Added function & tests for getting user profile (#65)

test_vx.py

@@ -7,6 +7,7 @@ import msgs
 from flask.testing import FlaskClient
 client = FlaskClient(twitfix.app)
 
+testUser="https://twitter.com/jack"
 testTextTweet="https://twitter.com/jack/status/20"
 testVideoTweet="https://twitter.com/Twitter/status/1263145271946551300"
 testMediaTweet="https://twitter.com/Twitter/status/1118295916874739714"
@@ -40,6 +41,12 @@ def test_textTweetExtract():
     assert 'extended_entities' not in tweet
     assert tweet["is_quote_status"]==False
 
+def test_textUserExtract():
+    user = twExtract.extractUser(testUser)
+    assert user["screen_name"]=="jack"
+    assert user["id"]==12
+    assert user["created_at"] == "Tue Mar 21 20:50:14 +0000 2006"
+
 def test_videoTweetExtract():
     tweet = twExtract.extractStatus(testVideoTweet)
     assert tweet["full_text"]==videoVNF_compare['description']

twExtract/__init__.py

@@ -7,7 +7,7 @@ from . import twExtractError
 
 guestToken=None
 pathregex = r"\w{1,15}\/(status|statuses)\/(\d{2,20})"
-
+userregex = r"^https?:\/\/(?:www\.)?twitter\.com\/(?:#!\/)?@?([^/?#]*)(?:[?#].*)?$"
 def getGuestToken():
     global guestToken
     if guestToken is None:
@@ -50,6 +50,21 @@ def extractStatus(url):
     except Exception as e:
         return extractStatus_fallback(url)
 
+def extractUser(url):
+    m = re.search(userregex, url)
+    if m is None:
+        raise twExtractError.TwExtractError(400, "Invalid URL")
+    screen_name = m.group(1)
+    # get guest token
+    guestToken = getGuestToken()
+    # get user
+    user = requests.get(f"https://api.twitter.com/1.1/users/show.json?screen_name={screen_name}",headers={"Authorization":"Bearer AAAAAAAAAAAAAAAAAAAAAPYXBAAAAAAACLXUNDekMxqa8h%2F40K4moUkGsoc%3DTYfbDKbT3jJPCEVnMYqilB28NHfOPqkca3qaAxGfsyKCs0wRbw", "x-guest-token":guestToken})
+    output = user.json()
+    if "errors" in output:
+        # pick the first error and create a twExtractError
+        error = output["errors"][0]
+        raise twExtractError.TwExtractError(error["code"], error["message"])
+    return output
 
 def lambda_handler(event, context):
     if ("queryStringParameters" not in event):