X-Git-Url: http://git.indexdata.com/?p=yaz-moved-to-github.git;a=blobdiff_plain;f=src%2Fcqltransform.c;h=6cd8c109d1013f97a4e5aa8ddefb3c065c438deb;hp=9cbe7e0b22a4801891600db2701069fd1d1df9f0;hb=52188379291c43d463d4ee6742cf81a700541723;hpb=fbe668f0fed0122f70d56c438e8055fa74c3f660 diff --git a/src/cqltransform.c b/src/cqltransform.c index 9cbe7e0..6cd8c10 100644 --- a/src/cqltransform.c +++ b/src/cqltransform.c @@ -1,11 +1,7 @@ -/* $Id: cqltransform.c,v 1.24 2006-10-05 16:12:23 adam Exp $ - Copyright (C) 1995-2005, Index Data ApS - Index Data Aps - -This file is part of the YAZ toolkit. - -See the file LICENSE. -*/ +/* This file is part of the YAZ toolkit. + * Copyright (C) 1995-2008 Index Data + * See the file LICENSE for details. + */ /** * \file cqltransform.c @@ -53,36 +49,48 @@ cql_transform_t cql_transform_open_FILE(FILE *f) { const char *cp_value_start; const char *cp_value_end; + const char *cp_pattern_start; const char *cp_pattern_end; const char *cp = line; - while (*cp && !strchr(" \t=\r\n#", *cp)) + + while (*cp && strchr(" \t", *cp)) + cp++; + cp_pattern_start = cp; + + while (*cp && !strchr(" \t\r\n=#", *cp)) cp++; cp_pattern_end = cp; - if (cp == line) + if (cp == cp_pattern_start) continue; - while (*cp && strchr(" \t\r\n", *cp)) + while (*cp && strchr(" \t", *cp)) cp++; if (*cp != '=') - continue; + { + *pp = 0; + cql_transform_close(ct); + return 0; + } cp++; while (*cp && strchr(" \t\r\n", *cp)) cp++; cp_value_start = cp; - if (!(cp_value_end = strchr(cp, '#'))) + cp_value_end = strchr(cp, '#'); + if (!cp_value_end) cp_value_end = strlen(line) + line; if (cp_value_end != cp_value_start && strchr(" \t\r\n", cp_value_end[-1])) cp_value_end--; *pp = (struct cql_prop_entry *) xmalloc (sizeof(**pp)); - (*pp)->pattern = (char *) xmalloc (cp_pattern_end - line + 1); - memcpy ((*pp)->pattern, line, cp_pattern_end - line); - (*pp)->pattern[cp_pattern_end-line] = 0; + (*pp)->pattern = (char *) xmalloc(cp_pattern_end-cp_pattern_start + 1); + memcpy ((*pp)->pattern, cp_pattern_start, + cp_pattern_end-cp_pattern_start); + (*pp)->pattern[cp_pattern_end-cp_pattern_start] = '\0'; - (*pp)->value = (char *) xmalloc (cp_value_end - cp_value_start + 1); + (*pp)->value = (char *) xmalloc (cp_value_end-cp_value_start + 1); if (cp_value_start != cp_value_end) memcpy ((*pp)->value, cp_value_start, cp_value_end-cp_value_start); - (*pp)->value[cp_value_end - cp_value_start] = 0; + (*pp)->value[cp_value_end - cp_value_start] = '\0'; pp = &(*pp)->next; } *pp = 0; @@ -173,6 +181,18 @@ int cql_pr_attr_uri(cql_transform_t ct, const char *category, { if (!res) res = cql_lookup_property(ct, category, prefix, eval); + /* we have some aliases for some relations unfortunately.. */ + if (!res && !prefix && !strcmp(category, "relation")) + { + if (!strcmp(val, "==")) + res = cql_lookup_property(ct, category, prefix, "exact"); + if (!strcmp(val, "=")) + res = cql_lookup_property(ct, category, prefix, "eq"); + if (!strcmp(val, "<=")) + res = cql_lookup_property(ct, category, prefix, "le"); + if (!strcmp(val, ">=")) + res = cql_lookup_property(ct, category, prefix, "ge"); + } if (!res) res = cql_lookup_property(ct, category, prefix, "*"); } @@ -183,6 +203,7 @@ int cql_pr_attr_uri(cql_transform_t ct, const char *category, const char *cp0 = res, *cp1; while ((cp1 = strchr(cp0, '='))) { + int i; while (*cp1 && *cp1 != ' ') cp1++; if (cp1 - cp0 >= sizeof(buf)) @@ -190,7 +211,19 @@ int cql_pr_attr_uri(cql_transform_t ct, const char *category, memcpy (buf, cp0, cp1 - cp0); buf[cp1-cp0] = 0; (*pr)("@attr ", client_data); - (*pr)(buf, client_data); + + for (i = 0; buf[i]; i++) + { + if (buf[i] == '*') + (*pr)(eval, client_data); + else + { + char tmp[2]; + tmp[0] = buf[i]; + tmp[1] = '\0'; + (*pr)(tmp, client_data); + } + } (*pr)(" ", client_data); cp0 = cp1; while (*cp0 == ' ') @@ -312,19 +345,30 @@ static int cql_pr_prox(cql_transform_t ct, struct cql_node *mods, * characters starting at `term', or a null pointer of there are * none -- like memchr(). */ -static const char *wcchar(const char *term, int length) +static const char *wcchar(int start, const char *term, int length) { - const char *best = 0; - const char *current; - char *whichp; - - for (whichp = "*?"; *whichp != '\0'; whichp++) { - current = (const char *) memchr(term, *whichp, length); - if (current != 0 && (best == 0 || current < best)) - best = current; + while (length > 0) + { + if (start || term[-1] != '\\') + if (strchr("*?", *term)) + return term; + term++; + length--; + start = 0; } + return 0; +} + - return best; +/* ### checks for CQL relation-name rather than Type-1 attribute */ +static int has_modifier(struct cql_node *cn, const char *name) { + struct cql_node *mod; + for (mod = cn->u.st.modifiers; mod != 0; mod = mod->u.st.modifiers) { + if (!strcmp(mod->u.st.index, name)) + return 1; + } + + return 0; } @@ -336,10 +380,12 @@ void emit_term(cql_transform_t ct, { int i; const char *ns = cn->u.st.index_uri; + int process_term = !has_modifier(cn, "regexp"); + char *z3958_mem = 0; assert(cn->which == CQL_NODE_ST); - if (length > 0) + if (process_term && length > 0) { if (length > 1 && term[0] == '^' && term[length-1] == '^') { @@ -368,35 +414,40 @@ void emit_term(cql_transform_t ct, } } - if (length > 0) + if (process_term && length > 0) { + const char *first_wc = wcchar(1, term, length); + const char *second_wc = first_wc ? + wcchar(0, first_wc+1, length-(first_wc-term)-1) : 0; + /* Check for well-known globbing patterns that represent * simple truncation attributes as expected by, for example, * Bath-compliant server. If we find such a pattern but * there's no mapping for it, that's fine: we just use a * general pattern-matching attribute. */ - if (length > 1 && term[0] == '*' && term[length-1] == '*' && - wcchar(term+1, length-2) == 0 && - cql_pr_attr(ct, "truncation", "both", 0, - pr, client_data, 0)) { + if (first_wc == term && second_wc == term + length-1 + && *first_wc == '*' && *second_wc == '*' + && cql_pr_attr(ct, "truncation", "both", 0, pr, client_data, 0)) + { term++; length -= 2; } - else if (term[0] == '*' && - wcchar(term+1, length-1) == 0 && - cql_pr_attr(ct, "truncation", "left", 0, - pr, client_data, 0)) { + else if (first_wc == term && second_wc == 0 && *first_wc == '*' + && cql_pr_attr(ct, "truncation", "left", 0, + pr, client_data, 0)) + { term++; length--; } - else if (term[length-1] == '*' && - wcchar(term, length-1) == 0 && - cql_pr_attr(ct, "truncation", "right", 0, - pr, client_data, 0)) { + else if (first_wc == term + length-1 && second_wc == 0 + && *first_wc == '*' + && cql_pr_attr(ct, "truncation", "right", 0, + pr, client_data, 0)) + { length--; } - else if (wcchar(term, length)) + else if (first_wc) { /* We have one or more wildcard characters, but not in a * way that can be dealt with using only the standard @@ -408,17 +459,22 @@ void emit_term(cql_transform_t ct, * supported". */ int i; - char *mem; cql_pr_attr(ct, "truncation", "z3958", 0, pr, client_data, 28); - mem = (char *) xmalloc(length+1); - for (i = 0; i < length; i++) { - if (term[i] == '*') mem[i] = '?'; - else if (term[i] == '?') mem[i] = '#'; - else mem[i] = term[i]; + z3958_mem = (char *) xmalloc(length+1); + for (i = 0; i < length; i++) + { + if (i > 0 && term[i-1] == '\\') + z3958_mem[i] = term[i]; + else if (term[i] == '*') + z3958_mem[i] = '?'; + else if (term[i] == '?') + z3958_mem[i] = '#'; + else + z3958_mem[i] = term[i]; } - mem[length] = '\0'; - term = mem; + z3958_mem[length] = '\0'; + term = z3958_mem; } else { /* No masking characters. Use "truncation.none" if given. */ @@ -445,22 +501,44 @@ void emit_term(cql_transform_t ct, for (i = 0; iu.st.extra_terms; + if (ne) + { + (*pr)("@", client_data); + (*pr)(op, client_data); + (*pr)(" ", client_data); + } + emit_term(ct, cn, cn->u.st.term, strlen(cn->u.st.term), + pr, client_data); + for (; ne; ne = ne->u.st.extra_terms) + { + if (ne->u.st.extra_terms) + { + (*pr)("@", client_data); + (*pr)(op, client_data); + (*pr)(" ", client_data); + } + emit_term(ct, cn, ne->u.st.term, strlen(ne->u.st.term), + pr, client_data); + } } void emit_wordlist(cql_transform_t ct, @@ -530,18 +608,7 @@ void cql_transform_r(cql_transform_t ct, } } cql_pr_attr(ct, "always", 0, 0, pr, client_data, 0); - if (cn->u.st.relation && !cql_strcmp(cn->u.st.relation, "=")) - cql_pr_attr(ct, "relation", "eq", "scr", - pr, client_data, 19); - else if (cn->u.st.relation && !cql_strcmp(cn->u.st.relation, "<=")) - cql_pr_attr(ct, "relation", "le", "scr", - pr, client_data, 19); - else if (cn->u.st.relation && !cql_strcmp(cn->u.st.relation, ">=")) - cql_pr_attr(ct, "relation", "ge", "scr", - pr, client_data, 19); - else - cql_pr_attr(ct, "relation", cn->u.st.relation, "eq", - pr, client_data, 19); + cql_pr_attr(ct, "relation", cn->u.st.relation, 0, pr, client_data, 19); cql_pr_attr(ct, "structure", cn->u.st.relation, 0, pr, client_data, 24); if (cn->u.st.relation && !cql_strcmp(cn->u.st.relation, "all")) @@ -554,8 +621,7 @@ void cql_transform_r(cql_transform_t ct, } else { - emit_term(ct, cn, cn->u.st.term, strlen(cn->u.st.term), - pr, client_data); + emit_terms(ct, cn, pr, client_data, "and"); } break; case CQL_NODE_BOOL: