From 6c50c7023580012c7118ab7bde4111b0ef7fa5ad Mon Sep 17 00:00:00 2001
From: "James R. Barlow" <jim@purplerock.ca>
Date: Fri, 11 May 2018 23:42:26 -0700
Subject: [PATCH] Try to optimize paletted images

---
 src/ocrmypdf/_optimize.py | 11 ++++++++++-
 src/ocrmypdf/leptonica.py |  3 +++
 src/ocrmypdf/pipeline.py  |  1 -
 3 files changed, 13 insertions(+), 2 deletions(-)

diff --git a/src/ocrmypdf/_optimize.py b/src/ocrmypdf/_optimize.py
index 0d27f49a..1deb6b91 100644
--- a/src/ocrmypdf/_optimize.py
+++ b/src/ocrmypdf/_optimize.py
@@ -125,7 +125,7 @@ def extract_image(*, doc, pike, root, log, image, xref, jbig2s,
         log.debug("Skipping small image, xref {}".format(xref))
         return False
 
-    bpc = image.get('/BitsPerComponent', 8)
+    bpc = int(image.get('/BitsPerComponent', 8))
     cs = image.get('/ColorSpace', '')
     w = int(image.Width)
     h = int(image.Height)
@@ -177,6 +177,15 @@ def extract_image(*, doc, pike, root, log, image, xref, jbig2s,
         raw_jpeg_data = raw_jpeg.read_raw_bytes()
         (root / '{:08d}.jpg'.format(xref)).write_bytes(raw_jpeg_data)
         jpegs.append(xref)
+    elif cs[0] == '/Indexed' \
+            and cs[1] in SIMPLE_COLORSPACES \
+            and options.optimize >= 3 \
+            and fitz:
+        # Try to improve on indexed images - these are far from low hanging
+        # fruit in most cases
+        pix = fitz.Pixmap(doc, xref)
+        pix.writePNG(make_img_name(root, xref), savealpha=False)
+        pngs.append(xref)        
     elif cs in SIMPLE_COLORSPACES and fitz:
         # For any 'inferior' filter including /FlateDecode we extract
         # and recode as /FlateDecode
diff --git a/src/ocrmypdf/leptonica.py b/src/ocrmypdf/leptonica.py
index 92e35db9..79e99877 100644
--- a/src/ocrmypdf/leptonica.py
+++ b/src/ocrmypdf/leptonica.py
@@ -514,6 +514,9 @@ class CompressedData:
             raise LeptonicaError("CompressedData.open")
         return CompressedData(p_compdata[0])
 
+    def __len__(self):
+        return self._compdata.nbytescomp
+
     def read(self):
         buf = ffi.buffer(self._compdata.datacomp, self._compdata.nbytescomp)
         return bytes(buf)
diff --git a/src/ocrmypdf/pipeline.py b/src/ocrmypdf/pipeline.py
index 90843b6a..96f3aa02 100644
--- a/src/ocrmypdf/pipeline.py
+++ b/src/ocrmypdf/pipeline.py
@@ -960,7 +960,6 @@ def weave_layers(
             procset = pdf_base.pages[0].Resources.ProcSet
             font = pdf_base.pages[0].Resources.Font.get(font_key)
 
-
     pdf_base.save(output_file)