public void highlight(
final Pattern searchText, final Pattern markingPattern, Color color, int pageNr) {
if (textCache == null || document == null) {
throw new IllegalArgumentException("TextCache was not initialized");
}
final List<PDPage> pages = document.getDocumentCatalog().getAllPages();
try {
boolean found = false;
final PDPage page = pages.get(pageNr - 1);
PDPageContentStream contentStream = new PDPageContentStream(document, page, true, true);
PDExtendedGraphicsState graphicsState = new PDExtendedGraphicsState();
graphicsState.setNonStrokingAlphaConstant(0.5f);
PDResources resources = page.findResources();
Map graphicsStateDictionary = resources.getGraphicsStates();
if (graphicsStateDictionary == null) {
// There is no graphics state dictionary in the resources dictionary, create one.
graphicsStateDictionary = new TreeMap();
}
graphicsStateDictionary.put("highlights", graphicsState);
resources.setGraphicsStates(graphicsStateDictionary);
for (Match searchMatch : textCache.match(pageNr, searchText)) {
if (textCache.match(searchMatch.positions, markingPattern).size() > 0) {
for (Match markingMatch : textCache.match(searchMatch.positions, markingPattern)) {
if (markupMatch(color, contentStream, markingMatch)) {
found = true;
}
}
} else {
System.out.println(
"Cannot highlight: " + markingPattern.pattern() + " on page " + (pageNr - 1));
}
if (found) {
break;
}
}
contentStream.close();
} catch (Exception e) {
e.printStackTrace();
} catch (Error e1) {
e1.printStackTrace();
throw e1;
}
}
public static void assumeNotHeadless() {
boolean headless = true;
try {
GraphicsEnvironment ge = GraphicsEnvironment.getLocalGraphicsEnvironment();
headless = ge.isHeadless();
} catch (Exception e) {
e.printStackTrace();
} catch (Error e) {
// Really not sure why this ever happens, maybe just jenkins issues
e.printStackTrace();
}
if (headless) {
System.out.println("You are trying to start a GUI in a headless environment. Aborting test");
}
org.junit.Assume.assumeTrue(!headless);
}
/*
* The following methods are overwritten from the PDTextStripper
*/
public void initialize(final PDDocument pdf) throws IOException {
try {
resetEngine();
document = pdf;
textCache = new TextCache();
if (getAddMoreFormatting()) {
setParagraphEnd(getLineSeparator());
setPageStart(getLineSeparator());
setArticleStart(getLineSeparator());
setArticleEnd(getLineSeparator());
}
startDocument(pdf);
processPages(pdf.getDocumentCatalog().getAllPages());
endDocument(pdf);
} catch (Exception e) {
e.printStackTrace();
} catch (Error e1) {
e1.printStackTrace();
}
}
public List<Match> match(List<TextPosition> textPositions, String text, final Pattern pattern) {
try {
final Matcher matcher = pattern.matcher(text);
final List<Match> matches = new ArrayList<Match>();
while (matcher.find()) {
final List<TextPosition> elements = textPositions.subList(matcher.start(), matcher.end());
matches.add(new Match(matcher.group(), elements));
}
return matches;
} catch (Error e) {
System.out.println("An error occurred while searching for: " + pattern.toString());
e.printStackTrace();
final List<Match> emptyList = new ArrayList<Match>();
return emptyList;
} catch (Exception e1) {
System.out.println("An exception occurred while seraching for: " + pattern.toString());
e1.printStackTrace();
final List<Match> emptyList = new ArrayList<Match>();
return emptyList;
}
}
