| 1 | package be.nikiroo.fanfix.supported; |
| 2 | |
| 3 | import java.io.IOException; |
| 4 | import java.io.InputStream; |
| 5 | import java.net.URL; |
| 6 | import java.util.AbstractMap; |
| 7 | import java.util.ArrayList; |
| 8 | import java.util.HashMap; |
| 9 | import java.util.List; |
| 10 | import java.util.Map; |
| 11 | import java.util.Map.Entry; |
| 12 | import java.util.TreeMap; |
| 13 | |
| 14 | import org.jsoup.nodes.Document; |
| 15 | |
| 16 | import be.nikiroo.fanfix.Instance; |
| 17 | import be.nikiroo.fanfix.bundles.Config; |
| 18 | import be.nikiroo.fanfix.data.MetaData; |
| 19 | import be.nikiroo.fanfix.data.Story; |
| 20 | import be.nikiroo.utils.IOUtils; |
| 21 | import be.nikiroo.utils.Image; |
| 22 | import be.nikiroo.utils.Progress; |
| 23 | |
| 24 | /** |
| 25 | * Support class for <a href="http://www.fimfiction.net/">FimFiction.net</a> |
| 26 | * stories, a website dedicated to My Little Pony. |
| 27 | * <p> |
| 28 | * This version uses the new, official API of FimFiction. |
| 29 | * |
| 30 | * @author niki |
| 31 | */ |
| 32 | class FimfictionApi extends BasicSupport { |
| 33 | private String oauth; |
| 34 | private String json; |
| 35 | |
| 36 | private Map<Integer, String> chapterNames; |
| 37 | private Map<Integer, String> chapterContents; |
| 38 | |
| 39 | public FimfictionApi() throws IOException { |
| 40 | if (Instance.getConfig().getBoolean( |
| 41 | Config.LOGIN_FIMFICTION_APIKEY_FORCE_HTML, false)) { |
| 42 | throw new IOException( |
| 43 | "Configuration is set to force HTML scrapping"); |
| 44 | } |
| 45 | |
| 46 | String oauth = Instance.getConfig().getString( |
| 47 | Config.LOGIN_FIMFICTION_APIKEY_TOKEN); |
| 48 | |
| 49 | if (oauth == null || oauth.isEmpty()) { |
| 50 | String clientId = Instance.getConfig().getString( |
| 51 | Config.LOGIN_FIMFICTION_APIKEY_CLIENT_ID) |
| 52 | + ""; |
| 53 | String clientSecret = Instance.getConfig().getString( |
| 54 | Config.LOGIN_FIMFICTION_APIKEY_CLIENT_SECRET) |
| 55 | + ""; |
| 56 | |
| 57 | if (clientId.trim().isEmpty() || clientSecret.trim().isEmpty()) { |
| 58 | throw new IOException("API key required for the beta API v2"); |
| 59 | } |
| 60 | |
| 61 | oauth = generateOAuth(clientId, clientSecret); |
| 62 | |
| 63 | Instance.getConfig().setString( |
| 64 | Config.LOGIN_FIMFICTION_APIKEY_TOKEN, oauth); |
| 65 | Instance.getConfig().updateFile(); |
| 66 | } |
| 67 | |
| 68 | this.oauth = oauth; |
| 69 | } |
| 70 | |
| 71 | @Override |
| 72 | protected Document loadDocument(URL source) throws IOException { |
| 73 | json = getJsonData(); |
| 74 | return null; |
| 75 | } |
| 76 | |
| 77 | @Override |
| 78 | public String getOAuth() { |
| 79 | return oauth; |
| 80 | } |
| 81 | |
| 82 | @Override |
| 83 | protected boolean isHtml() { |
| 84 | return true; |
| 85 | } |
| 86 | |
| 87 | /** |
| 88 | * Extract the full JSON data we will later use to build the {@link Story}. |
| 89 | * |
| 90 | * @return the data in a JSON format |
| 91 | * |
| 92 | * @throws IOException |
| 93 | * in case of I/O error |
| 94 | */ |
| 95 | private String getJsonData() throws IOException { |
| 96 | // extract the ID from: |
| 97 | // https://www.fimfiction.net/story/123456/name-of-story |
| 98 | String storyId = getKeyText(getSource().toString(), "/story/", null, |
| 99 | "/"); |
| 100 | |
| 101 | // Selectors, so to download all I need and only what I need |
| 102 | String storyContent = "fields[story]=title,description,date_published,cover_image"; |
| 103 | String authorContent = "fields[author]=name"; |
| 104 | String chapterContent = "fields[chapter]=chapter_number,title,content_html,authors_note_html"; |
| 105 | String includes = "author,chapters,tags"; |
| 106 | |
| 107 | String urlString = String.format( |
| 108 | "https://www.fimfiction.net/api/v2/stories/%s?" // |
| 109 | + "%s&%s&%s&" // |
| 110 | + "include=%s", // |
| 111 | storyId, // |
| 112 | storyContent, authorContent, chapterContent,// |
| 113 | includes); |
| 114 | |
| 115 | // URL params must be URL-encoded: "[ ]" <-> "%5B %5D" |
| 116 | urlString = urlString.replace("[", "%5B").replace("]", "%5D"); |
| 117 | |
| 118 | URL url = new URL(urlString); |
| 119 | InputStream jsonIn = Instance.getCache().open(url, this, false); |
| 120 | try { |
| 121 | return IOUtils.readSmallStream(jsonIn); |
| 122 | } finally { |
| 123 | jsonIn.close(); |
| 124 | } |
| 125 | } |
| 126 | |
| 127 | @Override |
| 128 | protected MetaData getMeta() throws IOException { |
| 129 | MetaData meta = new MetaData(); |
| 130 | |
| 131 | meta.setTitle(getKeyJson(json, 0, "type", "story", "title")); |
| 132 | meta.setAuthor(getKeyJson(json, 0, "type", "user", "name")); |
| 133 | meta.setDate(getKeyJson(json, 0, "type", "story", "date_published")); |
| 134 | meta.setTags(getTags()); |
| 135 | meta.setSource(getType().getSourceName()); |
| 136 | meta.setUrl(getSource().toString()); |
| 137 | meta.setPublisher(getType().getSourceName()); |
| 138 | meta.setUuid(getSource().toString()); |
| 139 | meta.setLuid(""); |
| 140 | meta.setLang("en"); |
| 141 | meta.setSubject("MLP"); |
| 142 | meta.setType(getType().toString()); |
| 143 | meta.setImageDocument(false); |
| 144 | |
| 145 | String coverImageLink = getKeyJson(json, 0, "type", "story", |
| 146 | "cover_image", "full"); |
| 147 | if (!coverImageLink.trim().isEmpty()) { |
| 148 | InputStream in = null; |
| 149 | try { |
| 150 | URL coverImageUrl = new URL(coverImageLink.trim()); |
| 151 | in = Instance.getCache().open(coverImageUrl, this, true); |
| 152 | meta.setCover(new Image(in)); |
| 153 | } finally { |
| 154 | in.close(); |
| 155 | } |
| 156 | } |
| 157 | |
| 158 | return meta; |
| 159 | } |
| 160 | |
| 161 | private List<String> getTags() { |
| 162 | List<String> tags = new ArrayList<String>(); |
| 163 | tags.add("MLP"); |
| 164 | |
| 165 | int pos = 0; |
| 166 | while (pos >= 0) { |
| 167 | pos = indexOfJsonAfter(json, pos, "type", "story_tag"); |
| 168 | if (pos >= 0) { |
| 169 | tags.add(getKeyJson(json, pos, "name").trim()); |
| 170 | } |
| 171 | } |
| 172 | |
| 173 | return tags; |
| 174 | } |
| 175 | |
| 176 | @Override |
| 177 | protected String getDesc() { |
| 178 | String desc = getKeyJson(json, 0, "type", "story", "description"); |
| 179 | return unbbcode(desc); |
| 180 | } |
| 181 | |
| 182 | @Override |
| 183 | protected List<Entry<String, URL>> getChapters(Progress pg) { |
| 184 | chapterNames = new TreeMap<Integer, String>(); |
| 185 | chapterContents = new TreeMap<Integer, String>(); |
| 186 | |
| 187 | int pos = 0; |
| 188 | while (pos >= 0) { |
| 189 | pos = indexOfJsonAfter(json, pos, "type", "chapter"); |
| 190 | if (pos >= 0) { |
| 191 | int posNumber = indexOfJsonAfter(json, pos, "chapter_number"); |
| 192 | int posComa = json.indexOf(",", posNumber); |
| 193 | final int number = Integer.parseInt(json.substring(posNumber, |
| 194 | posComa).trim()); |
| 195 | final String title = getKeyJson(json, pos, "title"); |
| 196 | String notes = getKeyJson(json, pos, "authors_note_html"); |
| 197 | String content = getKeyJson(json, pos, "content_html"); |
| 198 | |
| 199 | if (!notes.trim().isEmpty()) { |
| 200 | notes = "<br/>* * *<br/>" + notes; |
| 201 | } |
| 202 | |
| 203 | chapterNames.put(number, title); |
| 204 | chapterContents.put(number, content + notes); |
| 205 | } |
| 206 | } |
| 207 | |
| 208 | List<Entry<String, URL>> urls = new ArrayList<Entry<String, URL>>(); |
| 209 | for (String title : chapterNames.values()) { |
| 210 | urls.add(new AbstractMap.SimpleEntry<String, URL>(title, null)); |
| 211 | } |
| 212 | |
| 213 | return urls; |
| 214 | } |
| 215 | |
| 216 | @Override |
| 217 | protected String getChapterContent(URL source, int number, Progress pg) { |
| 218 | return chapterContents.get(number); |
| 219 | } |
| 220 | |
| 221 | @Override |
| 222 | protected boolean supports(URL url) { |
| 223 | return "fimfiction.net".equals(url.getHost()) |
| 224 | || "www.fimfiction.net".equals(url.getHost()); |
| 225 | } |
| 226 | |
| 227 | /** |
| 228 | * Generate a new token from the client ID and secret. |
| 229 | * <p> |
| 230 | * Note that those tokens are long-lived, and it would be badly seen to |
| 231 | * create a lot of them without due cause. |
| 232 | * <p> |
| 233 | * So, please cache and re-use them. |
| 234 | * |
| 235 | * @param clientId |
| 236 | * the client ID offered on FimFiction |
| 237 | * @param clientSecret |
| 238 | * the client secret that goes with it |
| 239 | * |
| 240 | * @return a new generated token linked to that client ID |
| 241 | * |
| 242 | * @throws IOException |
| 243 | * in case of I/O errors |
| 244 | */ |
| 245 | static private String generateOAuth(String clientId, String clientSecret) |
| 246 | throws IOException { |
| 247 | URL url = new URL("https://www.fimfiction.net/api/v2/token"); |
| 248 | Map<String, String> params = new HashMap<String, String>(); |
| 249 | params.put("client_id", clientId); |
| 250 | params.put("client_secret", clientSecret); |
| 251 | params.put("grant_type", "client_credentials"); |
| 252 | InputStream in = Instance.getCache().openNoCache(url, null, params, |
| 253 | null, null); |
| 254 | |
| 255 | String jsonToken = IOUtils.readSmallStream(in); |
| 256 | in.close(); |
| 257 | |
| 258 | // Extract token type and token from: { |
| 259 | // token_type = "Bearer", |
| 260 | // access_token = "xxxxxxxxxxxxxx" |
| 261 | // } |
| 262 | |
| 263 | String tokenType = getKeyText(jsonToken, "\"token_type\"", "\"", "\""); |
| 264 | String token = getKeyText(jsonToken, "\"access_token\"", "\"", "\""); |
| 265 | |
| 266 | return tokenType + " " + token; |
| 267 | } |
| 268 | |
| 269 | // afters: [name, value] pairs (or "" for any of them), can end without |
| 270 | // value |
| 271 | static private int indexOfJsonAfter(String json, int startAt, |
| 272 | String... afterKeys) { |
| 273 | ArrayList<String> afters = new ArrayList<String>(); |
| 274 | boolean name = true; |
| 275 | for (String key : afterKeys) { |
| 276 | if (key != null && !key.isEmpty()) { |
| 277 | afters.add("\"" + key + "\""); |
| 278 | } else { |
| 279 | afters.add("\""); |
| 280 | afters.add("\""); |
| 281 | } |
| 282 | |
| 283 | if (name) { |
| 284 | afters.add(":"); |
| 285 | } |
| 286 | |
| 287 | name = !name; |
| 288 | } |
| 289 | |
| 290 | return indexOfAfter(json, startAt, afters.toArray(new String[] {})); |
| 291 | } |
| 292 | |
| 293 | // afters: [name, value] pairs (or "" for any of them), can end without |
| 294 | // value but will then be empty, not NULL |
| 295 | static private String getKeyJson(String json, int startAt, |
| 296 | String... afterKeys) { |
| 297 | int pos = indexOfJsonAfter(json, startAt, afterKeys); |
| 298 | if (pos < 0) { |
| 299 | return ""; |
| 300 | } |
| 301 | |
| 302 | String result = ""; |
| 303 | String wip = json.substring(pos); |
| 304 | |
| 305 | pos = nextUnescapedQuote(wip, 0); |
| 306 | if (pos >= 0) { |
| 307 | wip = wip.substring(pos + 1); |
| 308 | pos = nextUnescapedQuote(wip, 0); |
| 309 | if (pos >= 0) { |
| 310 | result = wip.substring(0, pos); |
| 311 | } |
| 312 | } |
| 313 | |
| 314 | result = result.replace("\\t", "\t").replace("\\\"", "\""); |
| 315 | |
| 316 | return result; |
| 317 | } |
| 318 | |
| 319 | // next " but don't take \" into account |
| 320 | static private int nextUnescapedQuote(String result, int pos) { |
| 321 | while (pos >= 0) { |
| 322 | pos = result.indexOf("\"", pos); |
| 323 | if (pos == 0 || (pos > 0 && result.charAt(pos - 1) != '\\')) { |
| 324 | break; |
| 325 | } |
| 326 | |
| 327 | if (pos < result.length()) { |
| 328 | pos++; |
| 329 | } |
| 330 | } |
| 331 | |
| 332 | return pos; |
| 333 | } |
| 334 | |
| 335 | // quick & dirty filter |
| 336 | static private String unbbcode(String bbcode) { |
| 337 | String text = bbcode.replace("\\r\\n", "<br/>") // |
| 338 | .replace("[i]", "_").replace("[/i]", "_") // |
| 339 | .replace("[b]", "*").replace("[/b]", "*") // |
| 340 | .replaceAll("\\[[^\\]]*\\]", ""); |
| 341 | return text; |
| 342 | } |
| 343 | |
| 344 | /** |
| 345 | * Return the text between the key and the endKey (and optional subKey can |
| 346 | * be passed, in this case we will look for the key first, then take the |
| 347 | * text between the subKey and the endKey). |
| 348 | * |
| 349 | * @param in |
| 350 | * the input |
| 351 | * @param key |
| 352 | * the key to match (also supports "^" at start to say |
| 353 | * "only if it starts with" the key) |
| 354 | * @param subKey |
| 355 | * the sub key or NULL if none |
| 356 | * @param endKey |
| 357 | * the end key or NULL for "up to the end" |
| 358 | * @return the text or NULL if not found |
| 359 | */ |
| 360 | static private String getKeyText(String in, String key, String subKey, |
| 361 | String endKey) { |
| 362 | String result = null; |
| 363 | |
| 364 | String line = in; |
| 365 | if (line != null && line.contains(key)) { |
| 366 | line = line.substring(line.indexOf(key) + key.length()); |
| 367 | if (subKey == null || subKey.isEmpty() || line.contains(subKey)) { |
| 368 | if (subKey != null) { |
| 369 | line = line.substring(line.indexOf(subKey) |
| 370 | + subKey.length()); |
| 371 | } |
| 372 | if (endKey == null || line.contains(endKey)) { |
| 373 | if (endKey != null) { |
| 374 | line = line.substring(0, line.indexOf(endKey)); |
| 375 | result = line; |
| 376 | } |
| 377 | } |
| 378 | } |
| 379 | } |
| 380 | |
| 381 | return result; |
| 382 | } |
| 383 | |
| 384 | /** |
| 385 | * Return the first index after all the given "afters" have been found in |
| 386 | * the {@link String}, or -1 if it was not possible. |
| 387 | * |
| 388 | * @param in |
| 389 | * the input |
| 390 | * @param startAt |
| 391 | * start at this position in the string |
| 392 | * @param afters |
| 393 | * the sub-keys to find before checking for key/endKey |
| 394 | * |
| 395 | * @return the text or NULL if not found |
| 396 | */ |
| 397 | static private int indexOfAfter(String in, int startAt, String... afters) { |
| 398 | int pos = -1; |
| 399 | if (in != null && !in.isEmpty()) { |
| 400 | pos = startAt; |
| 401 | if (afters != null) { |
| 402 | for (int i = 0; pos >= 0 && i < afters.length; i++) { |
| 403 | String subKey = afters[i]; |
| 404 | if (!subKey.isEmpty()) { |
| 405 | pos = in.indexOf(subKey, pos); |
| 406 | if (pos >= 0) { |
| 407 | pos += subKey.length(); |
| 408 | } |
| 409 | } |
| 410 | } |
| 411 | } |
| 412 | } |
| 413 | |
| 414 | return pos; |
| 415 | } |
| 416 | } |