MuhammadSawalhy · July 25, 2022 19:03 · Jul 25, 2022 · Jul 25, 2022 · Mar 29, 2022 · Mar 29, 2022
diff --git a/my-story.md b/my-story.md
@@ -1,3 +1,5 @@
+## Plan A
+
 1. list images with command: `pdfimages -j -png file.pdf img`
 2. run **list-code-images.py** to find the dark theme code images
 3. invert these image to make them light theme code:
@@ -6,10 +8,17 @@
      convert $f -channel RGB -negate inversed/$f
    done
    ```
-4. find a way to replace images in a pdf with code, but at the
-   end I end up using PyMuPDF to invert dark theme code images
-   and save them in the same position using **replace-images.py**
+4. find a way to replace images in a pdf with code (but I gave up here)
+
+## Plan B
+
+I end up using PyMuPDF to invert dark theme code images and save them in the same position using **replace-images.py**.
+
+You will need to install these packages:
 
+```bash
+pip install fitz PyMuPDF
+```
 
 ------------
 

diff --git a/replace-images.py b/replace-images.py
@@ -1,20 +1,21 @@
 import fitz
-from rich import inspect
+from os import listdir
 # This creates the Document object doc
-doc: fitz.Document = fitz.open("file.pdf")
-for page in doc:
-    for img in page.get_images():
-        xref = img[-2]
-        xref_number = img[0]
-        pix = fitz.Pixmap(doc, xref)
-        bg_color = pix.pixel(pix.width - 1, int(pix.height / 2))
-        if bg_color == (12,12,12):
-            pix.invert_irect()
-            rect = page.get_image_bbox(xref)
-            page.insert_image(rect, pixmap=pix, keep_proportion=False)
 
-# doc.save(filename=r"file.new.pdf", clean=True)
-# doc.save(filename=r"file.new.pdf", clean=True, garbage=4)
-# without deflate_images=1 the file size is 112MB, but now it is just 12MB
-doc.save(filename=r"file.new.pdf", clean=True, deflate=4, deflate_images=1, deflate_fonts=1)
-doc.close()
+for file in listdir("./files"):
+    doc: fitz.Document = fitz.open(f"./files/{file}")
+    for page in doc:
+        for img in page.get_images(full=True):
+            xref = img[0]
+            pix = fitz.Pixmap(doc, xref)
+            bg_color = pix.pixel(pix.width - 1, int(pix.height / 2))
+            if bg_color == (12,12,12):
+                pix.invert_irect()
+                rect = page.get_image_bbox(img)
+                page.insert_image(rect, pixmap=pix, keep_proportion=False)
+
+    # doc.save(filename=r"file.new.pdf", clean=True)
+    # doc.save(filename=r"file.new.pdf", clean=True, garbage=4)
+    # without deflate_images=1 the file size is 112MB, but now it is just 12MB
+    doc.save(filename=f"./processed-files/{file}", clean=True, deflate=4, deflate_images=1, deflate_fonts=1)
+    doc.close()
diff --git a/my-story.md b/my-story.md
@@ -8,4 +8,11 @@
    ```
 4. find a way to replace images in a pdf with code, but at the
    end I end up using PyMuPDF to invert dark theme code images
-   and save them in the same position using **replace-images.py**
+   and save them in the same position using **replace-images.py**
+
+
+------------
+
+![image](https://user-images.githubusercontent.com/42011920/160681311-be009615-1e24-4a95-b847-78353bf53927.png)
+
+Alhamdulillah, all images replaced. This is an illusion of replacement, because the new images are placed in top of the old images. I think there exists some possible ways to really replace using [Document.update_object](https://pymupdf.readthedocs.io/en/latest/document.html#Document.update_object) or [Document.update_stream](https://pymupdf.readthedocs.io/en/latest/document.html#Document.update_stream) provided by PyMuPDF package.
diff --git a/replace-images.py b/replace-images.py
@@ -1,4 +1,3 @@
-
 import fitz
 from rich import inspect
 # This creates the Document object doc

diff --git a/list-code-images.py b/list-code-images.py
@@ -8,4 +8,4 @@
         x = math.floor(im.size[0]/2)
         px = im.getpixel((x,-1))
         if px == (12,12,12):
-            print(img)
+            print(img)
diff --git a/my-story.md b/my-story.md
@@ -0,0 +1,11 @@
+1. list images with command: `pdfimages -j -png file.pdf img`
+2. run **list-code-images.py** to find the dark theme code images
+3. invert these image to make them light theme code:
+   ```bash
+   for f in `cat file.code-images.txt`; do
+     convert $f -channel RGB -negate inversed/$f
+   done
+   ```
+4. find a way to replace images in a pdf with code, but at the
+   end I end up using PyMuPDF to invert dark theme code images
+   and save them in the same position using **replace-images.py**
diff --git a/replace-images.py b/replace-images.py
@@ -1,19 +1,21 @@
+
 import fitz
 from rich import inspect
 # This creates the Document object doc
 doc: fitz.Document = fitz.open("file.pdf")
 for page in doc:
     for img in page.get_images():
-        xref = img[0]
+        xref = img[-2]
+        xref_number = img[0]
         pix = fitz.Pixmap(doc, xref)
         bg_color = pix.pixel(pix.width - 1, int(pix.height / 2))
         if bg_color == (12,12,12):
             pix.invert_irect()
-            rect = page.get_image_bbox(img[-2])
+            rect = page.get_image_bbox(xref)
             page.insert_image(rect, pixmap=pix, keep_proportion=False)
 
 # doc.save(filename=r"file.new.pdf", clean=True)
 # doc.save(filename=r"file.new.pdf", clean=True, garbage=4)
+# without deflate_images=1 the file size is 112MB, but now it is just 12MB
 doc.save(filename=r"file.new.pdf", clean=True, deflate=4, deflate_images=1, deflate_fonts=1)
-
-doc.close()
+doc.close()
diff --git a/replace-images.py b/replace-images.py
@@ -12,6 +12,8 @@
             rect = page.get_image_bbox(img[-2])
             page.insert_image(rect, pixmap=pix, keep_proportion=False)
 
-doc.save(filename=r"file.new.pdf")
+# doc.save(filename=r"file.new.pdf", clean=True)
+# doc.save(filename=r"file.new.pdf", clean=True, garbage=4)
+doc.save(filename=r"file.new.pdf", clean=True, deflate=4, deflate_images=1, deflate_fonts=1)
 
 doc.close()
diff --git a/list-code-images.py b/list-code-images.py
@@ -0,0 +1,11 @@
+import math
+import glob
+from PIL import Image
+
+for img in glob.glob("images/*.png"):
+# for img in ["imageroot-014.png"]:
+    with Image.open(img) as im:
+        x = math.floor(im.size[0]/2)
+        px = im.getpixel((x,-1))
+        if px == (12,12,12):
+            print(img)
diff --git a/replace-images.py b/replace-images.py
@@ -0,0 +1,17 @@
+import fitz
+from rich import inspect
+# This creates the Document object doc
+doc: fitz.Document = fitz.open("file.pdf")
+for page in doc:
+    for img in page.get_images():
+        xref = img[0]
+        pix = fitz.Pixmap(doc, xref)
+        bg_color = pix.pixel(pix.width - 1, int(pix.height / 2))
+        if bg_color == (12,12,12):
+            pix.invert_irect()
+            rect = page.get_image_bbox(img[-2])
+            page.insert_image(rect, pixmap=pix, keep_proportion=False)
+
+doc.save(filename=r"file.new.pdf")
+
+doc.close()