X-Git-Url: http://git.indexdata.com/?p=yaz-moved-to-github.git;a=blobdiff_plain;f=src%2Fcqltransform.c;h=f17af20bf8777e80452242656f322d5442b57ae0;hp=9b2e18d0f460e278e3e0a292997e63f0ec56cf03;hb=b70c72a5a3914502cbdd17588e74e6c74bd282b6;hpb=8dde2679aeaae3fbe0cbee6c891ba07249245837 diff --git a/src/cqltransform.c b/src/cqltransform.c index 9b2e18d..f17af20 100644 --- a/src/cqltransform.c +++ b/src/cqltransform.c @@ -1,79 +1,255 @@ -/* $Id: cqltransform.c,v 1.20 2006-03-10 17:18:09 mike 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) 1995-2012 Index Data + * See the file LICENSE for details. + */ /** * \file cqltransform.c * \brief Implements CQL transform (CQL to RPN conversion). + * + * Evaluation order of rules: + * + * always + * relation + * structure + * position + * truncation + * index + * relationModifier */ +#if HAVE_CONFIG_H +#include +#endif +#include #include #include -#include +#include #include +#include +#include +#include +#include +#include +#include +#include struct cql_prop_entry { char *pattern; char *value; + Z_AttributeList attr_list; struct cql_prop_entry *next; }; struct cql_transform_t_ { struct cql_prop_entry *entry; + yaz_tok_cfg_t tok_cfg; int error; char *addinfo; + WRBUF w; + NMEM nmem; }; + +cql_transform_t cql_transform_create(void) +{ + cql_transform_t ct = (cql_transform_t) xmalloc(sizeof(*ct)); + ct->tok_cfg = yaz_tok_cfg_create(); + ct->w = wrbuf_alloc(); + ct->error = 0; + ct->addinfo = 0; + ct->entry = 0; + ct->nmem = nmem_create(); + return ct; +} + +static int cql_transform_parse_tok_line(cql_transform_t ct, + const char *pattern, + yaz_tok_parse_t tp) +{ + int ae_num = 0; + Z_AttributeElement *ae[20]; + int ret = 0; /* 0=OK, != 0 FAIL */ + int t; + t = yaz_tok_move(tp); + + while (t == YAZ_TOK_STRING && ae_num < 20) + { + WRBUF type_str = wrbuf_alloc(); + WRBUF set_str = 0; + Z_AttributeElement *elem = 0; + const char *value_str = 0; + /* attset type=value OR type=value */ + + elem = (Z_AttributeElement *) nmem_malloc(ct->nmem, sizeof(*elem)); + elem->attributeSet = 0; + ae[ae_num] = elem; + wrbuf_puts(ct->w, yaz_tok_parse_string(tp)); + wrbuf_puts(type_str, yaz_tok_parse_string(tp)); + t = yaz_tok_move(tp); + if (t == YAZ_TOK_EOF) + { + wrbuf_destroy(type_str); + if (set_str) + wrbuf_destroy(set_str); + break; + } + if (t == YAZ_TOK_STRING) + { + wrbuf_puts(ct->w, " "); + wrbuf_puts(ct->w, yaz_tok_parse_string(tp)); + set_str = type_str; + + elem->attributeSet = + yaz_string_to_oid_nmem(yaz_oid_std(), CLASS_ATTSET, + wrbuf_cstr(set_str), ct->nmem); + + type_str = wrbuf_alloc(); + wrbuf_puts(type_str, yaz_tok_parse_string(tp)); + t = yaz_tok_move(tp); + } + elem->attributeType = nmem_intdup(ct->nmem, 0); + if (sscanf(wrbuf_cstr(type_str), ODR_INT_PRINTF, elem->attributeType) + != 1) + { + wrbuf_destroy(type_str); + if (set_str) + wrbuf_destroy(set_str); + yaz_log(YLOG_WARN, "Expected numeric attribute type"); + ret = -1; + break; + } + + wrbuf_destroy(type_str); + if (set_str) + wrbuf_destroy(set_str); + + if (t != '=') + { + yaz_log(YLOG_WARN, "Expected = after after attribute type"); + ret = -1; + break; + } + t = yaz_tok_move(tp); + if (t != YAZ_TOK_STRING) /* value */ + { + yaz_log(YLOG_WARN, "Missing attribute value"); + ret = -1; + break; + } + value_str = yaz_tok_parse_string(tp); + if (yaz_isdigit(*value_str)) + { + elem->which = Z_AttributeValue_numeric; + elem->value.numeric = + nmem_intdup(ct->nmem, atoi(value_str)); + } + else + { + Z_ComplexAttribute *ca = (Z_ComplexAttribute *) + nmem_malloc(ct->nmem, sizeof(*ca)); + elem->which = Z_AttributeValue_complex; + elem->value.complex = ca; + ca->num_list = 1; + ca->list = (Z_StringOrNumeric **) + nmem_malloc(ct->nmem, sizeof(Z_StringOrNumeric *)); + ca->list[0] = (Z_StringOrNumeric *) + nmem_malloc(ct->nmem, sizeof(Z_StringOrNumeric)); + ca->list[0]->which = Z_StringOrNumeric_string; + ca->list[0]->u.string = nmem_strdup(ct->nmem, value_str); + ca->num_semanticAction = 0; + ca->semanticAction = 0; + } + wrbuf_puts(ct->w, "="); + wrbuf_puts(ct->w, yaz_tok_parse_string(tp)); + t = yaz_tok_move(tp); + wrbuf_puts(ct->w, " "); + ae_num++; + } + if (ret == 0) /* OK? */ + { + struct cql_prop_entry **pp = &ct->entry; + while (*pp) + pp = &(*pp)->next; + *pp = (struct cql_prop_entry *) xmalloc(sizeof(**pp)); + (*pp)->pattern = xstrdup(pattern); + (*pp)->value = xstrdup(wrbuf_cstr(ct->w)); + + (*pp)->attr_list.num_attributes = ae_num; + if (ae_num == 0) + (*pp)->attr_list.attributes = 0; + else + { + (*pp)->attr_list.attributes = (Z_AttributeElement **) + nmem_malloc(ct->nmem, + ae_num * sizeof(Z_AttributeElement *)); + memcpy((*pp)->attr_list.attributes, ae, + ae_num * sizeof(Z_AttributeElement *)); + } + (*pp)->next = 0; + + if (0) + { + ODR pr = odr_createmem(ODR_PRINT); + Z_AttributeList *alp = &(*pp)->attr_list; + odr_setprint(pr, yaz_log_file()); + z_AttributeList(pr, &alp, 0, 0); + odr_setprint(pr, 0); + odr_destroy(pr); + } + } + return ret; +} + +int cql_transform_define_pattern(cql_transform_t ct, const char *pattern, + const char *value) +{ + int r; + yaz_tok_parse_t tp = yaz_tok_parse_buf(ct->tok_cfg, value); + yaz_tok_cfg_single_tokens(ct->tok_cfg, "="); + r = cql_transform_parse_tok_line(ct, pattern, tp); + yaz_tok_parse_destroy(tp); + return r; +} + cql_transform_t cql_transform_open_FILE(FILE *f) { + cql_transform_t ct = cql_transform_create(); char line[1024]; - cql_transform_t ct = (cql_transform_t) xmalloc (sizeof(*ct)); - struct cql_prop_entry **pp = &ct->entry; - ct->error = 0; - ct->addinfo = 0; + yaz_tok_cfg_single_tokens(ct->tok_cfg, "="); + while (fgets(line, sizeof(line)-1, f)) { - const char *cp_value_start; - const char *cp_value_end; - const char *cp_pattern_end; - const char *cp = line; - while (*cp && !strchr(" \t=\r\n#", *cp)) - cp++; - cp_pattern_end = cp; - if (cp == line) - continue; - while (*cp && strchr(" \t\r\n", *cp)) - cp++; - if (*cp != '=') - continue; - cp++; - while (*cp && strchr(" \t\r\n", *cp)) - cp++; - cp_value_start = cp; - if (!(cp_value_end = strchr(cp, '#'))) - 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)->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 = &(*pp)->next; + yaz_tok_parse_t tp = yaz_tok_parse_buf(ct->tok_cfg, line); + int t; + wrbuf_rewind(ct->w); + t = yaz_tok_move(tp); + if (t == YAZ_TOK_STRING) + { + char * pattern = xstrdup(yaz_tok_parse_string(tp)); + t = yaz_tok_move(tp); + if (t != '=') + { + yaz_tok_parse_destroy(tp); + cql_transform_close(ct); + return 0; + } + if (cql_transform_parse_tok_line(ct, pattern, tp)) + { + yaz_tok_parse_destroy(tp); + cql_transform_close(ct); + return 0; + } + xfree(pattern); + } + else if (t != YAZ_TOK_EOF) + { + yaz_tok_parse_destroy(tp); + cql_transform_close(ct); + return 0; + } + yaz_tok_parse_destroy(tp); } - *pp = 0; return ct; } @@ -86,14 +262,16 @@ void cql_transform_close(cql_transform_t ct) while (pe) { struct cql_prop_entry *pe_next = pe->next; - xfree (pe->pattern); - xfree (pe->value); - xfree (pe); + xfree(pe->pattern); + xfree(pe->value); + xfree(pe); pe = pe_next; } - if (ct->addinfo) - xfree (ct->addinfo); - xfree (ct); + xfree(ct->addinfo); + yaz_tok_cfg_destroy(ct->tok_cfg); + wrbuf_destroy(ct->w); + nmem_destroy(ct->nmem); + xfree(ct); } cql_transform_t cql_transform_open_fname(const char *fname) @@ -107,6 +285,78 @@ cql_transform_t cql_transform_open_fname(const char *fname) return ct; } +#if 0 +struct Z_AttributeElement { + Z_AttributeSetId *attributeSet; /* OPT */ + int *attributeType; + int which; + union { + int *numeric; + Z_ComplexAttribute *complex; +#define Z_AttributeValue_numeric 1 +#define Z_AttributeValue_complex 2 + } value; +}; +#endif + +static int compare_attr(Z_AttributeElement *a, Z_AttributeElement *b) +{ + ODR odr_a = odr_createmem(ODR_ENCODE); + ODR odr_b = odr_createmem(ODR_ENCODE); + int len_a, len_b; + char *buf_a, *buf_b; + int ret; + + z_AttributeElement(odr_a, &a, 0, 0); + z_AttributeElement(odr_b, &b, 0, 0); + + buf_a = odr_getbuf(odr_a, &len_a, 0); + buf_b = odr_getbuf(odr_b, &len_b, 0); + + ret = yaz_memcmp(buf_a, buf_b, len_a, len_b); + + odr_destroy(odr_a); + odr_destroy(odr_b); + return ret; +} + +const char *cql_lookup_reverse(cql_transform_t ct, + const char *category, + Z_AttributeList *attributes) +{ + struct cql_prop_entry *e; + size_t clen = strlen(category); + for (e = ct->entry; e; e = e->next) + { + if (!strncmp(e->pattern, category, clen)) + { + /* category matches.. See if attributes in pattern value + are all listed in actual attributes */ + int i; + for (i = 0; i < e->attr_list.num_attributes; i++) + { + /* entry attribute */ + Z_AttributeElement *e_ae = e->attr_list.attributes[i]; + int j; + for (j = 0; j < attributes->num_attributes; j++) + { + /* actual attribute */ + Z_AttributeElement *a_ae = attributes->attributes[j]; + int r = compare_attr(e_ae, a_ae); + if (r == 0) + break; + } + if (j == attributes->num_attributes) + break; /* i was not found at all.. try next pattern */ + + } + if (i == e->attr_list.num_attributes) + return e->pattern + clen; + } + } + return 0; +} + static const char *cql_lookup_property(cql_transform_t ct, const char *pat1, const char *pat2, const char *pat3) @@ -115,16 +365,16 @@ static const char *cql_lookup_property(cql_transform_t ct, struct cql_prop_entry *e; if (pat1 && pat2 && pat3) - sprintf (pattern, "%.39s.%.39s.%.39s", pat1, pat2, pat3); + sprintf(pattern, "%.39s.%.39s.%.39s", pat1, pat2, pat3); else if (pat1 && pat2) - sprintf (pattern, "%.39s.%.39s", pat1, pat2); + sprintf(pattern, "%.39s.%.39s", pat1, pat2); else if (pat1 && pat3) - sprintf (pattern, "%.39s.%.39s", pat1, pat3); + sprintf(pattern, "%.39s.%.39s", pat1, pat3); else if (pat1) - sprintf (pattern, "%.39s", pat1); + sprintf(pattern, "%.39s", pat1); else return 0; - + for (e = ct->entry; e; e = e->next) { if (!cql_strcmp(e->pattern, pattern)) @@ -142,11 +392,11 @@ int cql_pr_attr_uri(cql_transform_t ct, const char *category, const char *res = 0; const char *eval = val ? val : default_val; const char *prefix = 0; - + if (uri) { struct cql_prop_entry *e; - + for (e = ct->entry; e; e = e->next) if (!memcmp(e->pattern, "set.", 4) && e->value && !strcmp(e->value, uri)) @@ -161,6 +411,18 @@ 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, "*"); } @@ -171,14 +433,27 @@ 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)) + if (cp1 - cp0 >= (ptrdiff_t) sizeof(buf)) break; - memcpy (buf, cp0, cp1 - cp0); + 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 == ' ') @@ -209,136 +484,313 @@ int cql_pr_attr(cql_transform_t ct, const char *category, } -/* Returns location of first wildcard character in the `length' - * characters starting at `term', or a null pointer of there are - * none -- like memchr(). - */ -static const char *wcchar(const char *term, int length) +static void cql_pr_int(int val, + void (*pr)(const char *buf, void *client_data), + void *client_data) +{ + char buf[21]; /* enough characters to 2^64 */ + sprintf(buf, "%d", val); + (*pr)(buf, client_data); + (*pr)(" ", client_data); +} + + +static int cql_pr_prox(cql_transform_t ct, struct cql_node *mods, + void (*pr)(const char *buf, void *client_data), + void *client_data) { - 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; + int exclusion = 0; + int distance = -1; + int ordered = 0; + int proxrel = 2; /* less than or equal */ + int unit = 2; /* word */ + + while (mods) + { + const char *name = mods->u.st.index; + const char *term = mods->u.st.term; + const char *relation = mods->u.st.relation; + + if (!strcmp(name, "distance")) { + distance = strtol(term, (char**) 0, 0); + if (!strcmp(relation, "=")) + proxrel = 3; + else if (!strcmp(relation, ">")) + proxrel = 5; + else if (!strcmp(relation, "<")) + proxrel = 1; + else if (!strcmp(relation, ">=")) + proxrel = 4; + else if (!strcmp(relation, "<=")) + proxrel = 2; + else if (!strcmp(relation, "<>")) + proxrel = 6; + else + { + ct->error = YAZ_SRW_UNSUPP_PROX_RELATION; + ct->addinfo = xstrdup(relation); + return 0; + } + } + else if (!strcmp(name, "ordered")) + ordered = 1; + else if (!strcmp(name, "unordered")) + ordered = 0; + else if (!strcmp(name, "unit")) + { + if (!strcmp(term, "word")) + unit = 2; + else if (!strcmp(term, "sentence")) + unit = 3; + else if (!strcmp(term, "paragraph")) + unit = 4; + else if (!strcmp(term, "element")) + unit = 8; + else + { + ct->error = YAZ_SRW_UNSUPP_PROX_UNIT; + ct->addinfo = xstrdup(term); + return 0; + } + } + else + { + ct->error = YAZ_SRW_UNSUPP_BOOLEAN_MODIFIER; + ct->addinfo = xstrdup(name); + return 0; + } + mods = mods->u.st.modifiers; } - return best; + if (distance == -1) + distance = (unit == 2) ? 1 : 0; + + cql_pr_int(exclusion, pr, client_data); + cql_pr_int(distance, pr, client_data); + cql_pr_int(ordered, pr, client_data); + cql_pr_int(proxrel, pr, client_data); + (*pr)("k ", client_data); + cql_pr_int(unit, pr, client_data); + + return 1; +} + +/* ### 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; } -void emit_term(cql_transform_t ct, - const char *term, int length, - void (*pr)(const char *buf, void *client_data), - void *client_data) +static void emit_term(cql_transform_t ct, + struct cql_node *cn, + const char *term, int length, + void (*pr)(const char *buf, void *client_data), + void *client_data) { int i; - if (length > 0) + const char *ns = cn->u.st.index_uri; + int z3958_mode = 0; + int process_term = 1; + + if (has_modifier(cn, "regexp")) + process_term = 0; + else if (cql_lookup_property(ct, "truncation", 0, "cql")) { - if (length > 1 && term[0] == '^' && term[length-1] == '^') + process_term = 0; + cql_pr_attr(ct, "truncation", "cql", 0, + pr, client_data, YAZ_SRW_MASKING_CHAR_UNSUPP); + } + assert(cn->which == CQL_NODE_ST); + + if (process_term) + { + unsigned anchor = 0; + unsigned trunc = 0; + for (i = 0; i < length; i++) + { + if (term[i] == '\\' && i < length - 1) + i++; + else + { + switch (term[i]) + { + case '^': + if (i == 0) + anchor |= 1; + else if (i == length - 1) + anchor |= 2; + break; + case '*': + if (i == 0) + trunc |= 1; + else if (i == length - 1) + trunc |= 2; + else + z3958_mode = 1; + break; + case '?': + z3958_mode = 1; + break; + } + } + } + if (anchor == 3) { cql_pr_attr(ct, "position", "firstAndLast", 0, - pr, client_data, 32); + pr, client_data, YAZ_SRW_ANCHORING_CHAR_IN_UNSUPP_POSITION); term++; length -= 2; } - else if (term[0] == '^') + else if (anchor == 1) { cql_pr_attr(ct, "position", "first", 0, - pr, client_data, 32); + pr, client_data, YAZ_SRW_ANCHORING_CHAR_IN_UNSUPP_POSITION); term++; length--; } - else if (term[length-1] == '^') + else if (anchor == 2) { cql_pr_attr(ct, "position", "last", 0, - pr, client_data, 32); + pr, client_data, YAZ_SRW_ANCHORING_CHAR_IN_UNSUPP_POSITION); length--; } else { cql_pr_attr(ct, "position", "any", 0, - pr, client_data, 32); + pr, client_data, YAZ_SRW_ANCHORING_CHAR_IN_UNSUPP_POSITION); + } + if (z3958_mode == 0) + { + if (trunc == 3 && cql_pr_attr(ct, "truncation", + "both", 0, pr, client_data, 0)) + { + term++; + length -= 2; + } + else if (trunc == 1 && cql_pr_attr(ct, "truncation", + "left", 0, pr, client_data, 0)) + { + term++; + length--; + } + else if (trunc == 2 && cql_pr_attr(ct, "truncation", "right", 0, + pr, client_data, 0)) + { + length--; + } + else if (trunc) + z3958_mode = 1; + else + cql_pr_attr(ct, "truncation", "none", 0, + pr, client_data, 0); } + if (z3958_mode) + cql_pr_attr(ct, "truncation", "z3958", 0, + pr, client_data, YAZ_SRW_MASKING_CHAR_UNSUPP); } - - if (length > 0) + if (ns) { + cql_pr_attr_uri(ct, "index", ns, + cn->u.st.index, "serverChoice", + pr, client_data, YAZ_SRW_UNSUPP_INDEX); + } + if (cn->u.st.modifiers) { - /* 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)) { - term++; - length -= 2; - } - else if (term[0] == '*' && - wcchar(term+1, length-1) == 0 && - 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)) { - length--; + struct cql_node *mod = cn->u.st.modifiers; + for (; mod; mod = mod->u.st.modifiers) + { + cql_pr_attr(ct, "relationModifier", mod->u.st.index, 0, + pr, client_data, YAZ_SRW_UNSUPP_RELATION_MODIFIER); } - else if (wcchar(term, length)) + } + (*pr)("\"", client_data); + if (process_term) + for (i = 0; i < length; i++) { - /* We have one or more wildcard characters, but not in a - * way that can be dealt with using only the standard - * left-, right- and both-truncation attributes. We need - * to translate the pattern into a Z39.58-type pattern, - * which has been supported in BIB-1 since 1996. If - * there's no configuration element for "truncation.z3958" - * we indicate this as error 28 "Masking character not - * 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]; + char x[2]; /* temp buffer */ + if (term[i] == '\\' && i < length - 1) + { + i++; + if (strchr("\"\\", term[i])) + pr("\\", client_data); + if (z3958_mode && strchr("#?", term[i])) + pr("\\\\", client_data); /* double \\ to survive PQF parse */ + x[0] = term[i]; + x[1] = '\0'; + pr(x, client_data); + } + else if (z3958_mode && term[i] == '*') + { + pr("?", client_data); + if (i < length - 1 && yaz_isdigit(term[i+1])) + pr("\\\\", client_data); /* dbl \\ to survive PQF parse */ + } + else if (z3958_mode && term[i] == '?') + { + pr("#", client_data); + } + else + { + if (term[i] == '\"') + pr("\\", client_data); + if (z3958_mode && strchr("#?", term[i])) + pr("\\\\", client_data); /* dbl \\ to survive PQF parse */ + x[0] = term[i]; + x[1] = '\0'; + pr(x, client_data); } - mem[length] = '\0'; - term = mem; } - else { - /* No masking characters. Use "truncation.none" if given. */ - cql_pr_attr(ct, "truncation", "none", 0, - pr, client_data, 0); + else + { + for (i = 0; i < length; i++) + { + char x[2]; + x[0] = term[i]; + x[1] = '\0'; + pr(x, client_data); } } + (*pr)("\" ", client_data); +} - (*pr)("\"", client_data); - for (i = 0; iu.st.extra_terms; + if (ne) { - char buf[2]; - buf[0] = term[i]; - buf[1] = 0; - (*pr)(buf, client_data); + (*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); } - (*pr)("\" ", client_data); } -void emit_wordlist(cql_transform_t ct, - struct cql_node *cn, - void (*pr)(const char *buf, void *client_data), - void *client_data, - const char *op) +static void emit_wordlist(cql_transform_t ct, + struct cql_node *cn, + void (*pr)(const char *buf, void *client_data), + void *client_data, + const char *op) { const char *cp0 = cn->u.st.term; const char *cp1; @@ -354,7 +806,7 @@ void emit_wordlist(cql_transform_t ct, (*pr)("@", client_data); (*pr)(op, client_data); (*pr)(" ", client_data); - emit_term(ct, last_term, last_length, pr, client_data); + emit_term(ct, cn, last_term, last_length, pr, client_data); } last_term = cp0; if (cp1) @@ -364,7 +816,7 @@ void emit_wordlist(cql_transform_t ct, cp0 = cp1; } if (last_term) - emit_term(ct, last_term, last_length, pr, client_data); + emit_term(ct, cn, last_term, last_length, pr, client_data); } void cql_transform_r(cql_transform_t ct, @@ -373,6 +825,7 @@ void cql_transform_r(cql_transform_t ct, void *client_data) { const char *ns; + struct cql_node *mods; if (!cn) return; @@ -395,70 +848,53 @@ void cql_transform_r(cql_transform_t ct, { if (!ct->error) { - ct->error = 15; + ct->error = YAZ_SRW_UNSUPP_CONTEXT_SET; ct->addinfo = 0; } } 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); - if (cn->u.st.modifiers) - { - struct cql_node *mod = cn->u.st.modifiers; - for (; mod; mod = mod->u.st.modifiers) - { - cql_pr_attr(ct, "relationModifier", mod->u.st.index, 0, - pr, client_data, 20); - } - } + cql_pr_attr(ct, "relation", cn->u.st.relation, 0, pr, client_data, + YAZ_SRW_UNSUPP_RELATION); cql_pr_attr(ct, "structure", cn->u.st.relation, 0, - pr, client_data, 24); - if (ns) { - cql_pr_attr_uri(ct, "index", ns, - cn->u.st.index, "serverChoice", - pr, client_data, 16); - } + pr, client_data, YAZ_SRW_UNSUPP_COMBI_OF_RELATION_AND_TERM); if (cn->u.st.relation && !cql_strcmp(cn->u.st.relation, "all")) - { emit_wordlist(ct, cn, pr, client_data, "and"); - } else if (cn->u.st.relation && !cql_strcmp(cn->u.st.relation, "any")) - { emit_wordlist(ct, cn, pr, client_data, "or"); - } else - { - emit_term(ct, 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: (*pr)("@", client_data); (*pr)(cn->u.boolean.value, client_data); (*pr)(" ", client_data); + mods = cn->u.boolean.modifiers; + if (!strcmp(cn->u.boolean.value, "prox")) + { + if (!cql_pr_prox(ct, mods, pr, client_data)) + return; + } + else if (mods) + { + /* Boolean modifiers other than on proximity not supported */ + ct->error = YAZ_SRW_UNSUPP_BOOLEAN_MODIFIER; + ct->addinfo = xstrdup(mods->u.st.index); + return; + } cql_transform_r(ct, cn->u.boolean.left, pr, client_data); cql_transform_r(ct, cn->u.boolean.right, pr, client_data); break; - + case CQL_NODE_SORT: + cql_transform_r(ct, cn->u.sort.search, pr, client_data); + break; default: fprintf(stderr, "Fatal: impossible CQL node-type %d\n", cn->which); abort(); } } -int cql_transform(cql_transform_t ct, - struct cql_node *cn, +int cql_transform(cql_transform_t ct, struct cql_node *cn, void (*pr)(const char *buf, void *client_data), void *client_data) { @@ -466,8 +902,7 @@ int cql_transform(cql_transform_t ct, NMEM nmem = nmem_create(); ct->error = 0; - if (ct->addinfo) - xfree (ct->addinfo); + xfree(ct->addinfo); ct->addinfo = 0; for (e = ct->entry; e ; e = e->next) @@ -477,7 +912,7 @@ int cql_transform(cql_transform_t ct, else if (!cql_strcmp(e->pattern, "set")) cql_apply_prefix(nmem, cn, 0, e->value); } - cql_transform_r (ct, cn, pr, client_data); + cql_transform_r(ct, cn, pr, client_data); nmem_destroy(nmem); return ct->error; } @@ -498,6 +933,16 @@ int cql_transform_buf(cql_transform_t ct, struct cql_node *cn, info.max = max; info.buf = out; r = cql_transform(ct, cn, cql_buf_write_handler, &info); + if (info.off < 0) { + /* Attempt to write past end of buffer. For some reason, this + SRW diagnostic is deprecated, but it's so perfect for our + purposes that it would be stupid not to use it. */ + char numbuf[30]; + ct->error = YAZ_SRW_TOO_MANY_CHARS_IN_QUERY; + sprintf(numbuf, "%ld", (long) info.max); + ct->addinfo = xstrdup(numbuf); + return -1; + } if (info.off >= 0) info.buf[info.off] = '\0'; return r; @@ -508,9 +953,18 @@ int cql_transform_error(cql_transform_t ct, const char **addinfo) *addinfo = ct->addinfo; return ct->error; } + +void cql_transform_set_error(cql_transform_t ct, int error, const char *addinfo) +{ + xfree(ct->addinfo); + ct->addinfo = addinfo ? xstrdup(addinfo) : 0; + ct->error = error; +} + /* * Local variables: * c-basic-offset: 4 + * c-file-style: "Stroustrup" * indent-tabs-mode: nil * End: * vim: shiftwidth=4 tabstop=8 expandtab