BACK-1720: implement HEAD requests for metadata fetcher

offchain/metadata/adapters/arweave.py

@@ -60,7 +60,7 @@ def parse_ar_url(self, url: str) -> str:
         return url
 
     async def gen_send(self, url: str, sess: httpx.AsyncClient(), *args, **kwargs) -> httpx.Response:  # type: ignore[no-untyped-def, valid-type]  # noqa: E501
-        """Format and send async request to ARWeave host.
+        """Format and send an async `GET` request to ARWeave host at parsed url.
 
         Args:
             url (str): url to send request to
@@ -72,7 +72,7 @@ async def gen_send(self, url: str, sess: httpx.AsyncClient(), *args, **kwargs) -
         return await sess.get(self.parse_ar_url(url), timeout=self.timeout, follow_redirects=True)  # type: ignore[no-any-return]  # noqa: E501
 
     def send(self, request: PreparedRequest, *args, **kwargs) -> Response:  # type: ignore[no-untyped-def]  # noqa: E501
-        """Format and send request to ARWeave host.
+        """Format and send a `GET` request to ARWeave host at parsed url.
 
         Args:
             request (PreparedRequest): incoming request
@@ -83,3 +83,15 @@ def send(self, request: PreparedRequest, *args, **kwargs) -> Response:  # type:
         request.url = self.parse_ar_url(request.url)  # type: ignore[arg-type]
         kwargs["timeout"] = self.timeout
         return super().send(request, *args, **kwargs)
+
+    async def gen_head(self, url: str, sess: httpx.AsyncClient(), *args, **kwargs) -> httpx.Response:  # type: ignore[no-untyped-def, valid-type]  # noqa: E501
+        """Format and send an async `HEAD` request to ARWeave host at parsed url.
+
+        Args:
+            url (str): url to send request to
+            sess (httpx.AsyncClient()): async client
+
+        Returns:
+            httpx.Response: response from ARWeave host.
+        """
+        return await sess.head(self.parse_ar_url(url), timeout=self.timeout, follow_redirects=True)  # type: ignore[no-any-return]  # noqa: E501

offchain/metadata/adapters/base_adapter.py

@@ -14,7 +14,22 @@ def __init__(self, *args, **kwargs):  # type: ignore[no-untyped-def]
         super().__init__()
 
     async def gen_send(self, url: str, *args, **kwargs) -> httpx.Response:  # type: ignore[no-untyped-def]  # noqa: E501
-        """Format and send async request to url host.
+        """
+        Format and send an async `GET` request to url host.
+        Abstract method, implemented in subclasses.
+
+        Args:
+            url (str): url to send request to
+
+        Returns:
+            httpx.Response: response from host.
+        """
+        raise NotImplementedError
+
+    async def gen_head(self, url: str, *args, **kwargs) -> httpx.Response:  # type: ignore[no-untyped-def]  # noqa: E501
+        """
+        Format and send an async `HEAD` request to url host.
+        Abstract method, implemented in subclasses.
 
         Args:
             url (str): url to send request to
@@ -40,7 +55,7 @@ def __init__(  # type: ignore[no-untyped-def]
         super().__init__(pool_connections, pool_maxsize, max_retries, pool_block)
 
     async def gen_send(self, url: str, sess: httpx.AsyncClient(), *args, **kwargs) -> httpx.Response:  # type: ignore[no-untyped-def, valid-type]  # noqa: E501
-        """Format and send async request to url host.
+        """Format and send an async `GET` request to url host.
 
         Args:
             url (str): url to send request to
@@ -50,6 +65,18 @@ async def gen_send(self, url: str, sess: httpx.AsyncClient(), *args, **kwargs) -
         """
         return await sess.get(url, follow_redirects=True)  # type: ignore[no-any-return]
 
+    async def gen_head(self, url: str, sess: httpx.AsyncClient(), *args, **kwargs) -> httpx.Response:  # type: ignore[no-untyped-def, valid-type]  # noqa: E501
+        """Format and send an async `HEAD` request to url host.
+
+        Args:
+            url (str): url to send request to
+            sess (httpx.AsyncClient()): async client
+
+        Returns:
+            httpx.Response: response from host.
+        """
+        return await sess.head(url, follow_redirects=True)  # type: ignore[no-any-return]
+
 
 Adapter = Union[BaseAdapter, HTTPAdapter]
 

offchain/metadata/adapters/data_uri.py

@@ -1,4 +1,5 @@
 import base64
+from email.message import Message
 from urllib.request import urlopen
 
 import httpx
@@ -28,7 +29,7 @@ def __init__(self, *args, **kwargs):  # type: ignore[no-untyped-def]
         super().__init__(*args, **kwargs)  # type: ignore[no-untyped-call]
 
     async def gen_send(self, url: str, *args, **kwargs) -> httpx.Response:  # type: ignore[no-untyped-def]  # noqa: E501
-        """Handle async data uri request.
+        """Handle async data uri `GET` request.
 
         Args:
             url (str): url
@@ -44,7 +45,7 @@ async def gen_send(self, url: str, *args, **kwargs) -> httpx.Response:  # type:
         return response
 
     def send(self, request: PreparedRequest, *args, **kwargs):  # type: ignore[no-untyped-def]  # noqa: E501
-        """Handle data uri request.
+        """Handle data uri `GET` request.
 
         Args:
             request (PreparedRequest): incoming request
@@ -66,5 +67,25 @@ def send(self, request: PreparedRequest, *args, **kwargs):  # type: ignore[no-un
         finally:
             return newResponse
 
+    async def gen_head(self, url: str, *args, **kwargs) -> httpx.Response:  # type: ignore[no-untyped-def]  # noqa: E501
+        """Handle async data uri `HEAD` request.
+
+        Args:
+            url (str): url
+
+        Returns:
+            httpx.Response: encoded data uri response.
+        """
+        response_headers = {}
+        with urlopen(url) as r:
+            message: Message = r.info()
+            response_headers = dict(message._headers)  # type: ignore[attr-defined]
+        response = httpx.Response(
+            status_code=200,
+            headers=response_headers,
+            request=httpx.Request(method="HEAD", url=url),
+        )
+        return response
+
     def close(self):  # type: ignore[no-untyped-def]
         self.response.close()

offchain/metadata/adapters/ipfs.py

@@ -97,7 +97,7 @@ def make_request_url(self, request_url: str, gateway: Optional[str] = None) -> s
         return build_request_url(gateway=gateway, request_url=request_url)
 
     async def gen_send(self, url: str, sess: httpx.AsyncClient(), *args, **kwargs) -> httpx.Response:  # type: ignore[no-untyped-def, valid-type]  # noqa: E501
-        """Format and send async request to IPFS host.
+        """Format and send an async `GET` request to IPFS host.
 
         Args:
             url (str): url to send request to
@@ -121,3 +121,15 @@ def send(self, request: PreparedRequest, *args, **kwargs) -> Response:  # type:
 
         kwargs["timeout"] = self.timeout
         return super().send(request, *args, **kwargs)
+
+    async def gen_head(self, url: str, sess: httpx.AsyncClient(), *args, **kwargs) -> httpx.Response:  # type: ignore[no-untyped-def, valid-type]  # noqa: E501
+        """Format and send an async `HEAD` request to IPFS host.
+
+        Args:
+            url (str): url to send request to
+            sess (httpx.AsyncClient()): async client session
+
+        Returns:
+            httpx.Response: response from IPFS host.
+        """
+        return await sess.head(self.make_request_url(url), timeout=self.timeout, follow_redirects=True)  # type: ignore[no-any-return]  # noqa: E501

offchain/metadata/fetchers/metadata_fetcher.py

@@ -63,7 +63,7 @@ def _head(self, uri: str):  # type: ignore[no-untyped-def]
     def _get(self, uri: str):  # type: ignore[no-untyped-def]
         return self.sess.get(uri, timeout=self.timeout, allow_redirects=True)
 
-    async def _gen(self, uri: str) -> httpx.Response:
+    async def _gen(self, uri: str, method: Optional[str] = "GET") -> httpx.Response:
         from offchain.metadata.pipelines.metadata_pipeline import (
             DEFAULT_ADAPTER_CONFIGS,
         )
@@ -78,13 +78,21 @@ async def _gen(self, uri: str) -> httpx.Response:
                 adapter = adapter_config.adapter_cls(
                     host_prefixes=adapter_config.host_prefixes, **adapter_config.kwargs
                 )
-                return await adapter.gen_send(
-                    url=uri, timeout=self.timeout, sess=self.async_sess
-                )
+                if method == "HEAD":
+                    return await adapter.gen_head(
+                        url=uri, timeout=self.timeout, sess=self.async_sess
+                    )
+                else:
+                    return await adapter.gen_send(
+                        url=uri, timeout=self.timeout, sess=self.async_sess
+                    )
         return await self.async_sess.get(
             uri, timeout=self.timeout, follow_redirects=True
         )
 
+    async def _gen_head(self, uri: str) -> httpx.Response:
+        return await self._gen(uri=uri, method="HEAD")
+
     def fetch_mime_type_and_size(self, uri: str) -> Tuple[str, int]:
         """Fetch the mime type and size of the content at a given uri.
 
@@ -123,11 +131,10 @@ async def gen_fetch_mime_type_and_size(self, uri: str) -> Tuple[str, int]:
             tuple[str, int]: mime type and size
         """
         try:
-            # try skip head request
-            # res = await self._gen_head(uri)
-            # # For any error status, try a get
-            # if 300 <= res.status_code < 600:
-            res = await self._gen(uri)
+            res = await self._gen_head(uri)
+            # For any error status, try a get
+            if 300 <= res.status_code < 600:
+                res = await self._gen(uri)
             res.raise_for_status()
             headers = res.headers
             size = headers.get("content-length", 0)