mirror of
https://github.com/Frooodle/Stirling-PDF.git
synced 2025-09-08 17:51:20 +02:00
improve code formatting and enhance readability in auto-redact.html, RedactionService, TextDecodingHelper, TextEncodingHelper, and TextFinder
Signed-off-by: Balázs Szücs <bszucs1209@gmail.com>
This commit is contained in:
Balázs Szücs
parent
ebe17f4c93
commit
e64bbebfd5
@ -3,8 +3,6 @@ package stirling.software.SPDF.pdf;
|
||||
import java.io.IOException;
|
||||
import java.util.ArrayList;
|
||||
import java.util.List;
|
||||
import java.util.regex.Matcher;
|
||||
import java.util.regex.Pattern;
|
||||
|
||||
import org.apache.pdfbox.pdmodel.PDPage;
|
||||
import org.apache.pdfbox.text.PDFTextStripper;
|
||||
@ -71,26 +69,33 @@ public class TextFinder extends PDFTextStripper {
|
||||
super.endPage(page);
|
||||
return;
|
||||
}
|
||||
String regex = this.useRegex ? processedSearchTerm : "\\Q" + processedSearchTerm + "\\E";
|
||||
if (this.wholeWordSearch) {
|
||||
if (processedSearchTerm.length() == 1
|
||||
&& Character.isDigit(processedSearchTerm.charAt(0))) {
|
||||
regex = "(?<![\\w])(?<!\\d[\\.,])" + regex + "(?![\\w])(?![\\.,]\\d)";
|
||||
} else if (processedSearchTerm.length() == 1) {
|
||||
regex = "(?<![\\w])" + regex + "(?![\\w])";
|
||||
} else {
|
||||
regex = "\\b" + regex + "\\b";
|
||||
// Build patterns using unified utility for consistency
|
||||
List<java.util.regex.Pattern> patterns =
|
||||
stirling.software.SPDF.utils.text.TextFinderUtils.createOptimizedSearchPatterns(
|
||||
java.util.Collections.singleton(processedSearchTerm),
|
||||
this.useRegex,
|
||||
this.wholeWordSearch);
|
||||
java.util.regex.Matcher matcher = null;
|
||||
java.util.regex.Pattern activePattern = null;
|
||||
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;
|
||||
}
|
||||
}
|
||||
|
||||
Pattern pattern = Pattern.compile(regex, Pattern.CASE_INSENSITIVE | Pattern.UNICODE_CASE);
|
||||
Matcher matcher = pattern.matcher(text);
|
||||
if (activePattern == null) {
|
||||
super.endPage(page);
|
||||
return;
|
||||
}
|
||||
matcher = activePattern.matcher(text);
|
||||
|
||||
log.debug(
|
||||
"Searching for '{}' in page {} with regex '{}' (wholeWord: {}, useRegex: {})",
|
||||
"Searching for '{}' in page {} with pattern '{}' (wholeWord: {}, useRegex: {})",
|
||||
processedSearchTerm,
|
||||
getCurrentPageNo(),
|
||||
regex,
|
||||
activePattern,
|
||||
wholeWordSearch,
|
||||
useRegex);
|
||||
|
||||
|
||||
@ -86,7 +86,7 @@ public class RedactionService {
|
||||
private static final Set<String> TEXT_SHOWING_OPERATORS = Set.of("Tj", "TJ", "'", "\"");
|
||||
private static final COSString EMPTY_COS_STRING = new COSString("");
|
||||
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 Map<Integer, List<AggressiveSegMatch>> aggressiveSegMatches = null;
|
||||
private final CustomPDFDocumentFactory pdfDocumentFactory;
|
||||
@ -2409,11 +2409,9 @@ public class RedactionService {
|
||||
textSegments.indexOf(task.segment), Collections.emptyList());
|
||||
|
||||
if (task.segment.tokenIndex >= newTokens.size()) {
|
||||
|
||||
continue;
|
||||
}
|
||||
if (task.segment.getText() == null || task.segment.getText().isEmpty()) {
|
||||
|
||||
continue;
|
||||
}
|
||||
|
||||
@ -2430,19 +2428,22 @@ public class RedactionService {
|
||||
|
||||
private static String extractStringWithFallbacks(COSString cosString, PDFont font) {
|
||||
if (cosString == null) return "";
|
||||
|
||||
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();
|
||||
if (!text.trim().isEmpty() && !isGibberish(text)) return text;
|
||||
|
||||
// Fallback: try basic font-based extraction
|
||||
if (font != null) {
|
||||
String fontBasedText = tryFontBasedExtraction(cosString, font);
|
||||
if (fontBasedText != null && !isGibberish(fontBasedText)) return fontBasedText;
|
||||
}
|
||||
|
||||
String encodingFallback = tryEncodingFallbacks(cosString);
|
||||
if (encodingFallback != null && !isGibberish(encodingFallback)) return encodingFallback;
|
||||
|
||||
// Last resort: sanitized raw
|
||||
return sanitizeText(text);
|
||||
} catch (Exception e) {
|
||||
return "\uFFFD";
|
||||
|
||||
@ -123,9 +123,8 @@ public class TextDecodingHelper {
|
||||
}
|
||||
} catch (Exception ignored) {
|
||||
}
|
||||
if (ch == null || !isPrintable(ch)) {
|
||||
// Handle problematic character codes specifically
|
||||
ch = "<EFBFBD>";
|
||||
if (ch == null) {
|
||||
return null; // fail fast if undecodable via font tables
|
||||
}
|
||||
out.append(ch);
|
||||
i += consumed;
|
||||
@ -250,16 +249,8 @@ public class TextDecodingHelper {
|
||||
}
|
||||
|
||||
public String handleProblematicCharacterCode(int code, PDFont font) {
|
||||
if (code >= PROBLEMATIC_CODE_LOWER_BOUND && code <= PROBLEMATIC_CODE_UPPER_BOUND) {
|
||||
int adjustedCode = code - PROBLEMATIC_CODE_LOWER_BOUND;
|
||||
if (adjustedCode >= ASCII_LOWER_BOUND) {
|
||||
return String.valueOf((char) adjustedCode);
|
||||
}
|
||||
if (font != null && font.getName() != null && font.getName().contains("+")) {
|
||||
return mapSubsetCharacter(adjustedCode);
|
||||
}
|
||||
}
|
||||
return "<EFBFBD>";
|
||||
// For correctness, avoid speculative remapping. Return replacement char only when needed.
|
||||
return "\uFFFD";
|
||||
}
|
||||
|
||||
public String mapSubsetCharacter(int code) {
|
||||
@ -267,7 +258,8 @@ public class TextDecodingHelper {
|
||||
return String.valueOf((char) code);
|
||||
}
|
||||
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;
|
||||
}
|
||||
|
||||
@ -20,80 +20,44 @@ public class TextEncodingHelper {
|
||||
if (font == null || text == null) {
|
||||
return false;
|
||||
}
|
||||
|
||||
if (text.isEmpty()) {
|
||||
return true;
|
||||
}
|
||||
|
||||
try {
|
||||
byte[] encoded = font.encode(text);
|
||||
if (encoded.length > 0) {
|
||||
return true;
|
||||
// Strict: every code point must be encodable by the font
|
||||
for (int i = 0; i < text.length(); ) {
|
||||
int cp = text.codePointAt(i);
|
||||
String ch = new String(Character.toChars(cp));
|
||||
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 validateAsCodePointArray(font, text);
|
||||
return true;
|
||||
}
|
||||
|
||||
private boolean validateAsCodePointArray(PDFont font, String text) {
|
||||
if (text == null || text.isEmpty()) {
|
||||
return true;
|
||||
}
|
||||
|
||||
int totalCodePoints = 0;
|
||||
int successfulCodePoints = 0;
|
||||
|
||||
for (int i = 0; i < text.length(); ) {
|
||||
int codePoint = text.codePointAt(i);
|
||||
String charStr = new String(Character.toChars(codePoint));
|
||||
totalCodePoints++;
|
||||
|
||||
try {
|
||||
byte[] charEncoded = font.encode(charStr);
|
||||
if (charEncoded.length > 0) {
|
||||
try {
|
||||
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) {
|
||||
}
|
||||
if (charEncoded == null || charEncoded.length == 0) {
|
||||
return false;
|
||||
}
|
||||
} catch (Exception e) {
|
||||
try {
|
||||
if (canDecodeCharacter(font, charStr)) {
|
||||
successfulCodePoints++;
|
||||
}
|
||||
} catch (Exception e2) {
|
||||
if (isBasicCharacter(codePoint)) {
|
||||
successfulCodePoints++;
|
||||
}
|
||||
}
|
||||
return false;
|
||||
}
|
||||
|
||||
i += Character.charCount(codePoint);
|
||||
}
|
||||
|
||||
if (totalCodePoints == 0) {
|
||||
return true;
|
||||
}
|
||||
|
||||
double successRate = (double) successfulCodePoints / totalCodePoints;
|
||||
return successRate >= 0.1;
|
||||
return true;
|
||||
}
|
||||
|
||||
private boolean canDecodeCharacter(PDFont font, String charStr) {
|
||||
@ -128,26 +92,17 @@ public class TextEncodingHelper {
|
||||
if (font == null || text == null) {
|
||||
return false;
|
||||
}
|
||||
|
||||
if (text.isEmpty()) {
|
||||
return true;
|
||||
}
|
||||
|
||||
if (isSimpleCharacter(text)) {
|
||||
try {
|
||||
font.encode(text);
|
||||
font.getStringWidth(text);
|
||||
return true;
|
||||
} catch (Exception e) {
|
||||
try {
|
||||
return canHandleText(font, text);
|
||||
} catch (Exception e2) {
|
||||
return false;
|
||||
}
|
||||
}
|
||||
// Strict: removable only if we can encode every codepoint and measure width
|
||||
if (!canEncodeCharacters(font, text)) return false;
|
||||
try {
|
||||
font.getStringWidth(text);
|
||||
return true;
|
||||
} catch (Exception e) {
|
||||
return false;
|
||||
}
|
||||
|
||||
return isTextFullyRemovable(font, text);
|
||||
}
|
||||
|
||||
private boolean canHandleText(PDFont font, String text) {
|
||||
@ -197,68 +152,14 @@ public class TextEncodingHelper {
|
||||
}
|
||||
|
||||
public boolean isTextFullyRemovable(PDFont font, String text) {
|
||||
if (font == null || text == null) {
|
||||
return false;
|
||||
}
|
||||
|
||||
if (text.isEmpty()) {
|
||||
return true;
|
||||
}
|
||||
|
||||
if (font == null || text == null) return false;
|
||||
if (text.isEmpty()) return true;
|
||||
if (!canEncodeCharacters(font, text)) return false;
|
||||
try {
|
||||
if (!canEncodeCharacters(font, text)) {
|
||||
return false;
|
||||
}
|
||||
|
||||
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;
|
||||
|
||||
float width = font.getStringWidth(text);
|
||||
return width >= 0;
|
||||
} catch (Exception e) {
|
||||
try {
|
||||
return canHandleText(font, text);
|
||||
} catch (Exception e2) {
|
||||
return false;
|
||||
}
|
||||
return false;
|
||||
}
|
||||
}
|
||||
|
||||
@ -381,45 +282,19 @@ public class TextEncodingHelper {
|
||||
}
|
||||
|
||||
public boolean fontSupportsCharacter(PDFont font, String character) {
|
||||
if (font == null || character == null) {
|
||||
return false;
|
||||
}
|
||||
|
||||
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) {
|
||||
}
|
||||
|
||||
if (font == null || character == null) return false;
|
||||
if (character.isEmpty()) return true;
|
||||
for (int i = 0; i < character.length(); ) {
|
||||
int codePoint = character.codePointAt(i);
|
||||
if (isBasicCharacter(codePoint)) {
|
||||
i += Character.charCount(codePoint);
|
||||
continue;
|
||||
int cp = character.codePointAt(i);
|
||||
String ch = new String(Character.toChars(cp));
|
||||
try {
|
||||
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;
|
||||
}
|
||||
|
||||
|
||||
@ -3,321 +3,324 @@
|
||||
xmlns:th="https://www.thymeleaf.org">
|
||||
|
||||
<head>
|
||||
<th:block th:insert="~{fragments/common :: head(title=#{autoRedact.title}, header=#{autoRedact.header})}"></th:block>
|
||||
<style>
|
||||
.redaction-options-group {
|
||||
margin-bottom: 1rem;
|
||||
}
|
||||
<th:block th:insert="~{fragments/common :: head(title=#{autoRedact.title}, header=#{autoRedact.header})}"></th:block>
|
||||
<style>
|
||||
.redaction-options-group {
|
||||
margin-bottom: 1rem;
|
||||
}
|
||||
|
||||
.form-text.text-muted {
|
||||
color: #6c757d !important;
|
||||
}
|
||||
.form-text.text-muted {
|
||||
color: #6c757d !important;
|
||||
}
|
||||
|
||||
.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);
|
||||
outline: 2px solid #0d6efd;
|
||||
outline-offset: 2px;
|
||||
}
|
||||
.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);
|
||||
outline: 2px solid #0d6efd;
|
||||
outline-offset: 2px;
|
||||
}
|
||||
|
||||
.form-check-input:checked {
|
||||
background-color: #0d6efd;
|
||||
border-color: #0d6efd;
|
||||
}
|
||||
.form-check-input:checked {
|
||||
background-color: #0d6efd;
|
||||
border-color: #0d6efd;
|
||||
}
|
||||
|
||||
/* OCR language list styling */
|
||||
#languages {
|
||||
max-height: 400px;
|
||||
overflow-y: auto;
|
||||
border: 1px solid var(--md-sys-color-surface-3);
|
||||
border-radius: 5px;
|
||||
padding: 10px;
|
||||
}
|
||||
/* OCR language list styling */
|
||||
#languages {
|
||||
max-height: 400px;
|
||||
overflow-y: auto;
|
||||
border: 1px solid var(--md-sys-color-surface-3);
|
||||
border-radius: 5px;
|
||||
padding: 10px;
|
||||
}
|
||||
|
||||
/* Better visibility for selected redaction option */
|
||||
.redaction-options-group .form-check {
|
||||
border: 1px solid var(--md-sys-color-surface-3);
|
||||
border-radius: 8px;
|
||||
padding: 10px 12px;
|
||||
transition: border-color .15s ease, background-color .15s ease, box-shadow .15s ease;
|
||||
}
|
||||
/* Better visibility for selected redaction option */
|
||||
.redaction-options-group .form-check {
|
||||
border: 1px solid var(--md-sys-color-surface-3);
|
||||
border-radius: 8px;
|
||||
padding: 10px 12px;
|
||||
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 {
|
||||
background-color: var(--md-sys-color-surface-1);
|
||||
}
|
||||
.redaction-options-group .form-check:hover {
|
||||
background-color: var(--md-sys-color-surface-1);
|
||||
}
|
||||
|
||||
.redaction-options-group .form-check.selected {
|
||||
border-color: #0d6efd;
|
||||
background-color: rgba(13,110,253,0.06);
|
||||
box-shadow: 0 0 0 2px rgba(13,110,253,0.1) inset;
|
||||
}
|
||||
.redaction-options-group .form-check.selected {
|
||||
border-color: #0d6efd;
|
||||
background-color: rgba(13,110,253,0.06);
|
||||
box-shadow: 0 0 0 2px rgba(13,110,253,0.1) inset;
|
||||
}
|
||||
|
||||
.redaction-options-group .form-check .form-check-label {
|
||||
font-weight: 600;
|
||||
}
|
||||
.redaction-options-group .form-check .form-check-label {
|
||||
font-weight: 600;
|
||||
}
|
||||
|
||||
.redaction-options-group small.form-text {
|
||||
margin-left: 1.8rem; /* align with radio */
|
||||
}
|
||||
</style>
|
||||
.redaction-options-group small.form-text {
|
||||
margin-left: 1.8rem; /* align with radio */
|
||||
}
|
||||
</style>
|
||||
</head>
|
||||
|
||||
<body>
|
||||
<div id="page-container">
|
||||
<div id="content-wrap">
|
||||
<th:block th:insert="~{fragments/navbar.html :: navbar}"></th:block>
|
||||
<br><br>
|
||||
<div class="container">
|
||||
<div class="row justify-content-center">
|
||||
<div class="col-md-6 bg-card">
|
||||
<div class="tool-header">
|
||||
<svg class="material-symbols-rounded tool-header-icon security">
|
||||
<use xlink:href="/images/redact-auto.svg#icon-redact-auto"></use>
|
||||
</svg>
|
||||
<span class="tool-header-text" id="form-title" th:text="#{autoRedact.header}"></span>
|
||||
</div>
|
||||
<form aria-labelledby="form-title" enctype="multipart/form-data" id="autoRedactForm"
|
||||
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 id="content-wrap">
|
||||
<th:block th:insert="~{fragments/navbar.html :: navbar}"></th:block>
|
||||
<br><br>
|
||||
<div class="container">
|
||||
<div class="row justify-content-center">
|
||||
<div class="col-md-6 bg-card">
|
||||
<div class="tool-header">
|
||||
<svg class="material-symbols-rounded tool-header-icon security">
|
||||
<use xlink:href="/images/redact-auto.svg#icon-redact-auto"></use>
|
||||
</svg>
|
||||
<span class="tool-header-text" id="form-title" th:text="#{autoRedact.header}"></span>
|
||||
</div>
|
||||
<form aria-labelledby="form-title" enctype="multipart/form-data" id="autoRedactForm"
|
||||
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 class="mb-3">
|
||||
<label class="form-label" for="listOfText" th:text="#{autoRedact.textsToRedactLabel}"></label>
|
||||
<textarea class="form-control" id="listOfText" name="listOfText" required rows="4"
|
||||
th:placeholder="#{autoRedact.textsToRedactPlaceholder}"></textarea>
|
||||
</div>
|
||||
<div class="mb-3">
|
||||
<label class="form-label" for="listOfText" th:text="#{autoRedact.textsToRedactLabel}"></label>
|
||||
<textarea class="form-control" id="listOfText" name="listOfText" required rows="4"
|
||||
th:placeholder="#{autoRedact.textsToRedactPlaceholder}"></textarea>
|
||||
</div>
|
||||
|
||||
<div class="mb-3">
|
||||
<div class="form-check">
|
||||
<input class="form-check-input" id="useRegex" name="useRegex" type="checkbox">
|
||||
<label class="form-check-label" for="useRegex" th:text="#{autoRedact.useRegexLabel}"></label>
|
||||
</div>
|
||||
<div class="form-check">
|
||||
<input class="form-check-input" id="wholeWordSearch" name="wholeWordSearch" type="checkbox">
|
||||
<label class="form-check-label" for="wholeWordSearch" th:text="#{autoRedact.wholeWordSearchLabel}"></label>
|
||||
</div>
|
||||
</div>
|
||||
<div class="mb-3">
|
||||
<div class="form-check">
|
||||
<input class="form-check-input" id="useRegex" name="useRegex" type="checkbox">
|
||||
<label class="form-check-label" for="useRegex" th:text="#{autoRedact.useRegexLabel}"></label>
|
||||
</div>
|
||||
<div class="form-check">
|
||||
<input class="form-check-input" id="wholeWordSearch" name="wholeWordSearch" type="checkbox">
|
||||
<label class="form-check-label" for="wholeWordSearch" th:text="#{autoRedact.wholeWordSearchLabel}"></label>
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div class="redaction-options-group">
|
||||
<label class="form-label fw-bold mb-3" th:text="#{autoRedact.redactionStyleLabel}"></label>
|
||||
<div class="form-check mb-2">
|
||||
<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>
|
||||
<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 class="form-check mb-2">
|
||||
<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>
|
||||
<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 class="form-check mb-3">
|
||||
<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>
|
||||
<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 class="redaction-options-group">
|
||||
<label class="form-label fw-bold mb-3" th:text="#{autoRedact.redactionStyleLabel}"></label>
|
||||
<div class="form-check mb-2">
|
||||
<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>
|
||||
<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 class="form-check mb-2">
|
||||
<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>
|
||||
<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 class="form-check mb-3">
|
||||
<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>
|
||||
<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 class="form-check">
|
||||
<input aria-describedby="guarantee-desc" class="form-check-input" id="guaranteeRedaction" name="convertPDFToImage" type="checkbox">
|
||||
<label class="form-check-label" for="guaranteeRedaction" th:text="#{autoRedact.pdfImageLabel}">PDF image</label>
|
||||
<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>
|
||||
</div>
|
||||
<!-- PDF Image checkbox - moved outside redaction-options-group to use plain styling -->
|
||||
<div class="mb-3">
|
||||
<div class="form-check">
|
||||
<input aria-describedby="guarantee-desc" class="form-check-input" id="convertPDFToImage" name="convertPDFToImage" type="checkbox">
|
||||
<label class="form-check-label" for="convertPDFToImage" th:text="#{autoRedact.convertPDFToImageLabel}">Convert PDF to PDF-Image</label>
|
||||
</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">
|
||||
<label class="form-label" for="defaultColor" th:text="#{autoRedact.colorLabel}"></label>
|
||||
<select class="form-select" id="defaultColor" name="defaultColor" onchange="handleColorChange(this.value)">
|
||||
<option th:text="#{black}" value="#000000">Black</option>
|
||||
<option th:text="#{white}" value="#FFFFFF">White</option>
|
||||
<option th:text="#{red}" value="#FF0000">Red</option>
|
||||
<option th:text="#{green}" value="#00FF00">Green</option>
|
||||
<option th:text="#{blue}" value="#0000FF">Blue</option>
|
||||
<option th:text="#{custom}" value="custom">Custom...</option>
|
||||
</select>
|
||||
</div>
|
||||
<div class="mb-3">
|
||||
<label class="form-label" for="defaultColor" th:text="#{autoRedact.colorLabel}"></label>
|
||||
<select class="form-select" id="defaultColor" name="defaultColor" onchange="handleColorChange(this.value)">
|
||||
<option th:text="#{black}" value="#000000">Black</option>
|
||||
<option th:text="#{white}" value="#FFFFFF">White</option>
|
||||
<option th:text="#{red}" value="#FF0000">Red</option>
|
||||
<option th:text="#{green}" value="#00FF00">Green</option>
|
||||
<option th:text="#{blue}" value="#0000FF">Blue</option>
|
||||
<option th:text="#{custom}" value="custom">Custom...</option>
|
||||
</select>
|
||||
</div>
|
||||
|
||||
<div class="mb-3" id="customColorContainer" style="display: none;">
|
||||
<label class="form-label" for="customColor">Custom Color (Hex)</label>
|
||||
<input class="form-control" id="customColor" name="redactColor" placeholder="#FF00FF" type="text">
|
||||
</div>
|
||||
<div class="mb-3" id="customColorContainer" style="display: none;">
|
||||
<label class="form-label" for="customColor">Custom Color (Hex)</label>
|
||||
<input class="form-control" id="customColor" name="redactColor" placeholder="#FF00FF" type="text">
|
||||
</div>
|
||||
|
||||
<div class="mb-3">
|
||||
<label class="form-label" for="customPadding" th:text="#{autoRedact.customPaddingLabel}"></label>
|
||||
<input class="form-control" id="customPadding" max="1" min="0" name="customPadding"
|
||||
step="0.1" type="number" value="0.1">
|
||||
</div>
|
||||
<div class="mb-3">
|
||||
<label class="form-label" for="customPadding" th:text="#{autoRedact.customPaddingLabel}"></label>
|
||||
<input class="form-control" id="customPadding" max="1" min="0" name="customPadding"
|
||||
step="0.1" type="number" value="0.1">
|
||||
</div>
|
||||
|
||||
<br>
|
||||
<br>
|
||||
|
||||
<div class="mb-3" th:if="${#lists.size(languages) > 0}">
|
||||
<label class="form-label" for="languages">OCR Languages</label>
|
||||
<div id="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" />
|
||||
<label th:attr="data-lang-code=${language}, data-lang-name=#{'lang.' + language}"
|
||||
<div class="mb-3" th:if="${#lists.size(languages) > 0}">
|
||||
<label class="form-label" for="languages">OCR Languages</label>
|
||||
<div id="languages">
|
||||
<div class="form-check" th:each="language, iterStat : ${languages}">
|
||||
<input class="form-check-input" onchange="handleLangSelection()" required th:checked="${language == 'eng'}" th:id="${'language-' + language}" th:name="languages" th:value="${language}" type="checkbox" />
|
||||
<label class="form-check-label" th:attr="data-lang-code=${language}, data-lang-name=#{'lang.' + language}"
|
||||
th:for="${'language-' + language}"
|
||||
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>
|
||||
<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>
|
||||
<th:block th:insert="~{fragments/footer.html :: footer}"></th:block>
|
||||
</div>
|
||||
<th:block th:insert="~{fragments/footer.html :: footer}"></th:block>
|
||||
</div>
|
||||
<script th:inline="javascript">
|
||||
function handleColorChange(selectedValue) {
|
||||
const container = document.getElementById('customColorContainer');
|
||||
const input = document.getElementById('customColor');
|
||||
if (selectedValue === "custom") {
|
||||
container.style.display = 'block';
|
||||
if (!input.value) {
|
||||
input.value = '#000000';
|
||||
}
|
||||
} else {
|
||||
container.style.display = 'none';
|
||||
input.value = selectedValue;
|
||||
}
|
||||
function handleColorChange(selectedValue) {
|
||||
const container = document.getElementById('customColorContainer');
|
||||
const input = document.getElementById('customColor');
|
||||
if (selectedValue === "custom") {
|
||||
container.style.display = 'block';
|
||||
if (!input.value) {
|
||||
input.value = '#000000';
|
||||
}
|
||||
} else {
|
||||
container.style.display = 'none';
|
||||
input.value = selectedValue;
|
||||
}
|
||||
}
|
||||
|
||||
function handleLangSelection() {
|
||||
let checkboxes = document.getElementsByName("languages");
|
||||
let selected = false;
|
||||
for (let i = 0; i < checkboxes.length; i++) {
|
||||
if (checkboxes[i].checked) {
|
||||
selected = true;
|
||||
checkboxes[i].setAttribute('required', 'false');
|
||||
}
|
||||
}
|
||||
if (selected) {
|
||||
for (let i = 0; i < checkboxes.length; i++) {
|
||||
checkboxes[i].removeAttribute('required');
|
||||
}
|
||||
} else {
|
||||
for (let i = 0; i < checkboxes.length; i++) {
|
||||
checkboxes[i].setAttribute('required', 'true');
|
||||
}
|
||||
}
|
||||
}
|
||||
function handleLangSelection() {
|
||||
let checkboxes = document.getElementsByName("languages");
|
||||
let selected = false;
|
||||
for (let i = 0; i < checkboxes.length; i++) {
|
||||
if (checkboxes[i].checked) {
|
||||
selected = true;
|
||||
checkboxes[i].setAttribute('required', 'false');
|
||||
}
|
||||
}
|
||||
if (selected) {
|
||||
for (let i = 0; i < checkboxes.length; i++) {
|
||||
checkboxes[i].removeAttribute('required');
|
||||
}
|
||||
} else {
|
||||
for (let i = 0; i < checkboxes.length; i++) {
|
||||
checkboxes[i].setAttribute('required', 'true');
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// Language translations map populated by Thymeleaf for available OCR languages
|
||||
const languageTranslations = {};
|
||||
/*[# th:each="lang : ${languages}"]*/
|
||||
languageTranslations['[(${lang})]'] = /*[[#{${'lang.' + lang}}]]*/'[(${lang})]';
|
||||
/*[/]*/
|
||||
// Language translations map populated by Thymeleaf for available OCR languages
|
||||
const languageTranslations = {};
|
||||
/*[# th:each="lang : ${languages}"]*/
|
||||
languageTranslations['[(${lang})]'] = /*[[#{${'lang.' + lang}}]]*/'[(${lang})]';
|
||||
/*[/]*/
|
||||
|
||||
const localeToTesseract = {
|
||||
'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',
|
||||
'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',
|
||||
'sk': 'slk', 'id': 'ind', 'th': 'tha', 'sl': 'slv'
|
||||
};
|
||||
const localeToTesseract = {
|
||||
'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',
|
||||
'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',
|
||||
'sk': 'slk', 'id': 'ind', 'th': 'tha', 'sl': 'slv'
|
||||
};
|
||||
|
||||
function getTranslatedLanguageName(shortCode) {
|
||||
// Use Thymeleaf-provided map; fall back to code when translation missing
|
||||
const name = languageTranslations[shortCode];
|
||||
if (name && !/^\?{2,}.+\?{2,}$/.test(name)) return name;
|
||||
return shortCode;
|
||||
}
|
||||
function getTranslatedLanguageName(shortCode) {
|
||||
// Use Thymeleaf-provided map; fall back to code when translation missing
|
||||
const name = languageTranslations[shortCode];
|
||||
if (name && !/^\?{2,}.+\?{2,}$/.test(name)) return name;
|
||||
return shortCode;
|
||||
}
|
||||
|
||||
function prioritizeLanguages() {
|
||||
const languageContainer = document.getElementById('languages');
|
||||
if (!languageContainer) return;
|
||||
const formChecks = Array.from(languageContainer.getElementsByClassName('form-check'));
|
||||
if (formChecks.length === 0) return;
|
||||
formChecks.forEach(element => {
|
||||
const label = element.querySelector('label');
|
||||
if (label) {
|
||||
const langCode = label.getAttribute('for').split('-')[1];
|
||||
// Always set from translations map; gracefully falls back to code
|
||||
label.textContent = getTranslatedLanguageName(langCode);
|
||||
}
|
||||
});
|
||||
const browserLanguage = document.documentElement.lang || navigator.language || navigator.userLanguage;
|
||||
const uiLanguage = document.documentElement.getAttribute('data-language') || browserLanguage;
|
||||
const primaryLanguageCode = (uiLanguage || '').split(/[-_]/)[0].toLowerCase();
|
||||
const tesseractPrimaryCode = localeToTesseract[primaryLanguageCode];
|
||||
const priorityLanguages = [];
|
||||
if (tesseractPrimaryCode) priorityLanguages.push(tesseractPrimaryCode);
|
||||
if (tesseractPrimaryCode !== 'eng') priorityLanguages.push('eng');
|
||||
const sortedElements = formChecks.sort((a, b) => {
|
||||
const aInput = a.querySelector('input');
|
||||
const bInput = b.querySelector('input');
|
||||
if (!aInput || !bInput) return 0;
|
||||
const aLangCode = aInput.id.split('-')[1];
|
||||
const bLangCode = bInput.id.split('-')[1];
|
||||
const aIsPriority = priorityLanguages.includes(aLangCode);
|
||||
const bIsPriority = priorityLanguages.includes(bLangCode);
|
||||
if (aIsPriority && !bIsPriority) return -1;
|
||||
if (!aIsPriority && bIsPriority) return 1;
|
||||
if (aIsPriority && bIsPriority) {
|
||||
return priorityLanguages.indexOf(aLangCode) - priorityLanguages.indexOf(bLangCode);
|
||||
}
|
||||
return getTranslatedLanguageName(aLangCode).localeCompare(getTranslatedLanguageName(bLangCode));
|
||||
});
|
||||
languageContainer.innerHTML = '';
|
||||
sortedElements.forEach(element => languageContainer.appendChild(element));
|
||||
}
|
||||
function prioritizeLanguages() {
|
||||
const languageContainer = document.getElementById('languages');
|
||||
if (!languageContainer) return;
|
||||
const formChecks = Array.from(languageContainer.getElementsByClassName('form-check'));
|
||||
if (formChecks.length === 0) return;
|
||||
formChecks.forEach(element => {
|
||||
const label = element.querySelector('label');
|
||||
if (label) {
|
||||
const langCode = label.getAttribute('for').split('-')[1];
|
||||
// Always set from translations map; gracefully falls back to code
|
||||
label.textContent = getTranslatedLanguageName(langCode);
|
||||
}
|
||||
});
|
||||
const browserLanguage = document.documentElement.lang || navigator.language || navigator.userLanguage;
|
||||
const uiLanguage = document.documentElement.getAttribute('data-language') || browserLanguage;
|
||||
const primaryLanguageCode = (uiLanguage || '').split(/[-_]/)[0].toLowerCase();
|
||||
const tesseractPrimaryCode = localeToTesseract[primaryLanguageCode];
|
||||
const priorityLanguages = [];
|
||||
if (tesseractPrimaryCode) priorityLanguages.push(tesseractPrimaryCode);
|
||||
if (tesseractPrimaryCode !== 'eng') priorityLanguages.push('eng');
|
||||
const sortedElements = formChecks.sort((a, b) => {
|
||||
const aInput = a.querySelector('input');
|
||||
const bInput = b.querySelector('input');
|
||||
if (!aInput || !bInput) return 0;
|
||||
const aLangCode = aInput.id.split('-')[1];
|
||||
const bLangCode = bInput.id.split('-')[1];
|
||||
const aIsPriority = priorityLanguages.includes(aLangCode);
|
||||
const bIsPriority = priorityLanguages.includes(bLangCode);
|
||||
if (aIsPriority && !bIsPriority) return -1;
|
||||
if (!aIsPriority && bIsPriority) return 1;
|
||||
if (aIsPriority && bIsPriority) {
|
||||
return priorityLanguages.indexOf(aLangCode) - priorityLanguages.indexOf(bLangCode);
|
||||
}
|
||||
return getTranslatedLanguageName(aLangCode).localeCompare(getTranslatedLanguageName(bLangCode));
|
||||
});
|
||||
languageContainer.innerHTML = '';
|
||||
sortedElements.forEach(element => languageContainer.appendChild(element));
|
||||
}
|
||||
|
||||
document.addEventListener('DOMContentLoaded', function () {
|
||||
const redactionModeRadios = document.querySelectorAll('input[name="redactionMode"]');
|
||||
const aggressiveModeHidden = document.getElementById('aggressiveMode');
|
||||
const guaranteeRedactionCheckbox = document.getElementById('guaranteeRedaction');
|
||||
const defaultColor = document.getElementById('defaultColor');
|
||||
document.addEventListener('DOMContentLoaded', function () {
|
||||
const redactionModeRadios = document.querySelectorAll('input[name="redactionMode"]');
|
||||
const aggressiveModeHidden = document.getElementById('aggressiveMode');
|
||||
const convertPDFToImageCheckbox = document.getElementById('convertPDFToImage');
|
||||
const defaultColor = document.getElementById('defaultColor');
|
||||
|
||||
function updateMode() {
|
||||
const selectedMode = document.querySelector('input[name="redactionMode"]:checked');
|
||||
if (selectedMode) {
|
||||
// Set aggressive mode for delete text option
|
||||
aggressiveModeHidden.value = selectedMode.value === 'aggressive' ? 'true' : 'false';
|
||||
const selectedMode = document.querySelector('input[name="redactionMode"]:checked');
|
||||
if (selectedMode) {
|
||||
// Set aggressive mode for delete text option
|
||||
aggressiveModeHidden.value = selectedMode.value === 'aggressive' ? 'true' : 'false';
|
||||
|
||||
// Handle PDF image checkbox based on selection
|
||||
if (selectedMode.value === 'visual') {
|
||||
// Visual mode automatically enables PDF image for maximum security
|
||||
guaranteeRedactionCheckbox.checked = true;
|
||||
} else {
|
||||
// Delete Text and Keep Layout modes disable PDF image
|
||||
guaranteeRedactionCheckbox.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');
|
||||
}
|
||||
// Handle PDF image checkbox based on selection
|
||||
if (selectedMode.value === 'visual') {
|
||||
// Visual mode automatically enables PDF image for maximum security
|
||||
convertPDFToImageCheckbox.checked = true;
|
||||
} else {
|
||||
// Delete Text and Keep Layout modes disable PDF image
|
||||
convertPDFToImageCheckbox.checked = false;
|
||||
}
|
||||
|
||||
redactionModeRadios.forEach(radio => {
|
||||
radio.addEventListener('change', updateMode);
|
||||
});
|
||||
// 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');
|
||||
}
|
||||
}
|
||||
|
||||
if (defaultColor) {
|
||||
handleColorChange(defaultColor.value);
|
||||
const customColorInput = document.getElementById('customColor');
|
||||
if (defaultColor.value !== 'custom') {
|
||||
customColorInput.value = defaultColor.value;
|
||||
}
|
||||
}
|
||||
redactionModeRadios.forEach(radio => {
|
||||
radio.addEventListener('change', updateMode);
|
||||
});
|
||||
|
||||
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
|
||||
prioritizeLanguages();
|
||||
|
||||
// Handle pre-selected English language
|
||||
handleLangSelection();
|
||||
});
|
||||
});
|
||||
</script>
|
||||
</body>
|
||||
</html>
|
||||
Loading…
Reference in New Issue
Block a user