X-Git-Url: https://git.mdrn.pl/librarian.git/blobdiff_plain/dcecf2226f792e051a230bb5fc68501ce7e06474..77004f7537a472419bf9701c8522a774a0c29245:/src/librarian/epubcheck.py

diff --git a/src/librarian/epubcheck.py b/src/librarian/epubcheck.py
new file mode 100644
index 0000000..00bfb13
--- /dev/null
+++ b/src/librarian/epubcheck.py
@@ -0,0 +1,33 @@
+import json
+import re
+import subprocess
+import zipfile
+
+
+def epubcheck(filename):
+    p = subprocess.run(
+        [
+            'epubcheck', '-q',
+            '-j', '-',
+            filename
+        ],
+        capture_output=True
+    )
+    output = json.loads(p.stdout)
+    epub = zipfile.ZipFile(filename)
+    messages = output.get('messages', [])
+    for message in messages:
+        for loc in message.get('locations', []):
+            if loc['path'].startswith('EPUB/part'):
+                with epub.open(loc['path']) as zfile:
+                    text = zfile.read().decode('utf-8')
+                line = text.split('\n')[loc['line'] - 1][:loc['column'] - 1:]
+                debug = re.findall(r' data-debug="(\d+):(\d+)', line)
+                if debug:
+                    debug = debug[-1]
+                    loc['wl_chunk'] = int(debug[0])
+                    loc['wl_line'] = int(debug[1])
+    return messages
+
+            
+