Fix inconsistent invalid URL handling (#1832)

Author: rodrigo.nogueira

httpx/_client.py

@@ -394,6 +394,8 @@ def _merge_url(self, url: URL | str) -> URL:
         to create the URL used for the outgoing request.
         """
         merge_url = URL(url)
+        if merge_url.scheme and not merge_url.host:
+            raise InvalidURL(f"Invalid URL '{url}': has scheme but missing host")
         if merge_url.is_relative_url:
             # To merge URLs we always append to the base URL. To get this
             # behaviour correct we always ensure the base URL ends in a '/'

tests/client/test_async_client.py

@@ -32,7 +32,9 @@ async def test_get(server):
 @pytest.mark.anyio
 async def test_get_invalid_url(server, url):
     async with httpx.AsyncClient() as client:
-        with pytest.raises((httpx.UnsupportedProtocol, httpx.LocalProtocolError)):
+        with pytest.raises(
+            (httpx.UnsupportedProtocol, httpx.LocalProtocolError, httpx.InvalidURL)
+        ):
             await client.get(url)
 
 

tests/client/test_client.py

@@ -40,10 +40,27 @@ def test_get(server):
 )
 def test_get_invalid_url(server, url):
     with httpx.Client() as client:
-        with pytest.raises((httpx.UnsupportedProtocol, httpx.LocalProtocolError)):
+        with pytest.raises(
+            (httpx.UnsupportedProtocol, httpx.LocalProtocolError, httpx.InvalidURL)
+        ):
             client.get(url)
 
 
+def test_get_invalid_url_with_scheme_no_host():
+    """
+    Regression test for: https://github.com/encode/httpx/issues/1832
+    URLs with scheme but no host should raise InvalidURL.
+    """
+    with httpx.Client() as client:
+        with pytest.raises(httpx.InvalidURL) as exc:
+            client.get("https:/google.com")
+        assert "has scheme but missing host" in str(exc.value)
+
+        with pytest.raises(httpx.InvalidURL) as exc:
+            client.get("https:///google.com")
+        assert "has scheme but missing host" in str(exc.value)
+
+
 def test_build_request(server):
     url = server.url.copy_with(path="/echo_headers")
     headers = {"Custom-header": "value"}