From 3795d6720f88692101ba13c537776b79283b4d0a Mon Sep 17 00:00:00 2001
From: "James R. Barlow" <jim@purplerock.ca>
Date: Fri, 23 Mar 2018 13:39:16 -0700
Subject: [PATCH] Try out pymupdf merging

With garbage collection it reduces waste on the worst case file.
That's nice. 1 MB -> 105 MB -> 1.5 MB.

Indicates really problem is using PyPDF2 to watermark.

Currently hacked into --output-type pdf.
---
 ocrmypdf/pipeline.py | 25 ++++++++++++++++++++++++-
 1 file changed, 24 insertions(+), 1 deletion(-)

diff --git a/ocrmypdf/pipeline.py b/ocrmypdf/pipeline.py
index a5682feb..c2cc39f7 100644
--- a/ocrmypdf/pipeline.py
+++ b/ocrmypdf/pipeline.py
@@ -978,6 +978,29 @@ def merge_pages_qpdf(
                log=log)
 
 
+def merge_pages_mupdf(
+        input_files_groups,
+        output_file,
+        log,
+        context):
+    options = context.get_options()
+
+    pdf_pages, metadata_file = _merge_pages_common(
+        input_files_groups, output_file, log, context)
+
+    import fitz
+    doc = fitz.Document()
+    
+    for pdf_page in pdf_pages:
+        page = fitz.open(pdf_page)
+        doc.insertPDF(page)
+
+    metadata = fitz.open(metadata_file)
+    doc.setToC(metadata.getToC())
+    doc.setMetadata(metadata.metadata)
+    doc.save(output_file, garbage=4)
+
+
 def merge_sidecars(
         input_files_groups,
         output_file,
@@ -1254,7 +1277,7 @@ def build_pipeline(options, work_folder, log, context):
     task_merge_pages_ghostscript.active_if(options.output_type.startswith('pdfa'))
 
     task_merge_pages_qpdf = main_pipeline.merge(
-        task_func=merge_pages_qpdf,
+        task_func=merge_pages_mupdf,
         input=[task_combine_layers,
                task_render_hocr_debug_page,
                task_skip_page,