From c2f516a281d5a2b9c8c2cc3d58dd055be7dde5ec Mon Sep 17 00:00:00 2001
From: xenofem <xenofem@xeno.science>
Date: Tue, 6 Feb 2024 23:52:59 -0500
Subject: [PATCH] add subcommand to show page counts and image sizes of an
 extracted work's files

---
 dlibrary/dlibrary.py | 26 ++++++++++++++++++++++++++
 1 file changed, 26 insertions(+)

diff --git a/dlibrary/dlibrary.py b/dlibrary/dlibrary.py
index faaa066..3d313b5 100755
--- a/dlibrary/dlibrary.py
+++ b/dlibrary/dlibrary.py
@@ -527,6 +527,28 @@ def manual_collate(args):
             print(f'Unknown file type {path}, stopping')
             return
 
+def fmt_size(s):
+    return f'{s[0]}x{s[1]}px'
+
+def analyze(args):
+    extract_dir = args.destdir / 'extract'
+    files = descendant_files_ignore(extract_dir / args.work_id)
+    files.sort()
+
+    for f in files:
+        print(f'{relpath(f, extract_dir)}', end='')
+        if f.suffix.lower() in IMAGE_FILE_EXTENSIONS:
+            size = standalone_image_size(f)
+            print(f'\t{fmt_size(size)}')
+        elif f.suffix.lower() == '.pdf':
+            sizes = pdf_image_sizes(f)
+            if len(sizes) == 0:
+                print(f'\tContains no images')
+            else:
+                print(f'\t{len(sizes)} images, median {fmt_size(median(sizes))}, min {fmt_size(min(sizes))}, max {fmt_size(max(sizes))}')
+        else:
+            print()
+
 def metadata(args):
     con = sqlite3.connect(args.destdir / 'meta.db')
     cur = con.cursor()
@@ -813,6 +835,10 @@ parser_manual_collate.add_argument(
 )
 parser_manual_collate.set_defaults(func=manual_collate)
 
+parser_analyze = subparsers.add_parser('analyze', help='analyze an extracted folder to assist in collation')
+parser_analyze.add_argument('work_id')
+parser_analyze.set_defaults(func=analyze)
+
 parser_metadata = subparsers.add_parser('metadata', help='view or modify metadata for a work')
 parser_metadata.add_argument('work_id')
 parser_metadata.add_argument(