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