private TikaImageExtractingParser(RenderingContext renderingContext) {
this.renderingContext = renderingContext;
// Our expected types
types = new HashSet<MediaType>();
types.add(MediaType.image("bmp"));
types.add(MediaType.image("gif"));
types.add(MediaType.image("jpg"));
types.add(MediaType.image("jpeg"));
types.add(MediaType.image("png"));
types.add(MediaType.image("tiff"));
// Are images going in the same place as the HTML?
if (renderingContext.getParamWithDefault(PARAM_IMAGES_SAME_FOLDER, false)) {
RenditionLocation location =
resolveRenditionLocation(
renderingContext.getSourceNode(),
renderingContext.getDefinition(),
renderingContext.getDestinationNode());
imgFolder = location.getParentRef();
if (logger.isDebugEnabled()) {
logger.debug("Using imgFolder: " + imgFolder);
}
}
}
/*
If Tesseract is found, test we retrieve the proper number of supporting Parsers.
*/
@Test
public void offersTypesIfFound() throws Exception {
TesseractOCRParser parser = new TesseractOCRParser();
DefaultParser defaultParser = new DefaultParser();
ParseContext parseContext = new ParseContext();
MediaType png = MediaType.image("png");
// Assuming that Tesseract is on the path, we should find 5 Parsers that support PNG.
assumeTrue(canRun());
assertEquals(5, parser.getSupportedTypes(parseContext).size());
assertTrue(parser.getSupportedTypes(parseContext).contains(png));
// DefaultParser will now select the TesseractOCRParser.
assertEquals(
TesseractOCRParser.class, defaultParser.getParsers(parseContext).get(png).getClass());
}
/*
Check that if Tesseract is not found, the TesseractOCRParser claims to not support
any file types. So, the standard image parser is called instead.
*/
@Test
public void offersNoTypesIfNotFound() throws Exception {
TesseractOCRParser parser = new TesseractOCRParser();
DefaultParser defaultParser = new DefaultParser();
MediaType png = MediaType.image("png");
// With an invalid path, will offer no types
TesseractOCRConfig invalidConfig = new TesseractOCRConfig();
invalidConfig.setTesseractPath("/made/up/path");
ParseContext parseContext = new ParseContext();
parseContext.set(TesseractOCRConfig.class, invalidConfig);
// No types offered
assertEquals(0, parser.getSupportedTypes(parseContext).size());
// And DefaultParser won't use us
assertEquals(ImageParser.class, defaultParser.getParsers(parseContext).get(png).getClass());
}