From 93cef78450bdfb95921b0e5173c9d616c6e29806 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Mike=20F=C3=A4hrmann?= <mike_faehrmann@web.de>
Date: Wed, 24 Nov 2021 23:34:46 +0100
Subject: [PATCH] [gelbooru] workaround pagination limits

Gelbooru only allows to retrieve the latest 20k posts for a tag search.
Add 'id:<N' to the search tags to work around that limitation, where N
is the ID of the last retrieved post.

http://gelbooru.me/index.php?page=forum&s=view&id=1467
---
 gallery_dl/extractor/gelbooru_v02.py | 15 ++++++++++++++-
 1 file changed, 14 insertions(+), 1 deletion(-)

diff --git a/gallery_dl/extractor/gelbooru_v02.py b/gallery_dl/extractor/gelbooru_v02.py
index e09e190d..a42a2025 100644
--- a/gallery_dl/extractor/gelbooru_v02.py
+++ b/gallery_dl/extractor/gelbooru_v02.py
@@ -27,8 +27,21 @@ class GelbooruV02Extractor(booru.BooruExtractor):
         params["pid"] = self.page_start
         params["limit"] = self.per_page
 
+        post = None
         while True:
-            root = self._api_request(params)
+            try:
+                root = self._api_request(params)
+            except ElementTree.ParseError:
+                if "tags" not in params or post is None:
+                    raise
+                taglist = [tag for tag in params["tags"].split()
+                           if not tag.startswith("id:<")]
+                taglist.append("id:<" + str(post.attrib["id"]))
+                params["tags"] = " ".join(taglist)
+                params["pid"] = 0
+                continue
+
+            post = None
             for post in root:
                 yield post.attrib