From 8a6e20860524a177d051a21312b8cf6bb1b8154b Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Mike=20F=C3=A4hrmann?= <mike_faehrmann@web.de>
Date: Mon, 29 Jul 2024 11:19:10 +0200
Subject: [PATCH] [zerochan] fix 'Invalid control character' errors (#5892)

---
 gallery_dl/extractor/zerochan.py | 10 +++++++++-
 test/results/zerochan.py         | 11 +++++++++++
 2 files changed, 20 insertions(+), 1 deletion(-)

diff --git a/gallery_dl/extractor/zerochan.py b/gallery_dl/extractor/zerochan.py
index 49b06a77..126ef494 100644
--- a/gallery_dl/extractor/zerochan.py
+++ b/gallery_dl/extractor/zerochan.py
@@ -12,6 +12,7 @@ from .booru import BooruExtractor
 from ..cache import cache
 from .. import text, util, exception
 import collections
+import re
 
 BASE_PATTERN = r"(?:https?://)?(?:www\.)?zerochan\.net"
 
@@ -92,7 +93,14 @@ class ZerochanExtractor(BooruExtractor):
 
     def _parse_entry_api(self, entry_id):
         url = "{}/{}?json".format(self.root, entry_id)
-        item = self.request(url).json()
+        text = self.request(url).text
+        try:
+            item = util.json_loads(text)
+        except ValueError as exc:
+            if " control character " not in str(exc):
+                raise
+            text = re.sub(r"[\x00-\x1f\x7f]", "", text)
+            item = util.json_loads(text)
 
         data = {
             "id"      : item["id"],
diff --git a/test/results/zerochan.py b/test/results/zerochan.py
index f1788c2b..ec8eddf6 100644
--- a/test/results/zerochan.py
+++ b/test/results/zerochan.py
@@ -177,4 +177,15 @@ __tests__ = (
     "width"    : 750,
 },
 
+{
+    "#url"     : "https://www.zerochan.net/1395035",
+    "#comment" : "Invalid control character '\r' in 'source' field (#5892)",
+    "#category": ("booru", "zerochan", "image"),
+    "#class"   : zerochan.ZerochanImageExtractor,
+    "#auth"    : True,
+    "#options" : {"metadata": True},
+
+    "source": "http://www.youtube.com/watch?v=0vodqkGPxt8",
+},
+
 )