From 7e040086b69c1156c8175d6eb43d7429c079da41 Mon Sep 17 00:00:00 2001
From: Tobias Gruetzmacher <tobias-git@23.gs>
Date: Mon, 28 Sep 2020 13:11:34 +0200
Subject: [PATCH] Try to inform the user about geo-blocks

Instead of letting the crawler run into "random" error messages, throw a
specific "geoblocked" exception instead.
---
 dosagelib/plugins/s.py           | 12 ++++++++++++
 dosagelib/plugins/smackjeeves.py |  3 +++
 dosagelib/scraper.py             |  9 +++++++++
 tests/test_modules.py            | 20 +++++++++++++++++++-
 4 files changed, 43 insertions(+), 1 deletion(-)

diff --git a/dosagelib/plugins/s.py b/dosagelib/plugins/s.py
index 548a87858..e4d003426 100644
--- a/dosagelib/plugins/s.py
+++ b/dosagelib/plugins/s.py
@@ -6,6 +6,8 @@
 from re import compile, escape, IGNORECASE, sub
 from os.path import splitext
 
+from requests.exceptions import HTTPError
+
 from ..scraper import _BasicScraper, _ParserScraper
 from ..helpers import indirectStarter, bounceStarter, joinPathPartsNamer
 from ..util import tagre
@@ -385,6 +387,16 @@ class SoloLeveling(_ParserScraper):
         self.imageUrls = [self.imageUrlModifier(x, data) for x in self.imageUrls]
         return self.imageUrls
 
+    def getPage(self, url):
+        try:
+            return super().getPage(url)
+        except HTTPError as e:
+            # CloudFlare WAF
+            if e.response.status_code == 403 and '1020' in e.response.text:
+                self.geoblocked()
+            else:
+                raise e
+
     def getPrevUrl(self, url, data):
         return self.stripUrl % str(int(url.strip('/').rsplit('-', 1)[-1]) - 1)
 
diff --git a/dosagelib/plugins/smackjeeves.py b/dosagelib/plugins/smackjeeves.py
index 78d16b176..69a13c553 100644
--- a/dosagelib/plugins/smackjeeves.py
+++ b/dosagelib/plugins/smackjeeves.py
@@ -29,6 +29,9 @@ class SmackJeeves(_ParserScraper):
         response = self.session.post(self.apiBase + 'articleList',
             params={'titleNo': self._comicid})
         response.raise_for_status()
+        if ('text/html' in response.headers['content-type'] and
+                'available in your area' in response.text):
+            self.geoblocked()
         return response.json()['result']['list'][self.lastid]['articleUrl']
 
     def fetchUrls(self, url, data, urlsearch):
diff --git a/dosagelib/scraper.py b/dosagelib/scraper.py
index 33b37790c..23f1d2b5c 100644
--- a/dosagelib/scraper.py
+++ b/dosagelib/scraper.py
@@ -32,6 +32,11 @@ from .xml import NS
 ARCHIVE_ORG_URL = re.compile(r'https?://web\.archive\.org/web/[^/]*/')
 
 
+class GeoblockedException(IOError):
+    def __init__(self):
+        super().__init__(f'It seems your current location is geo-blocked.')
+
+
 class Scraper(object):
     '''Base class for all comic scraper, but without a specific scrape
     implementation.'''
@@ -346,6 +351,10 @@ class Scraper(object):
                     pass
         return lang
 
+    def geoblocked(self):
+        """Helper method to indicate that the user is most probably geo-blocked."""
+        raise GeoblockedException()
+
 
 class _BasicScraper(Scraper):
     """
diff --git a/tests/test_modules.py b/tests/test_modules.py
index 0fa85f4bb..b08346231 100644
--- a/tests/test_modules.py
+++ b/tests/test_modules.py
@@ -7,7 +7,9 @@ import responses
 
 import dosagelib.cmd
 import httpmocks
-
+from dosagelib.plugins.s import SoloLeveling
+from dosagelib.plugins.smackjeeves import SmackJeeves
+from dosagelib.scraper import GeoblockedException
 
 def cmd(*options):
     """'Fake' run dosage with given options."""
@@ -39,3 +41,19 @@ class TestModules(object):
 
         cmd('--basepath', str(tmpdir), 'CalvinAndHobbesEnEspanol')
         cmd('--basepath', str(tmpdir), 'CalvinAndHobbesEnEspanol:2012/07/22')
+
+    @responses.activate
+    def test_smackjeeves_geoblock(self, tmpdir):
+        responses.add(responses.POST, re.compile('https://www.smackjeeves.com/api/.*'),
+            'is not currently available in your area', content_type='text/html')
+
+        with pytest.raises(GeoblockedException):
+            next(SmackJeeves.getmodules()[0].getStrips(1))
+
+    @responses.activate
+    def test_sololeveling_geoblock(self, tmpdir):
+        responses.add(responses.GET, 'https://w1.sololeveling.net/',
+            '<span>1020</span>', status=403)
+
+        with pytest.raises(GeoblockedException):
+            next(SoloLeveling.getmodules()[0].getStrips(1))