         {
00248             if (len >= (sizeof(line)-1))
00249                 len = sizeof(line)-1;
00250             memcpy(line, cp1, len);
00251             line[len] = '\0';
00252             ccl_qual_line(bibset, line);
00253         }
00254         if (!*cp2)
00255             break;
00256         cp1 = cp2+1;
00257     }
00258 }
00259 
00260 void ccl_qual_line(CCL_bibset bibset, char *line)
00261 {
00262     int  no_scan = 0;
00263     char qual_name[128];
00264     char *cp1, *cp = line;
00265     
00266     if (*cp == '#')
00267         return;        /* ignore lines starting with # */
00268     if (sscanf (cp, "%100s%n", qual_name, &no_scan) < 1)
00269         return;        /* also ignore empty lines */
00270     cp += no_scan;
00271     cp1 = strchr(cp, '#');
00272     if (cp1)
00273         *cp1 = '\0';
00274     ccl_qual_fitem (bibset, cp, qual_name);
00275 }
00276 
00277 /*
00278  * ccl_qual_file: Read bibset definition from file.
00279  * bibset:  Bibset
00280  * inf:     FILE pointer.
00281  *
00282  * Each line format is:
00283  *  <name> <t>=<v> <t>=<v> ....
00284  *  Where <name> is name of qualifier;
00285  *  <t>=<v> is a attribute definition pair where <t> is one of: 
00286  *     u(use), r(relation), p(position), t(truncation), c(completeness) 
00287  *     or plain integer.
00288  *  <v> is an integer or special pseudo-value.
00289  */
00290 void ccl_qual_file (CCL_bibset bibset, FILE *inf)
00291 {
00292     char line[256];
00293 
00294     while (fgets (line, 255, inf))
00295         ccl_qual_line(bibset, line);
00296 }
00297 
00298 int ccl_qual_fname (CCL_bibset bibset, const char *fname)
00299 {
00300     FILE *inf;
00301     inf = fopen (fname, "r");
00302     if (!inf)
00303         return -1;
00304     ccl_qual_file (bibset, inf);
00305     fclose (inf);
00306     return 0;
00307 }
00308 /*
00309  * Local variables:
00310  * c-basic-offset: 4
00311  * indent-tabs-mode: nil
00312  * End:
00313  * vim: shiftwidth=4 tabstop=8 expandtab
00314  */
00315