X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=util%2Fcharmap.c;h=3f5abdf97c074e6aa6b6b03445a01827b7382806;hb=b8844c65622c12fb105bf7fc6bab2fc237af7c14;hp=112aba29641a6c889e3f9dc0589be7f970c06e58;hpb=003ed9e50b40a26c54bf79cab71bd26f7957b870;p=idzebra-moved-to-github.git diff --git a/util/charmap.c b/util/charmap.c index 112aba2..3f5abdf 100644 --- a/util/charmap.c +++ b/util/charmap.c @@ -4,7 +4,22 @@ * Sebastian Hammer, Adam Dickmeiss * * $Log: charmap.c,v $ - * Revision 1.1 1996-05-31 09:07:18 quinn + * 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 * * @@ -16,12 +31,17 @@ */ #include +#include +#include #include #include #include #include +#define CHR_MAXSTR 1024 +#define CHR_MAXEQUIV 32 + const char *CHR_UNKNOWN = "\001"; const char *CHR_SPACE = "\002"; const char *CHR_BASE = "\003"; @@ -36,7 +56,16 @@ struct chr_t_entry 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. @@ -51,7 +80,13 @@ static chr_t_entry *set_map_string(chr_t_entry *root, char *from, int len, 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 *) xstrdup(to); + else if ((char*) to != CHR_SPACE) + logf(LOG_WARN, "Character map overlap"); + } else { if (!root->children) @@ -62,9 +97,10 @@ static chr_t_entry *set_map_string(chr_t_entry *root, char *from, int len, for (i = 0; i < 256; i++) root->children[i] = 0; } - root->children[(unsigned char) *from] = + if (!(root->children[(unsigned char) *from] = set_map_string(root->children[(unsigned char) *from], from + 1, - len - 1, to); + len - 1, to))) + return 0; } return root; } @@ -94,6 +130,40 @@ int chr_map_chrs(chr_t_entry *t, char **from, int len, int *read, char **to, return i; } +#if 1 + +static chr_t_entry *find_entry(chr_t_entry *t, char **from, int len) +{ + chr_t_entry *res; + + if (t->children && t->children[(unsigned char) **from]) + { + char *pos = *from; + + (*from)++; + if ((res = find_entry(t->children[(unsigned char) *pos], + from, len - 1))) + return res; + /* no match */ + *from = pos; + } + /* no children match. use ourselves */ + return t; +} + +char **chr_map_input(chr_t_entry *t, char **from, int len) +{ + static char *buf[2] = {0, 0}; + chr_t_entry *res; + + if (!(res = find_entry(t, from, len))) + abort(); + buf[0] = (char *) res->target; + return buf; +} + +#else + char **chr_map_input(chr_t_entry *t, char **from, int len) { static char *buf[2] = {0, 0}, str[2] = {0, 0}; @@ -127,6 +197,8 @@ char **chr_map_input(chr_t_entry *t, char **from, int len) /* return (char*) t->target; */ } +#endif + static unsigned char prim(char **s) { unsigned char c; @@ -180,10 +252,36 @@ static void fun_addspace(char *s, void *data, int num) tab->input = set_map_string(tab->input, s, strlen(s), (char*) CHR_SPACE); } +/* + * Create a string containing the mapped characters provided. + */ +static void fun_mkstring(char *s, void *data, int num) +{ + chrwork *arg = data; + char **res, *p = s; + + res = chr_map_input(arg->map->input, &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(char *s, void *data, int num) +{ + chrwork *arg = data; + + assert(arg->map->input); + set_map_string(arg->map->input, s, strlen(s), arg->string); +} + static int scan_string(char *s, void (*fun)(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) { @@ -212,7 +310,22 @@ static int scan_string(char *s, void (*fun)(char *c, void *data, int num), } 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'; @@ -238,27 +351,21 @@ chrmaptab *chr_read_maptab(char *name) res->input = xmalloc(sizeof(*res->input)); res->input->target = (unsigned char*) CHR_UNKNOWN; res->input->equiv = 0; -#if 0 +#if 1 res->input->children = xmalloc(sizeof(res->input) * 256); for (i = 0; i < 256; i++) { res->input->children[i] = xmalloc(sizeof(*res->input)); res->input->children[i]->children = 0; - res->input->children[i]->target = CHR_UNKNOWN; + 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; @@ -312,13 +419,38 @@ chrmaptab *chr_read_maptab(char *name) 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