X-Git-Url: http://git.nikiroo.be/?p=nikiroo-utils.git;a=blobdiff_plain;f=src%2Fbe%2Fnikiroo%2Ffanfix%2Fsupported%2FEpub.java;h=82af11855fdc299226fe2f52ca536f17a07d1d00;hp=bd5c7191a86fe687942a6411c4291ff9010ad060;hb=8d59ce0748baeeea0458bab49716ab4543aae439;hpb=dea6331397544cde6cc3d2f58dd6355c57679cea diff --git a/src/be/nikiroo/fanfix/supported/Epub.java b/src/be/nikiroo/fanfix/supported/Epub.java index bd5c719..82af118 100644 --- a/src/be/nikiroo/fanfix/supported/Epub.java +++ b/src/be/nikiroo/fanfix/supported/Epub.java @@ -1,23 +1,23 @@ 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.URISyntaxException; 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 org.jsoup.nodes.Document; + import be.nikiroo.fanfix.Instance; import be.nikiroo.fanfix.data.MetaData; import be.nikiroo.utils.IOUtils; -import be.nikiroo.utils.ImageUtils; -import be.nikiroo.utils.MarkableFileInputStream; -import be.nikiroo.utils.Progress; +import be.nikiroo.utils.Image; +import be.nikiroo.utils.StringUtils; +import be.nikiroo.utils.streams.MarkableFileInputStream; /** * Support class for EPUB files created with this program (as we need some @@ -26,150 +26,208 @@ import be.nikiroo.utils.Progress; * @author niki */ class Epub extends InfoText { - private File tmp; - protected MetaData meta; + private MetaData meta; + private File tmpDir; + private String desc; private URL fakeSource; private InputStream fakeIn; - @Override - public String getSourceName() { - return "epub"; + public File getSourceFileOriginal() { + return super.getSourceFile(); } @Override - protected boolean supports(URL url) { - if (url.getPath().toLowerCase().endsWith(".epub")) { - return true; + protected File getSourceFile() { + try { + return new File(fakeSource.toURI()); + } catch (URISyntaxException e) { + Instance.getTraceHandler() + .error(new IOException( + "Cannot get the source file from the info-text URL", + e)); } - return false; - } - - @Override - protected MetaData getMeta(URL source, InputStream in) throws IOException { - return meta; + return null; } @Override - protected String getDesc(URL source, InputStream in) throws IOException { + protected InputStream getInput() { if (fakeIn != null) { - fakeIn.reset(); - return super.getDesc(fakeSource, fakeIn); + try { + fakeIn.reset(); + } catch (IOException e) { + Instance.getTraceHandler() + .error(new IOException( + "Cannot reset the Epub Text stream", e)); + } + + return fakeIn; } return null; } @Override - protected List> getChapters(URL source, InputStream in, - Progress pg) throws IOException { - if (fakeIn != null) { - fakeIn.reset(); - return super.getChapters(fakeSource, fakeIn, pg); - } - - return null; + protected boolean supports(URL url) { + return url.getPath().toLowerCase().endsWith(".epub"); } @Override - 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; + protected MetaData getMeta() throws IOException { + return meta; } @Override - 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"); - File tmpInfo = new File(tmp + ".info"); - fakeSource = tmp.toURI().toURL(); - BufferedImage cover = null; - - for (ZipEntry entry = zipIn.getNextEntry(); entry != null; entry = zipIn - .getNextEntry()) { - if (!entry.isDirectory() - && entry.getName().startsWith(getDataPrefix())) { - String entryLName = entry.getName().toLowerCase(); - - boolean imageEntry = false; - for (String ext : getImageExt(false)) { - if (entryLName.endsWith(ext)) { - imageEntry = true; + protected Document loadDocument(URL source) throws IOException { + super.loadDocument(source); // prepares super.getSourceFile() and + // super.getInput() + + InputStream in = super.getInput(); + ZipInputStream zipIn = null; + try { + zipIn = new ZipInputStream(in); + tmpDir = Instance.getTempFiles().createTempDir( + "fanfic-reader-parser"); + File tmp = new File(tmpDir, "file.txt"); + File tmpInfo = new File(tmpDir, "file.info"); + + fakeSource = tmp.toURI().toURL(); + Image cover = null; + + String url; + try { + url = getSource().toURI().toURL().toString(); + } catch (URISyntaxException e1) { + url = getSource().toString(); + } + String title = null; + String author = null; + + for (ZipEntry entry = zipIn.getNextEntry(); entry != null; entry = zipIn + .getNextEntry()) { + if (!entry.isDirectory() + && entry.getName().startsWith(getDataPrefix())) { + String entryLName = entry.getName().toLowerCase(); + + boolean imageEntry = false; + for (String ext : bsImages.getImageExt(false)) { + if (entryLName.endsWith(ext)) { + imageEntry = true; + } } - } - if (entry.getName().equals(getDataPrefix() + "version")) { - // Nothing to do for now ("first" - // version is 3.0) - } else if (entryLName.endsWith(".info")) { - // Info file - IOUtils.write(zipIn, tmpInfo); - } else if (imageEntry) { - // Cover - if (getCover()) { - try { - cover = ImageUtils.fromStream(zipIn); - } catch (Exception e) { - Instance.syserr(e); + if (entry.getName().equals(getDataPrefix() + "version")) { + // Nothing to do for now ("first" + // version is 3.0) + } else if (entryLName.endsWith(".info")) { + // Info file + IOUtils.write(zipIn, tmpInfo); + } else if (imageEntry) { + // Cover + if (getCover()) { + try { + cover = new Image(zipIn); + } catch (Exception e) { + Instance.getTraceHandler().error(e); + } + } + } else if (entry.getName().equals(getDataPrefix() + "URL")) { + 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")) { + 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); } - } else if (entry.getName().equals(getDataPrefix() + "URL")) { - // Do nothing - } else if (entry.getName().equals(getDataPrefix() + "SUMMARY")) { - // Do nothing - } else { - // Hopefully the data file - IOUtils.write(zipIn, tmp); } } - } - if (requireInfo() && (!tmp.exists() || !tmpInfo.exists())) { - throw new IOException( - "file not supported (maybe not created with this program or corrupt)"); - } + if (requireInfo() && (!tmp.exists() || !tmpInfo.exists())) { + throw new IOException( + "file not supported (maybe not created with this program or corrupt)"); + } - if (tmp.exists()) { - this.fakeIn = new MarkableFileInputStream(new FileInputStream(tmp)); - } + if (tmp.exists()) { + this.fakeIn = new MarkableFileInputStream(tmp); + } + + if (tmpInfo.exists()) { + meta = InfoReader.readMeta(tmpInfo, true); + tmpInfo.delete(); + } else { + if (title == null || title.isEmpty()) { + title = getSourceFileOriginal().getName(); + if (title.toLowerCase().endsWith(".cbz")) { + title = title.substring(0, title.length() - 4); + } + title = URLDecoder.decode(title, "UTF-8").trim(); + } - if (tmpInfo.exists()) { - meta = InfoReader.readMeta(tmpInfo, true); - if (cover != null) { - meta.setCover(cover); + meta = new MetaData(); + meta.setLang("en"); + meta.setTags(new ArrayList()); + meta.setSource(getType().getSourceName()); + meta.setUuid(url); + meta.setUrl(url); + meta.setTitle(title); + meta.setAuthor(author); + meta.setImageDocument(isImagesDocumentByDefault()); } - tmpInfo.delete(); - } else { - meta = new MetaData(); - meta.setUuid(source.toString()); - meta.setLang("EN"); - meta.setTags(new ArrayList()); - meta.setSource(getSourceName()); - meta.setUrl(source.toString()); - } - } - @Override - protected void close() throws IOException { - if (tmp != null && tmp.exists()) { - if (!tmp.delete()) { - tmp.deleteOnExit(); + if (meta.getCover() == null) { + if (cover != null) { + meta.setCover(cover); + } else { + meta.setCover(InfoReader + .getCoverByName(getSourceFileOriginal().toURI() + .toURL())); + } + } + } finally { + if (zipIn != null) { + zipIn.close(); + } + if (in != null) { + in.close(); } } - tmp = null; + return null; + } - if (fakeIn != null) { - fakeIn.close(); + @Override + protected void close() { + if (tmpDir != null) { + IOUtils.deltree(tmpDir); } + tmpDir = null; + super.close(); } @@ -184,4 +242,8 @@ class Epub extends InfoText { protected boolean getCover() { return true; } + + protected boolean isImagesDocumentByDefault() { + return false; + } }