private ExtractedImage createImageObject(URL imageUrl, PlanarImage planarImage, int priority)
throws IOException {
ByteArrayOutputStream imagesBytes = new ByteArrayOutputStream();
String extension = FileUtils.extractExtension(imageUrl.toString(), "jpg");
ImageIO.write(planarImage, extension, imagesBytes);
return new ExtractedImage(
CryptographyUtils.sha1(imageUrl.toString()), imagesBytes.toByteArray(), priority);
}
public ExtractorContext extractContent() {
HtmlParseData htmlParseData = (HtmlParseData) page.getParseData();
ExtractorContext ctx = new ExtractorContext(page, htmlParseData.getHtml());
ctx.setMetaTags(MetadataExtractor.INST.extractMetaTags(ctx.getDoc()));
String language = ctx.getMetaTags().get(MetadataExtractor.CONTENT_LANGUAGE_TAG);
// skip not english pages
if ((language != null && !ENGLISH_US.equals(language.toLowerCase()))) {
LOG.info("Skipping " + ctx.getUrl() + " not ENGLISH");
return null;
}
if (ctx.getUrl().contains(YOUTUBE)) {
String keyWords = ctx.getMetaTags().get(MetadataExtractor.KEYWORDS_TAG);
if (keyWords == null) {
LOG.info("Skipping " + ctx.getUrl() + " no KEYWORDS");
return null;
}
keyWords = keyWords.toLowerCase();
if (!keyWords.contains(SAP)) {
LOG.info("Skipping " + ctx.getUrl() + " no SAP in KEYWORDS");
return null;
}
}
Document clearedHtml = cleanHtml(ctx);
if (clearedHtml.text().length() < MIN_CONTENT_LENGTH) {
LOG.info("Skipping " + ctx.getUrl() + " less then " + MIN_CONTENT_LENGTH + " characters");
return null;
}
// skip pages with not English content
if (!LanguageDetector.INST.isEnglish(clearedHtml.text())) {
LOG.info("Skipping " + ctx.getUrl() + " not ENGLISH");
return null;
}
NavigableSet<ExtractedImage> images = MostRelevantImageExtractor.INST.extract(ctx);
String uniqueFileName = CryptographyUtils.sha1(clearedHtml.text());
PageObject pageObject = new PageObject();
pageObject.setId(uniqueFileName);
pageObject.setTitle(ctx.getTitle());
pageObject.setContent(clearedHtml.html());
pageObject.setUrl(page.getWebURL().getURL());
pageObject.setHtmlCleanerName(ctx.getCleaner().getName());
List<TextBlock> blocks = TextBlocksExtractor.INST.extract(clearedHtml);
if (!blocks.isEmpty()) {
pageObject.setMainBlock(blocks.get(0));
blocks.remove(0);
pageObject.addBlocks(blocks);
}
// set title from First text block if NULL or EMPTY.
if ((pageObject.getTitle() == null || "".equals(pageObject.getTitle().trim()))
&& !pageObject.getBlocks().isEmpty()) {
pageObject.setTitle(pageObject.getBlocks().get(0).getTitle());
}
pageObject.addAllKeywords(BaseClassifier.INST.classify(ctx));
if (!images.isEmpty()) {
pageObject.addImages(images);
}
ctx.setPageObject(pageObject);
return ctx;
}