detect when directory names are just image resolutions

dlibrary/dlibrary.py

@@ -32,7 +32,7 @@ DLSITE_ID_REGEX = re.compile('^[BR]J[0-9]+$')
 FANZA_ID_REGEX = re.compile('^d_[0-9]+$')
 FAKKU_ID_REGEX = re.compile('.*_FAKKU$')
 
-HI_RES_REGEX = re.compile('高解像度|原寸', re.I)
+HI_RES_REGEX = re.compile('高解像度|原寸|大サイズ', re.I)
 NO_TONE_REGEX = re.compile('トーン(効果)?[な無]し|グレースケール', re.I)
 TONE_REGEX = re.compile('トーン(版|(効果)?[有あ]り)', re.I)
 COLOR_REGEX = re.compile('カラー', re.I)
@@ -42,6 +42,7 @@ IMAGE_QUALITY_REGEXES = [
     { 'better': NO_TONE_REGEX, 'worse': TONE_REGEX },
     { 'better': COLOR_REGEX, 'worse': MONOCHROME_REGEX },
 ]
+IMAGE_RESOLUTION_REGEX = re.compile('^(?P<x>[0-9]+)x(?P<y>[0-9]+)$')
 
 LANGUAGE_REGEXES = {
     'en_US': re.compile('english|英語', re.I),
@@ -441,20 +442,25 @@ class Collator:
         if select_language is not False:
             return select_language
 
-        if len(srcs) == 2 and all(src.is_dir() for src in srcs):
+        dirs = [src for src in srcs if src.is_dir()]
+        non_dirs = [src for src in srcs if not src.is_dir()]
+        if len(dirs) == 2 and len(descendant_files_ignore(dirs[0], self.exclude)) == len(descendant_files_ignore(dirs[1], self.exclude)):
+            resolution_matches = [IMAGE_RESOLUTION_REGEX.match(nname(src)) for src in dirs]
+            if all(resolution_matches):
+                pairs = [(int(m.group('x')), int(m.group('y'))) for m in resolution_matches]
+                for i in range(2):
+                    if pairs[i][0] > pairs[1-i][0] and pairs[i][1] > pairs[1-i][1]:
+                        return self.collate_from_paths([dirs[i]] + non_dirs)
             for quality in IMAGE_QUALITY_REGEXES:
                 def a_not_b(a, b, src):
                     if a in quality:
                         return quality[a].search(nname(src))
                     else:
                         return not quality[b].search(nname(src))
-                better_srcs = [src for src in srcs if a_not_b('better', 'worse', src)]
-                worse_srcs = [src for src in srcs if a_not_b('worse', 'better', src)]
-                if len(better_srcs) == 1 and len(worse_srcs) == 1 and better_srcs[0] != worse_srcs[0]:
-                    better = better_srcs[0]
-                    worse = worse_srcs[0]
-                    if len(descendant_files_ignore(better, self.exclude)) == len(descendant_files_ignore(worse, self.exclude)):
-                        return self.collate_from_paths([better])
+                better_dirs = [src for src in dirs if a_not_b('better', 'worse', src)]
+                worse_dirs = [src for src in dirs if a_not_b('worse', 'better', src)]
+                if len(better_dirs) == 1 and len(worse_dirs) == 1 and better_dirs[0] != worse_dirs[0]:
+                    return self.collate_from_paths(better_dirs + non_dirs)
 
         images_vs_pdf = self.try_collate_images_vs_pdf(srcs)
         if images_vs_pdf is not False: