slight link extraction refactor

.gitignore

@@ -3,5 +3,6 @@
 *.mp4
 *.pyc
 *.secret
+.cache
 .vscode
 __pycache__

Pipfile

@@ -16,3 +16,4 @@ youtube-dl = "==2017.11.26"
 [dev-packages]
 
 "e1839a8" = {path = ".", editable = true}
+pytest = "*"

Pipfile.lock

@@ -1,7 +1,7 @@
 {
     "_meta": {
         "hash": {
-            "sha256": "fc88391d1ef7a5701ec70beb432898e855e039105cb787bf9dd91ba3aa35c604"
+            "sha256": "971f24d8efade8b46ac288d607d6200a9ecab3a67a81b3f717d38baf720be05d"
         },
         "host-environment-markers": {
             "implementation_name": "cpython",
@@ -153,6 +153,13 @@
         }
     },
     "develop": {
+        "attrs": {
+            "hashes": [
+                "sha256:a17a9573a6f475c99b551c0e0a812707ddda1ec9653bed04c13841404ed6f450",
+                "sha256:1c7960ccfd6a005cd9f7ba884e6316b5e430a3f1a6c37c5f87d8b43f83b54ec9"
+            ],
+            "version": "==17.4.0"
+        },
         "certifi": {
             "hashes": [
                 "sha256:14131608ad2fd56836d33a71ee60fa1c82bc9d2c8d98b7bdbc631fe1b3cd1296",
@@ -232,6 +239,26 @@
             ],
             "version": "==1.2.1"
         },
+        "pluggy": {
+            "hashes": [
+                "sha256:7f8ae7f5bdf75671a718d2daf0a64b7885f74510bcd98b1a0bb420eb9a9d0cff"
+            ],
+            "version": "==0.6.0"
+        },
+        "py": {
+            "hashes": [
+                "sha256:8cca5c229d225f8c1e3085be4fcf306090b00850fefad892f9d96c7b6e2f310f",
+                "sha256:ca18943e28235417756316bfada6cd96b23ce60dd532642690dcfdaba988a76d"
+            ],
+            "version": "==1.5.2"
+        },
+        "pytest": {
+            "hashes": [
+                "sha256:b84878865558194630c6147f44bdaef27222a9f153bbd4a08908b16bf285e0b1",
+                "sha256:53548280ede7818f4dc2ad96608b9f08ae2cc2ca3874f2ceb6f97e3583f25bc4"
+            ],
+            "version": "==3.3.2"
+        },
         "python-dateutil": {
             "hashes": [
                 "sha256:95511bae634d69bc7329ba55e646499a842bc4ec342ad54a8cdb65645a0aad3c",

fediplay.py

@@ -85,8 +85,8 @@ def stream(api_base_url):
 def extract_tags(toot):
     return [tag['name'] for tag in toot['tags']]
 
-def has_external_link_class(class_string):
+def link_is_internal(link):
-    classes = class_string.split(' ')
+    classes = link.attrib.get('class', '').split(' ')
     if classes:
         return 'mention' in classes
 
@@ -95,7 +95,7 @@ def has_external_link_class(class_string):
 def extract_links(toot):
     html = HTML(toot['content'])
     all_links = html.cssselect('a')
-    return [link.attrib['href'] for link in all_links if not has_external_link_class(link.attrib.get('class', ''))]
+    return [link.attrib['href'] for link in all_links if not link_is_internal(link)]
 
 def main():
     from getpass import getpass

test_fediplay.py

@@ -0,0 +1,8 @@
+import fediplay
+
+def test_extract_links():
+    toot = {
+        'content': "<p><a href=\"https://cybre.space/tags/nowplaying\" class=\"mention hashtag\" rel=\"tag\">#<span>nowplaying</span></a> <a href=\"https://cybre.space/tags/fediplay\" class=\"mention hashtag\" rel=\"tag\">#<span>fediplay</span></a> Grimes ft. Janelle Mon\u00e1e - Venus Fly <a href=\"https://www.youtube.com/watch?v=eTLTXDHrgtw\" rel=\"nofollow noopener\" target=\"_blank\"><span class=\"invisible\">https://www.</span><span class=\"ellipsis\">youtube.com/watch?v=eTLTXDHrgt</span><span class=\"invisible\">w</span></a></p>"
+    }
+    urls = fediplay.extract_links(toot)
+    assert urls == ['https://www.youtube.com/watch?v=eTLTXDHrgtw']