[issuu] fix 'user' extractor

gallery_dl/extractor/issuu.py

@@ -54,26 +54,30 @@ class IssuuPublicationExtractor(IssuuBase, GalleryExtractor):
 class IssuuUserExtractor(IssuuBase, Extractor):
     """Extractor for all publications of a user/publisher"""
     subcategory = "user"
-    pattern = r"(?:https?://)?issuu\.com/([^/?#]+)/?$"
+    pattern = r"(?:https?://)?issuu\.com/([^/?#]+)(?:/(\d*))?$"
     example = "https://issuu.com/USER"
 
-    def __init__(self, match):
-        Extractor.__init__(self, match)
-        self.user = match.group(1)
-
     def items(self):
-        url = "{}/call/profile/v1/documents/{}".format(self.root, self.user)
+        user, pnum = self.groups
-        params = {"offset": 0, "limit": "25"}
+        base = self.root + "/" + user
+        pnum = text.parse_int(pnum, 1)
 
         while True:
-            data = self.request(url, params=params).json()
+            url = base + "/" + str(pnum) if pnum > 1 else base
-
+            try:
-            for publication in data["items"]:
+                html = self.request(url).text
-                publication["url"] = "{}/{}/docs/{}".format(
+                data = util.json_loads(text.unescape(text.extr(
-                    self.root, self.user, publication["uri"])
+                    html, '</main></div><script data-json="', '" id="')))
-                publication["_extractor"] = IssuuPublicationExtractor
+                docs = data["docs"]
-                yield Message.Queue, publication["url"], publication
+            except Exception as exc:
-
+                self.log.debug("", exc_info=exc)
-            if not data["hasMore"]:
                 return
-            params["offset"] += data["limit"]
+
+            for publication in docs:
+                url = self.root + "/" + publication["uri"]
+                publication["_extractor"] = IssuuPublicationExtractor
+                yield Message.Queue, url, publication
+
+            if len(docs) < 48:
+                return
+            pnum += 1

test/results/issuu.py

@@ -10,7 +10,6 @@ from gallery_dl.extractor import issuu
 __tests__ = (
 {
     "#url"     : "https://issuu.com/issuu/docs/motions-1-2019/",
-    "#category": ("", "issuu", "publication"),
     "#class"   : issuu.IssuuPublicationExtractor,
     "#pattern" : r"https://image.isu.pub/190916155301-\w+/jpg/page_\d+.jpg",
     "#count"   : 36,
@@ -37,10 +36,15 @@ __tests__ = (
 
 {
     "#url"     : "https://issuu.com/issuu",
-    "#category": ("", "issuu", "user"),
     "#class"   : issuu.IssuuUserExtractor,
     "#pattern" : issuu.IssuuPublicationExtractor.pattern,
-    "#count"   : "> 25",
+    "#count"   : range(100, 150),
+},
+
+{
+    "#url"     : "https://issuu.com/issuu/3",
+    "#class"   : issuu.IssuuUserExtractor,
+    "#count"   : range(4, 40),
 },
 
 )