Mirrors/Stirling-PDF
Mirrors/Stirling-PDF
1
0
Fork 0
mirror of https://github.com/Frooodle/Stirling-PDF.git synced 2025-09-08 17:51:20 +02:00
Code Issues Packages Projects Releases Wiki Activity

improve code formatting and enhance readability in auto-redact.html, RedactionService, TextDecodingHelper, TextEncodingHelper, and TextFinder

Browse Source 
Signed-off-by: Balázs Szücs <bszucs1209@gmail.com>
This commit is contained in:
Balázs Szücs 2025-09-03 00:16:30 +02:00
parent ebe17f4c93
commit e64bbebfd5
5 changed files with 344 additions and 468 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

37
app/core/src/main/java/stirling/software/SPDF/pdf/TextFinder.java
View File

@ -3,8 +3,6 @@ package stirling.software.SPDF.pdf;
import java.io.IOException; import java.io.IOException;
import java.util.ArrayList; import java.util.ArrayList;
import java.util.List; import java.util.List;
import java.util.regex.Matcher;
import java.util.regex.Pattern;
import org.apache.pdfbox.pdmodel.PDPage; import org.apache.pdfbox.pdmodel.PDPage;
import org.apache.pdfbox.text.PDFTextStripper; import org.apache.pdfbox.text.PDFTextStripper;
@ -71,26 +69,33 @@ public class TextFinder extends PDFTextStripper {
super.endPage(page); super.endPage(page);
return; return;
} }
String regex = this.useRegex ? processedSearchTerm : "\\Q" + processedSearchTerm + "\\E"; // Build patterns using unified utility for consistency
if (this.wholeWordSearch) { List<java.util.regex.Pattern> patterns =
if (processedSearchTerm.length() == 1 stirling.software.SPDF.utils.text.TextFinderUtils.createOptimizedSearchPatterns(
&& Character.isDigit(processedSearchTerm.charAt(0))) { java.util.Collections.singleton(processedSearchTerm),
regex = "(?<![\\w])(?<!\\d[\\.,])" + regex + "(?![\\w])(?![\\.,]\\d)"; this.useRegex,
} else if (processedSearchTerm.length() == 1) { this.wholeWordSearch);
regex = "(?<![\\w])" + regex + "(?![\\w])"; java.util.regex.Matcher matcher = null;
} else { java.util.regex.Pattern activePattern = null;
regex = "\\b" + regex + "\\b"; for (java.util.regex.Pattern p : patterns) {
matcher = p.matcher(text);
if (matcher
.find()) { // prime by checking has at least one match; we will re-iterate below
activePattern = p;
break;
} }
} }
if (activePattern == null) {
Pattern pattern = Pattern.compile(regex, Pattern.CASE_INSENSITIVE | Pattern.UNICODE_CASE); super.endPage(page);
Matcher matcher = pattern.matcher(text); return;
}
matcher = activePattern.matcher(text);
log.debug( log.debug(
"Searching for '{}' in page {} with regex '{}' (wholeWord: {}, useRegex: {})", "Searching for '{}' in page {} with pattern '{}' (wholeWord: {}, useRegex: {})",
processedSearchTerm, processedSearchTerm,
getCurrentPageNo(), getCurrentPageNo(),
regex, activePattern,
wholeWordSearch, wholeWordSearch,
useRegex); useRegex);

19
app/core/src/main/java/stirling/software/SPDF/service/RedactionService.java
View File

@ -86,7 +86,7 @@ public class RedactionService {
private static final Set<String> TEXT_SHOWING_OPERATORS = Set.of("Tj", "TJ", "'", "\""); private static final Set<String> TEXT_SHOWING_OPERATORS = Set.of("Tj", "TJ", "'", "\"");
private static final COSString EMPTY_COS_STRING = new COSString(""); private static final COSString EMPTY_COS_STRING = new COSString("");
private static final int MAX_SWEEPS = 3; private static final int MAX_SWEEPS = 3;
private static final Pattern PATTERN = Pattern.compile(".*(hoepap|temp|generated).*"); private static final Pattern PATTERN = Pattern.compile(".*(placeholder|temp|generated).*");
private boolean aggressiveMode = false; private boolean aggressiveMode = false;
private Map<Integer, List<AggressiveSegMatch>> aggressiveSegMatches = null; private Map<Integer, List<AggressiveSegMatch>> aggressiveSegMatches = null;
private final CustomPDFDocumentFactory pdfDocumentFactory; private final CustomPDFDocumentFactory pdfDocumentFactory;
@ -2409,11 +2409,9 @@ public class RedactionService {
textSegments.indexOf(task.segment), Collections.emptyList()); textSegments.indexOf(task.segment), Collections.emptyList());
if (task.segment.tokenIndex >= newTokens.size()) { if (task.segment.tokenIndex >= newTokens.size()) {
continue; continue;
} }
if (task.segment.getText() == null || task.segment.getText().isEmpty()) { if (task.segment.getText() == null || task.segment.getText().isEmpty()) {
continue; continue;
} }
@ -2430,19 +2428,22 @@ public class RedactionService {
private static String extractStringWithFallbacks(COSString cosString, PDFont font) { private static String extractStringWithFallbacks(COSString cosString, PDFont font) {
if (cosString == null) return ""; if (cosString == null) return "";
try { try {
// Prefer font-guided decoding for correctness
if (font != null) {
String enhanced =
TextDecodingHelper.decodeCharactersEnhanced(font, cosString.getBytes());
if (enhanced != null && !isGibberish(enhanced)) return enhanced;
}
// Fallback to COSString raw string if it seems valid
String text = cosString.getString(); String text = cosString.getString();
if (!text.trim().isEmpty() && !isGibberish(text)) return text; if (!text.trim().isEmpty() && !isGibberish(text)) return text;
// Fallback: try basic font-based extraction
if (font != null) { if (font != null) {
String fontBasedText = tryFontBasedExtraction(cosString, font); String fontBasedText = tryFontBasedExtraction(cosString, font);
if (fontBasedText != null && !isGibberish(fontBasedText)) return fontBasedText; if (fontBasedText != null && !isGibberish(fontBasedText)) return fontBasedText;
} }
// Last resort: sanitized raw
String encodingFallback = tryEncodingFallbacks(cosString);
if (encodingFallback != null && !isGibberish(encodingFallback)) return encodingFallback;
return sanitizeText(text); return sanitizeText(text);
} catch (Exception e) { } catch (Exception e) {
return "\uFFFD"; return "\uFFFD";

20
app/core/src/main/java/stirling/software/SPDF/utils/text/TextDecodingHelper.java
View File

@ -123,9 +123,8 @@ public class TextDecodingHelper {
} }
} catch (Exception ignored) { } catch (Exception ignored) {
} }
if (ch == null || !isPrintable(ch)) { if (ch == null) {
// Handle problematic character codes specifically return null; // fail fast if undecodable via font tables
ch = "<EFBFBD>";
} }
out.append(ch); out.append(ch);
i += consumed; i += consumed;
@ -250,16 +249,8 @@ public class TextDecodingHelper {
} }
public String handleProblematicCharacterCode(int code, PDFont font) { public String handleProblematicCharacterCode(int code, PDFont font) {
if (code >= PROBLEMATIC_CODE_LOWER_BOUND && code <= PROBLEMATIC_CODE_UPPER_BOUND) { // For correctness, avoid speculative remapping. Return replacement char only when needed.
int adjustedCode = code - PROBLEMATIC_CODE_LOWER_BOUND; return "\uFFFD";
if (adjustedCode >= ASCII_LOWER_BOUND) {
return String.valueOf((char) adjustedCode);
}
if (font != null && font.getName() != null && font.getName().contains("+")) {
return mapSubsetCharacter(adjustedCode);
}
}
return "<EFBFBD>";
} }
public String mapSubsetCharacter(int code) { public String mapSubsetCharacter(int code) {
@ -267,7 +258,8 @@ public class TextDecodingHelper {
return String.valueOf((char) code); return String.valueOf((char) code);
} }
if (code >= EXTENDED_ASCII_LOWER_BOUND && code <= EXTENDED_ASCII_UPPER_BOUND) { if (code >= EXTENDED_ASCII_LOWER_BOUND && code <= EXTENDED_ASCII_UPPER_BOUND) {
return String.valueOf((char) (code - 128)); // Do not alter code point arbitrarily; extended ASCII maps directly for correctness.
return String.valueOf((char) code);
} }
return null; return null;
} }

205
app/core/src/main/java/stirling/software/SPDF/utils/text/TextEncodingHelper.java
View File

@ -20,80 +20,44 @@ public class TextEncodingHelper {
if (font == null || text == null) { if (font == null || text == null) {
return false; return false;
} }
if (text.isEmpty()) { if (text.isEmpty()) {
return true; return true;
} }
// Strict: every code point must be encodable by the font
try { for (int i = 0; i < text.length(); ) {
byte[] encoded = font.encode(text); int cp = text.codePointAt(i);
if (encoded.length > 0) { String ch = new String(Character.toChars(cp));
return true; try {
byte[] encoded = font.encode(ch);
if (encoded == null || encoded.length == 0) {
return false;
}
} catch (Exception ex) {
return false;
} }
} catch (Exception e) { i += Character.charCount(cp);
} }
return true;
return validateAsCodePointArray(font, text);
} }
private boolean validateAsCodePointArray(PDFont font, String text) { private boolean validateAsCodePointArray(PDFont font, String text) {
if (text == null || text.isEmpty()) { if (text == null || text.isEmpty()) {
return true; return true;
} }
int totalCodePoints = 0;
int successfulCodePoints = 0;
for (int i = 0; i < text.length(); ) { for (int i = 0; i < text.length(); ) {
int codePoint = text.codePointAt(i); int codePoint = text.codePointAt(i);
String charStr = new String(Character.toChars(codePoint)); String charStr = new String(Character.toChars(codePoint));
totalCodePoints++;
try { try {
byte[] charEncoded = font.encode(charStr); byte[] charEncoded = font.encode(charStr);
if (charEncoded.length > 0) { if (charEncoded == null || charEncoded.length == 0) {
try { return false;
float charWidth = font.getStringWidth(charStr);
if (charWidth >= 0) {
successfulCodePoints++;
}
} catch (Exception e) {
try {
if (canDecodeCharacter(font, charStr)) {
successfulCodePoints++;
}
} catch (Exception e2) {
}
}
} else {
try {
if (canDecodeCharacter(font, charStr)) {
successfulCodePoints++;
}
} catch (Exception e) {
}
} }
} catch (Exception e) { } catch (Exception e) {
try { return false;
if (canDecodeCharacter(font, charStr)) {
successfulCodePoints++;
}
} catch (Exception e2) {
if (isBasicCharacter(codePoint)) {
successfulCodePoints++;
}
}
} }
i += Character.charCount(codePoint); i += Character.charCount(codePoint);
} }
return true;
if (totalCodePoints == 0) {
return true;
}
double successRate = (double) successfulCodePoints / totalCodePoints;
return successRate >= 0.1;
} }
private boolean canDecodeCharacter(PDFont font, String charStr) { private boolean canDecodeCharacter(PDFont font, String charStr) {
@ -128,26 +92,17 @@ public class TextEncodingHelper {
if (font == null || text == null) { if (font == null || text == null) {
return false; return false;
} }
if (text.isEmpty()) { if (text.isEmpty()) {
return true; return true;
} }
// Strict: removable only if we can encode every codepoint and measure width
if (isSimpleCharacter(text)) { if (!canEncodeCharacters(font, text)) return false;
try { try {
font.encode(text); font.getStringWidth(text);
font.getStringWidth(text); return true;
return true; } catch (Exception e) {
} catch (Exception e) { return false;
try {
return canHandleText(font, text);
} catch (Exception e2) {
return false;
}
}
} }
return isTextFullyRemovable(font, text);
} }
private boolean canHandleText(PDFont font, String text) { private boolean canHandleText(PDFont font, String text) {
@ -197,68 +152,14 @@ public class TextEncodingHelper {
} }
public boolean isTextFullyRemovable(PDFont font, String text) { public boolean isTextFullyRemovable(PDFont font, String text) {
if (font == null || text == null) { if (font == null || text == null) return false;
return false; if (text.isEmpty()) return true;
} if (!canEncodeCharacters(font, text)) return false;
if (text.isEmpty()) {
return true;
}
try { try {
if (!canEncodeCharacters(font, text)) { float width = font.getStringWidth(text);
return false; return width >= 0;
}
try {
float width = font.getStringWidth(text);
if (width < 0) {
return false;
}
} catch (Exception e) {
try {
if (!canCalculateTextWidth(font, text)) {
return false;
}
} catch (Exception e2) {
return false;
}
}
try {
if (font.getFontDescriptor() == null) {
try {
return canHandleWithoutDescriptor(font, text);
} catch (Exception e) {
return false;
}
}
} catch (Exception e) {
try {
return canHandleWithoutDescriptor(font, text);
} catch (Exception e2) {
return false;
}
}
try {
font.getFontDescriptor().getFontBoundingBox();
} catch (Exception e) {
try {
return canHandleWithoutBoundingBox(font, text);
} catch (Exception e2) {
return false;
}
}
return true;
} catch (Exception e) { } catch (Exception e) {
try { return false;
return canHandleText(font, text);
} catch (Exception e2) {
return false;
}
} }
} }
@ -381,45 +282,19 @@ public class TextEncodingHelper {
} }
public boolean fontSupportsCharacter(PDFont font, String character) { public boolean fontSupportsCharacter(PDFont font, String character) {
if (font == null || character == null) { if (font == null || character == null) return false;
return false; if (character.isEmpty()) return true;
}
if (character.isEmpty()) {
return true;
}
try {
byte[] encoded = font.encode(character);
if (encoded.length > 0) {
try {
float width = font.getStringWidth(character);
if (width >= 0) {
return true;
}
} catch (Exception e) {
}
return true;
}
} catch (Exception e) {
}
try {
if (canDecodeCharacter(font, character)) {
return true;
}
} catch (Exception e) {
}
for (int i = 0; i < character.length(); ) { for (int i = 0; i < character.length(); ) {
int codePoint = character.codePointAt(i); int cp = character.codePointAt(i);
if (isBasicCharacter(codePoint)) { String ch = new String(Character.toChars(cp));
i += Character.charCount(codePoint); try {
continue; byte[] encoded = font.encode(ch);
if (encoded == null || encoded.length == 0) return false;
} catch (Exception e) {
return false;
} }
return false; i += Character.charCount(cp);
} }
return true; return true;
} }

529
app/core/src/main/resources/templates/security/auto-redact.html
View File

@ -3,321 +3,324 @@
xmlns:th="https://www.thymeleaf.org"> xmlns:th="https://www.thymeleaf.org">
<head> <head>
<th:block th:insert="~{fragments/common :: head(title=#{autoRedact.title}, header=#{autoRedact.header})}"></th:block> <th:block th:insert="~{fragments/common :: head(title=#{autoRedact.title}, header=#{autoRedact.header})}"></th:block>
<style> <style>
.redaction-options-group { .redaction-options-group {
margin-bottom: 1rem; margin-bottom: 1rem;
} }
.form-text.text-muted { .form-text.text-muted {
color: #6c757d !important; color: #6c757d !important;
} }
.btn-primary:focus, .form-check-input:focus, .form-control:focus, .form-select:focus { .btn-primary:focus, .form-check-input:focus, .form-control:focus, .form-select:focus {
box-shadow: 0 0 0 0.2rem rgba(13, 110, 253, 0.25); box-shadow: 0 0 0 0.2rem rgba(13, 110, 253, 0.25);
outline: 2px solid #0d6efd; outline: 2px solid #0d6efd;
outline-offset: 2px; outline-offset: 2px;
} }
.form-check-input:checked { .form-check-input:checked {
background-color: #0d6efd; background-color: #0d6efd;
border-color: #0d6efd; border-color: #0d6efd;
} }
/* OCR language list styling */ /* OCR language list styling */
#languages { #languages {
max-height: 400px; max-height: 400px;
overflow-y: auto; overflow-y: auto;
border: 1px solid var(--md-sys-color-surface-3); border: 1px solid var(--md-sys-color-surface-3);
border-radius: 5px; border-radius: 5px;
padding: 10px; padding: 10px;
} }
/* Better visibility for selected redaction option */ /* Better visibility for selected redaction option */
.redaction-options-group .form-check { .redaction-options-group .form-check {
border: 1px solid var(--md-sys-color-surface-3); border: 1px solid var(--md-sys-color-surface-3);
border-radius: 8px; border-radius: 8px;
padding: 10px 12px; padding: 10px 12px;
transition: border-color .15s ease, background-color .15s ease, box-shadow .15s ease; transition: border-color .15s ease, background-color .15s ease, box-shadow .15s ease;
} }
.redaction-options-group .form-check + .form-check { margin-top: .5rem; } .redaction-options-group .form-check + .form-check { margin-top: .5rem; }
.redaction-options-group .form-check:hover { .redaction-options-group .form-check:hover {
background-color: var(--md-sys-color-surface-1); background-color: var(--md-sys-color-surface-1);
} }
.redaction-options-group .form-check.selected { .redaction-options-group .form-check.selected {
border-color: #0d6efd; border-color: #0d6efd;
background-color: rgba(13,110,253,0.06); background-color: rgba(13,110,253,0.06);
box-shadow: 0 0 0 2px rgba(13,110,253,0.1) inset; box-shadow: 0 0 0 2px rgba(13,110,253,0.1) inset;
} }
.redaction-options-group .form-check .form-check-label { .redaction-options-group .form-check .form-check-label {
font-weight: 600; font-weight: 600;
} }
.redaction-options-group small.form-text { .redaction-options-group small.form-text {
margin-left: 1.8rem; /* align with radio */ margin-left: 1.8rem; /* align with radio */
} }
</style> </style>
</head> </head>
<body> <body>
<div id="page-container"> <div id="page-container">
<div id="content-wrap"> <div id="content-wrap">
<th:block th:insert="~{fragments/navbar.html :: navbar}"></th:block> <th:block th:insert="~{fragments/navbar.html :: navbar}"></th:block>
<br><br> <br><br>
<div class="container"> <div class="container">
<div class="row justify-content-center"> <div class="row justify-content-center">
<div class="col-md-6 bg-card"> <div class="col-md-6 bg-card">
<div class="tool-header"> <div class="tool-header">
<svg class="material-symbols-rounded tool-header-icon security"> <svg class="material-symbols-rounded tool-header-icon security">
<use xlink:href="/images/redact-auto.svg#icon-redact-auto"></use> <use xlink:href="/images/redact-auto.svg#icon-redact-auto"></use>
</svg> </svg>
<span class="tool-header-text" id="form-title" th:text="#{autoRedact.header}"></span> <span class="tool-header-text" id="form-title" th:text="#{autoRedact.header}"></span>
</div> </div>
<form aria-labelledby="form-title" enctype="multipart/form-data" id="autoRedactForm" <form aria-labelledby="form-title" enctype="multipart/form-data" id="autoRedactForm"
method="post" th:action="@{'api/v1/security/auto-redact'}"> method="post" th:action="@{'api/v1/security/auto-redact'}">
<div th:replace="~{fragments/common :: fileSelector(name='fileInput', multipleInputsForSingleRequest=false, disableMultipleFiles=true, accept='application/pdf')}"></div> <div th:replace="~{fragments/common :: fileSelector(name='fileInput', multipleInputsForSingleRequest=false, disableMultipleFiles=true, accept='application/pdf')}"></div>
<div class="mb-3"> <div class="mb-3">
<label class="form-label" for="listOfText" th:text="#{autoRedact.textsToRedactLabel}"></label> <label class="form-label" for="listOfText" th:text="#{autoRedact.textsToRedactLabel}"></label>
<textarea class="form-control" id="listOfText" name="listOfText" required rows="4" <textarea class="form-control" id="listOfText" name="listOfText" required rows="4"
th:placeholder="#{autoRedact.textsToRedactPlaceholder}"></textarea> th:placeholder="#{autoRedact.textsToRedactPlaceholder}"></textarea>
</div> </div>
<div class="mb-3"> <div class="mb-3">
<div class="form-check"> <div class="form-check">
<input class="form-check-input" id="useRegex" name="useRegex" type="checkbox"> <input class="form-check-input" id="useRegex" name="useRegex" type="checkbox">
<label class="form-check-label" for="useRegex" th:text="#{autoRedact.useRegexLabel}"></label> <label class="form-check-label" for="useRegex" th:text="#{autoRedact.useRegexLabel}"></label>
</div> </div>
<div class="form-check"> <div class="form-check">
<input class="form-check-input" id="wholeWordSearch" name="wholeWordSearch" type="checkbox"> <input class="form-check-input" id="wholeWordSearch" name="wholeWordSearch" type="checkbox">
<label class="form-check-label" for="wholeWordSearch" th:text="#{autoRedact.wholeWordSearchLabel}"></label> <label class="form-check-label" for="wholeWordSearch" th:text="#{autoRedact.wholeWordSearchLabel}"></label>
</div> </div>
</div> </div>
<div class="redaction-options-group"> <div class="redaction-options-group">
<label class="form-label fw-bold mb-3" th:text="#{autoRedact.redactionStyleLabel}"></label> <label class="form-label fw-bold mb-3" th:text="#{autoRedact.redactionStyleLabel}"></label>
<div class="form-check mb-2"> <div class="form-check mb-2">
<input aria-describedby="visual-desc" checked class="form-check-input" id="visualImage" name="redactionMode" type="radio" value="visual"> <input aria-describedby="visual-desc" checked class="form-check-input" id="visualImage" name="redactionMode" type="radio" value="visual">
<label class="form-check-label" for="visualImage" th:text="#{autoRedact.visualRedactionLabel}">Visual</label> <label class="form-check-label" for="visualImage" th:text="#{autoRedact.visualRedactionLabel}">Visual</label>
<small class="form-text text-muted d-block mt-1" id="visual-desc" th:text="#{autoRedact.visualRedactionDescription}">Converts to image with visual redactions for maximum security.</small> <small class="form-text text-muted d-block mt-1" id="visual-desc" th:text="#{autoRedact.visualRedactionDescription}">Converts to image with visual redactions for maximum security.</small>
</div> </div>
<div class="form-check mb-2"> <div class="form-check mb-2">
<input aria-describedby="delete-desc" class="form-check-input" id="deleteText" name="redactionMode" type="radio" value="aggressive"> <input aria-describedby="delete-desc" class="form-check-input" id="deleteText" name="redactionMode" type="radio" value="aggressive">
<label class="form-check-label" for="deleteText" th:text="#{autoRedact.deleteTextLabel}">Remove Text</label> <label class="form-check-label" for="deleteText" th:text="#{autoRedact.deleteTextLabel}">Remove Text</label>
<small class="form-text text-muted d-block mt-1" id="delete-desc" th:text="#{autoRedact.deleteTextDescription}">Removes text completely, allowing the surrounding content to shift. This may change the document's original appearance.</small> <small class="form-text text-muted d-block mt-1" id="delete-desc" th:text="#{autoRedact.deleteTextDescription}">Removes text completely, allowing the surrounding content to shift. This may change the document's original appearance.</small>
</div> </div>
<div class="form-check mb-3"> <div class="form-check mb-3">
<input aria-describedby="keep-desc" class="form-check-input" id="keepLayout" name="redactionMode" type="radio" value="moderate"> <input aria-describedby="keep-desc" class="form-check-input" id="keepLayout" name="redactionMode" type="radio" value="moderate">
<label class="form-check-label" for="keepLayout" th:text="#{autoRedact.keepLayoutLabel}">Remove Text & Cover (Preserve Layout)</label> <label class="form-check-label" for="keepLayout" th:text="#{autoRedact.keepLayoutLabel}">Remove Text & Cover (Preserve Layout)</label>
<small class="form-text text-muted d-block mt-1" id="keep-desc" th:text="#{autoRedact.keepLayoutDescription}">Removes the underlying text and places a redaction box in its place, preserving the document's original layout.</small> <small class="form-text text-muted d-block mt-1" id="keep-desc" th:text="#{autoRedact.keepLayoutDescription}">Removes the underlying text and places a redaction box in its place, preserving the document's original layout.</small>
</div> </div>
</div>
<div class="form-check"> <!-- PDF Image checkbox - moved outside redaction-options-group to use plain styling -->
<input aria-describedby="guarantee-desc" class="form-check-input" id="guaranteeRedaction" name="convertPDFToImage" type="checkbox"> <div class="mb-3">
<label class="form-check-label" for="guaranteeRedaction" th:text="#{autoRedact.pdfImageLabel}">PDF image</label> <div class="form-check">
<small class="form-text text-muted d-block mt-1" id="guarantee-desc" th:text="#{autoRedact.pdfImageDescription}">For maximum security, uses an image-based method to ensure text is unrecoverable. May slightly affect document quality.</small> <input aria-describedby="guarantee-desc" class="form-check-input" id="convertPDFToImage" name="convertPDFToImage" type="checkbox">
</div> <label class="form-check-label" for="convertPDFToImage" th:text="#{autoRedact.convertPDFToImageLabel}">Convert PDF to PDF-Image</label>
</div> </div>
<small class="form-text text-muted d-block mt-1" id="guarantee-desc" th:text="#{autoRedact.pdfImageDescription}">For maximum security, uses an image-based method to ensure text is unrecoverable. May slightly affect document quality.</small>
</div>
<br> <br>
<div class="mb-3"> <div class="mb-3">
<label class="form-label" for="defaultColor" th:text="#{autoRedact.colorLabel}"></label> <label class="form-label" for="defaultColor" th:text="#{autoRedact.colorLabel}"></label>
<select class="form-select" id="defaultColor" name="defaultColor" onchange="handleColorChange(this.value)"> <select class="form-select" id="defaultColor" name="defaultColor" onchange="handleColorChange(this.value)">
<option th:text="#{black}" value="#000000">Black</option> <option th:text="#{black}" value="#000000">Black</option>
<option th:text="#{white}" value="#FFFFFF">White</option> <option th:text="#{white}" value="#FFFFFF">White</option>
<option th:text="#{red}" value="#FF0000">Red</option> <option th:text="#{red}" value="#FF0000">Red</option>
<option th:text="#{green}" value="#00FF00">Green</option> <option th:text="#{green}" value="#00FF00">Green</option>
<option th:text="#{blue}" value="#0000FF">Blue</option> <option th:text="#{blue}" value="#0000FF">Blue</option>
<option th:text="#{custom}" value="custom">Custom...</option> <option th:text="#{custom}" value="custom">Custom...</option>
</select> </select>
</div> </div>
<div class="mb-3" id="customColorContainer" style="display: none;"> <div class="mb-3" id="customColorContainer" style="display: none;">
<label class="form-label" for="customColor">Custom Color (Hex)</label> <label class="form-label" for="customColor">Custom Color (Hex)</label>
<input class="form-control" id="customColor" name="redactColor" placeholder="#FF00FF" type="text"> <input class="form-control" id="customColor" name="redactColor" placeholder="#FF00FF" type="text">
</div> </div>
<div class="mb-3"> <div class="mb-3">
<label class="form-label" for="customPadding" th:text="#{autoRedact.customPaddingLabel}"></label> <label class="form-label" for="customPadding" th:text="#{autoRedact.customPaddingLabel}"></label>
<input class="form-control" id="customPadding" max="1" min="0" name="customPadding" <input class="form-control" id="customPadding" max="1" min="0" name="customPadding"
step="0.1" type="number" value="0.1"> step="0.1" type="number" value="0.1">
</div> </div>
<br> <br>
<div class="mb-3" th:if="${#lists.size(languages) > 0}"> <div class="mb-3" th:if="${#lists.size(languages) > 0}">
<label class="form-label" for="languages">OCR Languages</label> <label class="form-label" for="languages">OCR Languages</label>
<div id="languages"> <div id="languages">
<div class="form-check" th:each="language, iterStat : ${languages}"> <div class="form-check" th:each="language, iterStat : ${languages}">
<input onchange="handleLangSelection()" required th:checked="${language == 'eng'}" th:id="${'language-' + language}" th:name="languages" th:value="${language}" type="checkbox" /> <input class="form-check-input" onchange="handleLangSelection()" required th:checked="${language == 'eng'}" th:id="${'language-' + language}" th:name="languages" th:value="${language}" type="checkbox" />
<label th:attr="data-lang-code=${language}, data-lang-name=#{'lang.' + language}" <label class="form-check-label" th:attr="data-lang-code=${language}, data-lang-name=#{'lang.' + language}"
th:for="${'language-' + language}" th:for="${'language-' + language}"
th:text="${language}"></label> th:text="${language}"></label>
</div>
</div>
<small class="form-text text-muted" id="ocr-desc">Used when OCR restoration is needed</small>
</div>
<input id="aggressiveMode" name="aggressiveMode" type="hidden" value="false">
<div class="mb-3 text-center">
<button class="btn btn-primary" id="submitBtn" th:text="#{autoRedact.submitButton}" type="submit"></button>
</div>
</form>
</div> </div>
</div>
<small class="form-text text-muted" id="ocr-desc">Used when OCR restoration is needed</small>
</div> </div>
<input id="aggressiveMode" name="aggressiveMode" type="hidden" value="false">
<div class="mb-3 text-center">
<button class="btn btn-primary" id="submitBtn" th:text="#{autoRedact.submitButton}" type="submit"></button>
</div>
</form>
</div> </div>
</div>
</div> </div>
<th:block th:insert="~{fragments/footer.html :: footer}"></th:block> </div>
<th:block th:insert="~{fragments/footer.html :: footer}"></th:block>
</div> </div>
<script th:inline="javascript"> <script th:inline="javascript">
function handleColorChange(selectedValue) { function handleColorChange(selectedValue) {
const container = document.getElementById('customColorContainer'); const container = document.getElementById('customColorContainer');
const input = document.getElementById('customColor'); const input = document.getElementById('customColor');
if (selectedValue === "custom") { if (selectedValue === "custom") {
container.style.display = 'block'; container.style.display = 'block';
if (!input.value) { if (!input.value) {
input.value = '#000000'; input.value = '#000000';
} }
} else { } else {
container.style.display = 'none'; container.style.display = 'none';
input.value = selectedValue; input.value = selectedValue;
}
} }
}
function handleLangSelection() { function handleLangSelection() {
let checkboxes = document.getElementsByName("languages"); let checkboxes = document.getElementsByName("languages");
let selected = false; let selected = false;
for (let i = 0; i < checkboxes.length; i++) { for (let i = 0; i < checkboxes.length; i++) {
if (checkboxes[i].checked) { if (checkboxes[i].checked) {
selected = true; selected = true;
checkboxes[i].setAttribute('required', 'false'); checkboxes[i].setAttribute('required', 'false');
} }
} }
if (selected) { if (selected) {
for (let i = 0; i < checkboxes.length; i++) { for (let i = 0; i < checkboxes.length; i++) {
checkboxes[i].removeAttribute('required'); checkboxes[i].removeAttribute('required');
} }
} else { } else {
for (let i = 0; i < checkboxes.length; i++) { for (let i = 0; i < checkboxes.length; i++) {
checkboxes[i].setAttribute('required', 'true'); checkboxes[i].setAttribute('required', 'true');
} }
} }
} }
// Language translations map populated by Thymeleaf for available OCR languages // Language translations map populated by Thymeleaf for available OCR languages
const languageTranslations = {}; const languageTranslations = {};
/*[# th:each="lang : ${languages}"]*/ /*[# th:each="lang : ${languages}"]*/
languageTranslations['[(${lang})]'] = /*[[#{${'lang.' + lang}}]]*/'[(${lang})]'; languageTranslations['[(${lang})]'] = /*[[#{${'lang.' + lang}}]]*/'[(${lang})]';
/*[/]*/ /*[/]*/
const localeToTesseract = { const localeToTesseract = {
'en': 'eng', 'fr': 'fra', 'de': 'deu', 'es': 'spa', 'it': 'ita', 'pt': 'por', 'ru': 'rus', 'en': 'eng', 'fr': 'fra', 'de': 'deu', 'es': 'spa', 'it': 'ita', 'pt': 'por', 'ru': 'rus',
'zh': 'chi_sim', 'ja': 'jpn', 'ko': 'kor', 'ar': 'ara', 'hi': 'hin', 'nl': 'nld', 'cs': 'ces', 'zh': 'chi_sim', 'ja': 'jpn', 'ko': 'kor', 'ar': 'ara', 'hi': 'hin', 'nl': 'nld', 'cs': 'ces',
'pl': 'pol', 'tr': 'tur', 'uk': 'ukr', 'vi': 'vie', 'sv': 'swe', 'no': 'nor', 'fi': 'fin', 'pl': 'pol', 'tr': 'tur', 'uk': 'ukr', 'vi': 'vie', 'sv': 'swe', 'no': 'nor', 'fi': 'fin',
'da': 'dan', 'el': 'ell', 'he': 'heb', 'hu': 'hun', 'bg': 'bul', 'ro': 'ron', 'hr': 'hrv', 'da': 'dan', 'el': 'ell', 'he': 'heb', 'hu': 'hun', 'bg': 'bul', 'ro': 'ron', 'hr': 'hrv',
'sk': 'slk', 'id': 'ind', 'th': 'tha', 'sl': 'slv' 'sk': 'slk', 'id': 'ind', 'th': 'tha', 'sl': 'slv'
}; };
function getTranslatedLanguageName(shortCode) { function getTranslatedLanguageName(shortCode) {
// Use Thymeleaf-provided map; fall back to code when translation missing // Use Thymeleaf-provided map; fall back to code when translation missing
const name = languageTranslations[shortCode]; const name = languageTranslations[shortCode];
if (name && !/^\?{2,}.+\?{2,}$/.test(name)) return name; if (name && !/^\?{2,}.+\?{2,}$/.test(name)) return name;
return shortCode; return shortCode;
} }
function prioritizeLanguages() { function prioritizeLanguages() {
const languageContainer = document.getElementById('languages'); const languageContainer = document.getElementById('languages');
if (!languageContainer) return; if (!languageContainer) return;
const formChecks = Array.from(languageContainer.getElementsByClassName('form-check')); const formChecks = Array.from(languageContainer.getElementsByClassName('form-check'));
if (formChecks.length === 0) return; if (formChecks.length === 0) return;
formChecks.forEach(element => { formChecks.forEach(element => {
const label = element.querySelector('label'); const label = element.querySelector('label');
if (label) { if (label) {
const langCode = label.getAttribute('for').split('-')[1]; const langCode = label.getAttribute('for').split('-')[1];
// Always set from translations map; gracefully falls back to code // Always set from translations map; gracefully falls back to code
label.textContent = getTranslatedLanguageName(langCode); label.textContent = getTranslatedLanguageName(langCode);
} }
}); });
const browserLanguage = document.documentElement.lang || navigator.language || navigator.userLanguage; const browserLanguage = document.documentElement.lang || navigator.language || navigator.userLanguage;
const uiLanguage = document.documentElement.getAttribute('data-language') || browserLanguage; const uiLanguage = document.documentElement.getAttribute('data-language') || browserLanguage;
const primaryLanguageCode = (uiLanguage || '').split(/[-_]/)[0].toLowerCase(); const primaryLanguageCode = (uiLanguage || '').split(/[-_]/)[0].toLowerCase();
const tesseractPrimaryCode = localeToTesseract[primaryLanguageCode]; const tesseractPrimaryCode = localeToTesseract[primaryLanguageCode];
const priorityLanguages = []; const priorityLanguages = [];
if (tesseractPrimaryCode) priorityLanguages.push(tesseractPrimaryCode); if (tesseractPrimaryCode) priorityLanguages.push(tesseractPrimaryCode);
if (tesseractPrimaryCode !== 'eng') priorityLanguages.push('eng'); if (tesseractPrimaryCode !== 'eng') priorityLanguages.push('eng');
const sortedElements = formChecks.sort((a, b) => { const sortedElements = formChecks.sort((a, b) => {
const aInput = a.querySelector('input'); const aInput = a.querySelector('input');
const bInput = b.querySelector('input'); const bInput = b.querySelector('input');
if (!aInput || !bInput) return 0; if (!aInput || !bInput) return 0;
const aLangCode = aInput.id.split('-')[1]; const aLangCode = aInput.id.split('-')[1];
const bLangCode = bInput.id.split('-')[1]; const bLangCode = bInput.id.split('-')[1];
const aIsPriority = priorityLanguages.includes(aLangCode); const aIsPriority = priorityLanguages.includes(aLangCode);
const bIsPriority = priorityLanguages.includes(bLangCode); const bIsPriority = priorityLanguages.includes(bLangCode);
if (aIsPriority && !bIsPriority) return -1; if (aIsPriority && !bIsPriority) return -1;
if (!aIsPriority && bIsPriority) return 1; if (!aIsPriority && bIsPriority) return 1;
if (aIsPriority && bIsPriority) { if (aIsPriority && bIsPriority) {
return priorityLanguages.indexOf(aLangCode) - priorityLanguages.indexOf(bLangCode); return priorityLanguages.indexOf(aLangCode) - priorityLanguages.indexOf(bLangCode);
} }
return getTranslatedLanguageName(aLangCode).localeCompare(getTranslatedLanguageName(bLangCode)); return getTranslatedLanguageName(aLangCode).localeCompare(getTranslatedLanguageName(bLangCode));
}); });
languageContainer.innerHTML = ''; languageContainer.innerHTML = '';
sortedElements.forEach(element => languageContainer.appendChild(element)); sortedElements.forEach(element => languageContainer.appendChild(element));
} }
document.addEventListener('DOMContentLoaded', function () { document.addEventListener('DOMContentLoaded', function () {
const redactionModeRadios = document.querySelectorAll('input[name="redactionMode"]'); const redactionModeRadios = document.querySelectorAll('input[name="redactionMode"]');
const aggressiveModeHidden = document.getElementById('aggressiveMode'); const aggressiveModeHidden = document.getElementById('aggressiveMode');
const guaranteeRedactionCheckbox = document.getElementById('guaranteeRedaction'); const convertPDFToImageCheckbox = document.getElementById('convertPDFToImage');
const defaultColor = document.getElementById('defaultColor'); const defaultColor = document.getElementById('defaultColor');
function updateMode() { function updateMode() {
const selectedMode = document.querySelector('input[name="redactionMode"]:checked'); const selectedMode = document.querySelector('input[name="redactionMode"]:checked');
if (selectedMode) { if (selectedMode) {
// Set aggressive mode for delete text option // Set aggressive mode for delete text option
aggressiveModeHidden.value = selectedMode.value === 'aggressive' ? 'true' : 'false'; aggressiveModeHidden.value = selectedMode.value === 'aggressive' ? 'true' : 'false';
// Handle PDF image checkbox based on selection // Handle PDF image checkbox based on selection
if (selectedMode.value === 'visual') { if (selectedMode.value === 'visual') {
// Visual mode automatically enables PDF image for maximum security // Visual mode automatically enables PDF image for maximum security
guaranteeRedactionCheckbox.checked = true; convertPDFToImageCheckbox.checked = true;
} else { } else {
// Delete Text and Keep Layout modes disable PDF image // Delete Text and Keep Layout modes disable PDF image
guaranteeRedactionCheckbox.checked = false; convertPDFToImageCheckbox.checked = false;
}
// Highlight selected card
document.querySelectorAll('.redaction-options-group .form-check').forEach(div => div.classList.remove('selected'));
const parent = selectedMode.closest('.form-check');
if (parent) parent.classList.add('selected');
}
} }
redactionModeRadios.forEach(radio => { // Highlight selected card
radio.addEventListener('change', updateMode); document.querySelectorAll('.redaction-options-group .form-check').forEach(div => div.classList.remove('selected'));
}); const parent = selectedMode.closest('.form-check');
if (parent) parent.classList.add('selected');
}
}
if (defaultColor) { redactionModeRadios.forEach(radio => {
handleColorChange(defaultColor.value); radio.addEventListener('change', updateMode);
const customColorInput = document.getElementById('customColor'); });
if (defaultColor.value !== 'custom') {
customColorInput.value = defaultColor.value;
}
}
updateMode(); if (defaultColor) {
handleColorChange(defaultColor.value);
const customColorInput = document.getElementById('customColor');
if (defaultColor.value !== 'custom') {
customColorInput.value = defaultColor.value;
}
}
updateMode();
// Initialize language list ordering & labels // Initialize language list ordering & labels
prioritizeLanguages(); prioritizeLanguages();
// Handle pre-selected English language // Handle pre-selected English language
handleLangSelection(); handleLangSelection();
}); });
</script> </script>
</body> </body>
</html> </html>