moparisthebest
/
filebot
mirror of https://github.com/mitb-archive/filebot
1
0
Fork 0
Code Issues Releases Wiki Activity

* separate long-term caches that have different update frequencies

This commit is contained in:
Reinhard Pointner 2013-11-29 04:29:56 +00:00
parent e7668f2c5c
commit 198b8b0e06
5 changed files with 90 additions and 118 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

18
source/ehcache.xml
View File

@ -46,12 +46,26 @@
memoryStoreEvictionPolicy="LRU"
/>
<!--
Long-lived (2 months) persistent disk cache for web responses (that can be updated via If-Modified or If-None-Match)
-->
<cache name="web-datasource-lv3"
maxElementsInMemory="200"
maxElementsOnDisk="95000"
eternal="false"
timeToIdleSeconds="5256000"
timeToLiveSeconds="5256000"
overflowToDisk="true"
diskPersistent="true"
memoryStoreEvictionPolicy="LRU"
/>
<!--
Very long-lived cache (4 months) anime/series lists, movie index, etc
-->
<cache name="web-persistent-datasource"
maxElementsInMemory="200"
maxElementsOnDisk="95000"
maxElementsInMemory="50"
maxElementsOnDisk="5000"
eternal="false"
timeToIdleSeconds="10512000"
timeToLiveSeconds="10512000"

2
source/net/sourceforge/filebot/web/CachedXmlResource.java
View File

@ -23,7 +23,7 @@ public class CachedXmlResource extends AbstractCachedResource<String, String> {
@Override
protected Cache getCache() {
return CacheManager.getInstance().getCache("web-persistent-datasource");
return CacheManager.getInstance().getCache("web-datasource-lv3");
}
public Document getDocument() throws IOException {

2
source/net/sourceforge/filebot/web/ETagCachedResource.java
View File

@ -45,7 +45,7 @@ public abstract class ETagCachedResource<T extends Serializable> extends CachedR
@Override
protected Cache getCache() {
return CacheManager.getInstance().getCache("web-persistent-datasource");
return CacheManager.getInstance().getCache("web-datasource-lv3");
}
}

89
source/net/sourceforge/filebot/web/FanartTV.java
View File

@ -1,7 +1,5 @@
package net.sourceforge.filebot.web;
import static net.sourceforge.filebot.web.WebRequest.*;
import static net.sourceforge.tuned.XPathUtilities.*;
@ -25,47 +23,40 @@ import net.sourceforge.filebot.web.FanartTV.FanartDescriptor.FanartProperty;
import org.w3c.dom.Document;
import org.w3c.dom.Node;
public class FanartTV {
private String apikey;
public FanartTV(String apikey) {
this.apikey = apikey;
}
public List<FanartDescriptor> getSeriesArtwork(int tvdbid) throws Exception {
return getSeriesArtwork(String.valueOf(tvdbid), "all", 1, 2);
}
public List<FanartDescriptor> getSeriesArtwork(String id, String type, int sort, int limit) throws Exception {
return getArtwork("series", id, type, sort, limit);
}
public List<FanartDescriptor> getMovieArtwork(int tmdbid) throws Exception {
return getMovieArtwork(String.valueOf(tmdbid), "all", 1, 2);
}
public List<FanartDescriptor> getMovieArtwork(String id, String type, int sort, int limit) throws Exception {
return getArtwork("movie", id, type, sort, limit);
}
public List<FanartDescriptor> getArtwork(String category, String id, String type, int sort, int limit) throws Exception {
String resource = getResource(category, id, "xml", type, sort, limit);
// cache results
CachedResource<FanartDescriptor[]> data = new CachedResource<FanartDescriptor[]>(resource, FanartDescriptor[].class) {
@Override
public FanartDescriptor[] process(ByteBuffer data) throws Exception {
Document dom = getDocument(Charset.forName("UTF-8").decode(data).toString());
List<FanartDescriptor> fanart = new ArrayList<FanartDescriptor>();
for (Node node : selectNodes("//*[@url]", dom)) {
// e.g. <seasonthumb id="3481" url="http://fanart.tv/fanart/tv/70327/seasonthumb/3481/Buffy (6).jpg" lang="en" likes="0" season="6"/>
@ -79,61 +70,52 @@ public class FanartTV {
}
fanart.add(new FanartDescriptor(fields));
}
return fanart.toArray(new FanartDescriptor[0]);
}
@Override
protected Cache getCache() {
return CacheManager.getInstance().getCache("web-datasource");
return CacheManager.getInstance().getCache("web-datasource-lv2");
}
};
return Arrays.asList(data.get());
}
public String getResource(String category, String id, String format, String type, int sort, int limit) throws MalformedURLException {
// e.g. http://fanart.tv/webservice/series/780b986b22c35e6f7a134a2f392c2deb/70327/xml/all/1/2
return String.format("http://api.fanart.tv/webservice/%s/%s/%s/%s/%s/%s/%s", category, apikey, id, format, type, sort, limit);
}
public static class FanartDescriptor implements Serializable {
public static enum FanartProperty {
type, id, url, lang, likes, season, disc_type
}
protected Map<FanartProperty, String> fields;
protected FanartDescriptor() {
// used by serializer
}
protected FanartDescriptor(Map<FanartProperty, String> fields) {
this.fields = new EnumMap<FanartProperty, String>(fields);
}
public String get(Object key) {
return fields.get(FanartProperty.valueOf(key.toString()));
}
public String get(FanartProperty key) {
return fields.get(key);
}
public String getType() {
return fields.get(FanartProperty.type);
}
public Integer getId() {
try {
return new Integer(fields.get(FanartProperty.id));
@ -141,13 +123,11 @@ public class FanartTV {
return null;
}
}
public String getName() {
return new File(getUrl().getFile()).getName();
}
public URL getUrl() {
try {
return new URL(fields.get(FanartProperty.url).replaceAll(" ", "%20")); // work around server-side url encoding issues
@ -155,8 +135,7 @@ public class FanartTV {
return null;
}
}
public Integer getLikes() {
try {
return new Integer(fields.get(FanartProperty.likes));
@ -164,8 +143,7 @@ public class FanartTV {
return null;
}
}
public Locale getLanguage() {
try {
return new Locale(fields.get(FanartProperty.lang));
@ -173,8 +151,7 @@ public class FanartTV {
return null;
}
}
public Integer getSeason() {
try {
return new Integer(fields.get(FanartProperty.season));
@ -182,17 +159,15 @@ public class FanartTV {
return null;
}
}
public String getDiskType() {
return fields.get(FanartProperty.disc_type);
}
@Override
public String toString() {
return fields.toString();
}
}
}

97
source/net/sourceforge/filebot/web/IMDbClient.java
View File

@ -1,7 +1,5 @@
package net.sourceforge.filebot.web;
import static net.sourceforge.filebot.web.WebRequest.*;
import static net.sourceforge.tuned.XPathUtilities.*;
@ -38,59 +36,54 @@ import org.w3c.dom.Document;
import org.w3c.dom.Node;
import org.xml.sax.SAXException;
public class IMDbClient implements MovieIdentificationService {
private String host = "www.imdb.com";
@Override
public String getName() {
return "IMDb";
}
@Override
public Icon getIcon() {
return ResourceManager.getIcon("search.imdb");
}
protected int getImdbId(String link) {
Matcher matcher = Pattern.compile("tt(\\d{7})").matcher(link);
if (matcher.find()) {
return Integer.parseInt(matcher.group(1));
}
// pattern not found
throw new IllegalArgumentException(String.format("Cannot find imdb id: %s", link));
}
@Override
public List<Movie> searchMovie(String query, Locale locale) throws Exception {
Document dom = parsePage(new URL("http", host, "/find?s=tt&q=" + encode(query, false)));
// select movie links followed by year in parenthesis
List<Node> nodes = selectNodes("//TABLE[@class='findList']//TD/A[substring-after(substring-before(following::text(),')'),'(')]", dom);
List<Movie> results = new ArrayList<Movie>(nodes.size());
for (Node node : nodes) {
try {
String name = node.getTextContent().trim();
if (name.startsWith("\""))
continue;
String year = node.getNextSibling().getTextContent().trim().replaceFirst("^\\(I\\)", "").replaceAll("[\\p{Punct}\\p{Space}]+", ""); // remove non-number characters
String href = getAttribute("href", node);
results.add(new Movie(name, Integer.parseInt(year), getImdbId(href), -1));
} catch (Exception e) {
// ignore illegal movies (TV Shows, Videos, Video Games, etc)
}
}
// we might have been redirected to the movie page
if (results.isEmpty()) {
try {
@ -103,28 +96,26 @@ public class IMDbClient implements MovieIdentificationService {
// ignore, can't find movie
}
}
return results;
}
protected Movie scrapeMovie(Document dom, Locale locale) {
try {
int imdbid = getImdbId(selectString("//LINK[@rel='canonical']/@href", dom));
String title = selectString("//META[@property='og:title']/@content", dom);
Matcher titleMatcher = Pattern.compile("(.+)\\s\\((?i:tv.|video.)?(\\d{4})\\)$").matcher(title);
if (!titleMatcher.matches())
return null;
return new Movie(titleMatcher.group(1), Integer.parseInt(titleMatcher.group(2)), imdbid, -1);
} catch (Exception e) {
// ignore, we probably got redirected to an error page
return null;
}
}
@Override
public Movie getMovieDescriptor(int imdbid, Locale locale) throws Exception {
try {
@ -133,56 +124,50 @@ public class IMDbClient implements MovieIdentificationService {
return null; // illegal imdbid
}
}
protected Document parsePage(URL url) throws IOException, SAXException {
CachedPage page = new CachedPage(url) {
@Override
protected Reader openConnection(URL url) throws IOException {
URLConnection connection = url.openConnection();
// IMDb refuses default user agent (Java/1.6.0_12) => SPOOF GOOGLEBOT
connection.addRequestProperty("User-Agent", "Mozilla/5.0 (compatible; Googlebot/2.1; +http://www.google.com/bot.html)");
connection.addRequestProperty("From", "googlebot(at)googlebot.com");
connection.addRequestProperty("Accept", "*/*");
connection.addRequestProperty("X-Forwarded-For", "66.249.73.100"); // TRICK ANNOYING IMDB GEO-LOCATION LOCALIZATION
return getReader(connection);
}
};
return getHtmlDocument(page.get());
}
public String scrape(String imdbid, String xpath) throws IOException, SAXException {
return scrape(getMoviePageLink(getImdbId(imdbid)).toURL(), xpath); // helper for scraping data in user scripts
}
public String scrape(URL url, String xpath) throws IOException, SAXException {
return selectString(xpath, parsePage(url)); // helper for scraping data in user scripts
}
public URI getMoviePageLink(int imdbId) {
return URI.create(String.format("http://www.imdb.com/title/tt%07d/", imdbId));
}
@Override
public Map<File, Movie> getMovieDescriptors(Collection<File> movieFiles, Locale locale) throws Exception {
throw new UnsupportedOperationException();
}
@SuppressWarnings({ "unchecked", "rawtypes" })
public Map<String, String> getImdbApiData(Integer i, String t, String y, boolean tomatoes) throws IOException {
// e.g. http://www.imdbapi.com/?i=tt0379786&r=xml&tomatoes=true
String url = String.format("http://www.omdbapi.com/?i=%s&t=%s&y=%s&r=xml&tomatoes=%s", String.format(i == null ? "" : "tt%07d", i), t, y, tomatoes);
CachedResource<HashMap> data = new CachedResource<HashMap>(url, HashMap.class) {
@Override
public HashMap process(ByteBuffer data) throws Exception {
Document xml = getDocument(Charset.forName("UTF-8").decode(data).toString());
@ -192,26 +177,24 @@ public class IMDbClient implements MovieIdentificationService {
}
return attr;
}
@Override
protected Cache getCache() {
return CacheManager.getInstance().getCache("web-datasource");
return CacheManager.getInstance().getCache("web-datasource-lv2");
}
};
return data.get();
}
public MovieInfo getImdbApiMovieInfo(Movie movie) throws IOException {
Map<String, String> data = movie.getImdbId() > 0 ? getImdbApiData(movie.getImdbId(), "", "", false) : getImdbApiData(null, movie.getName(), String.valueOf(movie.getYear()), false);
// sanity check
if (!Boolean.parseBoolean(data.get("response"))) {
throw new IllegalArgumentException("Movie not found: " + data);
}
Map<MovieProperty, String> fields = new EnumMap<MovieProperty, String>(MovieProperty.class);
fields.put(MovieProperty.title, data.get("title"));
fields.put(MovieProperty.certification, data.get("rated"));
@ -221,7 +204,7 @@ public class IMDbClient implements MovieIdentificationService {
fields.put(MovieProperty.vote_count, data.get("imdbVotes").replaceAll("\\D", ""));
fields.put(MovieProperty.imdb_id, data.get("imdbID"));
fields.put(MovieProperty.poster_path, data.get("poster"));
// convert release date to yyyy-MM-dd
Date released = Date.parse(data.get("released"), "dd MMM yyyy");
if (released != null) {
@ -232,25 +215,25 @@ public class IMDbClient implements MovieIdentificationService {
fields.put(MovieProperty.release_date, year.format("yyyy-MM-dd"));
}
}
List<String> genres = new ArrayList<String>();
for (String it : data.get("genre").split(",")) {
genres.add(it.trim());
}
List<Person> actors = new ArrayList<Person>();
for (String it : data.get("actors").split(",")) {
actors.add(new Person(it.trim(), null, null));
}
for (String director : data.get("director").split(",")) {
actors.add(new Person(director, null, "Director"));
}
for (String writer : data.get("writer").split(",")) {
actors.add(new Person(writer, null, "Writer"));
}
return new MovieInfo(fields, genres, new ArrayList<String>(0), actors, new ArrayList<Trailer>(0));
}
}