-/* $Id: d1_absyn.c,v 1.16 2004-12-13 20:51:27 adam Exp $
- Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
- Index Data Aps
+/* $Id: d1_absyn.c,v 1.25 2006-05-19 13:49:33 adam Exp $
+ Copyright (C) 1995-2006
+ Index Data ApS
This file is part of the Zebra server.
#include <yaz/log.h>
#include <yaz/oid.h>
#include <idzebra/data1.h>
+#include <idzebra/recctrl.h>
#include <zebra_xpath.h>
#include <d1_absyn.h>
*/
-const char * mk_xpath_regexp (data1_handle dh, char *expr)
+static const char * mk_xpath_regexp (data1_handle dh, const char *expr)
{
- char *p = expr;
- char *pp;
- char *s;
+ const char *p = expr;
int abs = 1;
- int i;
- int j;
- int e=0;
- int is_predicate = 0;
-
- static char *stack[32];
- static char res[1024];
- char *r = "";
+ int e = 0;
+ char *stack[32];
+ char *res_p, *res = 0;
+ size_t res_size = 1;
- if (*p != '/') { return (""); }
+ if (*p != '/')
+ return ("");
p++;
- if (*p == '/') { abs=0; p++; }
-
- while (*p) {
- i=0;
- while (*p && !strchr("/",*p)) {
- i++; p++;
- }
- stack[e] = (char *) nmem_malloc (data1_nmem_get (dh), i+1);
+ if (*p == '/')
+ {
+ abs =0;
+ p++;
+ }
+ while (*p)
+ {
+ int is_predicate = 0;
+ char *s;
+ int i, j;
+ for (i = 0; *p && !strchr("/",*p); i++, p++)
+ ;
+ res_size += (i+3); /* we'll add / between later .. */
+ stack[e] = (char *) nmem_malloc(data1_nmem_get(dh), i+1);
s = stack[e];
- for (j=0; j< i; j++) {
- pp = p-i+j;
- if (*pp == '[') {
- is_predicate=1;
- }
- else if (*pp == ']') {
- is_predicate=0;
- }
- else {
- if (!is_predicate) {
- if (*pp == '*')
- *s++ = '.';
- *s++ = *pp;
+ for (j = 0; j < i; j++)
+ {
+ const char *pp = p-i+j;
+ if (*pp == '[')
+ is_predicate=1;
+ else if (*pp == ']')
+ is_predicate=0;
+ else
+ {
+ if (!is_predicate) {
+ if (*pp == '*')
+ *s++ = '.';
+ *s++ = *pp;
+ }
}
- }
}
*s = 0;
e++;
- if (*p) {p++;}
+ if (*p)
+ p++;
}
- e--; p = &res[0]; i=0;
- sprintf (p, "^"); p++;
- while (e >= 0) {
- /* !!! res size is not checked !!! */
- sprintf (p, "%s/",stack[e]);
- p += strlen(stack[e]) + 1;
- e--;
+ res_p = res = nmem_malloc(data1_nmem_get(dh), res_size + 10);
+
+ if (stack[e-1][0] == '@') /* path/@attr spec (leaf is attribute) */
+ strcpy(res_p, "/");
+ else
+ strcpy(res_p, "[^@]*/"); /* path .. (index all cdata below it) */
+ res_p = res_p + strlen(res_p);
+ while (--e >= 0) {
+ sprintf(res_p, "%s/", stack[e]);
+ res_p += strlen(stack[e]) + 1;
+ }
+ if (!abs)
+ {
+ sprintf(res_p, ".*");
+ res_p += 2;
}
- if (!abs) { sprintf (p, ".*"); p+=2; }
- sprintf (p, "$"); p++;
- r = nmem_strdup (data1_nmem_get (dh), res);
- yaz_log(YLOG_DEBUG,"Got regexp: %s",r);
- return (r);
+ sprintf (res_p, "$");
+ res_p++;
+ yaz_log(YLOG_DEBUG, "Got regexp: %s", res);
+ return res;
}
-/* *ostrich*
-
- added arg xpelement... when called from xelm context, it's 1, saying
- that ! means xpath, not element name as attribute name...
-
- pop, 2002-12-13
- */
-static int parse_termlists (data1_handle dh, data1_termlist ***tpp,
- char *cp, const char *file, int lineno,
- const char *element_name, data1_absyn *res,
- int xpelement)
+static int parse_termlists(data1_handle dh, data1_termlist ***tpp,
+ char *cp, const char *file, int lineno,
+ const char *element_name, data1_absyn *res,
+ int xpelement,
+ data1_attset *attset)
{
data1_termlist **tp = *tpp;
while(1)
nmem_malloc(data1_nmem_get(dh), sizeof(**tp));
(*tp)->next = 0;
- if (!xpelement) {
- if (*attname == '!')
+ if (*attname == '!')
+ {
+ if (!xpelement && element_name)
strcpy(attname, element_name);
- }
- if (!((*tp)->att = data1_getattbyname(dh, res->attset,
- attname))) {
- if ((!xpelement) || (*attname != '!')) {
- yaz_log(YLOG_WARN,
- "%s:%d: Couldn't find att '%s' in attset",
- file, lineno, attname);
- return -1;
- } else {
- (*tp)->att = 0;
+ else if (xpelement)
+ strcpy(attname, ZEBRA_XPATH_CDATA);
+ }
+ if (attset)
+ {
+ if (!data1_getattbyname(dh, attset, attname))
+ {
+ yaz_log(YLOG_WARN, "Index '%s' not found in attset(s)",
+ attname);
}
- }
-
+ }
+
+ (*tp)->index_name = nmem_strdup(data1_nmem_get(dh), attname);
+ assert (*(*tp)->index_name != '!');
if (r == 2 && (source = strchr(structure, ':')))
*source++ = '\0'; /* cut off structure .. */
else
return 0;
}
+/* quinn
+ * Converts a 'melm' field[$subfield] pattern to a simple xpath
+ */
+static int melm2xpath(char *melm, char *buf)
+{
+ char *dollar;
+ char *field = melm;
+ char *subfield;
+ char *fieldtype;
+ if ((dollar = strchr(melm, '$'))) {
+ *dollar = '\0';
+ subfield = ++dollar;
+ } else
+ subfield = "";
+ if (field[0] == '0' && field[1] == '0')
+ fieldtype = "controlfield";
+ else
+ fieldtype = "datafield";
+ sprintf(buf, "/*/%s[@tag=\"%s\"]", fieldtype, field);
+ if (*subfield)
+ sprintf(buf + strlen(buf), "/subfield[@code=\"%s\"]", subfield);
+ else if (field[0] != '0' || field[1] != '0')
+ strcat(buf, "/subfield");
+ yaz_log(YLOG_DEBUG, "Created xpath: '%s'", buf);
+ return 0;
+}
+
const char *data1_systag_lookup(data1_absyn *absyn, const char *tag,
const char *default_value)
{
{
data1_sub_elements *cur_elements = NULL;
data1_xpelement *cur_xpelement = NULL;
+ data1_attset *attset_list = data1_empty_attset(dh);
+ data1_attset_child **attset_childp = &attset_list->children;
data1_absyn *res = 0;
FILE *f;
data1_maptab **maptabp;
data1_marctab **marcp;
data1_termlist *all = 0;
- data1_attset_child **attset_childp;
data1_tagset **tagset_childp;
struct data1_systag **systagsp;
int level = 0;
systagsp = &res->systags;
tagset_childp = &res->tagset;
- res->attset = data1_empty_attset (dh);
- attset_childp = &res->attset->children;
-
res->varset = 0;
res->esetnames = 0;
esetpp = &res->esetnames;
res->sub_elements = NULL;
res->main_elements = NULL;
res->xp_elements = NULL;
-
+
while (f && (argc = read_absyn_line(f, &lineno, line, 512, argv, 50)))
{
char *cmd = *argv;
p = termlists;
if (*p != '-')
{
- assert (res->attset);
-
- if (parse_termlists (dh, &tp, p, file, lineno, name, res, 0))
+ if (parse_termlists (dh, &tp, p, file, lineno, name, res, 0,
+ attset_list))
{
fclose (f);
return 0;
pop, 2003-01-17
*/
- else if (!strcmp(cmd, "xelm")) {
+ else if (!strcmp(cmd, "xelm") || !strcmp(cmd, "melm")) {
int i;
char *p, *xpath_expr, *termlists;
const char *regexp;
struct DFA *dfa = dfa = dfa_init();
data1_termlist **tp;
+ char melm_xpath[128];
if (argc < 3)
{
yaz_log(YLOG_WARN, "%s:%d: Bad # of args to xelm", file, lineno);
continue;
}
- xpath_expr = argv[1];
+
+ if (!strcmp(cmd, "melm")) {
+ if (melm2xpath(argv[1], melm_xpath) < 0)
+ continue;
+ xpath_expr = melm_xpath;
+ } else {
+ xpath_expr = argv[1];
+ }
termlists = argv[2];
regexp = mk_xpath_regexp(dh, xpath_expr);
i = dfa_parse (dfa, ®exp);
p = termlists;
if (*p != '-')
{
- assert (res->attset);
-
if (parse_termlists (dh, &tp, p, file, lineno,
- xpath_expr, res, 1))
+ xpath_expr, res, 1, attset_list))
{
fclose (f);
return 0;
file, lineno);
continue;
}
- if (parse_termlists (dh, &tp, argv[1], file, lineno, 0, res, 0))
+ if (parse_termlists (dh, &tp, argv[1], file, lineno, 0, res, 0,
+ attset_list))
{
fclose (f);
return 0;
}
else if (!strcmp(cmd, "attset"))
{
- char *name;
- data1_attset *attset;
-
- if (argc != 2)
- {
- yaz_log(YLOG_WARN, "%s:%d: Bad # of args to attset",
- file, lineno);
- continue;
- }
- name = argv[1];
- if (!(attset = data1_get_attset (dh, name)))
- {
- yaz_log(YLOG_WARN, "%s:%d: Couldn't find attset %s",
- file, lineno, name);
- continue;
- }
- *attset_childp = (data1_attset_child *)
- nmem_malloc (data1_nmem_get(dh), sizeof(**attset_childp));
- (*attset_childp)->child = attset;
- (*attset_childp)->next = 0;
- attset_childp = &(*attset_childp)->next;
+ char *name;
+ data1_attset *attset;
+
+ if (argc != 2)
+ {
+ yaz_log(YLOG_WARN, "%s:%d: Bad # of args to attset",
+ file, lineno);
+ continue;
+ }
+ name = argv[1];
+ if (!(attset = data1_get_attset (dh, name)))
+ {
+ yaz_log(YLOG_WARN, "%s:%d: Couldn't find attset %s",
+ file, lineno, name);
+ continue;
+ }
+ *attset_childp = (data1_attset_child *)
+ nmem_malloc (data1_nmem_get(dh), sizeof(**attset_childp));
+ (*attset_childp)->child = attset;
+ (*attset_childp)->next = 0;
+ attset_childp = &(*attset_childp)->next;
}
else if (!strcmp(cmd, "tagset"))
{
yaz_log(YLOG_DEBUG, "%s: data1_read_absyn end", file);
return res;
}
+/*
+ * Local variables:
+ * c-basic-offset: 4
+ * indent-tabs-mode: nil
+ * End:
+ * vim: shiftwidth=4 tabstop=8 expandtab
+ */
+