/*
- * Copyright (c) 1995, Index Data.
+ * Copyright (c) 1995-2001, Index Data.
* See the file LICENSE for details.
* Sebastian Hammer, Adam Dickmeiss
*
* $Log: d1_read.c,v $
- * Revision 1.11 1996-07-06 19:58:35 quinn
+ * Revision 1.38 2001-03-27 23:06:21 adam
+ * Quotes and slashes may occur within attributes.
+ *
+ * Revision 1.37 2001/02/28 09:00:06 adam
+ * Fixed problem with stack overflow for very nested records.
+ *
+ * Revision 1.36 2001/02/21 13:46:53 adam
+ * C++ fixes.
+ *
+ * Revision 1.35 2000/12/05 14:44:25 adam
+ * Readers skips <! ...> sections.
+ *
+ * Revision 1.34 2000/12/05 10:06:23 adam
+ * Added support for null-data rules like <tag/>.
+ *
+ * Revision 1.33 2000/11/29 14:22:47 adam
+ * Implemented XML/SGML attributes for data1 so that d1_read reads them
+ * and d1_write generates proper attributes for XML/SGML records. Added
+ * register locking for threaded version.
+ *
+ * Revision 1.32 2000/01/06 11:25:59 adam
+ * Added case to prevent warning.
+ *
+ * Revision 1.31 1999/12/21 14:16:20 ian
+ * Changed retrieval module to allow data1 trees with no associated absyn.
+ * Also added a simple interface for extracting values from data1 trees using
+ * a string based tagpath.
+ *
+ * Revision 1.30 1999/11/30 13:47:12 adam
+ * Improved installation. Moved header files to include/yaz.
+ *
+ * Revision 1.29 1999/10/21 12:06:29 adam
+ * Retrieval module no longer uses ctype.h - functions.
+ *
+ * Revision 1.28 1999/10/21 09:50:33 adam
+ * SGML reader uses own isspace - it doesn't do 8-bit on WIN32!
+ *
+ * Revision 1.27 1999/08/27 09:40:32 adam
+ * Renamed logf function to yaz_log. Removed VC++ project files.
+ *
+ * Revision 1.26 1999/07/13 13:23:48 adam
+ * Non-recursive version of data1_read_node. data1_read_nodex reads
+ * stream of bytes (instead of buffer in memory).
+ *
+ * Revision 1.25 1999/04/20 09:56:48 adam
+ * Added 'name' paramter to encoder/decoder routines (typedef Odr_fun).
+ * Modified all encoders/decoders to reflect this change.
+ *
+ * Revision 1.24 1998/10/28 15:10:09 adam
+ * Added --with-yc option to configure. For the data1_node in data1.h:
+ * decreased size of localdata and removed member "line" which wasn't useful.
+ *
+ * Revision 1.23 1998/03/12 11:28:45 adam
+ * Fix: didn't set root member of tagged node in function.
+ * data1_add_insert_taggeddata.
+ *
+ * Revision 1.22 1998/03/05 08:15:32 adam
+ * Implemented data1_add_insert_taggeddata utility which is more flexible
+ * than data1_insert_taggeddata.
+ *
+ * Revision 1.21 1998/02/27 14:08:05 adam
+ * Added const to some char pointer arguments.
+ * Reworked data1_read_node so that it doesn't create a tree with
+ * pointers to original "SGML"-buffer.
+ *
+ * Revision 1.20 1998/02/11 11:53:35 adam
+ * Changed code so that it compiles as C++.
+ *
+ * Revision 1.19 1997/12/09 16:17:09 adam
+ * Fix bug regarding variants. Tags with prefix "var" was incorrectly
+ * interpreted as "start of variants". Now, only "var" indicates such
+ * start.
+ * Cleaned up data1_read_node so tag names and variant names are
+ * copied and not pointed to by the generated data1 tree. Data nodes
+ * still point to old buffer.
+ *
+ * Revision 1.18 1997/11/18 09:51:09 adam
+ * Removed element num_children from data1_node. Minor changes in
+ * data1 to Explain.
+ *
+ * Revision 1.17 1997/11/05 09:20:51 adam
+ * Minor change.
+ *
+ * Revision 1.16 1997/09/17 12:10:37 adam
+ * YAZ version 1.4.
+ *
+ * Revision 1.15 1997/09/05 09:50:57 adam
+ * Removed global data1_tabpath - uses data1_get_tabpath() instead.
+ *
+ * Revision 1.14 1997/05/14 06:54:04 adam
+ * C++ support.
+ *
+ * Revision 1.13 1996/10/29 13:35:38 adam
+ * Implemented data1_set_tabpath and data1_get_tabpath.
+ *
+ * Revision 1.12 1996/10/11 10:35:38 adam
+ * Fixed a bug that caused data1_read_node to core dump when no abstract
+ * syntax was defined in a "sgml"-record.
+ *
+ * Revision 1.11 1996/07/06 19:58:35 quinn
* System headerfiles gathered in yconfig
*
* Revision 1.10 1996/01/19 15:41:47 quinn
*
*/
-#include <ctype.h>
+#include <assert.h>
#include <stdio.h>
#include <stdlib.h>
-#include <xmalloc.h>
-#include <log.h>
-#include <data1.h>
-
-char *data1_tabpath = 0; /* global path for tables */
-
-void data1_set_tabpath(char *p)
-{ data1_tabpath = p; }
-
-#if 0
-static data1_node *freelist = 0;
-#endif
+#include <yaz/xmalloc.h>
+#include <yaz/log.h>
+#include <yaz/data1.h>
/*
* get the tag which is the immediate parent of this node (this may mean
* traversing intermediate things like variants and stuff.
*/
-data1_node *get_parent_tag(data1_node *n)
+data1_node *get_parent_tag (data1_handle dh, data1_node *n)
{
for (; n && n->which != DATA1N_root; n = n->parent)
if (n->which == DATA1N_tag)
return 0;
}
-data1_node *data1_mk_node(NMEM m)
+data1_node *data1_mk_node (data1_handle dh, NMEM m)
{
data1_node *r;
-#if 0
- if ((r = freelist))
- freelist = r->next;
- else
- if (!(r = xmalloc(sizeof(*r))))
- abort();
-#else
- r = nmem_malloc(m, sizeof(*r));
-#endif
+ r = (data1_node *)nmem_malloc(m, sizeof(*r));
r->next = r->child = r->last_child = r->parent = 0;
- r->num_children = 0;
r->destroy = 0;
return r;
}
-#if 0
-static void fr_node(data1_node *n)
+data1_node *data1_mk_node_type (data1_handle dh, NMEM m, int type)
{
- n->next = freelist;
- freelist = n;
-}
+ data1_node *r;
+
+ r = data1_mk_node(dh, m);
+ r->which = type;
+ switch(type)
+ {
+ case DATA1N_tag:
+ r->u.tag.tag = 0;
+ r->u.tag.element = 0;
+ r->u.tag.no_data_requested = 0;
+ r->u.tag.node_selected = 0;
+ r->u.tag.make_variantlist = 0;
+ r->u.tag.get_bytes = -1;
+#if DATA1_USING_XATTR
+ r->u.tag.attributes = 0;
#endif
+ break;
+ case DATA1N_root:
+ r->u.root.type = 0;
+ r->u.root.absyn = 0;
+ break;
+ case DATA1N_data:
+ r->u.data.data = 0;
+ r->u.data.len = 0;
+ r->u.data.what = 0;
+ r->u.data.formatted_text = 0;
+ break;
+ default:
+ logf (LOG_WARN, "data_mk_node_type. bad type = %d\n", type);
+ }
+ return r;
+}
-void data1_free_tree(data1_node *t)
+void data1_free_tree (data1_handle dh, data1_node *t)
{
data1_node *p = t->child, *pn;
while (p)
{
pn = p->next;
- data1_free_tree(p);
+ data1_free_tree (dh, p);
p = pn;
}
if (t->destroy)
(*t->destroy)(t);
-#if 0
- fr_node(t);
-#endif
}
-/*
- * Insert a tagged node into the record root as first child of the node at
- * which should be root or tag itself). Returns pointer to the data node,
- * which can then be modified.
- */
-data1_node *data1_insert_taggeddata(data1_node *root, data1_node *at,
- char *tagname, NMEM m)
+char *data1_insert_string (data1_handle dh, data1_node *res,
+ NMEM m, const char *str)
+{
+ int len = strlen(str);
+
+ if (len >= DATA1_LOCALDATA)
+ return nmem_strdup (m, str);
+ else
+ {
+ strcpy (res->lbuf, str);
+ return res->lbuf;
+ }
+}
+
+data1_node *data1_add_insert_taggeddata(data1_handle dh, data1_node *root,
+ data1_node *at,
+ const char *tagname, NMEM m,
+ int first_flag, int local_allowed)
{
- data1_node *tagn = data1_mk_node(m);
+ data1_node *partag = get_parent_tag (dh, at);
+ data1_node *tagn = data1_mk_node_type (dh, m, DATA1N_tag);
+ data1_element *e = NULL;
data1_node *datn;
- tagn->which = DATA1N_tag;
- tagn->line = -1;
- tagn->u.tag.tag = 0;
- tagn->u.tag.node_selected = 0;
- tagn->u.tag.make_variantlist = 0;
- tagn->u.tag.no_data_requested = 0;
- tagn->u.tag.get_bytes = -1;
- if (!(tagn->u.tag.element = data1_getelementbytagname(root->u.root.absyn,
- 0, tagname)))
- return 0;
- tagn->child = datn = data1_mk_node(m);
- tagn->num_children = 1;
+ tagn->u.tag.tag = data1_insert_string (dh, tagn, m, tagname);
+
+ if (partag)
+ e = partag->u.tag.element;
+ tagn->u.tag.element =
+ data1_getelementbytagname (dh, root->u.root.absyn, e, tagname);
+ if (!local_allowed && !tagn->u.tag.element)
+ return NULL;
+ tagn->last_child = tagn->child = datn = data1_mk_node_type (dh, m, DATA1N_data);
+ tagn->root = root;
datn->parent = tagn;
datn->root = root;
- datn->which = DATA1N_data;
datn->u.data.formatted_text = 0;
- tagn->next = at->child;
tagn->parent = at;
- at->child = tagn;
- at->num_children++;
+
+ if (first_flag)
+ {
+ tagn->next = at->child;
+ if (!tagn->next)
+ at->last_child = tagn;
+ at->child = tagn;
+ }
+ else
+ {
+ if (!at->child)
+ at->child = tagn;
+ else
+ {
+ assert (at->last_child);
+ at->last_child->next = tagn;
+ }
+ at->last_child = tagn;
+ }
return datn;
}
+data1_node *data1_add_taggeddata(data1_handle dh, data1_node *root,
+ data1_node *at,
+ const char *tagname, NMEM m)
+{
+ return data1_add_insert_taggeddata (dh, root, at, tagname, m, 0, 1);
+}
+
+
/*
- * Ugh. Sometimes functions just grow and grow on you. This one reads a
- * 'node' and its children.
+ * Insert a tagged node into the record root as first child of the node at
+ * which should be root or tag itself). Returns pointer to the data node,
+ * which can then be modified.
*/
-data1_node *data1_read_node(char **buf, data1_node *parent, int *line,
- data1_absyn *absyn, NMEM m)
+data1_node *data1_insert_taggeddata(data1_handle dh, data1_node *root,
+ data1_node *at,
+ const char *tagname, NMEM m)
{
- data1_node *res;
-
- while (**buf && isspace(**buf))
- {
- if (**buf == '\n')
- (*line)++;
- (*buf)++;
- }
- if (!**buf)
- return 0;
+ return data1_add_insert_taggeddata (dh, root, at, tagname, m, 1, 0);
+}
- if (**buf == '<') /* beginning of tag */
+#if DATA1_USING_XATTR
+data1_xattr *data1_read_xattr (data1_handle dh, NMEM m,
+ int (*get_byte)(void *fh), void *fh,
+ WRBUF wrbuf, int *ch)
+{
+ data1_xattr *p_first = 0;
+ data1_xattr **pp = &p_first;
+ int c = *ch;
+ for (;;)
{
- char *tag = (*buf) + 1;
- char *args = 0;
- char *t = tag;
- data1_node **pp;
- data1_element *elem = 0;
-
- for (; *t && *t != '>' && !isspace(*t); t++);
- if (*t != '>' && !isspace(*t))
+ data1_xattr *p;
+ int len;
+ while (c && d1_isspace(c))
+ c = (*get_byte)(fh);
+ if (!c || c == '>' || c == '/')
+ break;
+ *pp = p = (data1_xattr *) nmem_malloc (m, sizeof(*p));
+ p->next = 0;
+ pp = &p->next;
+ p->value = 0;
+
+ wrbuf_rewind(wrbuf);
+ while (c && c != '=' && c != '>' && c != '/' && !d1_isspace(c))
{
- logf(LOG_WARN, "d1: %d: Malformed tag", *line);
- return 0;
+ wrbuf_putc (wrbuf, c);
+ c = (*get_byte)(fh);
}
- if (isspace(*t)) /* the tag has arguments */
+ wrbuf_putc (wrbuf, '\0');
+ len = wrbuf_len(wrbuf);
+ p->name = (char*) nmem_malloc (m, len);
+ strcpy (p->name, wrbuf_buf(wrbuf));
+ if (c == '=')
{
- while (isspace(*t))
- t++;
- if (*t != '>')
+ c = (*get_byte)(fh);
+ if (c == '"')
{
- args = t;
- for (; *t && *t != '>'; t++);
- if (*t != '>' && !isspace(*t))
+ c = (*get_byte)(fh);
+ wrbuf_rewind(wrbuf);
+ while (c && c != '"')
{
- logf(LOG_WARN, "d1: %d: Malformed tag", *line);
- return 0;
- }
+ wrbuf_putc (wrbuf, c);
+ c = (*get_byte)(fh);
+ }
+ if (c)
+ c = (*get_byte)(fh);
}
+ else
+ {
+ wrbuf_rewind(wrbuf);
+ while (c && c != '>' && c != '/')
+ {
+ wrbuf_putc (wrbuf, c);
+ c = (*get_byte)(fh);
+ }
+ }
+ wrbuf_putc (wrbuf, '\0');
+ len = wrbuf_len(wrbuf);
+ p->value = (char*) nmem_malloc (m, len);
+ strcpy (p->value, wrbuf_buf(wrbuf));
}
+ }
+ *ch = c;
+ return p_first;
+}
+#endif
+
+/*
+ * Ugh. Sometimes functions just grow and grow on you. This one reads a
+ * 'node' and its children.
+ */
+data1_node *data1_read_nodex (data1_handle dh, NMEM m,
+ int (*get_byte)(void *fh), void *fh, WRBUF wrbuf)
+{
+ data1_absyn *absyn = 0;
+ data1_node *d1_stack[256];
+ data1_node *res;
+ int c;
+ int level = 0;
+ int line = 1;
- /*
- * if end-tag, see if we terminate parent. If so, consume and return.
- * Else, return.
- */
- *t = '\0';
- if (*tag == '/')
+ d1_stack[level] = 0;
+ c = (*get_byte)(fh);
+ while (1)
+ {
+ data1_node *parent = level ? d1_stack[level-1] : 0;
+ while (c != '\0' && d1_isspace(c))
{
- if (!parent)
- return 0;
- if (!*(tag +1) || (parent->which == DATA1N_root && !strcmp(tag + 1,
- parent->u.root.type)) ||
- (parent->which == DATA1N_tag && !strcmp(tag + 1,
- parent->u.tag.tag)))
+ if (c == '\n')
+ line++;
+ c = (*get_byte)(fh);
+ }
+ if (c == '\0')
+ break;
+
+ if (c == '<') /* beginning of tag */
+ {
+#if DATA1_USING_XATTR
+ data1_xattr *xattr;
+#endif
+ char tag[64];
+ char args[256];
+ int null_tag = 0;
+ int end_tag = 0;
+ size_t i = 0;
+
+ c = (*get_byte)(fh);
+ if (c == '/')
{
- *buf = t + 1;
- return 0;
+ end_tag = 1;
+ c = (*get_byte)(fh);
}
- else
+ else if (c == '!') /* tags/comments that we don't deal with yet */
{
- *t = '>';
- return 0;
+ while (c && c != '>')
+ c = (*get_byte)(fh);
+ if (c)
+ c = (*get_byte)(fh);
+ continue;
}
- }
-
- if (!absyn) /* parent node - what are we? */
- {
- if (!(absyn = data1_get_absyn(tag)))
+ while (c && c != '>' && c != '/' && !d1_isspace(c))
{
- logf(LOG_WARN, "Unable to acquire abstract syntax for '%s'",
- tag);
- return 0;
+ if (i < (sizeof(tag)-1))
+ tag[i++] = c;
+ c = (*get_byte)(fh);
}
- res = data1_mk_node(m);
- res->which = DATA1N_root;
- res->u.root.type = tag;
- res->u.root.absyn = absyn;
- res->root = res;
- *buf = t + 1;
- }
- else if (!strncmp(tag, "var", 3))
- {
- char class[DATA1_MAX_SYMBOL], type[DATA1_MAX_SYMBOL];
- data1_vartype *tp;
- int val_offset;
- data1_node *p;
-
- if (sscanf(args, "%s %s %n", class, type, &val_offset) != 2)
+ tag[i] = '\0';
+#if DATA1_USING_XATTR
+ xattr = data1_read_xattr (dh, m, get_byte, fh, wrbuf, &c);
+ args[0] = '\0';
+#else
+ while (d1_isspace(c))
+ c = (*get_byte)(fh);
+ for (i = 0; c && c != '>' && c != '/'; c = (*get_byte)(fh))
+ if (i < (sizeof(args)-1))
+ args[i++] = c;
+ args[i] = '\0';
+#endif
+ if (c == '/')
+ { /* <tag attrs/> or <tag/> */
+ null_tag = 1;
+ c = (*get_byte)(fh);
+ }
+ if (c != '>')
{
- logf(LOG_WARN, "Malformed variant triple at '%s'", tag);
+ yaz_log(LOG_WARN, "d1: %d: Malformed tag", line);
return 0;
}
- if (!(tp = data1_getvartypebyct(parent->root->u.root.absyn->varset,
- class, type)))
- return 0;
-
- /*
- * If we're the first variant in this group, create a parent var,
- * and insert it before the current variant.
- */
- if (parent->which != DATA1N_variant)
+ else
+ c = (*get_byte)(fh);
+
+ /* End tag? */
+ if (end_tag)
{
- res = data1_mk_node(m);
- res->which = DATA1N_variant;
- res->u.variant.type = 0;
- res->u.variant.value = 0;
- res->root = parent->root;
- *t = '>';
+ if (*tag == '\0')
+ --level; /* </> */
+ else
+ { /* </tag> */
+ int i = level;
+ while (i > 0)
+ {
+ parent = d1_stack[--i];
+ if ((parent->which == DATA1N_root &&
+ !strcmp(tag, parent->u.root.type)) ||
+ (parent->which == DATA1N_tag &&
+ !strcmp(tag, parent->u.tag.tag)))
+ {
+ level = i;
+ break;
+ }
+ }
+ if (i != level)
+ {
+ yaz_log (LOG_WARN, "%d: no begin tag for %s",
+ line, tag);
+ break;
+ }
+ }
+ if (level == 0)
+ return d1_stack[0];
+ continue;
+ }
+ if (level == 0) /* root ? */
+ {
+ if (!(absyn = data1_get_absyn (dh, tag)))
+ {
+ yaz_log(LOG_WARN, "Unable to acquire abstract syntax " "for '%s'", tag);
+ /* It's now OK for a record not to have an absyn */
+ }
+ res = data1_mk_node_type (dh, m, DATA1N_root);
+ res->u.root.type = data1_insert_string (dh, res, m, tag);
+ res->u.root.absyn = absyn;
+ res->root = res;
}
- else
+ else if (!strcmp(tag, "var"))
{
+ char tclass[DATA1_MAX_SYMBOL], type[DATA1_MAX_SYMBOL];
+ data1_vartype *tp;
+ int val_offset;
+
+ if (sscanf(args, "%s %s %n", tclass, type, &val_offset) != 2)
+ {
+ yaz_log(LOG_WARN, "Malformed variant triple at '%s'", tag);
+ continue;
+ }
+ if (!(tp =
+ data1_getvartypebyct(dh,
+ parent->root->u.root.absyn->varset,
+ tclass, type)))
+ continue;
/*
- * now determine if one of our ancestor triples is of same type.
- * If so, we break here. This will make the parser unwind until
- * we become a sibling (alternate variant) to the aforementioned
- * triple. It stinks that we re-parse these tags on every
- * iteration of this. This is a function in need of a rewrite.
+ * If we're the first variant in this group, create a parent
+ * variant, and insert it before the current variant.
*/
- for (p = parent; p->which == DATA1N_variant; p = p->parent)
- if (p->u.variant.type == tp)
- {
- *t = '>';
- return 0;
- }
-
- res = data1_mk_node(m);
- res->which = DATA1N_variant;
+ if (parent->which != DATA1N_variant)
+ {
+ res = data1_mk_node (dh, m);
+ res->which = DATA1N_variant;
+ res->u.variant.type = 0;
+ res->u.variant.value = 0;
+ }
+ else
+ {
+ /*
+ * now determine if one of our ancestor triples is of
+ * same type. If so, we break here.
+ */
+ int i;
+ for (i = level-1; d1_stack[i]->which==DATA1N_variant; --i)
+ if (d1_stack[i]->u.variant.type == tp)
+ {
+ level = i;
+ break;
+ }
+ res = data1_mk_node (dh, m);
+ res->which = DATA1N_variant;
+ res->u.variant.type = tp;
+ res->u.variant.value =
+ data1_insert_string (dh, res, m, args + val_offset);
+ }
+ }
+ else /* tag.. acquire our element in the abstract syntax */
+ {
+ data1_node *partag = get_parent_tag (dh, parent);
+ data1_element *elem, *e = 0;
+ int localtag = 0;
+
+ if (parent->which == DATA1N_variant)
+ return 0;
+ if (partag)
+ if (!(e = partag->u.tag.element))
+ localtag = 1; /* our parent is a local tag */
+
+ elem = data1_getelementbytagname(dh, absyn, e, tag);
+ res = data1_mk_node_type (dh, m, DATA1N_tag);
+ res->u.tag.tag = data1_insert_string (dh, res, m, tag);
+ res->u.tag.element = elem;
+#if DATA1_USING_XATTR
+ res->u.tag.attributes = xattr;
+#endif
+ }
+ if (parent)
+ {
+ parent->last_child = res;
res->root = parent->root;
- res->u.variant.type = tp;
- res->u.variant.value = args + val_offset;
- *buf = t + 1;
}
+ res->parent = parent;
+ if (d1_stack[level])
+ d1_stack[level]->next = res;
+ else if (parent)
+ parent->child = res;
+ d1_stack[level] = res;
+ d1_stack[level+1] = 0;
+ if (level < 250 && !null_tag)
+ ++level;
}
- else /* tag.. acquire our element in the abstract syntax */
+ else /* != '<'... this is a body of text */
{
- data1_node *partag = get_parent_tag(parent);
- data1_element *e = 0;
- int localtag = 0;
-
- if (parent->which == DATA1N_variant)
+ const char *src;
+ char *dst;
+ int len, prev_char = 0;
+
+ if (level == 0)
{
- *t = '>';
- return 0;
+ c = (*get_byte)(fh);
+ continue;
}
- if (partag)
- if (!(e = partag->u.tag.element))
- localtag = 1; /* our parent is a local tag */
+ res = data1_mk_node_type (dh, m, DATA1N_data);
+ res->parent = parent;
+ res->u.data.what = DATA1I_text;
+ res->u.data.formatted_text = 0;
+ res->root = parent->root;
+ parent->last_child = res;
+ if (d1_stack[level])
+ d1_stack[level]->next = res;
+ else
+ parent->child = res;
+ d1_stack[level] = res;
+
+ wrbuf_rewind(wrbuf);
-#if 0
- if (!localtag && !(elem = data1_getelementbytagname(absyn,
- e, tag)) && (data1_gettagbyname(absyn->tagset, tag)))
+ while (c && c != '<')
{
- if (parent->which == DATA1N_root)
- logf(LOG_WARN, "Tag '%s' used out of context", tag);
- *t = '>';
- return 0;
+ wrbuf_putc (wrbuf, c);
+ c = (*get_byte)(fh);
}
-#else
- elem = data1_getelementbytagname(absyn, e, tag);
-#endif
- res = data1_mk_node(m);
- res->which = DATA1N_tag;
- res->u.tag.element = elem;
- res->u.tag.tag = tag;
- res->u.tag.node_selected = 0;
- res->u.tag.make_variantlist = 0;
- res->u.tag.no_data_requested = 0;
- res->u.tag.get_bytes = -1;
- res->root = parent->root;
- *buf = t + 1;
- }
-
- res->parent = parent;
- res->num_children = 0;
-
- pp = &res->child;
- /*
- * Read child nodes.
- */
- while ((*pp = data1_read_node(buf, res, line, absyn, m)))
- {
- res->last_child = *pp;
- res->num_children++;
- pp = &(*pp)->next;
- }
- }
- else /* != '<'... this is a body of text */
- {
- int len = 0;
- char *data = *buf, *pp = *buf;
-#if 0
- data1_node *partag = get_parent_tag(parent);
-#endif
+ len = wrbuf_len(wrbuf);
- /* Determine length and remove newlines/extra blanks */
- while (**buf && **buf != '<')
- {
- if (**buf == '\n')
- (*line)++;
- if (isspace(**buf))
+ /* use local buffer of nmem if too large */
+ if (len >= DATA1_LOCALDATA)
+ res->u.data.data = (char*) nmem_malloc (m, len);
+ else
+ res->u.data.data = res->lbuf;
+
+ /* read "data" and transfer while removing white space */
+ dst = res->u.data.data;
+ for (src = wrbuf_buf(wrbuf); --len >= 0; src++)
{
- *(pp++) = ' ';
- (*buf)++;
- while (isspace(**buf))
- (*buf)++;
+ if (*src == '\n')
+ line++;
+ if (d1_isspace (*src))
+ prev_char = ' ';
+ else
+ {
+ if (prev_char)
+ {
+ *dst++ = prev_char;
+ prev_char = 0;
+ }
+ *dst++ = *src;
+ }
}
- else
- *(pp++) = *((*buf)++);
- len++;
+ res->u.data.len = dst - res->u.data.data;
}
- while (isspace(data[len-1]))
- len--;
- res = data1_mk_node(m);
- res->parent = parent;
- res->which = DATA1N_data;
- res->u.data.what = DATA1I_text;
- res->u.data.len = len;
- res->u.data.data = data;
- res->u.data.formatted_text = 0;
- res->root = parent->root;
}
- return res;
+ return 0;
+}
+
+int getc_mem (void *fh)
+{
+ const char **p = (const char **) fh;
+ if (**p)
+ return *(*p)++;
+ return 0;
+}
+
+data1_node *data1_read_node (data1_handle dh, const char **buf, NMEM m)
+{
+ WRBUF wrbuf = wrbuf_alloc();
+ data1_node *node;
+
+ node = data1_read_nodex(dh, m, getc_mem, (void *) (buf), wrbuf);
+ wrbuf_free (wrbuf, 1);
+ return node;
}
/*
* Read a record in the native syntax.
*/
-data1_node *data1_read_record(int (*rf)(void *, char *, size_t), void *fh,
+data1_node *data1_read_record(data1_handle dh,
+ int (*rf)(void *, char *, size_t), void *fh,
NMEM m)
{
- static char *buf = 0;
- char *bp;
- static int size;
+ int *size;
+ char **buf = data1_get_read_buf (dh, &size);
+ const char *bp;
int rd = 0, res;
- int line = 0;
-
- if (!buf && !(buf = xmalloc(size = 4096)))
- abort();
-
+
+ if (!*buf)
+ *buf = (char *)xmalloc(*size = 4096);
+
for (;;)
{
- if (rd + 4096 > size && !(buf =xrealloc(buf, size *= 2)))
+ if (rd + 2048 >= *size && !(*buf =(char *)xrealloc(*buf, *size *= 2)))
abort();
- if ((res = (*rf)(fh, buf + rd, 4096)) <= 0)
+ if ((res = (*rf)(fh, *buf + rd, 2048)) <= 0)
{
if (!res)
{
- bp = buf;
- return data1_read_node(&bp, 0, &line, 0, m);
+ bp = *buf;
+ (*buf)[rd] = '\0';
+ return data1_read_node(dh, &bp, m);
}
else
return 0;
rd += res;
}
}
+
+data1_node *data1_read_sgml (data1_handle dh, NMEM m, const char *buf)
+{
+ const char *bp = buf;
+ return data1_read_node (dh, &bp, m);
+}
+
projects / yaz-moved-to-github.git / blobdiff