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

Refactor and fix: Update EML to PDF conversion to support only general fonts and improve attachment marker handling by removing emoji that might not interpreted by some systems.

Browse Source
This commit is contained in:
Balázs Szücs 2025-06-17 19:40:45 +02:00
parent b3a2bfbe71
commit f4a677f150
1 changed files with 37 additions and 66 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

103
common/src/main/java/stirling/software/common/util/EmlToPdf.java
View File

@ -43,12 +43,10 @@ import stirling.software.common.model.api.converters.EmlToPdfRequest;
@Slf4j @Slf4j
@UtilityClass @UtilityClass
public class EmlToPdf { public class EmlToPdf {
private static final class StyleConstants { private static final class StyleConstants {
// Font and layout constants // Font and layout constants
static final int DEFAULT_FONT_SIZE = 12; static final int DEFAULT_FONT_SIZE = 12;
static final String DEFAULT_FONT_FAMILY = static final String DEFAULT_FONT_FAMILY = "Helvetica, sans-serif";
"-apple-system, BlinkMacSystemFont, 'Segoe UI', Roboto, 'Helvetica Neue', Arial, sans-serif";
static final float DEFAULT_LINE_HEIGHT = 1.4f; static final float DEFAULT_LINE_HEIGHT = 1.4f;
static final String DEFAULT_ZOOM = "1.0"; static final String DEFAULT_ZOOM = "1.0";
@ -69,20 +67,15 @@ public class EmlToPdf {
static final int EML_CHECK_LENGTH = 8192; static final int EML_CHECK_LENGTH = 8192;
static final int MIN_HEADER_COUNT_FOR_VALID_EML = 2; static final int MIN_HEADER_COUNT_FOR_VALID_EML = 2;
private StyleConstants() { private StyleConstants() {}
// Utility class - prevent instantiation
}
} }
private static final class MimeConstants { private static final class MimeConstants {
static final Pattern MIME_ENCODED_PATTERN = static final Pattern MIME_ENCODED_PATTERN =
Pattern.compile("=\\?([^?]+)\\?([BbQq])\\?([^?]*)\\?="); Pattern.compile("=\\?([^?]+)\\?([BbQq])\\?([^?]*)\\?=");
static final String PAPERCLIP_EMOJI = "\uD83D\uDCCE"; // 📎 static final String ATTACHMENT_MARKER = "[@]";
static final String ATTACHMENT_ICON_PLACEHOLDER = "icon";
private MimeConstants() { private MimeConstants() {}
// Utility class - prevent instantiation
}
} }
private static final class FileSizeConstants { private static final class FileSizeConstants {
@ -90,9 +83,7 @@ public class EmlToPdf {
static final long BYTES_IN_MB = BYTES_IN_KB * 1024L; static final long BYTES_IN_MB = BYTES_IN_KB * 1024L;
static final long BYTES_IN_GB = BYTES_IN_MB * 1024L; static final long BYTES_IN_GB = BYTES_IN_MB * 1024L;
private FileSizeConstants() { private FileSizeConstants() {}
// Utility class - prevent instantiation
}
} }
// Cached Jakarta Mail availability check // Cached Jakarta Mail availability check
@ -169,7 +160,7 @@ public class EmlToPdf {
} }
} }
private static void validateEmlInput(byte[] emlBytes) throws IOException { private static void validateEmlInput(byte[] emlBytes) {
if (emlBytes == null || emlBytes.length == 0) { if (emlBytes == null || emlBytes.length == 0) {
throw new IllegalArgumentException("EML file is empty or null"); throw new IllegalArgumentException("EML file is empty or null");
} }
@ -205,7 +196,6 @@ public class EmlToPdf {
disableSanitize); disableSanitize);
} catch (IOException | InterruptedException e) { } catch (IOException | InterruptedException e) {
log.warn("Initial HTML to PDF conversion failed, trying with simplified HTML"); log.warn("Initial HTML to PDF conversion failed, trying with simplified HTML");
// Try with simplified HTML
String simplifiedHtml = simplifyHtmlContent(htmlContent); String simplifiedHtml = simplifyHtmlContent(htmlContent);
return FileToPdf.convertHtmlToPdf( return FileToPdf.convertHtmlToPdf(
weasyprintPath, weasyprintPath,
@ -256,7 +246,7 @@ public class EmlToPdf {
html.append("<html><head><meta charset=\"UTF-8\">\n"); html.append("<html><head><meta charset=\"UTF-8\">\n");
html.append("<title>").append(escapeHtml(subject)).append("</title>\n"); html.append("<title>").append(escapeHtml(subject)).append("</title>\n");
html.append("<style>\n"); html.append("<style>\n");
appendEnhancedStyles(html, request); appendEnhancedStyles(html);
html.append("</style>\n"); html.append("</style>\n");
html.append("</head><body>\n"); html.append("</head><body>\n");
@ -313,7 +303,7 @@ public class EmlToPdf {
// Show advanced features status if requested // Show advanced features status if requested
assert request != null; assert request != null;
if (request != null && request.getFileInput().isEmpty()) { if (request.getFileInput().isEmpty()) {
html.append("<div class=\"advanced-features-notice\">\n"); html.append("<div class=\"advanced-features-notice\">\n");
html.append( html.append(
"<p><em>Note: Some advanced features require Jakarta Mail dependencies.</em></p>\n"); "<p><em>Note: Some advanced features require Jakarta Mail dependencies.</em></p>\n");
@ -485,7 +475,7 @@ public class EmlToPdf {
attachmentInfo attachmentInfo
.append("<div class=\"attachment-item\">") .append("<div class=\"attachment-item\">")
.append("<span class=\"attachment-icon\">") .append("<span class=\"attachment-icon\">")
.append(MimeConstants.ATTACHMENT_ICON_PLACEHOLDER) .append(MimeConstants.ATTACHMENT_MARKER)
.append("</span> ") .append("</span> ")
.append("<span class=\"attachment-name\">") .append("<span class=\"attachment-name\">")
.append(escapeHtml(filename)) .append(escapeHtml(filename))
@ -659,7 +649,7 @@ public class EmlToPdf {
return processed; return processed;
} }
private static void appendEnhancedStyles(StringBuilder html, EmlToPdfRequest request) { private static void appendEnhancedStyles(StringBuilder html) {
int fontSize = StyleConstants.DEFAULT_FONT_SIZE; int fontSize = StyleConstants.DEFAULT_FONT_SIZE;
String textColor = StyleConstants.DEFAULT_TEXT_COLOR; String textColor = StyleConstants.DEFAULT_TEXT_COLOR;
String backgroundColor = StyleConstants.DEFAULT_BACKGROUND_COLOR; String backgroundColor = StyleConstants.DEFAULT_BACKGROUND_COLOR;
@ -841,7 +831,7 @@ public class EmlToPdf {
processMultipartAdvanced(messageContent, content, request); processMultipartAdvanced(messageContent, content, request);
} }
} catch (Exception e) { } catch (Exception e) {
log.warn("Error processing multipart content: {}", e.getMessage()); log.warn("Error processing content: {}", e.getMessage());
} }
} }
@ -979,7 +969,7 @@ public class EmlToPdf {
html.append("<html><head><meta charset=\"UTF-8\">\n"); html.append("<html><head><meta charset=\"UTF-8\">\n");
html.append("<title>").append(escapeHtml(content.getSubject())).append("</title>\n"); html.append("<title>").append(escapeHtml(content.getSubject())).append("</title>\n");
html.append("<style>\n"); html.append("<style>\n");
appendEnhancedStyles(html, request); appendEnhancedStyles(html);
html.append("</style>\n"); html.append("</style>\n");
html.append("</head><body>\n"); html.append("</head><body>\n");
@ -1036,7 +1026,7 @@ public class EmlToPdf {
.append(uniqueId) .append(uniqueId)
.append("\">") .append("\">")
.append("<span class=\"attachment-icon\">") .append("<span class=\"attachment-icon\">")
.append(MimeConstants.PAPERCLIP_EMOJI) .append(MimeConstants.ATTACHMENT_MARKER)
.append("</span> ") .append("</span> ")
.append("<span class=\"attachment-name\">") .append("<span class=\"attachment-name\">")
.append(escapeHtml(safeMimeDecode(attachment.getFilename()))) .append(escapeHtml(safeMimeDecode(attachment.getFilename())))
@ -1202,24 +1192,24 @@ public class EmlToPdf {
return; return;
} }
// 1. Find the screen position of all emoji anchors // 1. Find the screen position of all attachment markers
EmojiPositionFinder finder = new EmojiPositionFinder(); AttachmentMarkerPositionFinder finder = new AttachmentMarkerPositionFinder();
finder.setSortByPosition(true); // Process pages in order finder.setSortByPosition(true); // Process pages in order
finder.getText(document); finder.getText(document);
List<EmojiPosition> emojiPositions = finder.getPositions(); List<MarkerPosition> markerPositions = finder.getPositions();
// 2. Warn if the number of anchors and attachments don't match // 2. Warn if the number of markers and attachments don't match
if (emojiPositions.size() != attachments.size()) { if (markerPositions.size() != attachments.size()) {
log.warn( log.warn(
"Found {} emoji anchors, but there are {} attachments. Annotation count may be incorrect.", "Found {} attachment markers, but there are {} attachments. Annotation count may be incorrect.",
emojiPositions.size(), markerPositions.size(),
attachments.size()); attachments.size());
} }
// 3. Create an invisible annotation over each found emoji // 3. Create an invisible annotation over each found marker
int annotationsToAdd = Math.min(emojiPositions.size(), attachments.size()); int annotationsToAdd = Math.min(markerPositions.size(), attachments.size());
for (int i = 0; i < annotationsToAdd; i++) { for (int i = 0; i < annotationsToAdd; i++) {
EmojiPosition position = emojiPositions.get(i); MarkerPosition position = markerPositions.get(i);
EmailAttachment attachment = attachments.get(i); EmailAttachment attachment = attachments.get(i);
if (attachment.getEmbeddedFilename() != null) { if (attachment.getEmbeddedFilename() != null) {
@ -1354,7 +1344,6 @@ public class EmlToPdf {
} }
// MIME header decoding functionality for RFC 2047 encoded headers - moved to constants // MIME header decoding functionality for RFC 2047 encoded headers - moved to constants
private static String decodeMimeHeader(String encodedText) { private static String decodeMimeHeader(String encodedText) {
if (encodedText == null || encodedText.trim().isEmpty()) { if (encodedText == null || encodedText.trim().isEmpty()) {
return encodedText; return encodedText;
@ -1494,15 +1483,13 @@ public class EmlToPdf {
} }
@Data @Data
public static class EmojiPosition { public static class MarkerPosition {
private int pageIndex; private int pageIndex;
private float x; private float x;
private float y; private float y;
private String character; private String character;
public EmojiPosition() {} public MarkerPosition(int pageIndex, float x, float y, String character) {
public EmojiPosition(int pageIndex, float x, float y, String character) {
this.pageIndex = pageIndex; this.pageIndex = pageIndex;
this.x = x; this.x = x;
this.y = y; this.y = y;
@ -1510,14 +1497,15 @@ public class EmlToPdf {
} }
} }
public static class EmojiPositionFinder extends org.apache.pdfbox.text.PDFTextStripper { public static class AttachmentMarkerPositionFinder
@Getter private final List<EmojiPosition> positions = new ArrayList<>(); extends org.apache.pdfbox.text.PDFTextStripper {
@Getter private final List<MarkerPosition> positions = new ArrayList<>();
private int currentPageIndex; private int currentPageIndex;
private boolean sortByPosition; protected boolean sortByPosition;
private boolean isInAttachmentSection; private boolean isInAttachmentSection;
private boolean attachmentSectionFound; private boolean attachmentSectionFound;
public EmojiPositionFinder() throws IOException { public AttachmentMarkerPositionFinder() {
super(); super();
this.currentPageIndex = 0; this.currentPageIndex = 0;
this.sortByPosition = false; this.sortByPosition = false;
@ -1560,24 +1548,18 @@ public class EmlToPdf {
isInAttachmentSection = false; isInAttachmentSection = false;
} }
// Only look for emojis if we are in the attachment section // Only look for markers if we are in the attachment section
if (isInAttachmentSection) { if (isInAttachmentSection) {
// Look for paperclip emoji characters (U+1F4CE) String attachmentMarker = MimeConstants.ATTACHMENT_MARKER;
String paperclipEmoji = "\uD83D\uDCCE"; // 📎 Unicode representation for (int i = 0; (i = string.indexOf(attachmentMarker, i)) != -1; i++) {
if (i < textPositions.size()) {
for (int i = 0; i < string.length(); i++) {
// Check if we have a complete paperclip emoji at this position
if (i < string.length() - 1
&& string.substring(i, i + 2).equals(paperclipEmoji)
&& i < textPositions.size()) {
org.apache.pdfbox.text.TextPosition textPosition = textPositions.get(i); org.apache.pdfbox.text.TextPosition textPosition = textPositions.get(i);
EmojiPosition position = MarkerPosition position =
new EmojiPosition( new MarkerPosition(
currentPageIndex, currentPageIndex,
textPosition.getXDirAdj(), textPosition.getXDirAdj(),
textPosition.getYDirAdj(), textPosition.getYDirAdj(),
paperclipEmoji); attachmentMarker);
positions.add(position); positions.add(position);
} }
} }
@ -1589,16 +1571,5 @@ public class EmlToPdf {
public void setSortByPosition(boolean sortByPosition) { public void setSortByPosition(boolean sortByPosition) {
this.sortByPosition = sortByPosition; this.sortByPosition = sortByPosition;
} }
public boolean isSortByPosition() {
return sortByPosition;
}
public void reset() {
positions.clear();
currentPageIndex = 0;
isInAttachmentSection = false;
attachmentSectionFound = false;
}
} }
} }