X-Git-Url: http://git.indexdata.com/?p=yaz-moved-to-github.git;a=blobdiff_plain;f=src%2Fcql.y;h=cf6995e22468349188bfc311bb11e89a653c53ec;hp=8b3d3ee8116adeebede6791b99b434b8fdf2b3c2;hb=67263adf9c8efbd384b09ec71b98844ce7c7d2bf;hpb=4c7863aac66ff92802fcd5476f8b7ae1feb8ebdf diff --git a/src/cql.y b/src/cql.y index 8b3d3ee..cf6995e 100644 --- a/src/cql.y +++ b/src/cql.y @@ -1,27 +1,38 @@ -/* $Id: cql.y,v 1.4 2004-03-17 18:31:11 adam Exp $ - Copyright (C) 2002-2004 - Index Data Aps - -This file is part of the YAZ toolkit. - -See the file LICENSE. - - bison parser for CQL grammar. -*/ +/* This file is part of the YAZ toolkit. + * Copyright (C) 1995-2011 Index Data + * See the file LICENSE for details. + */ +/* bison parser for CQL grammar. */ %{ +/** + * \file cql.c + * \brief Implements CQL parser. + * + * This is a YACC parser, but since it must be reentrant, Bison is required. + * The original source file is cql.y. + */ +#if HAVE_CONFIG_H +#include +#endif #include #include #include -#include +#include #include #include #include - + + /** Node in the LALR parse tree. */ typedef struct { + /** Inhereted attribute: relation */ struct cql_node *rel; + /** Synthesized attribute: CQL node */ struct cql_node *cql; + /** string buffer with token */ char *buf; + /** length of token */ size_t len; + /** size of buffer (len <= size) */ size_t size; } token; @@ -45,20 +56,46 @@ See the file LICENSE. %} %pure_parser -%token TERM AND OR NOT PROX GE LE NE -%expect 9 +%token PREFIX_NAME SIMPLE_STRING AND OR NOT PROX GE LE NE EXACT SORTBY %% top: { - $$.rel = cql_node_mk_sc("cql.serverChoice", "scr", 0); + $$.rel = cql_node_mk_sc(((CQL_parser) parm)->nmem, + "cql.serverChoice", "=", 0); ((CQL_parser) parm)->top = 0; -} cqlQuery1 { +} cqlQuery1 sortby { cql_node_destroy($$.rel); - ((CQL_parser) parm)->top = $2.cql; + if ($3.cql) + { + $3.cql->u.sort.search = $2.cql; + ((CQL_parser) parm)->top = $3.cql; + } else { + ((CQL_parser) parm)->top = $2.cql; + } } ; +sortby: /* empty */ + { $$.cql = 0; } +| SORTBY sortSpec { + $$.cql = $2.cql; + }; + +sortSpec: sortSpec singleSpec { + $$.cql = $1.cql; + $$.cql->u.sort.next = $2.cql; + } +| singleSpec +{ + $$.cql = $1.cql; +}; + +singleSpec: index modifiers { + $$.cql = cql_node_mk_sort(((CQL_parser) parm)->nmem, $1.buf, $2.cql); + } +; + cqlQuery1: cqlQuery | cqlQuery error { cql_node_destroy($1.cql); @@ -66,13 +103,31 @@ cqlQuery1: cqlQuery } ; -cqlQuery: +cqlQuery: + scopedClause + | + '>' searchTerm '=' searchTerm { + $$.rel = $0.rel; + } cqlQuery { + $$.cql = cql_apply_prefix(((CQL_parser) parm)->nmem, + $6.cql, $2.buf, $4.buf); + } +| '>' searchTerm { + $$.rel = $0.rel; + } cqlQuery { + $$.cql = cql_apply_prefix(((CQL_parser) parm)->nmem, + $4.cql, 0, $2.buf); + } +; + +scopedClause: searchClause | - cqlQuery boolean modifiers { + scopedClause boolean modifiers { $$.rel = $0.rel; } searchClause { - struct cql_node *cn = cql_node_mk_boolean($2.buf); + struct cql_node *cn = cql_node_mk_boolean(((CQL_parser) parm)->nmem, + $2.buf); cn->u.boolean.modifiers = $3.cql; cn->u.boolean.left = $1.cql; @@ -90,48 +145,52 @@ searchClause: $$.cql = $3.cql; } | - searchTerm { - struct cql_node *st = cql_node_dup ($0.rel); - st->u.st.term = xstrdup($1.buf); +searchTerm extraTerms { + struct cql_node *st = cql_node_dup(((CQL_parser) parm)->nmem, $0.rel); + st->u.st.extra_terms = $2.cql; + st->u.st.term = nmem_strdup(((CQL_parser)parm)->nmem, $1.buf); $$.cql = st; } | index relation modifiers { - $$.rel = cql_node_mk_sc($1.buf, $2.buf, 0); + $$.rel = cql_node_mk_sc(((CQL_parser) parm)->nmem, $1.buf, $2.buf, 0); $$.rel->u.st.modifiers = $3.cql; } searchClause { $$.cql = $5.cql; cql_node_destroy($4.rel); } -| '>' searchTerm '=' searchTerm { - $$.rel = $0.rel; - } cqlQuery { - $$.cql = cql_apply_prefix($6.cql, $2.buf, $4.buf); - } -| '>' searchTerm { - $$.rel = $0.rel; - } cqlQuery { - $$.cql = cql_apply_prefix($4.cql, 0, $2.buf); - } ; -/* unary NOT search TERM here .. */ +extraTerms: +SIMPLE_STRING extraTerms { + struct cql_node *st = cql_node_mk_sc(((CQL_parser) parm)->nmem, + /* index */ 0, /* rel */ 0, $1.buf); + st->u.st.extra_terms = $2.cql; + $$.cql = st; +} +| +{ $$.cql = 0; } +; + + +/* unary NOT search SIMPLE_STRING here .. */ boolean: - AND | OR | NOT | PROX - ; + AND | OR | NOT | PROX ; modifiers: modifiers '/' searchTerm { - struct cql_node *mod = cql_node_mk_sc($3.buf, "=", 0); + struct cql_node *mod = cql_node_mk_sc(((CQL_parser)parm)->nmem, + $3.buf, 0, 0); mod->u.st.modifiers = $1.cql; $$.cql = mod; } | -modifiers '/' searchTerm mrelation searchTerm +modifiers '/' searchTerm relation_symbol searchTerm { - struct cql_node *mod = cql_node_mk_sc($3.buf, $4.buf, $5.buf); + struct cql_node *mod = cql_node_mk_sc(((CQL_parser)parm)->nmem, + $3.buf, $4.buf, $5.buf); mod->u.st.modifiers = $1.cql; $$.cql = mod; @@ -142,34 +201,29 @@ modifiers '/' searchTerm mrelation searchTerm } ; -mrelation: - '=' -| '>' -| '<' -| GE -| LE -| NE -; +relation: PREFIX_NAME | relation_symbol; -relation: +relation_symbol: '=' | '>' | '<' | GE | LE | NE -| TERM +| EXACT ; index: searchTerm; searchTerm: - TERM + SIMPLE_STRING +| PREFIX_NAME | AND | OR | NOT | PROX +| SORTBY ; %% @@ -179,16 +233,19 @@ int yyerror(char *s) return 0; } -/* - * bison lexer for CQL. +/** + * putb is a utility that puts one character to the string + * in current lexical token. This routine deallocates as + * necessary using NMEM. */ static void putb(YYSTYPE *lval, CQL_parser cp, int c) { if (lval->len+1 >= lval->size) { - char *nb = nmem_malloc(cp->nmem, (lval->size = lval->len * 2 + 20)); - memcpy (nb, lval->buf, lval->len); + char *nb = (char *) + nmem_malloc(cp->nmem, (lval->size = lval->len * 2 + 20)); + memcpy(nb, lval->buf, lval->len); lval->buf = nb; } if (c) @@ -197,6 +254,10 @@ static void putb(YYSTYPE *lval, CQL_parser cp, int c) } +/** + * yylex returns next token for Bison to be read. In this + * case one of the CQL terminals are returned. + */ int yylex(YYSTYPE *lval, void *vp) { CQL_parser cp = (CQL_parser) vp; @@ -205,7 +266,7 @@ int yylex(YYSTYPE *lval, void *vp) lval->rel = 0; lval->len = 0; lval->size = 10; - lval->buf = nmem_malloc(cp->nmem, lval->size); + lval->buf = (char *) nmem_malloc(cp->nmem, lval->size); lval->buf[0] = '\0'; do { @@ -214,12 +275,23 @@ int yylex(YYSTYPE *lval, void *vp) return 0; if (c == '\n') return 0; - } while (isspace(c)); + } while (yaz_isspace(c)); if (strchr("()=>') + if (c == '=') + { + c1 = cp->getbyte(cp->client_data); + if (c1 == '=') + { + putb(lval, cp, c1); + return EXACT; + } + else + cp->ungetbyte(c1, cp->client_data); + } + else if (c == '>') { c1 = cp->getbyte(cp->client_data); if (c1 == '=') @@ -253,36 +325,75 @@ int yylex(YYSTYPE *lval, void *vp) while ((c = cp->getbyte(cp->client_data)) != 0 && c != '"') { if (c == '\\') + { + putb(lval, cp, c); c = cp->getbyte(cp->client_data); - putb(lval, cp, c); + if (!c) + break; + } + putb(lval, cp, c); } putb(lval, cp, 0); + return SIMPLE_STRING; } else { - putb(lval, cp, c); - while ((c = cp->getbyte(cp->client_data)) != 0 && - !strchr(" \n()=<>/", c)) - { - if (c == '\\') - c = cp->getbyte(cp->client_data); - putb(lval, cp, c); - } + int relation_like = 0; + while (c != 0 && !strchr(" \n()=<>/", c)) + { + if (c == '.') + relation_like = 1; + if (c == '\\') + { + putb(lval, cp, c); + c = cp->getbyte(cp->client_data); + if (!c) + break; + } + putb(lval, cp, c); + c = cp->getbyte(cp->client_data); + } + putb(lval, cp, 0); #if YYDEBUG - printf ("got %s\n", lval->buf); + printf ("got %s\n", lval->buf); #endif - if (c != 0) - cp->ungetbyte(c, cp->client_data); - if (!strcmp(lval->buf, "and")) - return AND; - if (!strcmp(lval->buf, "or")) - return OR; - if (!strcmp(lval->buf, "not")) - return NOT; - if (!strncmp(lval->buf, "prox", 4)) - return PROX; + if (c != 0) + cp->ungetbyte(c, cp->client_data); + if (!cql_strcmp(lval->buf, "and")) + { + lval->buf = "and"; + return AND; + } + if (!cql_strcmp(lval->buf, "or")) + { + lval->buf = "or"; + return OR; + } + if (!cql_strcmp(lval->buf, "not")) + { + lval->buf = "not"; + return NOT; + } + if (!cql_strcmp(lval->buf, "prox")) + { + lval->buf = "prox"; + return PROX; + } + if (!cql_strcmp(lval->buf, "sortby")) + { + lval->buf = "sortby"; + return SORTBY; + } + if (!cql_strcmp(lval->buf, "all")) + relation_like = 1; + if (!cql_strcmp(lval->buf, "any")) + relation_like = 1; + if (!cql_strcmp(lval->buf, "adj")) + relation_like = 1; + if (relation_like) + return PREFIX_NAME; } - return TERM; + return SIMPLE_STRING; } @@ -295,8 +406,7 @@ int cql_parser_stream(CQL_parser cp, cp->getbyte = getbyte; cp->ungetbyte = ungetbyte; cp->client_data = client_data; - if (cp->top) - cql_node_destroy(cp->top); + cql_node_destroy(cp->top); cql_parse(cp); if (cp->top) return 0; @@ -305,7 +415,7 @@ int cql_parser_stream(CQL_parser cp, CQL_parser cql_parser_create(void) { - CQL_parser cp = (CQL_parser) xmalloc (sizeof(*cp)); + CQL_parser cp = (CQL_parser) xmalloc(sizeof(*cp)); cp->top = 0; cp->getbyte = 0; @@ -328,3 +438,12 @@ struct cql_node *cql_parser_result(CQL_parser cp) { return cp->top; } + +/* + * Local variables: + * c-basic-offset: 4 + * c-file-style: "Stroustrup" + * indent-tabs-mode: nil + * End: + * vim: shiftwidth=4 tabstop=8 expandtab + */