Use redirected URL, closes #7

Author: meyt

linkpreview/compose.py

@@ -11,7 +11,7 @@ def link_preview(
     if content is None:
         try:
             grabber = LinkGrabber()
-            content = grabber.get_content(url)
+            content, url = grabber.get_content(url)
         except InvalidMimeTypeError:
             content = ''
 

linkpreview/grabber.py

@@ -58,4 +58,4 @@ def get_content(self, url: str, headers: dict = None):
 
             content += chunk
 
-        return content
+        return content, r.url

tests/test_grabber.py

@@ -45,6 +45,20 @@ class FakeResponse(Response):
             headers={"content-length": "100000"},
         )
     )
+    httpserver.expect_request("/redirected").respond_with_response(
+        FakeResponse(
+            mimetype="text/html",
+            response=b"done!",
+        )
+    )
+    redirected_url = "http://%s:%s/redirected" % (httpserver.host, httpserver.port)
+    httpserver.expect_request("/redirection").respond_with_response(
+        FakeResponse(
+            mimetype="text/html",
+            headers={"location": redirected_url},
+            status=301
+        )
+    )
 
     # success
     grabber = LinkGrabber(maxsize=100)
@@ -80,3 +94,8 @@ class FakeResponse(Response):
     grabber = LinkGrabber()
     with pytest.raises(exceptions.InvalidContentError):
         grabber.get_content(httpserver.url_for("/badmime"))
+
+    # redirection
+    grabber = LinkGrabber()
+    content, url = grabber.get_content(httpserver.url_for("/redirection"))
+    assert url == redirected_url

tests/test_preview.py

@@ -2,6 +2,8 @@
 
 from pytest_httpserver import HTTPServer
 
+from werkzeug.wrappers.response import Response
+
 from linkpreview import Link, link_preview
 from linkpreview.preview import OpenGraph, TwitterCard, Schema, Generic
 
@@ -184,6 +186,18 @@ def test_link_preview(httpserver: HTTPServer):
         '{}',
         headers={"content-type": "application/json"},
     )
+    httpserver.expect_request("/redirected").respond_with_data(
+        get_sample("generic/h1-img.html"),
+        headers={"content-type": "text/html"},
+    )
+    redirected_url = "http://%s:%s/redirected" % (httpserver.host, httpserver.port)
+    httpserver.expect_request("/redirection").respond_with_response(
+        Response(
+            mimetype="text/html",
+            headers={"location": redirected_url},
+            status=301
+        )
+    )
 
     url = httpserver.url_for("/preview1")
     preview = link_preview(url)
@@ -235,3 +249,11 @@ def test_link_preview(httpserver: HTTPServer):
     assert preview.image is None
     assert preview.absolute_image is None
     assert preview.force_title == "Preview 3"
+
+    url = httpserver.url_for("/redirection")
+    preview = link_preview(url)
+    assert preview.link.url == redirected_url
+    assert preview.title == "This title is from the first h1 tag."
+    assert preview.description is None
+    assert preview.image == "http://localhost:8000/img/heck.jpg"
+    assert preview.absolute_image == "http://localhost:8000/img/heck.jpg"