Add script to check for unused images

We have a habbit of deleting pages and not thinking about the images that
are linked from them. Create a script that we can run as part of the CI
that checks the files in static/images/ are used on the site.

Signed-off-by: Martyn Welch <martyn.welch@collabora.com>

scripts/check_images.py

+#!/usr/bin/env python3
+
+# Check links to images.
+
+import markdown
+import os
+import re
+import sys
+
+def get_link(url):
+    link = url.group("link")
+
+    if not link[:8] == "/images/":
+        return
+
+    if not link[8:] in links:
+        links.append(link[8:])
+
+
+def parse_file(filename):
+    # print("%s: " % filename)
+    with open(filename, "r+") as file:
+        contents = file.read()
+
+        if not contents[0:3] == "+++":
+            return
+
+        fm = contents.split("+++")[1]
+        doc = contents.split("+++", 2)[2]
+
+        # Convert to HTML to simplify link detection
+        text = markdown.markdown(doc)
+
+        # From img tags
+        pattern = re.compile("src=\"(?P<link>.*?)\"")
+        doc = pattern.sub(get_link, text)
+
+        # We have some links to bits in images
+        pattern = re.compile("href=\"(?P<link>.*?)\"")
+        doc = pattern.sub(get_link, text)
+
+
+images = []
+
+# List of files in images folder
+for root, dirs, files in os.walk("%s/static/images" % (sys.argv[1])):
+    for file in files:
+        images.append(file)
+
+links = []
+
+# Parse aliases
+for root, dirs, files in os.walk("%s/content" % (sys.argv[1])):
+    for file in files:
+        if ".md" in file:
+            try:
+                parse_file("%s/%s" % (root, file))
+            except:
+                print("Failed to parse %s/%s" % (root, file))
+
+broken = []
+for link in links:
+    sys.stdout.flush()
+
+    if link in images:
+        images.remove(link)
+    else:
+        broken.append(link)
+
+print(f"Found {len(broken)} broken image links in {len(links)} tested:")
+broken.sort()
+for b in broken:
+    print(" ", b)
+
+print(f"Found {len(images)} unused images:")
+images.sort()
+for i in images:
+    print(" ", i)
+
+sys.exit(1 if broken or images else 0)