From 45a6aec7da1407243f9767c6ab0cff40efee72eb Mon Sep 17 00:00:00 2001
From: Nico Kemnitz <nico@zetta.ai>
Date: Thu, 10 Aug 2023 17:45:00 +0200
Subject: [PATCH] Verify Content-Encoding when querying Content-Length (#1320)

---
 fsspec/implementations/http.py            |  6 ++++--
 fsspec/implementations/tests/test_http.py | 13 +++++++++++++
 fsspec/tests/conftest.py                  | 14 +++++++++++++-
 3 files changed, 30 insertions(+), 3 deletions(-)

diff --git a/fsspec/implementations/http.py b/fsspec/implementations/http.py
index afd0c2664..5d118dcbd 100644
--- a/fsspec/implementations/http.py
+++ b/fsspec/implementations/http.py
@@ -195,7 +195,6 @@ async def _ls_real(self, url, detail=True, **kwargs):
             return list(sorted(out))
 
     async def _ls(self, url, detail=True, **kwargs):
-
         if self.use_listings_cache and url in self.dircache:
             out = self.dircache[url]
         else:
@@ -841,7 +840,10 @@ async def _file_info(url, session, size_policy="head", **kwargs):
         #                 or 'Accept-Ranges': 'none' (not 'bytes')
         #  to mean streaming only, no random access => return None
         if "Content-Length" in r.headers:
-            info["size"] = int(r.headers["Content-Length"])
+            # Some servers may choose to ignore Accept-Encoding and return
+            # compressed content, in which case the returned size is unreliable.
+            if r.headers.get("Content-Encoding", "identity") == "identity":
+                info["size"] = int(r.headers["Content-Length"])
         elif "Content-Range" in r.headers:
             info["size"] = int(r.headers["Content-Range"].split("/")[1])
 
diff --git a/fsspec/implementations/tests/test_http.py b/fsspec/implementations/tests/test_http.py
index d8bd64524..fb3a55cad 100644
--- a/fsspec/implementations/tests/test_http.py
+++ b/fsspec/implementations/tests/test_http.py
@@ -10,6 +10,7 @@
 
 import fsspec.asyn
 import fsspec.utils
+from fsspec.implementations.http import HTTPStreamFile
 from fsspec.tests.conftest import data, reset_files, server, win  # noqa: F401
 
 
@@ -280,6 +281,18 @@ def test_content_length_zero(server):
         assert f.read() == data
 
 
+def test_content_encoding_gzip(server):
+    h = fsspec.filesystem(
+        "http", headers={"give_length": "true", "gzip_encoding": "true"}
+    )
+    url = server + "/index/realfile"
+
+    with h.open(url, "rb") as f:
+        assert isinstance(f, HTTPStreamFile)
+        assert f.size is None
+        assert f.read() == data
+
+
 def test_download(server, tmpdir):
     h = fsspec.filesystem("http", headers={"give_length": "true", "head_ok": "true "})
     url = server + "/index/realfile"
diff --git a/fsspec/tests/conftest.py b/fsspec/tests/conftest.py
index fdaf03335..9fdf25b7a 100644
--- a/fsspec/tests/conftest.py
+++ b/fsspec/tests/conftest.py
@@ -1,4 +1,5 @@
 import contextlib
+import gzip
 import json
 import os
 import threading
@@ -76,7 +77,14 @@ def do_GET(self):
             if "use_206" in self.headers:
                 status = 206
         if "give_length" in self.headers:
-            response_headers = {"Content-Length": len(file_data)}
+            if "gzip_encoding" in self.headers:
+                file_data = gzip.compress(file_data)
+                response_headers = {
+                    "Content-Length": len(file_data),
+                    "Content-Encoding": "gzip",
+                }
+            else:
+                response_headers = {"Content-Length": len(file_data)}
             self._respond(status, response_headers, file_data)
         elif "give_range" in self.headers:
             self._respond(status, {"Content-Range": content_range}, file_data)
@@ -123,6 +131,10 @@ def do_HEAD(self):
             response_headers = {"Content-Length": len(file_data)}
             if "zero_length" in self.headers:
                 response_headers["Content-Length"] = 0
+            elif "gzip_encoding" in self.headers:
+                file_data = gzip.compress(file_data)
+                response_headers["Content-Encoding"] = "gzip"
+                response_headers["Content-Length"] = len(file_data)
 
             self._respond(200, response_headers)
         elif "give_range" in self.headers: