From c6c0b09e5607fc3a0f99ae0d4bdbc590e4c45b29 Mon Sep 17 00:00:00 2001
From: Albert S <mail@quitesimple.org>
Date: Sun, 14 May 2023 14:15:50 +0200
Subject: [PATCH] PdfProcessor: Extract outline from documents

---
 shared/pdfprocessor.cpp | 30 ++++++++++++++++++++++++++----
 shared/pdfprocessor.h   |  4 +++-
 2 files changed, 29 insertions(+), 5 deletions(-)

diff --git a/shared/pdfprocessor.cpp b/shared/pdfprocessor.cpp
index 6a40e31..34c84d5 100644
--- a/shared/pdfprocessor.cpp
+++ b/shared/pdfprocessor.cpp
@@ -5,9 +5,30 @@ PdfProcessor::PdfProcessor()
 {
 }
 
-QVector<PageData> PdfProcessor::process(const QByteArray &data) const
+QVector<DocumentOutlineEntry> PdfProcessor::createOutline(const QVector<Poppler::OutlineItem> &outlineItems) const
 {
-	QVector<PageData> result;
+	QVector<DocumentOutlineEntry> result;
+	for(const Poppler::OutlineItem &outlineItem : outlineItems)
+	{
+		DocumentOutlineEntry documentOutlineEntry;
+		documentOutlineEntry.text = outlineItem.name();
+		documentOutlineEntry.type = OUTLINE_DESTINATION_TYPE_PAGE;
+		if(!outlineItem.destination().isNull())
+		{
+			documentOutlineEntry.destinationPage = outlineItem.destination()->pageNumber();
+		}
+		if(outlineItem.hasChildren())
+		{
+			documentOutlineEntry.children = createOutline(outlineItem.children());
+		}
+		result.append(documentOutlineEntry);
+	}
+	return result;
+}
+
+DocumentProcessResult PdfProcessor::process(const QByteArray &data) const
+{
+	DocumentProcessResult result;
 	QScopedPointer<Poppler::Document> doc(Poppler::Document::loadFromData(data));
 	if(doc.isNull())
 	{
@@ -26,12 +47,13 @@ QVector<PageData> PdfProcessor::process(const QByteArray &data) const
 	for(auto i = 0; i < pagecount; i++)
 	{
 		QString text = doc->page(i)->text(entirePage);
-		result.append({static_cast<unsigned int>(i + 1), text});
+		result.pages.append({static_cast<unsigned int>(i + 1), text});
 		/*TODO: hack, so we can fts search several words over the whole document, not just pages.
 		 * this of course uses more space and should be solved differently.
 		 */
 		entire += text;
 	}
-	result.append({0, entire});
+	result.pages.append({0, entire});
+	result.outlines = createOutline(doc->outline());
 	return result;
 }
diff --git a/shared/pdfprocessor.h b/shared/pdfprocessor.h
index 4802879..f932f11 100644
--- a/shared/pdfprocessor.h
+++ b/shared/pdfprocessor.h
@@ -1,5 +1,6 @@
 #ifndef PDFPROCESSOR_H
 #define PDFPROCESSOR_H
+#include <poppler-qt5.h>
 #include "processor.h"
 class PdfProcessor : public Processor
 {
@@ -7,7 +8,8 @@ class PdfProcessor : public Processor
 	PdfProcessor();
 
   public:
-	QVector<PageData> process(const QByteArray &data) const override;
+	QVector<DocumentOutlineEntry> createOutline(const QVector<Poppler::OutlineItem> &outlineItems) const;
+	DocumentProcessResult process(const QByteArray &data) const override;
 };
 
 #endif // PDFPROCESSOR_H