Commit | Line | Data |
---|---|---|
08fe2e33 NR |
1 | package be.nikiroo.fanfix.supported; |
2 | ||
08fe2e33 NR |
3 | import java.io.File; |
4 | import java.io.FileInputStream; | |
5 | import java.io.IOException; | |
6 | import java.io.InputStream; | |
7445f856 | 7 | import java.net.URISyntaxException; |
08fe2e33 | 8 | import java.net.URL; |
778d8d85 | 9 | import java.net.URLDecoder; |
68686a37 | 10 | import java.util.ArrayList; |
08fe2e33 NR |
11 | import java.util.zip.ZipEntry; |
12 | import java.util.zip.ZipInputStream; | |
13 | ||
7445f856 NR |
14 | import org.jsoup.nodes.Document; |
15 | ||
08fe2e33 | 16 | import be.nikiroo.fanfix.Instance; |
68686a37 | 17 | import be.nikiroo.fanfix.data.MetaData; |
08fe2e33 | 18 | import be.nikiroo.utils.IOUtils; |
16a81ef7 | 19 | import be.nikiroo.utils.Image; |
67837328 | 20 | import be.nikiroo.utils.streams.MarkableFileInputStream; |
b7afbe42 | 21 | import be.nikiroo.utils.StringUtils; |
08fe2e33 NR |
22 | |
23 | /** | |
24 | * Support class for EPUB files created with this program (as we need some | |
25 | * metadata available in those we create). | |
26 | * | |
27 | * @author niki | |
28 | */ | |
68686a37 | 29 | class Epub extends InfoText { |
7445f856 | 30 | private MetaData meta; |
2aac79c7 | 31 | private File tmpDir; |
b7afbe42 | 32 | private String desc; |
08fe2e33 | 33 | |
68686a37 NR |
34 | private URL fakeSource; |
35 | private InputStream fakeIn; | |
08fe2e33 | 36 | |
7445f856 | 37 | public File getSourceFileOriginal() { |
298d405a | 38 | return super.getSourceFile(); |
08fe2e33 NR |
39 | } |
40 | ||
41 | @Override | |
7445f856 NR |
42 | protected File getSourceFile() { |
43 | try { | |
44 | return new File(fakeSource.toURI()); | |
45 | } catch (URISyntaxException e) { | |
46 | Instance.getTraceHandler() | |
47 | .error(new IOException( | |
48 | "Cannot get the source file from the info-text URL", | |
49 | e)); | |
08fe2e33 NR |
50 | } |
51 | ||
52 | return null; | |
53 | } | |
54 | ||
55 | @Override | |
7445f856 | 56 | protected InputStream getInput() { |
298d405a NR |
57 | if (fakeIn != null) { |
58 | try { | |
59 | fakeIn.reset(); | |
60 | } catch (IOException e) { | |
61 | Instance.getTraceHandler() | |
62 | .error(new IOException( | |
63 | "Cannot reset the Epub Text stream", e)); | |
64 | } | |
65 | ||
66 | return fakeIn; | |
67 | } | |
68 | ||
69 | return null; | |
08fe2e33 NR |
70 | } |
71 | ||
72 | @Override | |
7445f856 NR |
73 | protected boolean supports(URL url) { |
74 | return url.getPath().toLowerCase().endsWith(".epub"); | |
75 | } | |
08fe2e33 | 76 | |
7445f856 NR |
77 | @Override |
78 | protected MetaData getMeta() throws IOException { | |
79 | return meta; | |
08fe2e33 NR |
80 | } |
81 | ||
82 | @Override | |
7445f856 NR |
83 | protected Document loadDocument(URL source) throws IOException { |
84 | super.loadDocument(source); // prepares super.getSourceFile() and | |
85 | // super.getInput() | |
86 | ||
87 | InputStream in = super.getInput(); | |
88 | ZipInputStream zipIn = null; | |
89 | try { | |
90 | zipIn = new ZipInputStream(in); | |
91 | tmpDir = Instance.getTempFiles().createTempDir( | |
92 | "fanfic-reader-parser"); | |
93 | File tmp = new File(tmpDir, "file.txt"); | |
94 | File tmpInfo = new File(tmpDir, "file.info"); | |
95 | ||
96 | fakeSource = tmp.toURI().toURL(); | |
97 | Image cover = null; | |
98 | ||
99 | String url; | |
100 | try { | |
101 | url = getSource().toURI().toURL().toString(); | |
102 | } catch (URISyntaxException e1) { | |
103 | url = getSource().toString(); | |
104 | } | |
105 | String title = null; | |
106 | String author = null; | |
107 | ||
108 | for (ZipEntry entry = zipIn.getNextEntry(); entry != null; entry = zipIn | |
109 | .getNextEntry()) { | |
110 | if (!entry.isDirectory() | |
111 | && entry.getName().startsWith(getDataPrefix())) { | |
112 | String entryLName = entry.getName().toLowerCase(); | |
113 | ||
114 | boolean imageEntry = false; | |
115 | for (String ext : BasicSupportImages.getImageExt(false)) { | |
116 | if (entryLName.endsWith(ext)) { | |
117 | imageEntry = true; | |
118 | } | |
08fe2e33 | 119 | } |
08fe2e33 | 120 | |
7445f856 NR |
121 | if (entry.getName().equals(getDataPrefix() + "version")) { |
122 | // Nothing to do for now ("first" | |
123 | // version is 3.0) | |
124 | } else if (entryLName.endsWith(".info")) { | |
125 | // Info file | |
126 | IOUtils.write(zipIn, tmpInfo); | |
127 | } else if (imageEntry) { | |
128 | // Cover | |
129 | if (getCover()) { | |
130 | try { | |
131 | cover = new Image(zipIn); | |
132 | } catch (Exception e) { | |
133 | Instance.getTraceHandler().error(e); | |
134 | } | |
08fe2e33 | 135 | } |
7445f856 NR |
136 | } else if (entry.getName().equals(getDataPrefix() + "URL")) { |
137 | String[] descArray = StringUtils | |
138 | .unhtml(IOUtils.readSmallStream(zipIn)).trim() | |
139 | .split("\n"); | |
140 | if (descArray.length > 0) { | |
141 | url = descArray[0].trim(); | |
142 | } | |
143 | } else if (entry.getName().equals( | |
144 | getDataPrefix() + "SUMMARY")) { | |
145 | String[] descArray = StringUtils | |
146 | .unhtml(IOUtils.readSmallStream(zipIn)).trim() | |
147 | .split("\n"); | |
148 | int skip = 0; | |
149 | if (descArray.length > 1) { | |
150 | title = descArray[0].trim(); | |
151 | skip = 1; | |
152 | if (descArray.length > 2 | |
153 | && descArray[1].startsWith("©")) { | |
154 | author = descArray[1].substring(1).trim(); | |
155 | skip = 2; | |
156 | } | |
157 | } | |
158 | this.desc = ""; | |
159 | for (int i = skip; i < descArray.length; i++) { | |
160 | this.desc += descArray[i].trim() + "\n"; | |
b7afbe42 | 161 | } |
b7afbe42 | 162 | |
7445f856 NR |
163 | this.desc = this.desc.trim(); |
164 | } else { | |
165 | // Hopefully the data file | |
166 | IOUtils.write(zipIn, tmp); | |
167 | } | |
08fe2e33 NR |
168 | } |
169 | } | |
08fe2e33 | 170 | |
7445f856 NR |
171 | if (requireInfo() && (!tmp.exists() || !tmpInfo.exists())) { |
172 | throw new IOException( | |
173 | "file not supported (maybe not created with this program or corrupt)"); | |
174 | } | |
68686a37 | 175 | |
7445f856 | 176 | if (tmp.exists()) { |
67837328 | 177 | this.fakeIn = new MarkableFileInputStream(tmp); |
68686a37 | 178 | } |
7445f856 NR |
179 | |
180 | if (tmpInfo.exists()) { | |
181 | meta = InfoReader.readMeta(tmpInfo, true); | |
7445f856 NR |
182 | tmpInfo.delete(); |
183 | } else { | |
184 | if (title == null || title.isEmpty()) { | |
185 | title = getSourceFileOriginal().getName(); | |
186 | if (title.toLowerCase().endsWith(".cbz")) { | |
187 | title = title.substring(0, title.length() - 4); | |
188 | } | |
189 | title = URLDecoder.decode(title, "UTF-8").trim(); | |
778d8d85 | 190 | } |
778d8d85 | 191 | |
7445f856 NR |
192 | meta = new MetaData(); |
193 | meta.setLang("en"); | |
194 | meta.setTags(new ArrayList<String>()); | |
727108fe | 195 | meta.setSource(getType().getSourceName()); |
7445f856 NR |
196 | meta.setUuid(url); |
197 | meta.setUrl(url); | |
198 | meta.setTitle(title); | |
199 | meta.setAuthor(author); | |
200 | meta.setImageDocument(isImagesDocumentByDefault()); | |
201 | } | |
bb7021f2 NR |
202 | |
203 | if (meta.getCover() == null) { | |
204 | if (cover != null) { | |
205 | meta.setCover(cover); | |
206 | } else { | |
bb7021f2 NR |
207 | meta.setCover(InfoReader |
208 | .getCoverByName(getSourceFileOriginal().toURI() | |
209 | .toURL())); | |
210 | } | |
211 | } | |
7445f856 NR |
212 | } finally { |
213 | if (zipIn != null) { | |
214 | zipIn.close(); | |
215 | } | |
216 | if (in != null) { | |
217 | in.close(); | |
218 | } | |
08fe2e33 | 219 | } |
7445f856 NR |
220 | |
221 | return null; | |
08fe2e33 NR |
222 | } |
223 | ||
224 | @Override | |
0ffa4754 | 225 | protected void close() { |
2aac79c7 NR |
226 | if (tmpDir != null) { |
227 | IOUtils.deltree(tmpDir); | |
08fe2e33 NR |
228 | } |
229 | ||
2aac79c7 | 230 | tmpDir = null; |
dea63313 | 231 | |
68686a37 | 232 | super.close(); |
08fe2e33 NR |
233 | } |
234 | ||
235 | protected String getDataPrefix() { | |
236 | return "DATA/"; | |
237 | } | |
238 | ||
239 | protected boolean requireInfo() { | |
240 | return true; | |
241 | } | |
242 | ||
243 | protected boolean getCover() { | |
244 | return true; | |
245 | } | |
e4fa48a0 NR |
246 | |
247 | protected boolean isImagesDocumentByDefault() { | |
248 | return false; | |
249 | } | |
08fe2e33 | 250 | } |