static void
rss_read_copy (char *d, xml_node_t* n)
{
const char *p = (const char *) xml_get_string (n);
if (p)
strncpy (d, p, RSSMAXBUFSIZE)[RSSMAXBUFSIZE-1] = 0;
else
*d = 0;
}
bool pascalfull_dataset<Tdata>::process_xml(const string &xmlfile) {
string image_filename;
string image_fullname;
string obj_classname;
// parse xml file
try {
DomParser parser;
// parser.set_validate();
parser.parse_file(xmlfile);
if (parser) {
// initialize root node and list
const Node* pNode = parser.get_document()->get_root_node();
Node::NodeList list = pNode->get_children();
// get image filename
for(Node::NodeList::iterator iter = list.begin();
iter != list.end(); ++iter) {
if (!strcmp((*iter)->get_name().c_str(), "filename")) {
xml_get_string(*iter, image_filename);
iter = list.end(); iter--; // stop loop
}
}
image_fullname = imgroot;
image_fullname += image_filename;
// parse all objects in image
for(Node::NodeList::iterator iter = list.begin();
iter != list.end(); ++iter) {
if (!strcmp((*iter)->get_name().c_str(), "object")) {
Node::NodeList olist = (*iter)->get_children();
for(Node::NodeList::iterator oiter = olist.begin();
oiter != olist.end(); ++oiter) {
if (!strcmp((*oiter)->get_name().c_str(), "name")) {
xml_get_string(*oiter, obj_classname);
// if object's name matches an excluded class, stop this xml
if (find(exclude.begin(), exclude.end(),
obj_classname) != exclude.end())
return false;
}
}
}
}
}
} catch (const std::exception& ex) {
cerr << "error: Xml exception caught: " << ex.what() << endl;
return false;
} catch (const char *err) {
cerr << "error: " << err << endl;
return false;
}
// copy image into output directory
ostringstream cmd;
ostringstream tgt;
tgt << outdir << "/" << image_filename;
cmd << "cp " << image_fullname << " " << tgt.str();
if (std::system(cmd.str().c_str()))
cerr << "warning: failed to execute: " << cmd.str() << endl;
else {
cout << data_cnt << ": copied " << tgt.str() << endl;
data_cnt++;
}
return true;
}
static st_rss_t *
rss_open_atom (st_rss_t *rss, const char *encoding)
{
xml_doc_t *doc;
xml_node_t *node;
const char *p = NULL;
doc = xml_parse (rss->url, encoding);
if (!doc)
{
fprintf (stderr, "ERROR: cannot read %s\n", rss->url);
return NULL;
}
node = xml_get_rootnode (doc);
if (!node)
{
fprintf (stderr, "ERROR: empty document %s\n", rss->url);
xml_free (doc);
return NULL;
}
node = xml_get_childnode (node);
while (node && xml_is_empty_node (node))
node = xml_get_nextnode (node);
if (!node)
{
// fprintf (stderr, "");
return NULL;
}
while (node)
{
while (node && xml_is_empty_node (node))
node = xml_get_nextnode (node);
if (!node)
break;
if (!strcasecmp (xml_get_name (node), "title"))
rss_read_copy (rss->title, xml_get_childnode (node));
else if (!strcasecmp (xml_get_name (node), "description"))
rss_read_copy (rss->desc, xml_get_childnode (node));
// else if (!strcasecmp (xml_get_name (node), "link"))
// rss_read_copy (rss->url, xml_get_childnode (node));
else if (!strcasecmp (xml_get_name (node), "date") ||
!strcasecmp (xml_get_name (node), "pubDate") ||
!strcasecmp (xml_get_name (node), "dc:date") ||
!strcasecmp (xml_get_name (node), "modified") ||
!strcasecmp (xml_get_name (node), "updated"))
rss->date = strptime2 ((const char *) xml_get_string (xml_get_childnode (node)));
else if ((!strcasecmp (xml_get_name (node), "entry")))
{
xml_node_t *pnode = xml_get_childnode (node);
st_rss_item_t *item = &rss->item[rss->item_count];
// int found = 0;
char link[RSSMAXBUFSIZE];
*link = 0;
while (pnode)
{
while (pnode && xml_is_empty_node (pnode))
pnode = xml_get_nextnode (pnode);
if (!pnode)
break;
#ifdef DEBUG
printf ("%s\n", xml_get_name (pnode));
fflush (stdout);
#endif
if (!strcasecmp (xml_get_name (pnode), "title"))
{
rss_read_copy (item->title, xml_get_childnode (pnode));
// found = 1;
}
#if 0
else if (!strcasecmp (xml_get_name (pnode), "id"))
{
rss_read_copy (link, xml_get_childnode (pnode));
// found = 1;
}
#endif
else if (!strcasecmp (xml_get_name (pnode), "link") && (!(*link)))
{
#if 0
<link rel="alternate" type="text/html" href="http://edition.cnn.com/2006/POLITICS/11/01/kerry.remarks/"/>
#endif
p = (const char *) xml_get_value (pnode, "href");
if (p)
{
strncpy (link, p, RSSMAXBUFSIZE)[RSSMAXBUFSIZE-1] = 0;
// found = 1;
}
}
else if (!strcasecmp (xml_get_name (pnode), "content"))
{
rss_read_copy (item->desc, xml_get_childnode (pnode));
// found = 1;
}
else if (!strcasecmp (xml_get_name (pnode), "author"))
{
xml_node_t *tnode = xml_get_childnode (pnode);
if (!strcasecmp (xml_get_name (tnode), "name"))
rss_read_copy (item->user, xml_get_childnode (tnode));
// found = 1;
}
else if (!strcasecmp (xml_get_name (pnode), "modified") ||
!strcasecmp (xml_get_name (pnode), "updated"))
{
item->date = strptime2 ((const char *) xml_get_string (xml_get_childnode (pnode)));
// found = 1;
}
// else if (!strcasecmp (xml_get_name (pnode), "duration")) // HACK yt:duration
else if (stristr (xml_get_name (pnode), "duration")) // HACK yt:duration
{
p = (const char *) xml_get_value (pnode, "seconds");
if (p)
{
item->media.duration = strtol (p, NULL, 10);
// found = 1;
// break;
}
}
// else if (!strcasecmp (xml_get_name (pnode), "group")) // media:group
else if (stristr (xml_get_name (pnode), "group")) // media:group
#if 1
{
rss_open_rss_mrss (pnode, item);
}
#else
{
xml_node_t *tnode = xml_get_childnode (pnode);
while (tnode)
{
if (!tnode)
break;
// if (!strcasecmp (xml_get_name (tnode), "content")) // media:content
if (stristr (xml_get_name (tnode), "content")) // media:content
{
p = (const char *) xml_get_value (tnode, "duration");
if (p)
{
item->media.duration = strtol (p, NULL, 10);
// found = 1;
// break;
}
p = (const char *) xml_get_value (tnode, "filesize");
if (p)
{
item->media.filesize = strtol (p, NULL, 10);
// found = 1;
// break;
}
p = (const char *) xml_get_value (tnode, "width");
if (p)
{
item->media.width = strtol (p, NULL, 10);
// found = 1;
// break;
}
p = (const char *) xml_get_value (tnode, "height");
if (p)
{
item->media.height = strtol (p, NULL, 10);
// found = 1;
// break;
}
}
// else if (!strcasecmp (xml_get_name (tnode), "keywords")) // media:keywords
else if (stristr (xml_get_name (tnode), "keywords")) // media:keywords
{
rss_read_copy (item->media.keywords, xml_get_childnode (tnode));
}
// else if (!strcasecmp (xml_get_name (tnode), "thumbnail")) // media:thumbnail
else if (stristr (xml_get_name (tnode), "thumbnail")) // media:thumbnail
{
p = (const char *) xml_get_value (tnode, "url");
if (p)
if (!(item->media.thumbnail[0]))
{
strncpy (item->media.thumbnail, p, RSSMAXBUFSIZE)[RSSMAXBUFSIZE-1] = 0;
// found = 1;
// break;
}
}
tnode = xml_get_nextnode (tnode);
}
}
#endif
pnode = xml_get_nextnode (pnode);
}
if (*link)
strcpy (item->url, link);
rss->item_count++;
if (rss->item_count == RSSMAXITEM)
break;
}
static st_rss_t *
rss_open_rss (st_rss_t *rss, const char *encoding)
{
xml_doc_t *doc;
xml_node_t *node;
int rdf = 0;
doc = xml_parse (rss->url, encoding);
if (!doc)
{
fprintf (stderr, "ERROR: cannot read %s\n", rss->url);
return NULL;
}
node = xml_get_rootnode (doc);
if (!node)
{
fprintf (stderr, "ERROR: empty document %s\n", rss->url);
xml_free (doc);
return NULL;
}
// rdf?
// TODO: move this to rss_demux()
if (strcasecmp (xml_get_name (node), "rss") != 0 &&
!strcasecmp (xml_get_name (node), "rdf"))
rdf = 1;
node = xml_get_childnode (node);
while (node && xml_is_empty_node (node))
node = xml_get_nextnode (node);
if (!node)
{
// fprintf (stderr, "");
return NULL;
}
if (strcasecmp (xml_get_name (node), "channel"))
{
fprintf (stderr, "ERROR: bad document: did not immediately find the RSS element\n");
return NULL;
}
if (!rdf) // document is RSS
node = xml_get_childnode (node);
while (node)
{
while (node && xml_is_empty_node (node))
node = xml_get_nextnode (node);
if (!node)
break;
if (!strcasecmp (xml_get_name (node), "title"))
rss_read_copy (rss->title, xml_get_childnode (node));
else if (!strcasecmp (xml_get_name (node), "description"))
rss_read_copy (rss->desc, xml_get_childnode (node));
// else if (!strcasecmp (xml_get_name (node), "link"))
// rss_read_copy (rss->url, xml_get_childnode (node));
else if (!strcasecmp (xml_get_name (node), "date") ||
!strcasecmp (xml_get_name (node), "pubDate") ||
!strcasecmp (xml_get_name (node), "dc:date"))
rss->date = strptime2 ((const char *) xml_get_string (xml_get_childnode (node)));
else if (!strcasecmp (xml_get_name (node), "channel") && rdf)
{
xml_node_t *pnode = xml_get_childnode (node);
while (pnode)
{
if (!strcasecmp (xml_get_name (pnode), "title"))
rss_read_copy (rss->title, xml_get_childnode (pnode));
else if (!strcasecmp (xml_get_name (pnode), "description"))
rss_read_copy (rss->desc, xml_get_childnode (pnode));
else if (!strcasecmp (xml_get_name (pnode), "date") ||
!strcasecmp (xml_get_name (pnode), "pubDate") ||
!strcasecmp (xml_get_name (pnode), "dc:date"))
rss->date = strptime2 ((const char *) xml_get_string (xml_get_childnode (pnode)));
pnode = xml_get_nextnode (pnode);
}
}
else if (!strcasecmp (xml_get_name (node), "item") || !strcasecmp (xml_get_name (node), "entry"))
{
xml_node_t *pnode = xml_get_childnode (node);
st_rss_item_t *item = &rss->item[rss->item_count];
// int found = 0;
const char *p = NULL;
char link[RSSMAXBUFSIZE], guid[RSSMAXBUFSIZE];
printf ("SHIT");
fflush (stdout);
*link = *guid = 0;
while (pnode)
{
while (pnode && xml_is_empty_node (pnode))
pnode = xml_get_nextnode (pnode);
if (!pnode)
break;
#ifdef DEBUG
printf ("%s\n", xml_get_name (pnode));
fflush (stdout);
#endif
if (!strcasecmp (xml_get_name (pnode), "title"))
{
rss_read_copy (item->title, xml_get_childnode (pnode));
// found = 1;
}
else if (!strcasecmp (xml_get_name (pnode), "link"))
{
rss_read_copy (link, xml_get_childnode (pnode));
// found = 1;
}
#if 0
else if (!strcasecmp (xml_get_name (pnode), "enclosure"))
{
p = (const char *) xml_get_value (pnode, "url");
if (p)
{
strncpy (link, p, RSSMAXBUFSIZE)[RSSMAXBUFSIZE-1] = 0;
// found = 1;
}
}
#endif
else if (!strcasecmp (xml_get_name (pnode), "guid") && (!(*link)))
{
rss_read_copy (guid, xml_get_childnode (pnode));
// found = 1;
}
else if (!strcasecmp (xml_get_name (pnode), "description"))
{
rss_read_copy (item->desc, xml_get_childnode (pnode));
// found = 1;
}
else if (!strcasecmp (xml_get_name (pnode), "date") ||
!strcasecmp (xml_get_name (pnode), "pubDate") ||
!strcasecmp (xml_get_name (pnode), "dc:date") ||
!strcasecmp (xml_get_name (pnode), "cropDate"))
{
item->date = strptime2 ((const char *) xml_get_string (xml_get_childnode (pnode)));
// found = 1;
}
// else if (!strcasecmp (xml_get_name (pnode), "duration")) // HACK yt:duration
else if (stristr (xml_get_name (pnode), "duration")) // HACK yt:duration
{
p = (const char *) xml_get_value (pnode, "seconds");
if (p)
{
item->media.duration = strtol (p, NULL, 10);
// found = 1;
// break;
}
}
// else if (!strcasecmp (xml_get_name (pnode), "group")) // media:group
else if (stristr (xml_get_name (pnode), "group")) // media:group
{
rss_open_rss_mrss (pnode, item);
}
else if (!strcasecmp (xml_get_name (pnode), "author") ||
!strcasecmp (xml_get_name (pnode), "dc:creator") ||
!strcasecmp (xml_get_name (pnode), "creator"))
{
rss_read_copy (item->user, xml_get_childnode (pnode));
// found = 1;
}
#if 0
else
{
if (!found) // possibly malformed feed
break;
else
found = 0;
}
#endif
pnode = xml_get_nextnode (pnode);
}
// some feeds use the guid tag for the link
if (*link)
strcpy (item->url, link);
else if (*guid)
strcpy (item->url, guid);
else
*(item->url) = 0;
rss->item_count++;
if (rss->item_count == RSSMAXITEM)
break;
}
// rss->item_count++;
node = xml_get_nextnode (node);
}
#ifdef DEBUG
rss_st_rss_t_sanity_check (rss);
fflush (stdout);
#endif
return rss;
}
void pascal_dataset<Tdata>::count_sample(Node::NodeList &olist) {
uint difficult = 0, truncated = 0, occluded = 0;
std::string obj_classname, pose;
bool pose_found = false;
Node::NodeList::iterator oiter;
for(oiter = olist.begin(); oiter != olist.end(); ++oiter) {
if (!strcmp((*oiter)->get_name().c_str(), "difficult"))
difficult = xml_get_uint(*oiter);
else if (!strcmp((*oiter)->get_name().c_str(), "truncated"))
truncated = xml_get_uint(*oiter);
else if (!strcmp((*oiter)->get_name().c_str(), "occluded"))
occluded = xml_get_uint(*oiter);
else if (!strcmp((*oiter)->get_name().c_str(), "name"))
xml_get_string(*oiter, obj_classname);
else if (!strcmp((*oiter)->get_name().c_str(), "pose")) {
xml_get_string(*oiter, pose);
pose_found = true;
}
}
////////////////////////////////////////////////////////////////
// object
if (!usepartsonly) {
// add object's class to dataset
if (included_pascal(obj_classname, difficult, truncated, occluded)) {
if (usepose && pose_found) { // append pose to class name
obj_classname += "_";
obj_classname += pose;
}
if (included_pascal(obj_classname, difficult, truncated, occluded))
this->add_class(obj_classname);
}
}
// increment samples numbers
total_samples++;
if (difficult) total_difficult++;
if (truncated) total_truncated++;
if (occluded) total_occluded++;
if ((difficult && ignore_difficult)
|| (truncated && ignore_truncated)
|| (occluded && ignore_occluded))
total_ignored++;
////////////////////////////////////////////////////////////////
// parts
if (useparts || usepartsonly) {
std::string part_classname;
// add part's class to dataset
for(oiter = olist.begin();oiter != olist.end(); ++oiter) {
if (!strcmp((*oiter)->get_name().c_str(), "part")) {
// get part's name
Node::NodeList plist = (*oiter)->get_children();
for(Node::NodeList::iterator piter = plist.begin();
piter != plist.end(); ++piter) {
if (!strcmp((*piter)->get_name().c_str(), "name")) {
xml_get_string(*piter, part_classname);
// found a part and its name, add it
if (included_pascal(part_classname, difficult, truncated, occluded))
{
if (usepose && pose_found) { // append pose to class name
part_classname += "_";
part_classname += pose;
}
if (dataset<Tdata>::included(part_classname)) {
this->add_class(part_classname);
// increment samples numbers
this->total_samples++;
if (difficult) total_difficult++;
if (truncated) total_truncated++;
if (occluded) total_occluded++;
if ((difficult && ignore_difficult)
|| (truncated && ignore_truncated)
|| (occluded && ignore_occluded))
total_ignored++;
}
}
}
}
}
}
}
}
