lastfm service: fix artist info fetching

NEWS

@@ -1,4 +1,5 @@
 ncmpcpp-0.6.7 (????-??-??)
+* Fetching artist info from last.fm was fixed.
 
 ncmpcpp-0.6.6 (2015-09-07)
 * A typo in the example configuration file was fixed.

src/lastfm_service.cpp

@@ -116,9 +116,11 @@ Service::Result ArtistInfo::processData(const std::string &data)
 			rx.assign("<link rel=\"original\" href=\"(.*?)\"");
 			if (boost::regex_search(data, what, rx))
 			{
+				std::string url = what[1], wiki;
+				// unescape &amp;s
+				unescapeHtmlEntities(url);
 				// ...try to get the content of it...
-				std::string wiki;
-				CURLcode code = Curl::perform(wiki, what[1]);
+				CURLcode code = Curl::perform(wiki, url, "", true);
 				
 				if (code != CURLE_OK)
 				{
@@ -128,7 +130,7 @@ Service::Result ArtistInfo::processData(const std::string &data)
 				else
 				{
 					// ...and filter it to get the whole description.
-					rx.assign("<div id=\"wiki\">(.*?)</div>");
+					rx.assign("<div class=\"wiki\">(.*?)</div>");
 					if (boost::regex_search(wiki, what, rx))
 						desc = unescapeHtmlUtf8(what[1]);
 				}

src/utility/html.cpp

@@ -20,7 +20,6 @@
 
 #include <boost/algorithm/string/replace.hpp>
 #include "utility/html.h"
-//#include "utility/string.h"
 
 std::string unescapeHtmlUtf8(const std::string &data)
 {
@@ -51,18 +50,28 @@ std::string unescapeHtmlUtf8(const std::string &data)
 	return result;
 }
 
+void unescapeHtmlEntities(std::string &s)
+{
+	// well, at least some of them.
+	boost::replace_all(s, "&amp;", "&");
+	boost::replace_all(s, "&gt;", ">");
+	boost::replace_all(s, "&lt;", "<");
+	boost::replace_all(s, "&nbsp;", " ");
+	boost::replace_all(s, "&quot;", "\"");
+}
+
 void stripHtmlTags(std::string &s)
 {
 	bool erase = 0;
 	for (size_t i = s.find("<"); i != std::string::npos; i = s.find("<"))
 	{
 		size_t j = s.find(">", i)+1;
+		if (s.compare(i, j-i, "<p>") == 0 || s.compare(i, j-i, "</p>") == 0)
+			s.replace(i, j-i, "\n");
+		else
 			s.replace(i, j-i, "");
 	}
-	boost::replace_all(s, "&#039;", "'");
-	boost::replace_all(s, "&amp;", "&");
-	boost::replace_all(s, "&quot;", "\"");
-	boost::replace_all(s, "&nbsp;", " ");
+	unescapeHtmlEntities(s);
 	for (size_t i = 0; i < s.length(); ++i)
 	{
 		if (erase)

src/utility/html.h

@@ -24,7 +24,7 @@
 #include <string>
 
 std::string unescapeHtmlUtf8(const std::string &s);
-
+void unescapeHtmlEntities(std::string &s);
 void stripHtmlTags(std::string &s);
 
 #endif // NCMPCPP_UTILITY_HTML_H