From 6e041d07eb0a48de73612ffa3f7c7af1be236531 Mon Sep 17 00:00:00 2001
From: Andreas Stefl <stefl.andreas@gmail.com>
Date: Thu, 2 Jan 2025 10:56:01 +0100
Subject: [PATCH] page cache

---
 pdf2htmlEX/src/CoveredTextDetector.cc      | 14 ++++++------
 pdf2htmlEX/src/CoveredTextDetector.h       |  2 +-
 pdf2htmlEX/src/HTMLRenderer/HTMLRenderer.h |  5 +++++
 pdf2htmlEX/src/HTMLRenderer/general.cc     | 26 ++++++++++++++++++++++
 4 files changed, 39 insertions(+), 8 deletions(-)

diff --git a/pdf2htmlEX/src/CoveredTextDetector.cc b/pdf2htmlEX/src/CoveredTextDetector.cc
index 0792c528..f46817cb 100644
--- a/pdf2htmlEX/src/CoveredTextDetector.cc
+++ b/pdf2htmlEX/src/CoveredTextDetector.cc
@@ -14,7 +14,7 @@
 
 namespace pdf2htmlEX {
 
-CoveredTextDetector::CoveredTextDetector(Param & param): param(param)
+CoveredTextDetector::CoveredTextDetector(Param & param): param(&param)
 {
 }
 
@@ -41,10 +41,10 @@ void CoveredTextDetector::add_char_bbox_clipped(cairo_t *cairo, double * bbox, i
     char_pts_visible.push_back(pts_visible);
 
     // DCRH: Hide if no points are visible, or if some points are visible and correct_text_visibility == 2
-    if (pts_visible == 0 || param.correct_text_visibility == 2) {
+    if (pts_visible == 0 || param->correct_text_visibility == 2) {
         chars_covered.push_back(true);
-        if (pts_visible > 0 && param.correct_text_visibility == 2) {
-            param.actual_dpi = std::min(param.text_dpi, param.max_dpi); // Char partially covered so increase background resolution
+        if (pts_visible > 0 && param->correct_text_visibility == 2) {
+            param->actual_dpi = std::min(param->text_dpi, param->max_dpi); // Char partially covered so increase background resolution
         }
     } else {
         chars_covered.push_back(false);
@@ -98,13 +98,13 @@ printf("pts_visible=%x\n", pts_visible);
 printf("pts_visible=%x\n", pts_visible);
 #endif
             char_pts_visible[i] = pts_visible;
-            if (pts_visible == 0 || (pts_visible != (1|2|4|8) && param.correct_text_visibility == 2)) {
+            if (pts_visible == 0 || (pts_visible != (1|2|4|8) && param->correct_text_visibility == 2)) {
 #ifdef DEBUG
 printf("Char covered\n");
 #endif
                 chars_covered[i] = true;
-                if (pts_visible > 0 && param.correct_text_visibility == 2) { // Partially visible text => increase rendering DPI
-                    param.actual_dpi = std::min(param.text_dpi, param.max_dpi);
+                if (pts_visible > 0 && param->correct_text_visibility == 2) { // Partially visible text => increase rendering DPI
+                    param->actual_dpi = std::min(param->text_dpi, param->max_dpi);
                 }
             }
         } else {
diff --git a/pdf2htmlEX/src/CoveredTextDetector.h b/pdf2htmlEX/src/CoveredTextDetector.h
index 0f0506f3..2e664a7f 100644
--- a/pdf2htmlEX/src/CoveredTextDetector.h
+++ b/pdf2htmlEX/src/CoveredTextDetector.h
@@ -60,7 +60,7 @@ class CoveredTextDetector
     // x00, y00, x01, y01; x10, y10, x11, y11;...
     std::vector<double> char_bboxes;
     std::vector<int> char_pts_visible;
-    Param & param;
+    Param * param;
 };
 
 }
diff --git a/pdf2htmlEX/src/HTMLRenderer/HTMLRenderer.h b/pdf2htmlEX/src/HTMLRenderer/HTMLRenderer.h
index 984b1d7d..983962d1 100644
--- a/pdf2htmlEX/src/HTMLRenderer/HTMLRenderer.h
+++ b/pdf2htmlEX/src/HTMLRenderer/HTMLRenderer.h
@@ -381,6 +381,11 @@ struct HTMLRenderer : OutputDev
 
     CoveredTextDetector covered_text_detector;
     DrawingTracer tracer;
+
+    struct PageCache {
+        CoveredTextDetector covered_text_detector;
+    };
+    std::unordered_map<int, PageCache> page_cache;
 };
 
 } //namespace pdf2htmlEX
diff --git a/pdf2htmlEX/src/HTMLRenderer/general.cc b/pdf2htmlEX/src/HTMLRenderer/general.cc
index 07c22fd0..c2811238 100644
--- a/pdf2htmlEX/src/HTMLRenderer/general.cc
+++ b/pdf2htmlEX/src/HTMLRenderer/general.cc
@@ -183,12 +183,31 @@ void HTMLRenderer::process(PDFDoc *doc)
 
     post_process();
 
+    if (param.delay_background == 0)
+    {
+        bg_renderer = nullptr;
+        fallback_bg_renderer = nullptr;
+    }
+
     if(param.quiet == 0)
         cerr << endl;
 }
 
 bool HTMLRenderer::renderPage(PDFDoc *doc, int pageno)
 {
+    if (param.delay_background == 0)
+    {
+        return false;
+    }
+
+    if (page_cache.find(pageno) != page_cache.end())
+    {
+        cerr << "Page number " << pageno << " not found in page cache" << endl;
+        return false;
+    }
+
+    covered_text_detector = page_cache[pageno].covered_text_detector;
+
     if (bg_renderer->render_page(cur_doc, pageno))
     {
         return true;
@@ -209,6 +228,13 @@ void HTMLRenderer::setDefaultCTM(const double *ctm)
 
 void HTMLRenderer::startPage(int pageNum, GfxState *state, XRef * xref)
 {
+    if (param.delay_background && this->pageNum > 0)
+    {
+        page_cache[this->pageNum] = {
+            .covered_text_detector = covered_text_detector,
+        };
+    }
+
     covered_text_detector.reset();
     tracer.reset(state);