From c26300d752e4506a43c31ff40167443c168d1829 Mon Sep 17 00:00:00 2001
From: xenofem <xenofem@xeno.science>
Date: Tue, 28 May 2024 00:33:00 -0400
Subject: [PATCH] add best-effort sort option

---
 dlibrary/dlibrary.py | 24 ++++++++++++++++++++++++
 1 file changed, 24 insertions(+)

diff --git a/dlibrary/dlibrary.py b/dlibrary/dlibrary.py
index ae7b0c1..386230b 100755
--- a/dlibrary/dlibrary.py
+++ b/dlibrary/dlibrary.py
@@ -32,6 +32,8 @@ import requests
 NUMBER_REGEX = re.compile('[0-9０-９]+')
 ALPHABETIC_NUMBERING_REGEX = re.compile('^(?P<prefix>[^a-zａ-ｚ0-9０-９]*)((?P<letter>[a-zａ-ｚ])(?P<suffix>[^a-zａ-ｚ0-9０-９]*))?$', re.I)
 
+STRING_TOKENIZE_REGEX = re.compile('(?P<str>[^0-9０-９]+)|(?P<num>[0-9０-９]+)')
+
 EXTRA_NORMALIZATION_TABLE = str.maketrans({'\u301c': '\uff5e'})
 
 DLSITE_ID_REGEX = re.compile('^[BR]J[0-9]+$')
@@ -530,6 +532,10 @@ class Collator:
 
         if all(src.is_file() and is_image(src) for src in srcs):
             ordering = complete_prefix_number_ordering(srcs)
+            if not ordering and self.args.sort:
+                ordering = srcs.copy()
+                ordering.sort(key=best_effort_sort_key)
+                debug(f'Applying best-effort sort: {ordering}')
             if ordering:
                 print(f'Symlinking image files: {ordering[0]}...')
                 return self.link_ordered_files(ordering)
@@ -812,6 +818,19 @@ def nname(entry):
 def nstem(entry):
     return normalize_string(entry.stem)
 
+def best_effort_sort_key(entry):
+    result = []
+    for token in STRING_TOKENIZE_REGEX.finditer(nstem(entry)):
+        if token.lastgroup == 'num':
+            if len(result) == 0:
+                result.append('') # to prevent failed int/string comparisons against other files
+            result.append(int(token.group()))
+        else:
+            result.append(token.group())
+
+    debug(f"Tokenized {entry} as {result}")
+    return result
+
 def complete_prefix_number_ordering(entries):
     if len(entries) == 1:
         return entries
@@ -1460,6 +1479,11 @@ parser_collate.add_argument(
     action='store_true',
     help="only collate works that are explicitly specified",
 )
+parser_collate.add_argument(
+    '-s', '--sort',
+    action='store_true',
+    help="apply a best-effort sorting algorithm when the ordering of image files is unclear",
+)
 parser_collate.add_argument(
     '-p', '--pdf-strategy',
     choices=[