/* Parse the object tag corresponding to a list item.
*
* At this step we look for all of the "param" child tags, using this information
* to build up the information about the list item. When we reach the </object>
* tag we know that we've finished parsing this list item.
*/
static IndexItem *parse_index_sitemap_object(HHInfo *info, stream_t *stream)
{
strbuf_t node, node_name;
IndexItem *item;
strbuf_init(&node);
strbuf_init(&node_name);
item = heap_alloc_zero(sizeof(IndexItem));
item->nItems = 0;
item->items = heap_alloc_zero(0);
item->itemFlags = 0x11;
while(next_node(stream, &node)) {
get_node_name(&node, &node_name);
TRACE("%s\n", node.buf);
if(!strcasecmp(node_name.buf, "param")) {
parse_index_obj_node_param(item, node.buf, info->pCHMInfo->codePage);
}else if(!strcasecmp(node_name.buf, "/object")) {
break;
}else {
WARN("Unhandled tag! %s\n", node_name.buf);
}
strbuf_zero(&node);
}
strbuf_free(&node);
strbuf_free(&node_name);
return item;
}
static ContentItem *parse_hhc(HHInfo *info, IStream *str, ContentItem *hhc_root,
insert_type_t *insert_type)
{
stream_t stream;
strbuf_t node, node_name;
ContentItem *ret = NULL, *prev = NULL;
*insert_type = INSERT_NEXT;
strbuf_init(&node);
strbuf_init(&node_name);
stream_init(&stream, str);
while(next_node(&stream, &node)) {
get_node_name(&node, &node_name);
TRACE("%s\n", node.buf);
if(!strcasecmp(node_name.buf, "ul")) {
ContentItem *item = parse_ul(info, &stream, hhc_root);
prev = insert_item(prev, item, INSERT_CHILD);
if(!ret)
ret = prev;
*insert_type = INSERT_CHILD;
}
strbuf_zero(&node);
}
strbuf_free(&node);
strbuf_free(&node_name);
return ret;
}
void get_node_name(strbuf_t *node, strbuf_t *name)
{
const char *ptr = node->buf+1;
strbuf_zero(name);
while(*ptr != '>' && !isspace(*ptr))
ptr++;
strbuf_append(name, node->buf+1, ptr-node->buf-1);
strbuf_append(name, "", 1);
}
/*--------------------------------------------------------------------*/
void
strbuf_free (strbuf_t *buf)
/* Free the given string buffer <buf>.
* TODO: Not necessary once all strings are counted and with length?
*/
{
if (buf->buf)
xfree(buf->buf);
strbuf_zero(buf);
}
static ContentItem *parse_sitemap_object(HHInfo *info, stream_t *stream, ContentItem *hhc_root,
insert_type_t *insert_type)
{
strbuf_t node, node_name;
ContentItem *item;
*insert_type = INSERT_NEXT;
strbuf_init(&node);
strbuf_init(&node_name);
item = heap_alloc_zero(sizeof(ContentItem));
while(next_node(stream, &node)) {
get_node_name(&node, &node_name);
TRACE("%s\n", node.buf);
if(!strcasecmp(node_name.buf, "/object"))
break;
if(!strcasecmp(node_name.buf, "param"))
parse_obj_node_param(item, hhc_root, node.buf, info->pCHMInfo->codePage);
strbuf_zero(&node);
}
strbuf_free(&node);
strbuf_free(&node_name);
if(item->merge.chm_index) {
IStream *merge_stream;
merge_stream = GetChmStream(info->pCHMInfo, item->merge.chm_file, &item->merge);
if(merge_stream) {
item->child = parse_hhc(info, merge_stream, hhc_root, insert_type);
IStream_Release(merge_stream);
}else {
WARN("Could not get %s::%s stream\n", debugstr_w(item->merge.chm_file),
debugstr_w(item->merge.chm_file));
if(!item->name) {
free_content_item(item);
item = NULL;
}
}
}
return item;
}
/* Parse the HTML Help page corresponding to all of the Index items.
*
* At this high-level stage we locate out each HTML list item tag.
* Since there is no end-tag for the <LI> item, we must hope that
* the <LI> entry is parsed correctly or tags might get lost.
*
* Within each entry it is also possible to encounter an additional
* <UL> tag. When this occurs the tag indicates that the topics
* contained within it are related to the parent <LI> topic and
* should be inset by an indent.
*/
static void parse_hhindex(HHInfo *info, IStream *str, IndexItem *item)
{
stream_t stream;
strbuf_t node, node_name;
int indent_level = -1;
strbuf_init(&node);
strbuf_init(&node_name);
stream_init(&stream, str);
while(next_node(&stream, &node)) {
get_node_name(&node, &node_name);
TRACE("%s\n", node.buf);
if(!strcasecmp(node_name.buf, "li")) {
IndexItem *new_item;
new_item = parse_li(info, &stream);
if(new_item && item->keyword && strcmpW(new_item->keyword, item->keyword) == 0) {
int num_items = item->nItems;
item_realloc(item, num_items+1);
memcpy(&item->items[num_items], &new_item->items[0], sizeof(IndexSubItem));
heap_free(new_item->keyword);
heap_free(new_item->items);
heap_free(new_item);
} else if(new_item) {
item->next = new_item;
item->next->merge = item->merge;
item = item->next;
item->indentLevel = indent_level;
}
}else if(!strcasecmp(node_name.buf, "ul")) {
indent_level++;
}else if(!strcasecmp(node_name.buf, "/ul")) {
indent_level--;
}else {
WARN("Unhandled tag! %s\n", node_name.buf);
}
strbuf_zero(&node);
}
strbuf_free(&node);
strbuf_free(&node_name);
}
static ContentItem *parse_ul(HHInfo *info, stream_t *stream, ContentItem *hhc_root)
{
strbuf_t node, node_name;
ContentItem *ret = NULL, *prev = NULL, *new_item = NULL;
insert_type_t it;
strbuf_init(&node);
strbuf_init(&node_name);
while(next_node(stream, &node)) {
get_node_name(&node, &node_name);
TRACE("%s\n", node.buf);
if(!strcasecmp(node_name.buf, "object")) {
const char *ptr;
int len;
static const char sz_text_sitemap[] = "text/sitemap";
ptr = get_attr(node.buf, "type", &len);
if(ptr && len == sizeof(sz_text_sitemap)-1
&& !memcmp(ptr, sz_text_sitemap, len)) {
new_item = parse_sitemap_object(info, stream, hhc_root, &it);
prev = insert_item(prev, new_item, it);
if(!ret)
ret = prev;
}
}else if(!strcasecmp(node_name.buf, "ul")) {
new_item = parse_ul(info, stream, hhc_root);
insert_item(prev, new_item, INSERT_CHILD);
}else if(!strcasecmp(node_name.buf, "/ul")) {
break;
}
strbuf_zero(&node);
}
strbuf_free(&node);
strbuf_free(&node_name);
return ret;
}
/* Parse the HTML list item node corresponding to a specific help entry.
*
* At this stage we look for the only child tag we expect to find under
* the list item: the <OBJECT> tag. We also only expect to find object
* tags with the "type" attribute set to "text/sitemap".
*/
static IndexItem *parse_li(HHInfo *info, stream_t *stream)
{
strbuf_t node, node_name;
IndexItem *ret = NULL;
strbuf_init(&node);
strbuf_init(&node_name);
while(next_node(stream, &node)) {
get_node_name(&node, &node_name);
TRACE("%s\n", node.buf);
if(!strcasecmp(node_name.buf, "object")) {
const char *ptr;
int len;
static const char sz_text_sitemap[] = "text/sitemap";
ptr = get_attr(node.buf, "type", &len);
if(ptr && len == sizeof(sz_text_sitemap)-1
&& !memcmp(ptr, sz_text_sitemap, len)) {
ret = parse_index_sitemap_object(info, stream);
break;
}
}else {
WARN("Unhandled tag! %s\n", node_name.buf);
}
strbuf_zero(&node);
}
if(!ret)
FIXME("Failed to parse <li> tag!\n");
strbuf_free(&node);
strbuf_free(&node_name);
return ret;
}
/* Search the CHM storage stream (an HTML file) for the requested text.
*
* Before searching the HTML file all HTML tags are removed so that only
* the content of the document is scanned. If the search string is found
* then the title of the document is returned.
*/
static WCHAR *SearchCHM_File(IStorage *pStorage, const WCHAR *file, const char *needle)
{
char *buffer = heap_alloc(BLOCK_SIZE);
strbuf_t content, node, node_name;
IStream *temp_stream = NULL;
DWORD i, buffer_size = 0;
WCHAR *title = NULL;
BOOL found = FALSE;
stream_t stream;
HRESULT hres;
hres = IStorage_OpenStream(pStorage, file, NULL, STGM_READ, 0, &temp_stream);
if(FAILED(hres)) {
FIXME("Could not open '%s' stream: %08x\n", debugstr_w(file), hres);
goto cleanup;
}
strbuf_init(&node);
strbuf_init(&content);
strbuf_init(&node_name);
stream_init(&stream, temp_stream);
/* Remove all HTML formatting and record the title */
while(next_node(&stream, &node)) {
get_node_name(&node, &node_name);
if(next_content(&stream, &content) && content.len > 1)
{
char *text = &content.buf[1];
int textlen = content.len-1;
if(!strcasecmp(node_name.buf, "title"))
{
int wlen = MultiByteToWideChar(CP_ACP, 0, text, textlen, NULL, 0);
title = heap_alloc((wlen+1)*sizeof(WCHAR));
MultiByteToWideChar(CP_ACP, 0, text, textlen, title, wlen);
title[wlen] = 0;
}
buffer = heap_realloc(buffer, buffer_size + textlen + 1);
memcpy(&buffer[buffer_size], text, textlen);
buffer[buffer_size + textlen] = '\0';
buffer_size += textlen;
}
strbuf_zero(&node);
strbuf_zero(&content);
}
/* Convert the buffer to lower case for comparison against the
* requested text (already in lower case).
*/
for(i=0;i<buffer_size;i++)
buffer[i] = tolower(buffer[i]);
/* Search the decoded buffer for the requested text */
if(strstr(buffer, needle))
found = TRUE;
strbuf_free(&node);
strbuf_free(&content);
strbuf_free(&node_name);
cleanup:
heap_free(buffer);
if(temp_stream)
IStream_Release(temp_stream);
if(!found)
{
heap_free(title);
return NULL;
}
return title;
}
