DGNum/Stirling-PDF
14
0
Fork 0
Code Issues 4 Pull requests 1 Projects Releases Packages Wiki Activity

Update TextFinder.java (#980)

Browse source
This commit is contained in:
Anthony Stirling 2024-03-26 19:25:16 +00:00 committed by GitHub
parent 7e2d58b3e8
commit 2fff3083ae
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 21 additions and 10 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

31
src/main/java/stirling/software/SPDF/pdf/TextFinder.java
View file

@ -19,6 +19,16 @@ public class TextFinder extends PDFTextStripper {
private final boolean wholeWordSearch; private final boolean wholeWordSearch;
private final List<PDFText> textOccurrences = new ArrayList<>(); private final List<PDFText> textOccurrences = new ArrayList<>();
private class MatchInfo {
int startIndex;
int matchLength;
MatchInfo(int startIndex, int matchLength) {
this.startIndex = startIndex;
this.matchLength = matchLength;
}
}
public TextFinder(String searchText, boolean useRegex, boolean wholeWordSearch) public TextFinder(String searchText, boolean useRegex, boolean wholeWordSearch)
throws IOException { throws IOException {
this.searchText = searchText.toLowerCase(); this.searchText = searchText.toLowerCase();
@ -27,36 +37,37 @@ public class TextFinder extends PDFTextStripper {
setSortByPosition(true); setSortByPosition(true);
} }
private List<Integer> findOccurrencesInText(String searchText, String content) { private List<MatchInfo> findOccurrencesInText(String searchText, String content) {
List<Integer> indexes = new ArrayList<>(); List<MatchInfo> matches = new ArrayList<>();
Pattern pattern; Pattern pattern;
if (useRegex) { if (useRegex) {
// Use regex-based search // Use regex-based search
pattern = pattern =
wholeWordSearch wholeWordSearch
? Pattern.compile("(\\b|_|\\.)" + searchText + "(\\b|_|\\.)") ? Pattern.compile("\\b" + searchText + "\\b")
: Pattern.compile(searchText); : Pattern.compile(searchText);
} else { } else {
// Use normal text search // Use normal text search
pattern = pattern =
wholeWordSearch wholeWordSearch
? Pattern.compile( ? Pattern.compile("\\b" + Pattern.quote(searchText) + "\\b")
"(\\b|_|\\.)" + Pattern.quote(searchText) + "(\\b|_|\\.)")
: Pattern.compile(Pattern.quote(searchText)); : Pattern.compile(Pattern.quote(searchText));
} }
Matcher matcher = pattern.matcher(content); Matcher matcher = pattern.matcher(content);
while (matcher.find()) { while (matcher.find()) {
indexes.add(matcher.start()); matches.add(new MatchInfo(matcher.start(), matcher.end() - matcher.start()));
} }
return indexes; return matches;
} }
@Override @Override
protected void writeString(String text, List<TextPosition> textPositions) { protected void writeString(String text, List<TextPosition> textPositions) {
for (Integer index : findOccurrencesInText(searchText, text.toLowerCase())) { for (MatchInfo match : findOccurrencesInText(searchText, text.toLowerCase())) {
if (index + searchText.length() <= textPositions.size()) { int index = match.startIndex;
if (index + match.matchLength <= textPositions.size()) {
// Initial values based on the first character // Initial values based on the first character
TextPosition first = textPositions.get(index); TextPosition first = textPositions.get(index);
float minX = first.getX(); float minX = first.getX();
@ -65,7 +76,7 @@ public class TextFinder extends PDFTextStripper {
float maxY = first.getY() + first.getHeight(); float maxY = first.getY() + first.getHeight();
// Loop over the rest of the characters and adjust bounding box values // Loop over the rest of the characters and adjust bounding box values
for (int i = index; i < index + searchText.length(); i++) { for (int i = index; i < index + match.matchLength; i++) {
TextPosition position = textPositions.get(i); TextPosition position = textPositions.get(i);
minX = Math.min(minX, position.getX()); minX = Math.min(minX, position.getX());
minY = Math.min(minY, position.getY()); minY = Math.min(minY, position.getY());