From da80d3f3545e80325b9ce87ed3c352a7fd6cb553 Mon Sep 17 00:00:00 2001
From: "James R. Barlow" <jim@purplerock.ca>
Date: Mon, 7 May 2018 17:37:46 -0700
Subject: [PATCH] Add unconditional (for now) whiteout of text areas

---
 src/ocrmypdf/pdfinfo.py  | 11 ++++++++++-
 src/ocrmypdf/pipeline.py | 32 +++++++++++++++++++++++++++++---
 2 files changed, 39 insertions(+), 4 deletions(-)

diff --git a/src/ocrmypdf/pdfinfo.py b/src/ocrmypdf/pdfinfo.py
index eef720ee..bef0f60b 100644
--- a/src/ocrmypdf/pdfinfo.py
+++ b/src/ocrmypdf/pdfinfo.py
@@ -654,7 +654,7 @@ class PageInfo:
 
     @property
     def has_text(self):
-        return self._pageinfo['has_text']
+        return self._pageinfo['has_text']                
 
     @property
     def width_inches(self):
@@ -687,6 +687,15 @@ class PageInfo:
     def images(self):
         return self._pageinfo['images']
 
+    def get_textareas(self):
+        if not fitz:
+            raise NotImplementedError("no impl without fitz")
+        doc = fitz.open(self._infile)
+        page = doc[self._pageno]
+        text = page.getText('dict')
+        for block in text['blocks']:
+            yield block['bbox']
+
     @property
     def xres(self):
         return self._pageinfo.get('xres', None)
diff --git a/src/ocrmypdf/pipeline.py b/src/ocrmypdf/pipeline.py
index 9df1b032..9f659c66 100644
--- a/src/ocrmypdf/pipeline.py
+++ b/src/ocrmypdf/pipeline.py
@@ -583,11 +583,37 @@ def select_ocr_image(
         log,
         context):
     """Select the image we send for OCR. May not be the same as the display
-    image depending on preprocessing."""
+    image depending on preprocessing. This image will never be shown to the
+    user."""
 
-    # For the moment this is always the .pp-clean.png image
     image = infiles[0]
-    re_symlink(image, output_file, log)
+    pageinfo = get_pageinfo(image, context)
+
+    with Image.open(image) as im:
+        from PIL import ImageColor
+        from PIL import ImageDraw
+        from decimal import Decimal
+        white = ImageColor.getcolor('#ffffff', im.mode)
+        draw = ImageDraw.ImageDraw(im)
+
+        for bbox in pageinfo.get_textareas():
+            # Calculate resolution based on the image size and page dimensions
+            # without regard whatever resolution is in pageinfo (may differ or
+            # be None)
+            xres = im.width / pageinfo.width_inches
+            yres = im.height / pageinfo.height_inches
+            log.debug('calculated resolution %r %r', xres, yres)
+
+            pixcoords = [Decimal(bbox[0]) / Decimal(72) * xres,
+                         Decimal(bbox[1]) / Decimal(72) * yres,
+                         Decimal(bbox[2]) / Decimal(72) * xres,
+                         Decimal(bbox[3]) / Decimal(72) * yres]
+            pixcoords = [int(c) for c in pixcoords]
+            log.debug('blanking %r', pixcoords)
+            draw.rectangle(pixcoords, fill=white)
+        
+        del draw
+        im.save(output_file)
 
 
 def ocr_tesseract_hocr(