Fixed BongaCams, links parsing improved (#297)

* Fixed BongaCams, links parsing improved * Fixed tests
soxoj · Jan 8, 2022 · 5c05cfa · 5c05cfa
1 parent 3e884d4
commit 5c05cfa
Show file tree

Hide file tree

Showing 4 changed files with 13 additions and 7 deletions.
diff --git a/maigret/resources/data.json b/maigret/resources/data.json
@@ -2144,11 +2144,17 @@
                 "cz",
                 "webcam"
             ],
-            "checkType": "status_code",
+            "absenceStrs": [
+                "<link rel=\"preconnect\""
+            ],
+            "presenseStrs": [
+                "<title>Informa\u00e7\u00e3o e p\u00e1gina"
+            ],
+            "checkType": "message",
             "alexaRank": 30,
             "urlMain": "https://pt.bongacams.com",
             "url": "https://pt.bongacams.com/profile/{username}",
-            "usernameClaimed": "asuna-black",
+            "usernameClaimed": "Icehotangel",
             "usernameUnclaimed": "noonewouldeverusethis77777"
         },
         "Bookandreader": {
@@ -13126,7 +13132,7 @@
             "checkType": "response_url",
             "alexaRank": 82345,
             "urlMain": "http://sprashivai.ru",
-            "url": "http://sprashivai.ru/{username}?sl",
+            "url": "http://sprashivai.ru/{username}",
             "usernameClaimed": "red",
             "usernameUnclaimed": "noonewouldeverusethis7"
         },

diff --git a/maigret/utils.py b/maigret/utils.py
@@ -42,7 +42,7 @@ def enrich_link_str(link: str) -> str:
 
 
 class URLMatcher:
-    _HTTP_URL_RE_STR = "^https?://(www.)?(.+)$"
+    _HTTP_URL_RE_STR = "^https?://(www.|m.)?(.+)$"
     HTTP_URL_RE = re.compile(_HTTP_URL_RE_STR)
     UNSAFE_SYMBOLS = ".?"
 
@@ -66,7 +66,7 @@ def make_profile_url_regexp(self, url: str, username_regexp: str = ""):
         )
         regexp_str = self._HTTP_URL_RE_STR.replace("(.+)", url_regexp)
 
-        return re.compile(regexp_str)
+        return re.compile(regexp_str, re.IGNORECASE)
 
 
 def ascii_data_display(data: str) -> Any:

diff --git a/tests/test_sites.py b/tests/test_sites.py
@@ -115,7 +115,7 @@ def test_site_url_detector():
 
     assert (
         db.sites[0].url_regexp.pattern
-        == r'^https?://(www.)?forum\.amperka\.ru/members/\?username=(.+?)$'
+        == r'^https?://(www.|m.)?forum\.amperka\.ru/members/\?username=(.+?)$'
     )
     assert (
         db.sites[0].detect_username('http://forum.amperka.ru/members/?username=test')

diff --git a/tests/test_utils.py b/tests/test_utils.py
@@ -98,7 +98,7 @@ def test_url_make_profile_url_regexp():
         # ensure all combinations match pattern
         assert (
             URLMatcher.make_profile_url_regexp(url).pattern
-            == r'^https?://(www.)?flickr\.com/photos/(.+?)$'
+            == r'^https?://(www.|m.)?flickr\.com/photos/(.+?)$'
         )