X-Git-Url: http://git.indexdata.com/?p=yaz-moved-to-github.git;a=blobdiff_plain;f=src%2Fcqltransform.c;h=b2422c9ee80989cd9ecd4d9f3ec6be809bcaadcf;hp=dbdd8aad59794820180698fb60577c6ed11f8c2b;hb=1254f5913d639e30ac822165196170a9b53748ff;hpb=8e9d2cec97def1cd8b89b1fe427f1a232661ec6f diff --git a/src/cqltransform.c b/src/cqltransform.c index dbdd8aa..b2422c9 100644 --- a/src/cqltransform.c +++ b/src/cqltransform.c @@ -1,8 +1,7 @@ /* This file is part of the YAZ toolkit. - * Copyright (C) 1995-2008 Index Data + * Copyright (C) 1995-2012 Index Data * See the file LICENSE for details. */ - /** * \file cqltransform.c * \brief Implements CQL transform (CQL to RPN conversion). @@ -17,17 +16,20 @@ * index * relationModifier */ +#if HAVE_CONFIG_H +#include +#endif #include #include #include -#include -#include +#include #include #include #include #include #include +#include #include #include @@ -78,7 +80,7 @@ static int cql_transform_parse_tok_line(cql_transform_t ct, const char *value_str = 0; /* attset type=value OR type=value */ - elem = nmem_malloc(ct->nmem, sizeof(*elem)); + 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)); @@ -106,7 +108,7 @@ static int cql_transform_parse_tok_line(cql_transform_t ct, t = yaz_tok_move(tp); } elem->attributeType = nmem_intdup(ct->nmem, 0); - if (sscanf(wrbuf_cstr(type_str), "%d", elem->attributeType) + if (sscanf(wrbuf_cstr(type_str), ODR_INT_PRINTF, elem->attributeType) != 1) { wrbuf_destroy(type_str); @@ -135,7 +137,7 @@ static int cql_transform_parse_tok_line(cql_transform_t ct, break; } value_str = yaz_tok_parse_string(tp); - if (isdigit(*value_str)) + if (yaz_isdigit(*value_str)) { elem->which = Z_AttributeValue_numeric; elem->value.numeric = @@ -143,7 +145,8 @@ static int cql_transform_parse_tok_line(cql_transform_t ct, } else { - Z_ComplexAttribute *ca = nmem_malloc(ct->nmem, sizeof(*ca)); + Z_ComplexAttribute *ca = (Z_ComplexAttribute *) + nmem_malloc(ct->nmem, sizeof(*ca)); elem->which = Z_AttributeValue_complex; elem->value.complex = ca; ca->num_list = 1; @@ -176,7 +179,7 @@ static int cql_transform_parse_tok_line(cql_transform_t ct, (*pp)->attr_list.attributes = 0; else { - (*pp)->attr_list.attributes = + (*pp)->attr_list.attributes = (Z_AttributeElement **) nmem_malloc(ct->nmem, ae_num * sizeof(Z_AttributeElement *)); memcpy((*pp)->attr_list.attributes, ae, @@ -348,7 +351,7 @@ const char *cql_lookup_reverse(cql_transform_t ct, } if (i == e->attr_list.num_attributes) - return e->pattern; + return e->pattern + clen; } } return 0; @@ -433,7 +436,7 @@ int cql_pr_attr_uri(cql_transform_t ct, const char *category, int i; while (*cp1 && *cp1 != ' ') cp1++; - if (cp1 - cp0 >= sizeof(buf)) + if (cp1 - cp0 >= (ptrdiff_t) sizeof(buf)) break; memcpy(buf, cp0, cp1 - cp0); buf[cp1-cp0] = 0; @@ -605,17 +608,25 @@ static int has_modifier(struct cql_node *cn, const char *name) { } -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) +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; const char *ns = cn->u.st.index_uri; - int process_term = !has_modifier(cn, "regexp"); - char *z3958_mem = 0; + 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")) + { + 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 && length > 0) @@ -682,34 +693,12 @@ void emit_term(cql_transform_t ct, } 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 - * 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; + z3958_mode = 1; cql_pr_attr(ct, "truncation", "z3958", 0, pr, client_data, YAZ_SRW_MASKING_CHAR_UNSUPP); - 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]; - } - z3958_mem[length] = '\0'; - term = z3958_mem; } - else { + else + { /* No masking characters. Use "truncation.none" if given. */ cql_pr_attr(ct, "truncation", "none", 0, pr, client_data, 0); @@ -730,27 +719,54 @@ void emit_term(cql_transform_t ct, } } + /* produce only \-sequences if: + 1) the output is a Z39.58-trunc reserved character + 2) the output is a PQF reserved character (\\, \") + */ (*pr)("\"", client_data); - for (i = 0; i 0 && term[i-1] == '\\') + { + if (term[i] == '\"' || 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); + /* avoid ?n sequences output (n=[0-9]) because that has + different semantics than just a single ? in Z39.58 + */ + if (i < length - 1 && yaz_isdigit(term[i+1])) + pr("\\\\", client_data); /* double \\ to survive PQF parse */ + } + else if (z3958_mode && term[i] == '?') + pr("#", client_data); + else if (term[i] != '\\') + { + if (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); + } } (*pr)("\" ", client_data); - xfree(z3958_mem); } -void emit_terms(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_terms(cql_transform_t ct, + struct cql_node *cn, + void (*pr)(const char *buf, void *client_data), + void *client_data, + const char *op) { struct cql_node *ne = cn->u.st.extra_terms; if (ne) @@ -774,11 +790,11 @@ void emit_terms(cql_transform_t ct, } } -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; @@ -873,7 +889,9 @@ void cql_transform_r(cql_transform_t ct, 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(); @@ -909,7 +927,8 @@ int cql_transform_FILE(cql_transform_t ct, struct cql_node *cn, FILE *f) return cql_transform(ct, cn, cql_fputs, f); } -int cql_transform_buf(cql_transform_t ct, struct cql_node *cn, char *out, int max) +int cql_transform_buf(cql_transform_t ct, struct cql_node *cn, + char *out, int max) { struct cql_buf_write_info info; int r; @@ -949,6 +968,7 @@ void cql_transform_set_error(cql_transform_t ct, int error, const char *addinfo) /* * Local variables: * c-basic-offset: 4 + * c-file-style: "Stroustrup" * indent-tabs-mode: nil * End: * vim: shiftwidth=4 tabstop=8 expandtab