* Sebastian Hammer, Adam Dickmeiss
*
* $Log: zebramap.c,v $
- * Revision 1.12 1999-02-12 13:29:25 adam
+ * Revision 1.17 1999-09-08 12:13:21 adam
+ * Fixed minor bug "replace"-mappings. Removed some logging messages.
+ *
+ * Revision 1.16 1999/09/07 07:19:21 adam
+ * Work on character mapping. Implemented replace rules.
+ *
+ * Revision 1.15 1999/05/26 07:49:14 adam
+ * C++ compilation.
+ *
+ * Revision 1.14 1999/02/19 10:37:40 adam
+ * Minor fix.
+ *
+ * Revision 1.13 1999/02/18 15:01:04 adam
+ * Structure=key uses register type 0.
+ *
+ * Revision 1.12 1999/02/12 13:29:25 adam
* Implemented position-flag for registers.
*
* Revision 1.11 1998/10/13 20:09:19 adam
#define ZEBRA_MAP_TYPE_SORT 1
#define ZEBRA_MAP_TYPE_INDEX 2
+#define ZEBRA_REPLACE_ANY 300
+
+struct zm_token {
+ int *token_from;
+ char *token_to;
+ int token_min;
+ struct zm_token *next;
+};
+
struct zebra_map {
unsigned reg_id;
int completeness;
chrmaptab maptab;
const char *maptab_name;
struct zebra_map *next;
+ struct zm_token *replace_tokens;
};
struct zebra_maps {
char temp_map_str[2];
const char *temp_map_ptr[2];
struct zebra_map **lookup_array;
+ WRBUF wrbuf_1, wrbuf_2;
};
void zebra_maps_close (ZebraMaps zms)
chrmaptab_destroy (zm->maptab);
zm = zm->next;
}
+ wrbuf_free (zms->wrbuf_1, 1);
+ wrbuf_free (zms->wrbuf_2, 1);
nmem_destroy (zms->nmem);
xfree (zms);
}
zm = &zms->map_list;
else
zm = &(*zm)->next;
- *zm = nmem_malloc (zms->nmem, sizeof(**zm));
+ *zm = (struct zebra_map *) nmem_malloc (zms->nmem, sizeof(**zm));
(*zm)->reg_id = argv[1][0];
(*zm)->maptab_name = NULL;
(*zm)->maptab = NULL;
(*zm)->type = ZEBRA_MAP_TYPE_INDEX;
(*zm)->completeness = 0;
(*zm)->positioned = 1;
+ (*zm)->replace_tokens = 0;
}
else if (!yaz_matchstr (argv[0], "sort") && argc == 2)
{
zm = &zms->map_list;
else
zm = &(*zm)->next;
- *zm = nmem_malloc (zms->nmem, sizeof(**zm));
+ *zm = (struct zebra_map *) nmem_malloc (zms->nmem, sizeof(**zm));
(*zm)->reg_id = argv[1][0];
(*zm)->maptab_name = NULL;
(*zm)->type = ZEBRA_MAP_TYPE_SORT;
(*zm)->maptab = NULL;
(*zm)->completeness = 0;
(*zm)->positioned = 0;
+ (*zm)->replace_tokens = 0;
}
else if (zm && !yaz_matchstr (argv[0], "charmap") && argc == 2)
{
if ((*zm)->type == ZEBRA_MAP_TYPE_SORT)
(*zm)->u.sort.entry_size = atoi (argv[1]);
}
+ else if (zm && !yaz_matchstr (argv[0], "replace") && argc >= 2)
+ {
+ struct zm_token *token = nmem_malloc (zms->nmem, sizeof(*token));
+ token->next = (*zm)->replace_tokens;
+ (*zm)->replace_tokens = token;
+ token->token_from = 0;
+ if (argc >= 2)
+ {
+ char *cp = argv[1];
+ int *dp = token->token_from = (int *)
+ nmem_malloc (zms->nmem, (1+strlen(cp))*sizeof(int));
+ while (*cp)
+ if (*cp == '$')
+ {
+ *dp++ = ' ';
+ cp++;
+ }
+ else if (*cp == '.')
+ {
+ *dp++ = ZEBRA_REPLACE_ANY;
+ cp++;
+ }
+ else
+ *dp++ = zebra_prim(&cp);
+ *dp = '\0';
+ }
+ if (argc >= 3)
+ {
+ char *cp = argv[2];
+ char *dp = token->token_to =
+ nmem_malloc (zms->nmem, strlen(cp)+1);
+ while (*cp)
+ if (*cp == '$')
+ {
+ *dp++ = ' ';
+ cp++;
+ }
+ else
+ *dp++ = zebra_prim(&cp);
+ *dp = '\0';
+ }
+ else
+ token->token_to = 0;
+ }
}
if (zm)
(*zm)->next = NULL;
static void zms_map_handle (void *p, const char *name, const char *value)
{
- ZebraMaps zms = p;
+ ZebraMaps zms = (ZebraMaps) p;
zebra_map_read (zms, value);
}
ZebraMaps zebra_maps_open (Res res)
{
- ZebraMaps zms = xmalloc (sizeof(*zms));
+ ZebraMaps zms = (ZebraMaps) xmalloc (sizeof(*zms));
int i;
zms->nmem = nmem_create ();
- zms->tabpath = nmem_strdup (zms->nmem, res_get (res, "profilePath"));
+ zms->tabpath = nmem_strdup (zms->nmem,
+ res_get_def (res, "profilePath", "."));
zms->map_list = NULL;
zms->temp_map_str[0] = '\0';
zms->temp_map_ptr[0] = zms->temp_map_str;
zms->temp_map_ptr[1] = NULL;
- zms->lookup_array =
+ zms->lookup_array = (struct zebra_map**)
nmem_malloc (zms->nmem, sizeof(*zms->lookup_array)*256);
for (i = 0; i<256; i++)
zms->lookup_array[i] = 0;
if (!res || !res_trav (res, "index", zms, zms_map_handle))
zebra_map_read (zms, "default.idx");
+
+ zms->wrbuf_1 = wrbuf_alloc();
+ zms->wrbuf_2 = wrbuf_alloc();
return zms;
}
struct zebra_map *zm = zebra_map_get (zms, reg_id);
if (!zm)
{
- zm = nmem_malloc (zms->nmem, sizeof(*zm));
+ zm = (struct zebra_map *) nmem_malloc (zms->nmem, sizeof(*zm));
logf (LOG_WARN, "Unknown register type: %c", reg_id);
zm->reg_id = reg_id;
return zms->temp_map_ptr;
}
+#if 0
+int zebra_maps_input_tokens (ZebraMaps zms, unsigned reg_id,
+ const char *input_str, int input_len,
+ WRBUF wrbuf)
+{
+ chrmaptab maptab = zebra_charmap_get (zms, reg_id);
+ int len[4];
+ char *str[3];
+ int input_i = 0;
+ int first = 1;
+ const char **out;
+
+ if (!maptab)
+ {
+ wrbuf_write (wrbuf, input_str, input_len);
+ return -1;
+ }
+ str[0] = " ";
+ len[0] = 1;
+ str[1] = input_str;
+ len[1] = input_len;
+ str[2] = " ";
+ len[2] = 1;
+ len[3] = -1;
+
+ out = chr_map_input (maptab, str, len);
+ while (len[1] > 0)
+ {
+ while (out && *out && **out == *CHR_SPACE)
+ out = chr_map_input (maptab, str, len);
+ }
+}
+#endif
+
const char *zebra_maps_output(ZebraMaps zms, unsigned reg_id,
const char **from)
{
- chrmaptab maptab;
- unsigned char i = (unsigned char) **from;
- static char buf[2] = {0,0};
-
- maptab = zebra_charmap_get (zms, reg_id);
- if (maptab)
- return chr_map_output (maptab, from, 1);
- (*from)++;
- buf[0] = i;
- return buf;
+ chrmaptab maptab = zebra_charmap_get (zms, reg_id);
+ if (!maptab)
+ return 0;
+ return chr_map_output (maptab, from, 1);
}
case -1:
case 1: /* phrase */
case 2: /* word */
- case 3: /* key */
case 108: /* string */
*search_type = "phrase";
break;
*reg_id = 'u';
*search_type = "phrase";
break;
+ case 3: /* key */
+ *reg_id = '0';
+ *search_type = "phrase";
+ break;
default:
return -1;
}
return 0;
}
+
+int zebra_replace_sub(ZebraMaps zms, unsigned reg_id, const char *ex_list,
+ const char *input_str, int input_len, WRBUF wrbuf);
+
+WRBUF zebra_replace(ZebraMaps zms, unsigned reg_id, const char *ex_list,
+ const char *input_str, int input_len)
+{
+ struct zebra_map *zm = zebra_map_get (zms, reg_id);
+
+ wrbuf_rewind(zms->wrbuf_1);
+ wrbuf_write(zms->wrbuf_1, input_str, input_len);
+ if (!zm->replace_tokens)
+ return zms->wrbuf_1;
+
+#if 0
+ logf (LOG_LOG, "in:%.*s:", wrbuf_len(zms->wrbuf_1),
+ wrbuf_buf(zms->wrbuf_1));
+#endif
+ for (;;)
+ {
+ if (!zebra_replace_sub(zms, reg_id, ex_list, wrbuf_buf(zms->wrbuf_1),
+ wrbuf_len(zms->wrbuf_1), zms->wrbuf_2))
+ return zms->wrbuf_2;
+ if (!zebra_replace_sub(zms, reg_id, ex_list, wrbuf_buf(zms->wrbuf_2),
+ wrbuf_len(zms->wrbuf_2), zms->wrbuf_1))
+ return zms->wrbuf_1;
+ }
+ return 0;
+}
+
+int zebra_replace_sub(ZebraMaps zms, unsigned reg_id, const char *ex_list,
+ const char *input_str, int input_len, WRBUF wrbuf)
+{
+ int i = -1;
+ int no_replaces = 0;
+ struct zebra_map *zm = zebra_map_get (zms, reg_id);
+
+ wrbuf_rewind(wrbuf);
+ for (i = -1; i <= input_len; )
+ {
+ struct zm_token *token;
+ char replace_string[128];
+ int replace_out;
+ int replace_in = 0;
+
+ for (token = zm->replace_tokens; !replace_in && token;
+ token = token->next)
+ {
+ int j = 0;
+ int replace_done = 0;
+ replace_out = 0;
+ for (;; j++)
+ {
+ int c;
+ if (!token->token_from[j])
+ {
+ replace_in = j;
+ break;
+ }
+ if (ex_list && strchr (ex_list, token->token_from[j]))
+ break;
+ if (i+j < 0 || j+i >= input_len)
+ c = ' ';
+ else
+ c = tolower(input_str[j+i]);
+ if (token->token_from[j] == ZEBRA_REPLACE_ANY)
+ {
+ if (c == ' ')
+ break;
+ replace_string[replace_out++] = c;
+ }
+ else
+ {
+ if (c != token->token_from[j])
+ break;
+ if (!replace_done)
+ {
+ const char *cp = token->token_to;
+ replace_done = 1;
+ for (; cp && *cp; cp++)
+ replace_string[replace_out++] = *cp;
+ }
+ }
+ }
+ }
+ if (!replace_in)
+ {
+ if (i >= 0 && i < input_len)
+ wrbuf_putc(wrbuf, input_str[i]);
+ i++;
+ }
+ else
+ {
+ no_replaces++;
+ if (replace_out)
+ wrbuf_write(wrbuf, replace_string, replace_out);
+ i += replace_in;
+ }
+ }
+#if 0
+ logf (LOG_LOG, "out:%.*s:", wrbuf_len(wrbuf), wrbuf_buf(wrbuf));
+#endif
+ return no_replaces;
+}