/*
- * Copyright (C) 1994, Index Data I/S
+ * Copyright (C) 1996-1998, Index Data
* All rights reserved.
* Sebastian Hammer, Adam Dickmeiss
*
* $Log: charmap.c,v $
- * Revision 1.1 1996-05-31 09:07:18 quinn
+ * Revision 1.15 1999-05-26 07:49:14 adam
+ * C++ compilation.
+ *
+ * Revision 1.14 1998/10/13 20:09:18 adam
+ * Changed call to readconf_line.
+ *
+ * Revision 1.13 1997/10/27 14:33:06 adam
+ * Moved towards generic character mapping depending on "structure"
+ * field in abstract syntax file. Fixed a few memory leaks. Fixed
+ * bug with negative integers when doing searches with relational
+ * operators.
+ *
+ * Revision 1.12 1997/09/05 15:30:11 adam
+ * Changed prototype for chr_map_input - added const.
+ * Added support for C++, headers uses extern "C" for public definitions.
+ *
+ * Revision 1.11 1997/09/05 09:52:32 adam
+ * Extra argument added to function chr_read_maptab (tab path).
+ *
+ * Revision 1.10 1997/07/01 13:01:08 adam
+ * Bug fix in routine find_entry: didn't take into account the len arg.
+ *
+ * Revision 1.9 1996/10/29 13:48:14 adam
+ * Updated to use zebrautl.h instead of alexutil.h.
+ *
+ * Revision 1.8 1996/10/18 12:39:23 adam
+ * Uses LOG_DEBUG instead of LOG_WARN for "Character map overlap".
+ *
+ * Revision 1.7 1996/06/06 12:08:56 quinn
+ * Fixed bug.
+ *
+ * Revision 1.6 1996/06/04 13:28:00 quinn
+ * More work on charmapping
+ *
+ * Revision 1.5 1996/06/04 08:32:15 quinn
+ * Moved default keymap to keychars.c
+ *
+ * Revision 1.4 1996/06/03 16:32:13 quinn
+ * Temporary bug-fix
+ *
+ * Revision 1.3 1996/06/03 15:17:46 quinn
+ * Fixed bug.
+ *
+ * Revision 1.2 1996/06/03 10:15:09 quinn
+ * Fixed bug in mapping function.
+ *
+ * Revision 1.1 1996/05/31 09:07:18 quinn
* Work on character-set handling
*
*
*/
#include <ctype.h>
+#include <string.h>
+#include <assert.h>
-#include <alexutil.h>
#include <yaz-util.h>
#include <charmap.h>
-#include <tpath.h>
+
+#define CHR_MAXSTR 1024
+#define CHR_MAXEQUIV 32
+
+int chr_map_chrs(chr_t_entry *t, char **from, int len,
+ int *read, char **to, int max);
const char *CHR_UNKNOWN = "\001";
const char *CHR_SPACE = "\002";
const char *CHR_BASE = "\003";
-extern char *data1_tabpath;
+struct chrmaptab_info
+{
+ chr_t_entry *input; /* mapping table for input data */
+ chr_t_entry *query_equiv; /* mapping table for queries */
+ unsigned char *output[256]; /* return mapping - for display of registers */
+ int base_uppercase; /* Start of upper-case ordinals */
+ char **tmp_buf;
+ NMEM nmem;
+};
/*
* Character map trie node.
chr_t_entry **children; /* array of children */
unsigned char *target; /* target for this node, if any */
unsigned char *equiv; /* equivalent to, or sumthin */
-} t_entry;
+};
+
+/*
+ * General argument structure for callback functions (internal use only)
+ */
+typedef struct chrwork
+{
+ chrmaptab map;
+ char string[CHR_MAXSTR+1];
+} chrwork;
/*
* Add an entry to the character map.
*/
-static chr_t_entry *set_map_string(chr_t_entry *root, char *from, int len,
- char *to)
+static chr_t_entry *set_map_string(chr_t_entry *root, NMEM nmem,
+ const char *from, int len, char *to)
{
if (!root)
{
- root = xmalloc(sizeof(*root));
+ root = (chr_t_entry *) nmem_malloc(nmem, sizeof(*root));
root->children = 0;
root->target = 0;
}
if (!len)
- root->target = (unsigned char *) xstrdup(to);
+ {
+ if (!root->target || (char*) root->target == CHR_SPACE ||
+ (char*) root->target == CHR_UNKNOWN)
+ root->target = (unsigned char *) nmem_strdup(nmem, to);
+ else if ((char*) to != CHR_SPACE)
+ logf(LOG_DEBUG, "Character map overlap");
+ }
else
{
if (!root->children)
{
int i;
- root->children = xmalloc(sizeof(chr_t_entry*) * 256);
+ root->children = (chr_t_entry **)
+ nmem_malloc(nmem, sizeof(chr_t_entry*) * 256);
for (i = 0; i < 256; i++)
root->children[i] = 0;
}
- root->children[(unsigned char) *from] =
- set_map_string(root->children[(unsigned char) *from], from + 1,
- len - 1, to);
+ if (!(root->children[(unsigned char) *from] =
+ set_map_string(root->children[(unsigned char) *from], nmem,
+ from + 1, len - 1, to)))
+ return 0;
}
return root;
}
return i;
}
-char **chr_map_input(chr_t_entry *t, char **from, int len)
+
+static chr_t_entry *find_entry(chr_t_entry *t, const char **from, int len)
{
- static char *buf[2] = {0, 0}, str[2] = {0, 0};
- char *start = *from;
+ chr_t_entry *res;
- if (t)
+ if (len && t->children && t->children[(unsigned char) **from])
{
- while (len && t->children && t->children[(unsigned char) **from])
- {
- t = t->children[(unsigned char) **from];
- (*from)++;
- len--;
- }
- buf[0] = (char*) t->target;
- }
- else /* null mapping */
- {
- if (isalnum(**from))
- {
- str[0] = **from;
- buf[0] = str;
- }
- else if (isspace(**from))
- buf[0] = (char*) CHR_SPACE;
- else
- buf[0] = (char*) CHR_UNKNOWN;
- }
- if (start == *from)
+ const char *pos = *from;
+
(*from)++;
- return buf;
- /* return (char*) t->target; */
+ if ((res = find_entry(t->children[(unsigned char) *pos],
+ from, len - 1)))
+ return res;
+ /* no match */
+ *from = pos;
+ }
+ /* no children match. use ourselves, if we have a target */
+ return t->target ? t : 0;
+}
+
+const char **chr_map_input(chrmaptab maptab, const char **from, int len)
+{
+ chr_t_entry *t = maptab->input;
+ chr_t_entry *res;
+
+ if (!(res = find_entry(t, from, len)))
+ abort();
+ maptab->tmp_buf[0] = (char*) res->target;
+ maptab->tmp_buf[1] = NULL;
+ return (const char **) maptab->tmp_buf;
+}
+
+const char *chr_map_output(chrmaptab maptab, const char **from, int len)
+{
+ unsigned char c = ** (unsigned char **) from;
+ (*from)++;
+ return (const char*) maptab->output[c];
}
static unsigned char prim(char **s)
* Callback function.
* Add an entry to the value space.
*/
-static void fun_addentry(char *s, void *data, int num)
+static void fun_addentry(const char *s, void *data, int num)
{
- chrmaptab *tab = data;
+ chrmaptab tab = (chrmaptab) data;
char tmp[2];
tmp[0] = num; tmp[1] = '\0';
- tab->input = set_map_string(tab->input, s, strlen(s), tmp);
- tab->output[num + tab->base_uppercase] = (unsigned char *) xstrdup(s);
+ tab->input = set_map_string(tab->input, tab->nmem, s, strlen(s), tmp);
+ tab->output[num + tab->base_uppercase] =
+ (unsigned char *) nmem_strdup(tab->nmem, s);
}
/*
* Callback function.
* Add a space-entry to the value space.
*/
-static void fun_addspace(char *s, void *data, int num)
+static void fun_addspace(const char *s, void *data, int num)
{
- chrmaptab *tab = data;
- tab->input = set_map_string(tab->input, s, strlen(s), (char*) CHR_SPACE);
+ chrmaptab tab = (chrmaptab) data;
+ tab->input = set_map_string(tab->input, tab->nmem, s, strlen(s),
+ (char*) CHR_SPACE);
}
-static int scan_string(char *s, void (*fun)(char *c, void *data, int num),
- void *data, int *num)
+/*
+ * Create a string containing the mapped characters provided.
+ */
+static void fun_mkstring(const char *s, void *data, int num)
+{
+ chrwork *arg = (chrwork *) data;
+ const char **res, *p = s;
+
+ res = chr_map_input(arg->map, &s, strlen(s));
+ if (*res == (char*) CHR_UNKNOWN)
+ logf(LOG_WARN, "Map: '%s' has no mapping", p);
+ strncat(arg->string, *res, CHR_MAXSTR - strlen(arg->string));
+ arg->string[CHR_MAXSTR] = '\0';
+}
+
+/*
+ * Add a map to the string contained in the argument.
+ */
+static void fun_addmap(const char *s, void *data, int num)
+{
+ chrwork *arg = (chrwork *) data;
+
+ assert(arg->map->input);
+ set_map_string(arg->map->input, arg->map->nmem, s, strlen(s), arg->string);
+}
+
+static int scan_string(char *s,
+ void (*fun)(const char *c, void *data, int num),
+ void *data, int *num)
{
- unsigned char c, str[1024], begin, end;
+ unsigned char c, str[1024], begin, end, *p;
while (*s)
{
}
break;
case '[': s++; abort(); break;
- case '(': s++; abort(); break;
+ case '(':
+ p = (unsigned char*) ++s;
+ /* Find the end-marker, ignoring escapes */
+ do
+ {
+ if (!(p = (unsigned char*) strchr((char*) p, ')')))
+ {
+ logf(LOG_FATAL, "Missing ')' in string");
+ return -1;
+ }
+ }
+ while (*(p - 1) == '\\');
+ *p = 0;
+ (*fun)(s, data, num ? (*num)++ : 0);
+ s = (char*) p + 1;
+ break;
default:
c = prim(&s);
str[0] = c; str[1] = '\0';
return 0;
}
-chrmaptab *chr_read_maptab(char *name)
+chrmaptab chrmaptab_create(const char *tabpath, const char *name, int map_only)
{
FILE *f;
char line[512], *argv[50];
- chrmaptab *res = xmalloc(sizeof(*res));
+ chrmaptab res;
+ int lineno = 0;
int argc, num = (int) *CHR_BASE, i;
- if (!(f = yaz_path_fopen(data1_tabpath, name, "r")))
+ if (!(f = yaz_path_fopen(tabpath, name, "r")))
{
logf(LOG_WARN|LOG_ERRNO, "%s", name);
return 0;
}
- res = xmalloc(sizeof(*res));
- res->input = xmalloc(sizeof(*res->input));
+ res = (chrmaptab) xmalloc(sizeof(*res));
+ res->nmem = nmem_create ();
+ res->tmp_buf = (char **)
+ nmem_malloc (res->nmem, sizeof(*res->tmp_buf) * 100);
+ res->input = (chr_t_entry *) nmem_malloc(res->nmem, sizeof(*res->input));
res->input->target = (unsigned char*) CHR_UNKNOWN;
res->input->equiv = 0;
-#if 0
- res->input->children = xmalloc(sizeof(res->input) * 256);
+ res->input->children = (chr_t_entry **)
+ nmem_malloc(res->nmem, sizeof(res->input) * 256);
for (i = 0; i < 256; i++)
{
- res->input->children[i] = xmalloc(sizeof(*res->input));
+ res->input->children[i] = (chr_t_entry *)
+ nmem_malloc(res->nmem, sizeof(*res->input));
res->input->children[i]->children = 0;
- res->input->children[i]->target = CHR_UNKNOWN;
+ if (map_only)
+ {
+ res->input->children[i]->target = (unsigned char *)
+ nmem_malloc (res->nmem, 2 * sizeof(char));
+ res->input->children[i]->target[0] = i;
+ res->input->children[i]->target[1] = 0;
+ }
+ else
+ res->input->children[i]->target = (unsigned char*) CHR_UNKNOWN;
res->input->children[i]->equiv = 0;
}
-#else
- res->input->children = 0;
-#endif
res->query_equiv = 0;
- for (i = 0; i < 256; i++)
- {
- char *t = xmalloc(2);
-
- t[0] = i;
- t[1] = '\0';
- res->output[i] = (unsigned char*)t;
- }
+ for (i = *CHR_BASE; i < 256; i++)
+ res->output[i] = 0;
res->output[(int) *CHR_SPACE] = (unsigned char *) " ";
res->output[(int) *CHR_UNKNOWN] = (unsigned char*) "@";
res->base_uppercase = 0;
- while ((argc = readconf_line(f, line, 512, argv, 50)))
- if (!yaz_matchstr(argv[0], "lowercase"))
+ while ((argc = readconf_line(f, &lineno, line, 512, argv, 50)))
+ if (!map_only && !yaz_matchstr(argv[0], "lowercase"))
{
if (argc != 2)
{
res->output[(int) *CHR_UNKNOWN + num] = (unsigned char*) "@";
num = (int) *CHR_BASE;
}
- else if (!yaz_matchstr(argv[0], "uppercase"))
+ else if (!map_only && !yaz_matchstr(argv[0], "uppercase"))
{
if (!res->base_uppercase)
{
return 0;
}
}
- else if (!yaz_matchstr(argv[0], "space"))
+ else if (!map_only && !yaz_matchstr(argv[0], "space"))
{
if (argc != 2)
{
fclose(f);
return 0;
}
- if (scan_string(argv[1], fun_addspace, res, 0))
+ if (scan_string(argv[1], fun_addspace, res, 0) < 0)
{
logf(LOG_FATAL, "Bad space specification");
fclose(f);
return 0;
}
}
+ else if (!yaz_matchstr(argv[0], "map"))
+ {
+ chrwork buf;
+
+ if (argc != 3)
+ {
+ logf(LOG_FATAL, "charmap MAP directive requires 2 args");
+ fclose(f);
+ return 0;
+ }
+ buf.map = res;
+ buf.string[0] = '\0';
+ if (scan_string(argv[2], fun_mkstring, &buf, 0) < 0)
+ {
+ logf(LOG_FATAL, "Bad map target");
+ fclose(f);
+ return 0;
+ }
+ if (scan_string(argv[1], fun_addmap, &buf, 0) < 0)
+ {
+ logf(LOG_FATAL, "Bad map source");
+ fclose(f);
+ return 0;
+ }
+ }
else
{
-#if 0
- logf(LOG_WARN, "Syntax error at '%s' in %s", line, file);
- fclose(f);
- return 0;
-#endif
+ logf(LOG_WARN, "Syntax error at '%s' in %s", line, name);
}
fclose(f);
return res;
}
+
+void chrmaptab_destroy(chrmaptab tab)
+{
+ nmem_destroy (tab->nmem);
+ xfree (tab);
+}
+
+
projects / idzebra-moved-to-github.git / blobdiff