From 2fff3083ae9d608397582abbde86cff0e79d123d Mon Sep 17 00:00:00 2001
From: Anthony Stirling <77850077+Frooodle@users.noreply.github.com>
Date: Tue, 26 Mar 2024 19:25:16 +0000
Subject: [PATCH] Update TextFinder.java (#980)
---
.../software/SPDF/pdf/TextFinder.java | 31 +++++++++++++------
1 file changed, 21 insertions(+), 10 deletions(-)
diff --git a/src/main/java/stirling/software/SPDF/pdf/TextFinder.java b/src/main/java/stirling/software/SPDF/pdf/TextFinder.java
index cdfb55017..f9e339c2f 100644
--- a/src/main/java/stirling/software/SPDF/pdf/TextFinder.java
+++ b/src/main/java/stirling/software/SPDF/pdf/TextFinder.java
@@ -19,6 +19,16 @@ public class TextFinder extends PDFTextStripper {
private final boolean wholeWordSearch;
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)
throws IOException {
this.searchText = searchText.toLowerCase();
@@ -27,36 +37,37 @@ public class TextFinder extends PDFTextStripper {
setSortByPosition(true);
}
- private List<Integer> findOccurrencesInText(String searchText, String content) {
- List<Integer> indexes = new ArrayList<>();
+ private List<MatchInfo> findOccurrencesInText(String searchText, String content) {
+ List<MatchInfo> matches = new ArrayList<>();
+
Pattern pattern;
if (useRegex) {
// Use regex-based search
pattern =
wholeWordSearch
- ? Pattern.compile("(\\b|_|\\.)" + searchText + "(\\b|_|\\.)")
+ ? Pattern.compile("\\b" + searchText + "\\b")
: Pattern.compile(searchText);
} else {
// Use normal text search
pattern =
wholeWordSearch
- ? Pattern.compile(
- "(\\b|_|\\.)" + Pattern.quote(searchText) + "(\\b|_|\\.)")
+ ? Pattern.compile("\\b" + Pattern.quote(searchText) + "\\b")
: Pattern.compile(Pattern.quote(searchText));
}
Matcher matcher = pattern.matcher(content);
while (matcher.find()) {
- indexes.add(matcher.start());
+ matches.add(new MatchInfo(matcher.start(), matcher.end() - matcher.start()));
}
- return indexes;
+ return matches;
}
@Override
protected void writeString(String text, List<TextPosition> textPositions) {
- for (Integer index : findOccurrencesInText(searchText, text.toLowerCase())) {
- if (index + searchText.length() <= textPositions.size()) {
+ for (MatchInfo match : findOccurrencesInText(searchText, text.toLowerCase())) {
+ int index = match.startIndex;
+ if (index + match.matchLength <= textPositions.size()) {
// Initial values based on the first character
TextPosition first = textPositions.get(index);
float minX = first.getX();
@@ -65,7 +76,7 @@ public class TextFinder extends PDFTextStripper {
float maxY = first.getY() + first.getHeight();
// 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);
minX = Math.min(minX, position.getX());
minY = Math.min(minY, position.getY());