X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=retrieval%2Fd1_read.c;h=d2c886328a0325d0d3311f68a5e2c29bb48a2ca2;hb=3fef09b4aaeb051928abf940049d138e5d73e20d;hp=e164202c8e83300deab789089a9e9378cba926a9;hpb=9446211daf0869d342b5738d153d810230b7a299;p=yaz-moved-to-github.git diff --git a/retrieval/d1_read.c b/retrieval/d1_read.c index e164202..d2c8863 100644 --- a/retrieval/d1_read.c +++ b/retrieval/d1_read.c @@ -1,93 +1,24 @@ /* - * Copyright (c) 1995, Index Data. + * Copyright (c) 1995-2002, Index Data. * See the file LICENSE for details. * Sebastian Hammer, Adam Dickmeiss * - * $Log: d1_read.c,v $ - * Revision 1.6 1995-12-12 16:37:08 quinn - * Added destroy element to data1_node. - * - * Revision 1.5 1995/12/11 15:22:37 quinn - * Added last_child field to the node. - * Rewrote schema-mapping. - * - * Revision 1.4 1995/11/13 09:27:36 quinn - * Fiddling with the variant stuff. - * - * Revision 1.3 1995/11/01 16:34:57 quinn - * Making data1 look for tables in data1_tabpath - * - * Revision 1.2 1995/11/01 13:54:48 quinn - * Minor adjustments - * - * Revision 1.1 1995/11/01 11:56:09 quinn - * Added Retrieval (data management) functions en masse. - * - * Revision 1.14 1995/10/30 12:40:55 quinn - * Fixed a couple of bugs. - * - * Revision 1.13 1995/10/25 16:00:47 quinn - * USMARC support is now almost operational - * - * Revision 1.12 1995/10/16 14:02:55 quinn - * Changes to support element set names and espec1 - * - * Revision 1.11 1995/10/13 16:05:08 quinn - * Adding Espec1-processing - * - * Revision 1.10 1995/10/11 14:53:44 quinn - * Work on variants. - * - * Revision 1.9 1995/10/06 16:56:50 quinn - * Fixed ranked result. - * - * Revision 1.8 1995/10/06 16:44:13 quinn - * Work on attribute set mapping, etc. - * - * Revision 1.7 1995/10/06 12:58:35 quinn - * SUTRS support - * - * Revision 1.6 1995/10/04 09:29:49 quinn - * Adjustments to support USGS test data - * - * Revision 1.5 1995/10/03 17:56:43 quinn - * Fixing GRS code. - * - * Revision 1.4 1995/10/02 15:53:19 quinn - * Work - * - * Revision 1.3 1995/10/02 14:55:21 quinn - * *** empty log message *** - * - * Revision 1.2 1995/09/14 15:18:13 quinn - * Work - * - * Revision 1.1 1995/09/12 11:24:30 quinn - * Beginning to add code for structured records. - * - * + * $Id: d1_read.c,v 1.43 2002-05-21 07:43:16 adam Exp $ */ -#include +#include #include #include -#include -#include -#include - -char *data1_tabpath = 0; /* global path for tables */ - -void data1_set_tabpath(char *p) -{ data1_tabpath = p; } - -static data1_node *freelist = 0; +#include +#include +#include /* * 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) @@ -95,362 +26,672 @@ data1_node *get_parent_tag(data1_node *n) return 0; } -data1_node *data1_mk_node(void) +data1_node *data1_mk_node (data1_handle dh, NMEM m) { - data1_node *r; + return data1_mk_node2 (dh, m, DATA1N_root, 0); +} - if ((r = freelist)) - freelist = r->next; - else - if (!(r = xmalloc(sizeof(*r)))) - abort(); - r->next = r->child = r->last_child = r->parent = 0; - r->num_children = 0; - r->destroy = 0; - return r; +data1_node *data1_mk_node_type (data1_handle dh, NMEM m, int type) +{ + return data1_mk_node2 (dh, m, type, 0); } -static void fr_node(data1_node *n) +data1_node *data1_mk_node2 (data1_handle dh, NMEM m, int type, + data1_node *parent) { - n->next = freelist; - freelist = n; + data1_node *r; + + r = (data1_node *)nmem_malloc(m, sizeof(*r)); + r->next = r->child = r->last_child = 0; + r->destroy = 0; + + if (!parent) + { + r->root = r; + } + else + { + r->root = parent->root; + r->parent = parent; + if (!parent->child) + parent->child = parent->last_child = r; + else + parent->last_child->next = r; + parent->last_child = r; + } + 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; + r->u.tag.attributes = 0; + 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; + case DATA1N_variant: + r->u.variant.type = 0; + r->u.variant.value = 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); - fr_node(t); } +data1_node *data1_mk_root (data1_handle dh, NMEM nmem, const char *name) +{ + data1_absyn *absyn = data1_get_absyn (dh, name); + data1_node *res; + if (!absyn) + { + yaz_log(LOG_WARN, "Unable to acquire abstract syntax " "for '%s'", + name); + /* It's now OK for a record not to have an absyn */ + } + res = data1_mk_node2 (dh, nmem, DATA1N_root, 0); + res->u.root.type = data1_insert_string (dh, res, nmem, name); + res->u.root.absyn = absyn; + return res; +} + +data1_node *data1_mk_tag_n (data1_handle dh, NMEM nmem, + const char *tag, size_t len, const char **attr, + data1_node *at) +{ + data1_node *partag = get_parent_tag(dh, at); + data1_node *res = data1_mk_node2 (dh, nmem, DATA1N_tag, at); + data1_element *e = NULL; + data1_xattr **p; + + res->u.tag.tag = data1_insert_string_n (dh, res, nmem, tag, len); + + if (partag) + e = partag->u.tag.element; + res->u.tag.element = + data1_getelementbytagname (dh, at->root->u.root.absyn, + e, res->u.tag.tag); + p = &res->u.tag.attributes; + while (attr && *attr) + { + *p = (data1_xattr*) nmem_malloc (nmem, sizeof(**p)); + (*p)->name = nmem_strdup (nmem, *attr++); + (*p)->value = nmem_strdup (nmem, *attr++); + p = &(*p)->next; + } + *p = 0; + return res; +} + +data1_node *data1_mk_tag (data1_handle dh, NMEM nmem, + const char *tag, const char **attr, data1_node *at) +{ + return data1_mk_tag_n (dh, nmem, tag, strlen(tag), attr, at); +} + +data1_node *data1_search_tag (data1_handle dh, data1_node *n, + const char *tag) +{ + for (; n; n = n->next) + if (n->which == DATA1N_tag && n->u.tag.tag && + !yaz_matchstr (tag, n->u.tag.tag)) + { + return n; + } + return 0; +} + +data1_node *data1_mk_tag_uni (data1_handle dh, NMEM nmem, + const char *tag, data1_node *at) +{ + data1_node *node = data1_search_tag (dh, at->child, tag); + if (!node) + node = data1_mk_tag (dh, nmem, tag, 0 /* attr */, at); + else + node->child = node->last_child = 0; + return node; +} + + +data1_node *data1_mk_text_n (data1_handle dh, NMEM mem, + const char *buf, size_t len, data1_node *parent) +{ + data1_node *res = data1_mk_node2 (dh, mem, DATA1N_data, parent); + res->u.data.what = DATA1I_text; + res->u.data.len = len; + + res->u.data.data = data1_insert_string_n (dh, res, mem, buf, len); + return res; +} + +data1_node *data1_mk_text (data1_handle dh, NMEM mem, + const char *buf, data1_node *parent) +{ + return data1_mk_text_n (dh, mem, buf, strlen(buf), parent); +} + +data1_node *data1_mk_comment_n (data1_handle dh, NMEM mem, + const char *buf, size_t len, + data1_node *parent) +{ + data1_node *res = data1_mk_node2 (dh, mem, DATA1N_comment, parent); + res->u.data.what = DATA1I_text; + res->u.data.len = len; + + res->u.data.data = data1_insert_string_n (dh, res, mem, buf, len); + return res; +} + +data1_node *data1_mk_comment (data1_handle dh, NMEM mem, + const char *buf, data1_node *parent) +{ + return data1_mk_comment_n (dh, mem, buf, strlen(buf), parent); +} + +char *data1_insert_string_n (data1_handle dh, data1_node *res, + NMEM m, const char *str, size_t len) +{ + char *b; + if (len >= DATA1_LOCALDATA) + b = nmem_malloc (m, len+1); + else + b = res->lbuf; + memcpy (b, str, len); + b[len] = 0; + return b; +} + +char *data1_insert_string (data1_handle dh, data1_node *res, + NMEM m, const char *str) +{ + return data1_insert_string_n (dh, res, m, str, strlen(str)); +} + +static data1_node *data1_add_insert_taggeddata(data1_handle dh, + data1_node *at, + const char *tagname, NMEM m, + int local_allowed) +{ + data1_node *root = at->root; + data1_node *partag = get_parent_tag (dh, at); + data1_element *e = NULL; + data1_node *datn = 0; + data1_node *tagn = 0; + + if (partag) + e = partag->u.tag.element; + e = data1_getelementbytagname (dh, root->u.root.absyn, e, tagname); + if (local_allowed || e) + { + tagn = data1_mk_node2 (dh, m, DATA1N_tag, at); + tagn->u.tag.tag = data1_insert_string (dh, tagn, m, tagname); + tagn->u.tag.element = e; + datn = data1_mk_node2 (dh, m, DATA1N_data, tagn); + } + return datn; +} + +data1_node *data1_mk_tag_data(data1_handle dh, data1_node *at, + const char *tagname, NMEM m) +{ + return data1_add_insert_taggeddata (dh, at, tagname, m, 1); +} + + /* * 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) +data1_node *data1_mk_tag_data_wd(data1_handle dh, data1_node *at, + const char *tagname, NMEM m) { - data1_node *tagn = data1_mk_node(); - 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; - if (!(tagn->u.tag.element = data1_getelementbytagname(root->u.root.absyn, - 0, tagname))) - { - fr_node(tagn); + return data1_add_insert_taggeddata (dh, at, tagname, m, 0); +} + +data1_node *data1_insert_taggeddata (data1_handle dh, data1_node *root, + data1_node *at, const char *tagname, + NMEM m) +{ + return data1_add_insert_taggeddata (dh, at, tagname, m, 0); +} + +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, at, tagname, m, 1); +} + +data1_node *data1_mk_tag_data_int (data1_handle dh, data1_node *at, + const char *tag, int num, + NMEM nmem) +{ + data1_node *node_data; + + node_data = data1_mk_tag_data (dh, at, tag, nmem); + if (!node_data) return 0; + node_data->u.data.what = DATA1I_num; + node_data->u.data.data = node_data->lbuf; + sprintf (node_data->u.data.data, "%d", num); + node_data->u.data.len = strlen (node_data->u.data.data); + return node_data; +} + +data1_node *data1_mk_tag_data_oid (data1_handle dh, data1_node *at, + const char *tag, Odr_oid *oid, + NMEM nmem) +{ + data1_node *node_data; + char str[128], *p = str; + Odr_oid *ii; + + node_data = data1_mk_tag_data (dh, at, tag, nmem); + if (!node_data) + return 0; + + for (ii = oid; *ii >= 0; ii++) + { + if (ii != oid) + *p++ = '.'; + sprintf (p, "%d", *ii); + p += strlen (p); } - tagn->child = datn = data1_mk_node(); - tagn->num_children = 1; - datn->parent = tagn; - datn->root = root; - datn->which = DATA1N_data; - tagn->next = at->child; - tagn->parent = at; - at->child = tagn; - at->num_children++; - return datn; + node_data->u.data.what = DATA1I_oid; + node_data->u.data.len = strlen (str); + node_data->u.data.data = data1_insert_string (dh, node_data, nmem, str); + return node_data; } -/* - * Ugh. Sometimes functions just grow and grow on you. This one reads a - * 'node' and its children. - */ -data1_node *data1_read_node(char **buf, data1_node *parent, int *line, - data1_absyn *absyn) + +data1_node *data1_mk_tag_data_text (data1_handle dh, data1_node *at, + const char *tag, const char *str, + NMEM nmem) { - data1_node *res; + data1_node *node_data; + + node_data = data1_mk_tag_data (dh, at, tag, nmem); + if (!node_data) + return 0; + node_data->u.data.what = DATA1I_text; + node_data->u.data.len = strlen (str); + node_data->u.data.data = data1_insert_string (dh, node_data, nmem, str); + return node_data; +} - while (**buf && isspace(**buf)) + +data1_node *data1_mk_tag_data_text_uni (data1_handle dh, data1_node *at, + const char *tag, const char *str, + NMEM nmem) +{ + data1_node *node = data1_search_tag (dh, at->child, tag); + if (!node) + return data1_mk_tag_data_text (dh, at, tag, str, nmem); + else { - if (**buf == '\n') - (*line)++; - (*buf)++; + data1_node *node_data = node->child; + node_data->u.data.what = DATA1I_text; + node_data->u.data.len = strlen (str); + node_data->u.data.data = data1_insert_string (dh, node_data, + nmem, str); + node_data->child = node_data->last_child = 0; + return node_data; } - if (!**buf) - return 0; +} - if (**buf == '<') /* beginning of tag */ + +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; +} + +/* + * 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_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 */ + { + data1_xattr *xattr; + + 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(); - 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'; + xattr = data1_read_xattr (dh, m, get_byte, fh, wrbuf, &c); + args[0] = '\0'; + if (c == '/') + { /* or */ + 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) + { + if (*tag == '\0') + --level; /* */ + else + { /* */ + 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 ? */ { - res = data1_mk_node(); - res->which = DATA1N_variant; - res->u.variant.type = 0; - res->u.variant.value = 0; - res->root = parent->root; - *t = '>'; + res = data1_mk_root (dh, m, tag); } - 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(); - res->which = DATA1N_variant; - res->root = parent->root; - res->u.variant.type = tp; - res->u.variant.value = args + val_offset; - *buf = t + 1; + if (parent->which != DATA1N_variant) + { + res = data1_mk_node2 (dh, m, DATA1N_variant, parent); + } + 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_node2 (dh, m, DATA1N_variant, parent); + 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 */ + res = data1_mk_tag (dh, m, tag, 0 /* attr */, parent); + res->u.tag.attributes = xattr; + } + 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_node2 (dh, m, DATA1N_data, parent); + res->u.data.what = DATA1I_text; + res->u.data.formatted_text = 0; + 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(); - 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->root = parent->root; - *buf = t + 1; - } + len = wrbuf_len(wrbuf); - res->parent = parent; - res->num_children = 0; - - pp = &res->child; - /* - * Read child nodes. - */ - while ((*pp = data1_read_node(buf, res, line, absyn))) - { - 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 - - /* Determine length and remove newlines/extra blanks */ - while (**buf && **buf != '<') - { - if (**buf == '\n') - (*line)++; - if (isspace(**buf)) - { - *(pp++) = ' '; - (*buf)++; - while (isspace(**buf)) - (*buf)++; - } + /* use local buffer of nmem if too large */ + if (len >= DATA1_LOCALDATA) + res->u.data.data = (char*) nmem_malloc (m, len); else - *(pp++) = *((*buf)++); - len++; - } - while (isspace(data[len-1])) - len--; - res = data1_mk_node(); - res->parent = parent; - res->which = DATA1N_data; - res->u.data.what = DATA1I_text; - res->u.data.len = len; - res->u.data.data = data; - res->root = parent->root; - - /* - * if the parent is structured, we'll insert a 'wellKnown' marker - * in front of the data. - */ -#if 0 - if (partag->u.tag.element && partag->u.tag.element->tag->kind == - DATA1K_structured) - { - data1_node *wk = mk_node(); - static data1_element wk_element = { 0, 0, 0, 0, 0}; - - wk->parent = partag; - wk->root = partag->root; - wk->which = DATA1N_tag; - wk->u.tag.tag = 0; - /* - * get well-known tagdef if required. - */ - if (!wk_element.tag && !(wk_element.tag = - data1_gettagbynum(wk->root->u.root.absyn->tagset, 1, 19))) + 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++) + { + if (*src == '\n') + line++; + if (d1_isspace (*src)) + prev_char = ' '; + else { - logf(LOG_WARN, - "Failed to initialize 'wellknown' tag from tagsetM"); - return 0; + if (prev_char) + { + *dst++ = prev_char; + prev_char = 0; + } + *dst++ = *src; } - wk->u.tag.element = &wk_element; - wk->child = partag->child; - if (wk->child) - wk->child->parent = wk; - partag->child = wk; + } + res->u.data.len = dst - res->u.data.data; } -#endif } - 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)(int, char *, size_t), int fd) +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)(fd, buf + rd, 4096)) <= 0) + if ((res = (*rf)(fh, *buf + rd, 2048)) <= 0) { if (!res) { - bp = buf; - return data1_read_node(&bp, 0, &line, 0); + bp = *buf; + (*buf)[rd] = '\0'; + return data1_read_node(dh, &bp, m); } else return 0; @@ -458,3 +699,10 @@ data1_node *data1_read_record(int (*rf)(int, char *, size_t), int fd) 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); +} +