merge #5997: [batoto] improve chapter info regex (#5988)

… and add 'chapter_string' metadata

gallery_dl/extractor/batoto.py

@@ -51,28 +51,30 @@ class BatotoChapterExtractor(BatotoBase, ChapterExtractor):
         if not manga:
             manga = extr('link-hover">', "<")
             info = text.remove_html(extr('link-hover">', "</"))
+        info = text.unescape(info)
 
         match = re.match(
             r"(?:Volume\s+(\d+) )?"
-            r"\w+\s+(\d+)(.*)", info)
+            r"[Cc]hapter\s*(\d+)([\w.]*)", info)
         if match:
             volume, chapter, minor = match.groups()
-            title = text.remove_html(extr(
-                "selected>", "</option")).partition(" : ")[2]
+            title = text.unescape(text.remove_html(extr(
+                "selected>", "</option")).partition(" : ")[2])
         else:
             volume = chapter = 0
             minor = ""
             title = info
 
         return {
-            "manga"        : text.unescape(manga),
-            "manga_id"     : text.parse_int(manga_id),
-            "title"        : text.unescape(title),
-            "volume"       : text.parse_int(volume),
-            "chapter"      : text.parse_int(chapter),
-            "chapter_minor": minor,
-            "chapter_id"   : text.parse_int(self.chapter_id),
-            "date"         : text.parse_timestamp(extr(' time="', '"')[:-3]),
+            "manga"         : text.unescape(manga),
+            "manga_id"      : text.parse_int(manga_id),
+            "title"         : title,
+            "volume"        : text.parse_int(volume),
+            "chapter"       : text.parse_int(chapter),
+            "chapter_minor" : minor,
+            "chapter_string": info,
+            "chapter_id"    : text.parse_int(self.chapter_id),
+            "date"          : text.parse_timestamp(extr(' time="', '"')[:-3]),
         }
 
     def images(self, page):

test/results/batoto.py

@@ -59,6 +59,18 @@ __tests__ = (
     "volume"    : 0
 },
 
+{
+    "#url"     : "https://bato.to/title/90710-new-suitor-for-the-abandoned-wife/2089747-ch_76",
+    "#comment" : "duplicate info in chapter_minor / title (#5988)",
+    "#category": ("", "batoto", "chapter"),
+    "#class"   : batoto.BatotoChapterExtractor,
+
+    "chapter"      : 76,
+    "chapter_id"   : 2089747,
+    "chapter_minor": "",
+    "title"        : "Side Story 4 [END]",
+},
+
 {
     "#url"     : "https://bato.to/title/86408/1681030",
     "#category": ("", "batoto", "chapter"),