Commit | Line | Data |
---|---|---|
08fe2e33 NR |
1 | package be.nikiroo.fanfix.supported; |
2 | ||
08fe2e33 | 3 | import java.io.File; |
08fe2e33 NR |
4 | import java.io.IOException; |
5 | import java.io.InputStream; | |
7445f856 | 6 | import java.net.URISyntaxException; |
08fe2e33 | 7 | import java.net.URL; |
778d8d85 | 8 | import java.net.URLDecoder; |
68686a37 | 9 | import java.util.ArrayList; |
15da4d0a NR |
10 | import java.util.Arrays; |
11 | import java.util.Collections; | |
08fe2e33 NR |
12 | import java.util.zip.ZipEntry; |
13 | import java.util.zip.ZipInputStream; | |
14 | ||
7445f856 NR |
15 | import org.jsoup.nodes.Document; |
16 | ||
08fe2e33 | 17 | import be.nikiroo.fanfix.Instance; |
68686a37 | 18 | import be.nikiroo.fanfix.data.MetaData; |
08fe2e33 | 19 | import be.nikiroo.utils.IOUtils; |
16a81ef7 | 20 | import be.nikiroo.utils.Image; |
b7afbe42 | 21 | import be.nikiroo.utils.StringUtils; |
8d59ce07 | 22 | import be.nikiroo.utils.streams.MarkableFileInputStream; |
08fe2e33 NR |
23 | |
24 | /** | |
25 | * Support class for EPUB files created with this program (as we need some | |
26 | * metadata available in those we create). | |
27 | * | |
28 | * @author niki | |
29 | */ | |
68686a37 | 30 | class Epub extends InfoText { |
7445f856 | 31 | private MetaData meta; |
2aac79c7 | 32 | private File tmpDir; |
b7afbe42 | 33 | private String desc; |
08fe2e33 | 34 | |
68686a37 NR |
35 | private URL fakeSource; |
36 | private InputStream fakeIn; | |
08fe2e33 | 37 | |
7445f856 | 38 | public File getSourceFileOriginal() { |
298d405a | 39 | return super.getSourceFile(); |
08fe2e33 NR |
40 | } |
41 | ||
42 | @Override | |
7445f856 NR |
43 | protected File getSourceFile() { |
44 | try { | |
45 | return new File(fakeSource.toURI()); | |
46 | } catch (URISyntaxException e) { | |
076caecc NR |
47 | Instance.getInstance().getTraceHandler().error(new IOException( |
48 | "Cannot get the source file from the info-text URL", e)); | |
08fe2e33 NR |
49 | } |
50 | ||
51 | return null; | |
52 | } | |
53 | ||
54 | @Override | |
7445f856 | 55 | protected InputStream getInput() { |
298d405a NR |
56 | if (fakeIn != null) { |
57 | try { | |
58 | fakeIn.reset(); | |
59 | } catch (IOException e) { | |
076caecc NR |
60 | Instance.getInstance().getTraceHandler().error(new IOException( |
61 | "Cannot reset the Epub Text stream", e)); | |
298d405a NR |
62 | } |
63 | ||
64 | return fakeIn; | |
65 | } | |
66 | ||
67 | return null; | |
08fe2e33 NR |
68 | } |
69 | ||
70 | @Override | |
7445f856 NR |
71 | protected boolean supports(URL url) { |
72 | return url.getPath().toLowerCase().endsWith(".epub"); | |
73 | } | |
08fe2e33 | 74 | |
7445f856 NR |
75 | @Override |
76 | protected MetaData getMeta() throws IOException { | |
77 | return meta; | |
08fe2e33 NR |
78 | } |
79 | ||
80 | @Override | |
7445f856 NR |
81 | protected Document loadDocument(URL source) throws IOException { |
82 | super.loadDocument(source); // prepares super.getSourceFile() and | |
83 | // super.getInput() | |
84 | ||
85 | InputStream in = super.getInput(); | |
86 | ZipInputStream zipIn = null; | |
87 | try { | |
88 | zipIn = new ZipInputStream(in); | |
076caecc NR |
89 | tmpDir = Instance.getInstance().getTempFiles() |
90 | .createTempDir("fanfic-reader-parser"); | |
7445f856 NR |
91 | File tmp = new File(tmpDir, "file.txt"); |
92 | File tmpInfo = new File(tmpDir, "file.info"); | |
93 | ||
94 | fakeSource = tmp.toURI().toURL(); | |
95 | Image cover = null; | |
96 | ||
97 | String url; | |
98 | try { | |
99 | url = getSource().toURI().toURL().toString(); | |
100 | } catch (URISyntaxException e1) { | |
101 | url = getSource().toString(); | |
102 | } | |
103 | String title = null; | |
104 | String author = null; | |
105 | ||
076caecc NR |
106 | for (ZipEntry entry = zipIn |
107 | .getNextEntry(); entry != null; entry = zipIn | |
108 | .getNextEntry()) { | |
7445f856 NR |
109 | if (!entry.isDirectory() |
110 | && entry.getName().startsWith(getDataPrefix())) { | |
111 | String entryLName = entry.getName().toLowerCase(); | |
15da4d0a | 112 | entryLName = entryLName.substring(getDataPrefix().length()); |
7445f856 NR |
113 | |
114 | boolean imageEntry = false; | |
8d59ce07 | 115 | for (String ext : bsImages.getImageExt(false)) { |
7445f856 NR |
116 | if (entryLName.endsWith(ext)) { |
117 | imageEntry = true; | |
118 | } | |
08fe2e33 | 119 | } |
08fe2e33 | 120 | |
15da4d0a | 121 | if (entryLName.equals("version")) { |
7445f856 NR |
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 | |
15da4d0a | 129 | if (getCover() && cover == null) { |
7445f856 | 130 | try { |
0a264fbe NR |
131 | Image img = new Image(zipIn); |
132 | if (img.getSize() == 0) { | |
133 | img.close(); | |
134 | throw new IOException( | |
135 | "Empty image not accepted"); | |
136 | } | |
137 | cover = img; | |
7445f856 | 138 | } catch (Exception e) { |
076caecc NR |
139 | Instance.getInstance().getTraceHandler() |
140 | .error(e); | |
7445f856 | 141 | } |
08fe2e33 | 142 | } |
15da4d0a | 143 | } else if (entryLName.equals("url")) { |
7445f856 NR |
144 | String[] descArray = StringUtils |
145 | .unhtml(IOUtils.readSmallStream(zipIn)).trim() | |
146 | .split("\n"); | |
147 | if (descArray.length > 0) { | |
148 | url = descArray[0].trim(); | |
149 | } | |
15da4d0a | 150 | } else if (entryLName.endsWith(".desc")) { |
076caecc NR |
151 | // // For old files |
152 | // if (this.desc != null) { | |
153 | // this.desc = IOUtils.readSmallStream(zipIn).trim(); | |
154 | // } | |
15da4d0a | 155 | } else if (entryLName.equals("summary")) { |
7445f856 NR |
156 | String[] descArray = StringUtils |
157 | .unhtml(IOUtils.readSmallStream(zipIn)).trim() | |
158 | .split("\n"); | |
159 | int skip = 0; | |
160 | if (descArray.length > 1) { | |
161 | title = descArray[0].trim(); | |
162 | skip = 1; | |
163 | if (descArray.length > 2 | |
164 | && descArray[1].startsWith("©")) { | |
165 | author = descArray[1].substring(1).trim(); | |
166 | skip = 2; | |
167 | } | |
168 | } | |
076caecc NR |
169 | // this.desc = ""; |
170 | // for (int i = skip; i < descArray.length; i++) { | |
171 | // this.desc += descArray[i].trim() + "\n"; | |
172 | // } | |
173 | // | |
174 | // this.desc = this.desc.trim(); | |
7445f856 NR |
175 | } else { |
176 | // Hopefully the data file | |
177 | IOUtils.write(zipIn, tmp); | |
178 | } | |
08fe2e33 NR |
179 | } |
180 | } | |
08fe2e33 | 181 | |
15da4d0a | 182 | if (requireInfo() && !tmp.exists()) { |
7445f856 NR |
183 | throw new IOException( |
184 | "file not supported (maybe not created with this program or corrupt)"); | |
185 | } | |
68686a37 | 186 | |
7445f856 | 187 | if (tmp.exists()) { |
67837328 | 188 | this.fakeIn = new MarkableFileInputStream(tmp); |
68686a37 | 189 | } |
7445f856 NR |
190 | |
191 | if (tmpInfo.exists()) { | |
192 | meta = InfoReader.readMeta(tmpInfo, true); | |
7445f856 NR |
193 | tmpInfo.delete(); |
194 | } else { | |
195 | if (title == null || title.isEmpty()) { | |
196 | title = getSourceFileOriginal().getName(); | |
15da4d0a NR |
197 | String exts[] = new String[] {".epub", ".cbz"}; |
198 | for (String ext : exts) { | |
199 | if (title.toLowerCase().endsWith(ext)) { | |
200 | title = title.substring(0, | |
201 | title.length() - ext.length()); | |
202 | } | |
7445f856 NR |
203 | } |
204 | title = URLDecoder.decode(title, "UTF-8").trim(); | |
778d8d85 | 205 | } |
778d8d85 | 206 | |
7445f856 NR |
207 | meta = new MetaData(); |
208 | meta.setLang("en"); | |
15da4d0a | 209 | meta.setTags(Arrays.asList("[no_info]")); |
727108fe | 210 | meta.setSource(getType().getSourceName()); |
7445f856 NR |
211 | meta.setUuid(url); |
212 | meta.setUrl(url); | |
213 | meta.setTitle(title); | |
214 | meta.setAuthor(author); | |
215 | meta.setImageDocument(isImagesDocumentByDefault()); | |
15da4d0a NR |
216 | |
217 | InfoReader.completeMeta(tmp, meta); | |
7445f856 | 218 | } |
bb7021f2 NR |
219 | |
220 | if (meta.getCover() == null) { | |
221 | if (cover != null) { | |
222 | meta.setCover(cover); | |
223 | } else { | |
076caecc NR |
224 | meta.setCover(InfoReader.getCoverByName( |
225 | getSourceFileOriginal().toURI().toURL())); | |
bb7021f2 NR |
226 | } |
227 | } | |
7445f856 NR |
228 | } finally { |
229 | if (zipIn != null) { | |
230 | zipIn.close(); | |
231 | } | |
232 | if (in != null) { | |
233 | in.close(); | |
234 | } | |
08fe2e33 | 235 | } |
7445f856 NR |
236 | |
237 | return null; | |
08fe2e33 NR |
238 | } |
239 | ||
240 | @Override | |
0ffa4754 | 241 | protected void close() { |
2aac79c7 NR |
242 | if (tmpDir != null) { |
243 | IOUtils.deltree(tmpDir); | |
08fe2e33 NR |
244 | } |
245 | ||
2aac79c7 | 246 | tmpDir = null; |
dea63313 | 247 | |
68686a37 | 248 | super.close(); |
08fe2e33 NR |
249 | } |
250 | ||
251 | protected String getDataPrefix() { | |
252 | return "DATA/"; | |
253 | } | |
254 | ||
255 | protected boolean requireInfo() { | |
256 | return true; | |
257 | } | |
258 | ||
259 | protected boolean getCover() { | |
260 | return true; | |
261 | } | |
e4fa48a0 NR |
262 | |
263 | protected boolean isImagesDocumentByDefault() { | |
264 | return false; | |
265 | } | |
08fe2e33 | 266 | } |