Change BasicSupport to use jsoup
[nikiroo-utils.git] / src / be / nikiroo / fanfix / supported / Epub.java
index 82ebb2b0ea025bdeb77c88a323b193f1a266b9ed..9b06f202f965e82782f3acf98607d9bf6c37ac65 100644 (file)
@@ -1,22 +1,24 @@
 package be.nikiroo.fanfix.supported;
 
-import java.awt.image.BufferedImage;
 import java.io.File;
 import java.io.FileInputStream;
 import java.io.IOException;
 import java.io.InputStream;
 import java.net.URL;
+import java.net.URLDecoder;
+import java.util.ArrayList;
 import java.util.List;
 import java.util.Map.Entry;
 import java.util.zip.ZipEntry;
 import java.util.zip.ZipInputStream;
 
-import javax.imageio.ImageIO;
-
 import be.nikiroo.fanfix.Instance;
-import be.nikiroo.fanfix.bundles.Config;
+import be.nikiroo.fanfix.data.MetaData;
 import be.nikiroo.utils.IOUtils;
+import be.nikiroo.utils.Image;
 import be.nikiroo.utils.MarkableFileInputStream;
+import be.nikiroo.utils.Progress;
+import be.nikiroo.utils.StringUtils;
 
 /**
  * Support class for EPUB files created with this program (as we need some
@@ -24,16 +26,13 @@ import be.nikiroo.utils.MarkableFileInputStream;
  * 
  * @author niki
  */
-class Epub extends BasicSupport {
-       private InfoText base;
-       private URL fakeSource;
-
-       private File tmpCover;
-       private File tmpInfo;
+class Epub extends InfoText {
+       protected MetaData meta;
        private File tmp;
+       private String desc;
 
-       /** Only used by {@link Epub#getInput()} so it is always reset. */
-       private InputStream in;
+       private URL fakeSource;
+       private InputStream fakeIn;
 
        @Override
        public String getSourceName() {
@@ -50,145 +49,58 @@ class Epub extends BasicSupport {
        }
 
        @Override
-       protected boolean isHtml() {
-               if (tmpInfo.exists()) {
-                       return base.isHtml();
-               }
-
-               return false;
-       }
-
-       @Override
-       protected String getTitle(URL source, InputStream in) throws IOException {
-               if (tmpInfo.exists()) {
-                       return base.getTitle(fakeSource, getFakeInput());
-               }
-
-               return source.toString();
-       }
-
-       @Override
-       protected String getAuthor(URL source, InputStream in) throws IOException {
-               if (tmpInfo.exists()) {
-                       return base.getAuthor(fakeSource, getFakeInput());
-               }
-
-               return null;
-       }
-
-       @Override
-       protected String getDate(URL source, InputStream in) throws IOException {
-               if (tmpInfo.exists()) {
-                       return base.getDate(fakeSource, getFakeInput());
-               }
-
-               return null;
-       }
-
-       @Override
-       protected String getSubject(URL source, InputStream in) throws IOException {
-               if (tmpInfo.exists()) {
-                       return base.getSubject(fakeSource, getFakeInput());
-               }
-
-               return null;
+       protected MetaData getMeta(URL source, InputStream in) throws IOException {
+               return meta;
        }
 
        @Override
        protected String getDesc(URL source, InputStream in) throws IOException {
-               if (tmpInfo.exists()) {
-                       return base.getDesc(fakeSource, getFakeInput());
+               if (desc != null) {
+                       return desc;
                }
 
-               return null;
-       }
-
-       @Override
-       protected URL getCover(URL source, InputStream in) throws IOException {
-               if (tmpCover.exists()) {
-                       return tmpCover.toURI().toURL();
+               if (fakeIn != null) {
+                       fakeIn.reset();
+                       return super.getDesc(fakeSource, fakeIn);
                }
 
                return null;
        }
 
        @Override
-       protected List<Entry<String, URL>> getChapters(URL source, InputStream in)
-                       throws IOException {
-               if (tmpInfo.exists()) {
-                       return base.getChapters(fakeSource, getFakeInput());
+       protected List<Entry<String, URL>> getChapters(URL source, InputStream in,
+                       Progress pg) throws IOException {
+               if (fakeIn != null) {
+                       fakeIn.reset();
+                       return super.getChapters(fakeSource, fakeIn, pg);
                }
 
                return null;
        }
 
        @Override
-       protected String getChapterContent(URL source, InputStream in, int number)
-                       throws IOException {
-               if (tmpInfo.exists()) {
-                       return base.getChapterContent(fakeSource, getFakeInput(), number);
+       protected String getChapterContent(URL source, InputStream in, int number,
+                       Progress pg) throws IOException {
+               if (fakeIn != null) {
+                       fakeIn.reset();
+                       return super.getChapterContent(fakeSource, fakeIn, number, pg);
                }
 
                return null;
        }
 
        @Override
-       protected String getLang(URL source, InputStream in) throws IOException {
-               if (tmpInfo.exists()) {
-                       return base.getLang(fakeSource, getFakeInput());
-               }
-
-               return super.getLang(source, in);
-       }
-
-       @Override
-       protected String getPublisher(URL source, InputStream in)
-                       throws IOException {
-               if (tmpInfo.exists()) {
-                       return base.getPublisher(fakeSource, getFakeInput());
-               }
-
-               return super.getPublisher(source, in);
-       }
-
-       @Override
-       protected List<String> getTags(URL source, InputStream in)
-                       throws IOException {
-               if (tmpInfo.exists()) {
-                       return base.getTags(fakeSource, getFakeInput());
-               }
-
-               return super.getTags(source, in);
-       }
-
-       @Override
-       protected String getUuid(URL source, InputStream in) throws IOException {
-               if (tmpInfo.exists()) {
-                       return base.getUuid(fakeSource, getFakeInput());
-               }
-
-               return super.getUuid(source, in);
-       }
-
-       @Override
-       protected String getLuid(URL source, InputStream in) throws IOException {
-               if (tmpInfo.exists()) {
-                       return base.getLuid(fakeSource, getFakeInput());
-               }
-
-               return super.getLuid(source, in);
-       }
-
-       @Override
-       protected void preprocess(InputStream in) throws IOException {
+       protected void preprocess(URL source, InputStream in) throws IOException {
                // Note: do NOT close this stream, as it would also close "in"
                ZipInputStream zipIn = new ZipInputStream(in);
                tmp = File.createTempFile("fanfic-reader-parser_", ".tmp");
-               tmpInfo = new File(tmp + ".info");
-               tmpCover = File.createTempFile("fanfic-reader-parser_", ".tmp");
-
-               base = new InfoText();
+               File tmpInfo = new File(tmp + ".info");
                fakeSource = tmp.toURI().toURL();
+               Image cover = null;
+
+               String url = source.toString();
+               String title = null;
+               String author = null;
 
                for (ZipEntry entry = zipIn.getNextEntry(); entry != null; entry = zipIn
                                .getNextEntry()) {
@@ -213,18 +125,38 @@ class Epub extends BasicSupport {
                                        // Cover
                                        if (getCover()) {
                                                try {
-                                                       BufferedImage image = ImageIO.read(zipIn);
-                                                       ImageIO.write(image, Instance.getConfig()
-                                                                       .getString(Config.IMAGE_FORMAT_COVER)
-                                                                       .toLowerCase(), tmpCover);
+                                                       cover = new Image(zipIn);
                                                } catch (Exception e) {
-                                                       Instance.syserr(e);
+                                                       Instance.getTraceHandler().error(e);
                                                }
                                        }
                                } else if (entry.getName().equals(getDataPrefix() + "URL")) {
-                                       // Do nothing
+                                       String[] descArray = StringUtils
+                                                       .unhtml(IOUtils.readSmallStream(zipIn)).trim()
+                                                       .split("\n");
+                                       if (descArray.length > 0) {
+                                               url = descArray[0].trim();
+                                       }
                                } else if (entry.getName().equals(getDataPrefix() + "SUMMARY")) {
-                                       // Do nothing
+                                       String[] descArray = StringUtils
+                                                       .unhtml(IOUtils.readSmallStream(zipIn)).trim()
+                                                       .split("\n");
+                                       int skip = 0;
+                                       if (descArray.length > 1) {
+                                               title = descArray[0].trim();
+                                               skip = 1;
+                                               if (descArray.length > 2
+                                                               && descArray[1].startsWith("©")) {
+                                                       author = descArray[1].substring(1).trim();
+                                                       skip = 2;
+                                               }
+                                       }
+                                       this.desc = "";
+                                       for (int i = skip; i < descArray.length; i++) {
+                                               this.desc += descArray[i].trim() + "\n";
+                                       }
+
+                                       this.desc = this.desc.trim();
                                } else {
                                        // Hopefully the data file
                                        IOUtils.write(zipIn, tmp);
@@ -238,33 +170,54 @@ class Epub extends BasicSupport {
                }
 
                if (tmp.exists()) {
-                       this.in = new MarkableFileInputStream(new FileInputStream(tmp));
+                       this.fakeIn = new MarkableFileInputStream(new FileInputStream(tmp));
+               }
+
+               if (tmpInfo.exists()) {
+                       meta = InfoReader.readMeta(tmpInfo, true);
+                       if (cover != null) {
+                               meta.setCover(cover);
+                       }
+                       tmpInfo.delete();
+               } else {
+                       if (title == null || title.isEmpty()) {
+                               title = new File(source.getPath()).getName();
+                               if (title.toLowerCase().endsWith(".cbz")) {
+                                       title = title.substring(0, title.length() - 4);
+                               }
+                               title = URLDecoder.decode(title, "UTF-8").trim();
+                       }
+
+                       meta = new MetaData();
+                       meta.setLang("EN");
+                       meta.setTags(new ArrayList<String>());
+                       meta.setSource(getSourceName());
+                       meta.setUuid(url);
+                       meta.setUrl(url);
+                       meta.setTitle(title);
+                       meta.setAuthor(author);
                }
        }
 
        @Override
-       protected void close() throws IOException {
-               for (File file : new File[] { tmp, tmpInfo, tmpCover }) {
-                       if (file != null && file.exists()) {
-                               if (!file.delete()) {
-                                       file.deleteOnExit();
-                               }
+       protected void close() {
+               if (tmp != null && tmp.exists()) {
+                       if (!tmp.delete()) {
+                               tmp.deleteOnExit();
                        }
                }
 
                tmp = null;
-               tmpInfo = null;
-               tmpCover = null;
-               fakeSource = null;
 
-               try {
-                       if (in != null) {
-                               in.close();
+               if (fakeIn != null) {
+                       try {
+                               fakeIn.close();
+                       } catch (Exception e) {
+                               Instance.getTraceHandler().error(e);
                        }
-               } finally {
-                       in = null;
-                       base.close();
                }
+
+               super.close();
        }
 
        protected String getDataPrefix() {
@@ -278,17 +231,4 @@ class Epub extends BasicSupport {
        protected boolean getCover() {
                return true;
        }
-
-       /**
-        * Reset then return {@link Epub#in}.
-        * 
-        * @return {@link Epub#in}
-        * 
-        * @throws IOException
-        *             in case of I/O error
-        */
-       private InputStream getFakeInput() throws IOException {
-               in.reset();
-               return in;
-       }
 }