/* reads a PIE feed URL and returns a new channel structure (even if
the feed could not be read) */
static void pie_parse(feedParserCtxtPtr ctxt, xmlNodePtr cur) {
gchar *tmp2, *tmp = NULL, *tmp3;
NsHandler *nsh;
parseChannelTagFunc pf;
while(TRUE) {
if(xmlStrcmp(cur->name, BAD_CAST"feed")) {
g_string_append(ctxt->feed->parseErrors, "<p>Could not find Atom/Echo/PIE header!</p>");
break;
}
/* parse feed contents */
cur = cur->xmlChildrenNode;
while(cur) {
if(!cur->name || cur->type != XML_ELEMENT_NODE) {
cur = cur->next;
continue;
}
/* check namespace of this tag */
if(cur->ns) {
if((cur->ns->href && (nsh = (NsHandler *)g_hash_table_lookup(ns_pie_ns_uri_table, (gpointer)cur->ns->href))) ||
(cur->ns->prefix && (nsh = (NsHandler *)g_hash_table_lookup(pie_nstable, (gpointer)cur->ns->prefix)))) {
pf = nsh->parseChannelTag;
if(pf)
(*pf)(ctxt, cur);
cur = cur->next;
continue;
} else {
/*g_print("unsupported namespace \"%s\"\n", cur->ns->prefix);*/
}
} /* explicitly no following else !!! */
if(!xmlStrcmp(cur->name, BAD_CAST"title")) {
tmp = unhtmlize(pie_parse_content_construct(cur));
if(tmp) {
if(ctxt->title)
g_free(ctxt->title);
ctxt->title = tmp;
}
} else if(!xmlStrcmp(cur->name, BAD_CAST"link")) {
tmp = xml_get_attribute (cur, "href");
if(tmp) {
/* 0.3 link : rel, type and href attribute */
tmp2 = xml_get_attribute (cur, "rel");
if(tmp2 && g_str_equal(tmp2, "alternate"))
subscription_set_homepage (ctxt->subscription, tmp);
/* else
FIXME: Maybe do something with other links? */
g_free(tmp2);
g_free(tmp);
} else {
/* 0.2 link : element content is the link, or non-alternate link in 0.3 */
tmp = (gchar *)xmlNodeListGetString(ctxt->doc, cur->xmlChildrenNode, 1);
if(tmp) {
subscription_set_homepage (ctxt->subscription, tmp);
g_free(tmp);
}
}
/* parse feed author */
} else if(!xmlStrcmp(cur->name, BAD_CAST"author")) {
/* parse feed author */
tmp = parseAuthor(cur);
if(tmp) {
ctxt->subscription->metadata = metadata_list_append(ctxt->subscription->metadata, "author", tmp);
g_free(tmp);
}
} else if (!xmlStrcmp (cur->name, BAD_CAST"tagline")) {
tmp = pie_parse_content_construct (cur);
if (tmp) {
metadata_list_set (&ctxt->subscription->metadata, "description", tmp);
g_free (tmp);
}
} else if(!xmlStrcmp(cur->name, BAD_CAST"generator")) {
tmp = unhtmlize((gchar *)xmlNodeListGetString(ctxt->doc, cur->xmlChildrenNode, 1));
if(tmp && tmp[0] != '\0') {
tmp2 = xml_get_attribute (cur, "version");
if(tmp2) {
tmp3 = g_strdup_printf("%s %s", tmp, tmp2);
g_free(tmp);
g_free(tmp2);
tmp = tmp3;
}
tmp2 = xml_get_attribute (cur, "url");
if(tmp2) {
tmp3 = g_strdup_printf("<a href=\"%s\">%s</a>", tmp2, tmp);
g_free(tmp2);
g_free(tmp);
tmp = tmp3;
}
ctxt->subscription->metadata = metadata_list_append(ctxt->subscription->metadata, "feedgenerator", tmp);
}
g_free(tmp);
} else if(!xmlStrcmp(cur->name, BAD_CAST"copyright")) {
tmp = pie_parse_content_construct(cur);
if(tmp) {
ctxt->subscription->metadata = metadata_list_append(ctxt->subscription->metadata, "copyright", tmp);
g_free(tmp);
}
} else if(!xmlStrcmp(cur->name, BAD_CAST"modified")) { /* Modified was last used in IETF draft 02) */
tmp = (gchar *)xmlNodeListGetString(ctxt->doc, cur->xmlChildrenNode, 1);
if(tmp) {
ctxt->subscription->metadata = metadata_list_append(ctxt->subscription->metadata, "pubDate", tmp);
ctxt->feed->time = date_parse_ISO8601 (tmp);
g_free(tmp);
}
} else if(!xmlStrcmp(cur->name, BAD_CAST"updated")) { /* Updated was added in IETF draft 03 */
tmp = (gchar *)xmlNodeListGetString(ctxt->doc, cur->xmlChildrenNode, 1);
if(tmp) {
ctxt->subscription->metadata = metadata_list_append(ctxt->subscription->metadata, "pubDate", tmp);
ctxt->feed->time = date_parse_ISO8601 (tmp);
g_free(tmp);
}
} else if(!xmlStrcmp(cur->name, BAD_CAST"contributor")) {
tmp = parseAuthor(cur);
if(tmp) {
ctxt->subscription->metadata = metadata_list_append(ctxt->subscription->metadata, "contributor", tmp);
g_free(tmp);
}
} else if((!xmlStrcmp(cur->name, BAD_CAST"entry"))) {
ctxt->item = parseEntry(ctxt, cur);
if(ctxt->item) {
if(0 == ctxt->item->time)
ctxt->item->time = ctxt->feed->time;
ctxt->items = g_list_append(ctxt->items, ctxt->item);
}
}
cur = cur->next;
}
break;
}
}
/*
The follow are not used, but had been recognized:
"language", <---- Not in atom 0.2 or 0.3. We should use xml:lang
"lastBuildDate", <--- Where is this from?
"issued", <-- Not in the specs for feeds
"created", <---- Not in the specs for feeds
*/
gchar* pie_parse_content_construct(xmlNodePtr cur) {
gchar *mode, *type, *ret;
g_assert(NULL != cur);
ret = NULL;
/* determine encoding mode */
mode = xml_get_attribute (cur, "mode");
type = xml_get_attribute (cur, "type");
/* Modes are used in older versions of ATOM, including 0.3. It
does not exist in the newer IETF drafts.*/
if(NULL != mode) {
if(!strcmp(mode, "escaped")) {
gchar *tmp;
tmp = xhtml_extract (cur, 0, NULL);
if(NULL != tmp)
ret = tmp;
} else if(!strcmp(mode, "xml")) {
ret = xhtml_extract (cur, 1,NULL);
} else if(!strcmp(mode, "base64")) {
g_warning("Base64 encoded <content> in Atom feeds not supported!\n");
} else if(!strcmp(mode, "multipart/alternative")) {
if(NULL != cur->xmlChildrenNode)
ret = pie_parse_content_construct(cur->xmlChildrenNode);
}
g_free(mode);
} else {
/* some feeds don'ts specify a mode but a MIME type in the
type attribute... */
/* not sure what MIME types are necessary... */
/* This that need to be de-encoded and should not contain sub-tags.*/
if(NULL == type ||
!g_ascii_strcasecmp(type, "TEXT") ||
!strcmp(type, "text/plain")) {
gchar *tmp;
tmp = (gchar *)xmlNodeListGetString(cur->doc, cur->xmlChildrenNode, 1);
ret = g_markup_printf_escaped("<div xmlns=\"http://www.w3.org/1999/xhtml\"><pre>%s</pre></div>", tmp);
g_free(tmp);
/* Next are things that contain subttags */
} else if(!g_ascii_strcasecmp(type, "HTML") ||
!strcmp(type, "text/html")) {
ret = xhtml_extract (cur, 0,"http://default.base.com/");
} else if(/* HTML types */
!g_ascii_strcasecmp(type, "xhtml") ||
!strcmp(type, "application/xhtml+xml")) {
ret = xhtml_extract (cur, 1,"http://default.base.com/");
}
}
/* If the type was text, everything must be now escaped and
wrapped in pre tags.... Also, the atom 0.3 spec says that the
default type MUST be considered to be text/plain. The type tag
is required in 0.2.... */
//if (ret != NULL && (type == NULL || !strcmp(type, "text/plain") || !strcmp(type,"TEXT")))) {
g_free(type);
return ret;
}
/* method to parse standard tags for each item element */
itemPtr parseEntry(feedParserCtxtPtr ctxt, xmlNodePtr cur) {
xmlChar *xtmp;
gchar *tmp2, *tmp;
NsHandler *nsh;
parseItemTagFunc pf;
g_assert(NULL != cur);
ctxt->item = item_new();
cur = cur->xmlChildrenNode;
while(cur) {
if(!cur->name) {
g_warning("invalid XML: parser returns NULL value -> tag ignored!");
cur = cur->next;
continue;
}
/* check namespace of this tag */
if(cur->ns) {
if((cur->ns->href && (nsh = (NsHandler *)g_hash_table_lookup(ns_pie_ns_uri_table, (gpointer)cur->ns->href))) ||
(cur->ns->prefix && (nsh = (NsHandler *)g_hash_table_lookup(pie_nstable, (gpointer)cur->ns->prefix)))) {
if(NULL != (pf = nsh->parseItemTag))
(*pf)(ctxt, cur);
cur = cur->next;
continue;
} else {
/*g_print("unsupported namespace \"%s\"\n", cur->ns->prefix);*/
}
} /* explicitly no following else !!! */
if(!xmlStrcmp(cur->name, BAD_CAST"title")) {
if(NULL != (tmp = unhtmlize(pie_parse_content_construct(cur)))) {
item_set_title(ctxt->item, tmp);
g_free(tmp);
}
} else if(!xmlStrcmp(cur->name, BAD_CAST"link")) {
if(NULL != (tmp2 = xml_get_attribute(cur, "href"))) {
/* 0.3 link : rel, type and href attribute */
xtmp = xmlGetProp(cur, BAD_CAST"rel");
if(xtmp != NULL && !xmlStrcmp(xtmp, BAD_CAST"alternate"))
item_set_source(ctxt->item, tmp2);
/* else
FIXME: Maybe do something with other links? */
xmlFree(xtmp);
g_free(tmp2);
} else {
/* 0.2 link : element content is the link, or non-alternate link in 0.3 */
if(NULL != (tmp = (gchar *)xmlNodeListGetString(ctxt->doc, cur->xmlChildrenNode, 1))) {
item_set_source(ctxt->item, tmp);
g_free(tmp);
}
}
} else if(!xmlStrcmp(cur->name, BAD_CAST"author")) {
/* parse feed author */
tmp = parseAuthor(cur);
ctxt->item->metadata = metadata_list_append(ctxt->item->metadata, "author", tmp);
g_free(tmp);
} else if(!xmlStrcmp(cur->name, BAD_CAST"contributor")) {
/* parse feed contributors */
tmp = parseAuthor(cur);
ctxt->item->metadata = metadata_list_append(ctxt->item->metadata, "contributor", tmp);
g_free(tmp);
} else if(!xmlStrcmp(cur->name, BAD_CAST"id")) {
if(NULL != (tmp = (gchar *)xmlNodeListGetString(ctxt->doc, cur->xmlChildrenNode, 1))) {
item_set_id(ctxt->item, tmp);
g_free(tmp);
}
} else if(!xmlStrcmp(cur->name, BAD_CAST"issued")) {
/* FIXME: is <modified> or <issued> or <created> the time tag we want to display? */
if(NULL != (tmp = (gchar *)xmlNodeListGetString(ctxt->doc, cur->xmlChildrenNode, 1))) {
ctxt->item->time = date_parse_ISO8601 (tmp);
g_free(tmp);
}
} else if(!xmlStrcmp(cur->name, BAD_CAST"content")) {
/* <content> support */
if(NULL != (tmp = pie_parse_content_construct(cur))) {
item_set_description(ctxt->item, tmp);
g_free(tmp);
}
} else if(!xmlStrcmp(cur->name, BAD_CAST"summary")) {
/* <summary> can be used for short text descriptions, if there is no
<content> description we show the <summary> content */
if(!item_get_description(ctxt->item)) {
if(NULL != (tmp = pie_parse_content_construct(cur))) {
item_set_description(ctxt->item, tmp);
g_free(tmp);
}
}
} else if(!xmlStrcmp(cur->name, BAD_CAST"copyright")) {
if(NULL != (tmp = (gchar *)xmlNodeListGetString(ctxt->doc, cur->xmlChildrenNode, 1))) {
ctxt->item->metadata = metadata_list_append(ctxt->item->metadata, "copyright", tmp);
g_free(tmp);
}
}
cur = cur->next;
}
/* after parsing we fill the infos into the itemPtr structure */
ctxt->item->readStatus = FALSE;
return ctxt->item;
}
