From 7df2e3fdc01e66a9369a80d8396d4defcfb6ee75 Mon Sep 17 00:00:00 2001
From: Timmy Welch <timmy@narnian.us>
Date: Mon, 19 Sep 2022 20:24:42 -0700
Subject: [PATCH] Automatically crop black borders from covers

---
 comictaggerlib/ctoptions/file.py  |  7 +++++
 comictaggerlib/issueidentifier.py | 52 +++++++++++++++++++++++++++++--
 tests/issueidentifier_test.py     | 24 ++++++++++++++
 3 files changed, 80 insertions(+), 3 deletions(-)

diff --git a/comictaggerlib/ctoptions/file.py b/comictaggerlib/ctoptions/file.py
index 7791feb..0d6e4f7 100644
--- a/comictaggerlib/ctoptions/file.py
+++ b/comictaggerlib/ctoptions/file.py
@@ -42,6 +42,13 @@ def internal(parser: settngs.Manager) -> None:
 def identifier(parser: settngs.Manager) -> None:
     # identifier settings
     parser.add_setting("--series-match-identify-thresh", default=91, type=int, help="")
+    parser.add_setting(
+        "-b",
+        "--border-crop-percent",
+        default=10,
+        type=int,
+        help="ComicTagger will automatically add an additional cover that has any black borders cropped. If the difference in height is less than %(default)s%% the cover will not be cropped.",
+    )
     parser.add_setting(
         "--publisher-filter",
         default=["Panini Comics", "Abril", "Planeta DeAgostini", "Editorial Televisa", "Dino Comics"],
diff --git a/comictaggerlib/issueidentifier.py b/comictaggerlib/issueidentifier.py
index d919b1d..dade198 100644
--- a/comictaggerlib/issueidentifier.py
+++ b/comictaggerlib/issueidentifier.py
@@ -35,7 +35,7 @@ from comictalker.talkerbase import ComicTalker, TalkerError
 logger = logging.getLogger(__name__)
 
 try:
-    from PIL import Image
+    from PIL import Image, ImageChops
 
     pil_available = True
 except ImportError:
@@ -166,6 +166,48 @@ class IssueIdentifier:
 
         return cropped_image_data
 
+    # Adapted from https://stackoverflow.com/a/10616717/20629671
+    def crop_border(self, image_data: bytes, ratio: int) -> bytes | None:
+        im = Image.open(io.BytesIO(image_data))
+
+        # RGBA doesn't work????
+        tmp = im.convert("RGB")
+
+        bg = Image.new("RGB", tmp.size, "black")
+        diff = ImageChops.difference(tmp, bg)
+        diff = ImageChops.add(diff, diff, 2.0, -100)
+
+        bbox = diff.getbbox()
+
+        width_percent = 0
+        height_percent = 0
+
+        # If bbox is None that should mean it's solid black
+        if bbox:
+            width = bbox[2] - bbox[0]
+            height = bbox[3] - bbox[1]
+
+            # Convert to percent
+            width_percent = 100 - ((width / im.width) * 100)
+            height_percent = 100 - ((height / im.height) * 100)
+            logger.debug(
+                "Width: %s Height: %s, ratio: %s  %s ratio met: %s",
+                im.width,
+                im.height,
+                width_percent,
+                height_percent,
+                width_percent > ratio or height_percent > ratio,
+            )
+
+        # If there is a difference return the image otherwise return None
+        if width_percent > ratio or height_percent > ratio:
+            output = io.BytesIO()
+            im.crop(bbox).save(output, format="PNG")
+            cropped_image_data = output.getvalue()
+            output.close()
+            return cropped_image_data
+        return None
+
     def set_progress_callback(self, cb_func: Callable[[int, int], None]) -> None:
         self.callback = cb_func
 
@@ -308,8 +350,7 @@ class IssueIdentifier:
                     self.log_msg(score, False)
 
                 if score <= self.strong_score_thresh:
-                    # such a good score, we can quit now, since for sure we
-                    # have a winner
+                    # such a good score, we can quit now, since for sure we have a winner
                     done = True
                     break
             if done:
@@ -464,6 +505,11 @@ class IssueIdentifier:
             if narrow_cover_hash is not None:
                 hash_list.append(narrow_cover_hash)
 
+            cropped_border = self.crop_border(cover_image_data, self.options.identifier_border_crop_percent)
+            if cropped_border is not None:
+                hash_list.append(self.calculate_hash(cropped_border))
+                logger.info("Adding cropped cover to the hashlist")
+
             try:
                 image_url = issue.image_url
                 alt_urls = issue.alt_image_urls
diff --git a/tests/issueidentifier_test.py b/tests/issueidentifier_test.py
index 9a62d02..f8a8acf 100644
--- a/tests/issueidentifier_test.py
+++ b/tests/issueidentifier_test.py
@@ -1,6 +1,9 @@
 from __future__ import annotations
 
+import io
+
 import pytest
+from PIL import Image
 
 import comicapi.comicarchive
 import comicapi.issuestring
@@ -73,3 +76,24 @@ def test_search(cbz, options, comicvine_api):
     for r, e in zip(results, [cv_expected]):
         del r["url_image_hash"]
         assert r == e
+
+
+def test_crop_border(cbz, options, comicvine_api):
+    settings, definitions = options
+    ii = comictaggerlib.issueidentifier.IssueIdentifier(cbz, settings, comicvine_api)
+
+    # This creates a white square centered on a black background
+    bg = Image.new("RGBA", (100, 100), (0, 0, 0, 255))
+    fg = Image.new("RGBA", (50, 50), (255, 255, 255, 255))
+    bg.paste(fg, (bg.width // 2 - (fg.width // 2), bg.height // 2 - (fg.height // 2)))
+    output = io.BytesIO()
+    bg.save(output, format="PNG")
+    image_data = output.getvalue()
+    output.close()
+
+    cropped = ii.crop_border(image_data, 49)
+
+    im = Image.open(io.BytesIO(cropped))
+    assert im.width == fg.width
+    assert im.height == fg.height
+    assert list(im.getdata()) == list(fg.getdata())