Merge branch 'develop' into custom-auth-code-templates

.pre-commit-config.yaml

@@ -1,5 +1,5 @@
 repos:
   - repo: https://github.com/psf/black
-    rev: 23.12.1
+    rev: 24.1.1
     hooks:
       - id: black

CHANGELOG.md

@@ -7,6 +7,11 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ## [Unreleased]
 ### Fixed
 - Remove deprecation warnings due to usage of `utcnow` and `utcfromtimestamp`. Thanks to [`Raphael Krupinski`](https://github.com/rafalkrupinski).
+- `httpx_auth.AWS4Auth.default_include_headers` value kept growing in size every time a new `httpx_auth.AWS4Auth` instance was created with `security_token` parameter provided. Thanks to [`Miikka Koskinen`](https://github.com/miikka).
+
+### Changed
+- `httpx_auth.AWS4Auth.default_include_headers` is not available anymore, use `httpx_auth.AWS4Auth` `include_headers` parameter instead to change the list of included headers if the default does not fit your need ().
+- `httpx_auth.AWS4Auth` `include_headers` values will not be stripped anymore, meaning that you can now include headers prefixed and/or suffixed with blank spaces.
 
 ## [0.19.0] - 2024-01-09
 ### Added
@@ -114,7 +119,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Changed
 - `get_token` cache method now requires `on_missing_token` function args to be provided as kwargs instead of args.
-- `get_token` cache method now requires `on_missing_token` parameter to be provided as a non positional argument.
+- `get_token` cache method now requires `on_missing_token` parameter to be provided as a non-positional argument.
 - `get_token` cache method now expose `early_expiry` parameter, defaulting to 30 seconds.
 
 ### Fixed
@@ -166,13 +171,13 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ### Added
 - Still under development, subject to breaking changes without notice: `AWS4Auth` authentication class for AWS. Ported from [`requests-aws4auth`](https://github.com/sam-washington/requests-aws4auth) by [`Michael E. Martinka`](https://github.com/martinka).
 Note that a few changes were made:
-  - deprecated `amz_date` attribute has been removed.
-  - it is not possible to provide an `AWSSigningKey` instance, use explicit parameters instead.
-  - it is not possible to provide a `date`. It will default to now.
-  - it is not possible to provide `raise_invalid_date` parameter anymore as the date will always be valid.
+  - Deprecated `amz_date` attribute has been removed.
+  - It is not possible to provide an `AWSSigningKey` instance, use explicit parameters instead.
+  - It is not possible to provide a `date`. It will default to now.
+  - It is not possible to provide `raise_invalid_date` parameter anymore as the date will always be valid.
   - `include_hdrs` parameter was renamed into `include_headers`
   - `host` is not considered as a specific Amazon service anymore (no test specific code).
-  - Each request now has it's own signing key and x-amz-date. Meaning you can use the same auth instance for more than one request.
+  - Each request now has its own signing key and `x-amz-date`. Meaning you can use the same auth instance for more than one request.
   - `session_token` was renamed into `security_token` for consistency with the underlying name at Amazon.
 
 ## [0.3.0] - 2020-05-26

CONTRIBUTING.md

@@ -21,7 +21,7 @@ Before creating an issue please make sure that it was not already reported.
 
 1) Go to the *Issues* tab and click on the *New issue* button.
 2) Title should be a small sentence describing the request.
-3) The comment should contains as much information as possible
+3) The comment should contain as much information as possible
     * Actual behavior (including the version you used)
     * Expected behavior
     * Steps to reproduce
@@ -56,7 +56,7 @@ Before creating an issue please make sure that it was not already reported.
 1) Go to the *Pull requests* tab and click on the *New pull request* button.
 2) *base* should always be set to `develop` and it should be compared to your branch.
 3) Title should be a small sentence describing the request.
-3) The comment should contains as much information as possible
+4) The comment should contain as much information as possible
     * Actual behavior (before the new code)
     * Expected behavior (with the new code)
     * Steps to reproduce (with and without the new code to see the difference)

README.md

@@ -677,7 +677,7 @@ OAuth2.token_cache = JsonTokenFileCache('path/to/my_token_cache.json')
 
 ## AWS Signature v4
 
-Amazon Web Service Signature version 4 is implemented following [Amazon S3 documentation](https://docs.aws.amazon.com/AmazonS3/latest/API/sigv4-auth-using-authorization-header.html) and [request-aws4auth](https://github.com/sam-washington/requests-aws4auth).
+Amazon Web Service Signature version 4 is implemented following [Amazon S3 documentation](https://docs.aws.amazon.com/AmazonS3/latest/API/sigv4-auth-using-authorization-header.html) and [request-aws4auth 1.0.1](https://github.com/sam-washington/requests-aws4auth) (with some changes, see below).
 
 Use `httpx_auth.AWS4Auth` to configure this kind of authentication.
 
@@ -690,15 +690,28 @@ with httpx.Client() as client:
     client.get('http://s3-eu-west-1.amazonaws.com', auth=aws)
 ```
 
+Note that the following changes were made compared to `requests-aws4auth`:
+  - Each request now has its own signing key and `x-amz-date`. Meaning **you can use the same auth instance for more than one request**.
+  - `session_token` was renamed into `security_token` for consistency with the underlying name at Amazon.
+  - `include_hdrs` parameter was renamed into `include_headers`. When using this parameter:
+    - Provided values will not be stripped, [WYSIWYG](https://en.wikipedia.org/wiki/WYSIWYG).
+    - If multiple values are provided for a same header, the computation will be based on the value order you provided and value separated by `, `. Instead of ordered values separated by comma for `requests-aws4auth`.
+  - `amz_date` attribute has been removed.
+  - It is not possible to provide a `date`. It will default to now.
+  - It is not possible to provide an `AWSSigningKey` instance, use explicit parameters instead.
+  - It is not possible to provide `raise_invalid_date` parameter anymore as the date will always be valid.
+  - `host` is not considered as a specific Amazon service anymore (no test specific code).
+
 ### Parameters
 
-| Name             | Description                | Mandatory | Default value |
-|:-----------------|:---------------------------|:----------|:--------------|
-| `access_id`      | AWS access ID. | Mandatory | |
-| `secret_key`     | AWS secret access key. | Mandatory | |
-| `region`         | The region you are connecting to, as per [this list](http://docs.aws.amazon.com/general/latest/gr/rande.html#s3_region). For services which do not require a region (e.g. IAM), use us-east-1. | Mandatory | |
-| `service`        | The name of the service you are connecting to, as per [this list](http://docs.aws.amazon.com/general/latest/gr/rande.html). e.g. elasticbeanstalk. | Mandatory | |
-| `security_token` | Used for the `x-amz-security-token` header, for use with STS temporary credentials. | Optional | |
+| Name               | Description                                                                                                                                                                                                                | Mandatory  | Default value                                                                                                                    |
+|:-------------------|:---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:-----------|:---------------------------------------------------------------------------------------------------------------------------------|
+| `access_id`        | AWS access ID.                                                                                                                                                                                                             | Mandatory  |                                                                                                                                  |
+| `secret_key`       | AWS secret access key.                                                                                                                                                                                                     | Mandatory  |                                                                                                                                  |
+| `region`           | The region you are connecting to, as per [this list](http://docs.aws.amazon.com/general/latest/gr/rande.html#s3_region). For services which do not require a region (e.g. IAM), use us-east-1.                             | Mandatory  |                                                                                                                                  |
+| `service`          | The name of the service you are connecting to, as per [this list](http://docs.aws.amazon.com/general/latest/gr/rande.html). e.g. elasticbeanstalk.                                                                         | Mandatory  |                                                                                                                                  |
+| `security_token`   | Used for the `x-amz-security-token` header, for use with STS temporary credentials.                                                                                                                                        | Optional   |                                                                                                                                  |
+| `include_headers`  | Set of headers to include in the canonical and signed headers. Specific values are `x-amz-*` that matches any header starting with `x-amz-` (except for `x-amz-client-context`) and `*` that include every provided header. | Optional   | {"host", "content-type", "date", "x-amz-*"} if `security_token` is provided, `x-amz-security-token` is also included by default. |
 
 ## API key in header
 

httpx_auth/aws.py

@@ -2,6 +2,7 @@
 Provides code for AWSAuth ported to httpx from Sam Washington's requests-aws4auth
 https://github.com/sam-washington/requests-aws4auth
 """
+
 import hmac
 import hashlib
 import posixpath
@@ -21,8 +22,6 @@ class AWS4Auth(httpx.Auth):
 
     requires_request_body = True
 
-    default_include_headers = ["host", "content-type", "date", "x-amz-*"]
-
     def __init__(
         self, access_id: str, secret_key: str, region: str, service: str, **kwargs
     ):
@@ -37,6 +36,12 @@ def __init__(
         http://docs.aws.amazon.com/general/latest/gr/rande.html
         e.g. elasticbeanstalk.
         :param security_token: Used for the x-amz-security-token header, for use with STS temporary credentials.
+        :param include_headers: Set of headers to include in the canonical and signed headers.
+        {"host", "content-type", "date", "x-amz-*"} by default.
+        Note that if security_token is provided, x-amz-security-token is also included by default.
+        Specific values:
+        - "x-amz-*" matches any header starting with 'x-amz-' except for x-amz-client context.
+        - "*" will include every provided header.
         """
         self.secret_key = secret_key
         if not self.secret_key:
@@ -47,13 +52,14 @@ def __init__(
         self.service = service
 
         self.security_token = kwargs.get("security_token")
-        # TODO Check if we really need to be able to override this default ?
+
+        include_headers = {"host", "content-type", "date", "x-amz-*"}
         if self.security_token:
-            # TODO Avoid modifying shared variable
-            self.default_include_headers.append("x-amz-security-token")
-        self.include_headers = kwargs.get(
-            "include_headers", self.default_include_headers
-        )
+            include_headers.add("x-amz-security-token")
+
+        self.include_headers = {
+            header.lower() for header in kwargs.get("include_headers", include_headers)
+        }
 
     def auth_flow(
         self, request: httpx.Request
@@ -76,9 +82,7 @@ def auth_flow(
         if self.security_token:
             request.headers["x-amz-security-token"] = self.security_token
 
-        cano_headers, signed_headers = self._get_canonical_headers(
-            request, self.include_headers
-        )
+        cano_headers, signed_headers = self._get_canonical_headers(request)
         cano_req = self._get_canonical_request(request, cano_headers, signed_headers)
         sig_string = self._get_sig_string(request, cano_req, scope)
         sig_string = sig_string.encode("utf-8")
@@ -121,56 +125,31 @@ def _get_canonical_request(
         ]
         return "\n".join(req_parts)
 
-    @classmethod
-    def _get_canonical_headers(
-        cls, req: httpx.Request, include: List[str]
-    ) -> Tuple[str, str]:
+    def _get_canonical_headers(self, req: httpx.Request) -> Tuple[str, str]:
         """
         Generate the Canonical Headers section of the Canonical Request.
         Return the Canonical Headers and the Signed Headers strs as a tuple
         (canonical_headers, signed_headers).
-
-        :param include: List of headers to include in the canonical and signed
-        headers. It's primarily included to allow testing against
-        specific examples from Amazon. If omitted or None it
-        includes host, content-type and any header starting 'x-amz-'
-        except for x-amz-client context, which appears to break
-        mobile analytics auth if included. Except for the
-        x-amz-client-context exclusion these defaults are per the
-        AWS documentation.
         """
-        include = [x.lower() for x in include]
-        headers = req.headers.copy()
-        # Aggregate for upper/lowercase header name collisions in header names,
-        # AMZ requires values of colliding headers be concatenated into a
-        # single header with lowercase name.  Although this is not possible with
-        # Requests, since it uses a case-insensitive dict to hold headers, this
-        # is here just in case you duck type with a regular dict
-        cano_headers_dict = {}
-        for hdr, val in headers.items():
-            hdr = hdr.strip().lower()
-            val = cls._amz_norm_whitespace(val).strip()
-            if (
-                hdr in include
-                or "*" in include
-                or (
-                    "x-amz-*" in include
-                    and hdr.startswith("x-amz-")
-                    and not hdr == "x-amz-client-context"
-                )
+        included_headers = {}
+        for header, header_value in req.headers.items():
+            if (header or "*") in self.include_headers or (
+                "x-amz-*" in self.include_headers
+                and header.startswith("x-amz-")
+                # x-amz-client-context break mobile analytics auth if included
+                and not header == "x-amz-client-context"
             ):
-                vals = cano_headers_dict.setdefault(hdr, [])
-                vals.append(val)
-        # Flatten cano_headers dict to string and generate signed_headers
-        cano_headers = ""
-        signed_headers_list = []
-        for hdr in sorted(cano_headers_dict):
-            vals = cano_headers_dict[hdr]
-            val = ",".join(sorted(vals))
-            cano_headers += f"{hdr}:{val}\n"
-            signed_headers_list.append(hdr)
-        signed_headers = ";".join(signed_headers_list)
-        return cano_headers, signed_headers
+                included_headers[header] = _amz_norm_whitespace(header_value)
+
+        canonical_headers = ""
+        signed_headers = []
+        for header in sorted(included_headers):
+            signed_headers.append(header)
+            canonical_headers += f"{header}:{included_headers[header]}\n"
+
+        signed_headers = ";".join(signed_headers)
+
+        return canonical_headers, signed_headers
 
     @staticmethod
     def _get_sig_string(req: httpx.Request, cano_req: str, scope: str) -> str:
@@ -183,10 +162,9 @@ def _get_sig_string(req: httpx.Request, cano_req: str, scope: str) -> str:
         amz_date = req.headers["x-amz-date"]
         hsh = hashlib.sha256(cano_req.encode())
         sig_items = ["AWS4-HMAC-SHA256", amz_date, scope, hsh.hexdigest()]
-        sig_string = "\n".join(sig_items)
-        return sig_string
+        return "\n".join(sig_items)
 
-    def _amz_cano_path(self, path) -> str:
+    def _amz_cano_path(self, path: str) -> str:
         """
         Generate the canonical path as per AWS4 auth requirements.
         Not documented anywhere, determined from aws4_testsuite examples,
@@ -232,14 +210,6 @@ def _amz_cano_querystring(qs: str) -> str:
         qs = "&".join(sorted(qs_strings))
         return qs
 
-    @staticmethod
-    def _amz_norm_whitespace(text: str) -> str:
-        """
-        Replace runs of whitespace with a single space.
-        Ignore text enclosed in quotes.
-        """
-        return " ".join(shlex.split(text, posix=False))
-
 
 def generate_key(secret_key: str, region: str, service: str, date: str) -> bytes:
     init_key = f"AWS4{secret_key}".encode("utf-8")
@@ -251,3 +221,11 @@ def generate_key(secret_key: str, region: str, service: str, date: str) -> bytes
 
 def sign_sha256(signing_key: bytes, message: str) -> bytes:
     return hmac.new(signing_key, message.encode("utf-8"), hashlib.sha256).digest()
+
+
+def _amz_norm_whitespace(text: str) -> str:
+    """
+    Replace runs of whitespace with a single space.
+    Ignore text enclosed in quotes.
+    """
+    return " ".join(shlex.split(text, posix=False)).strip()

pyproject.toml

@@ -50,6 +50,8 @@ testing = [
     "time-machine==2.*",
     # Used to check coverage
     "pytest-cov==4.*",
+    # Used to run async tests
+    "pytest-asyncio==0.23.*",
 ]
 
 [tool.setuptools.packages.find]

tests/api_key/test_api_key_async.py

@@ -0,0 +1,58 @@
+import pytest
+from pytest_httpx import HTTPXMock
+import httpx
+
+
+import httpx_auth
+
+
+@pytest.mark.asyncio
+async def test_header_api_key_is_sent_in_x_api_key_by_default(httpx_mock: HTTPXMock):
+    auth = httpx_auth.HeaderApiKey("my_provided_api_key")
+
+    httpx_mock.add_response(
+        url="https://authorized_only",
+        method="GET",
+        match_headers={"X-API-Key": "my_provided_api_key"},
+    )
+
+    async with httpx.AsyncClient() as client:
+        await client.get("https://authorized_only", auth=auth)
+
+
+@pytest.mark.asyncio
+async def test_query_api_key_is_sent_in_api_key_by_default(httpx_mock: HTTPXMock):
+    auth = httpx_auth.QueryApiKey("my_provided_api_key")
+
+    httpx_mock.add_response(
+        url="https://authorized_only?api_key=my_provided_api_key", method="GET"
+    )
+
+    async with httpx.AsyncClient() as client:
+        await client.get("https://authorized_only", auth=auth)
+
+
+@pytest.mark.asyncio
+async def test_header_api_key_can_be_sent_in_a_custom_field_name(httpx_mock: HTTPXMock):
+    auth = httpx_auth.HeaderApiKey("my_provided_api_key", "X-API-HEADER-KEY")
+
+    httpx_mock.add_response(
+        url="https://authorized_only",
+        method="GET",
+        match_headers={"X-API-HEADER-KEY": "my_provided_api_key"},
+    )
+
+    async with httpx.AsyncClient() as client:
+        await client.get("https://authorized_only", auth=auth)
+
+
+@pytest.mark.asyncio
+async def test_query_api_key_can_be_sent_in_a_custom_field_name(httpx_mock: HTTPXMock):
+    auth = httpx_auth.QueryApiKey("my_provided_api_key", "X-API-QUERY-KEY")
+
+    httpx_mock.add_response(
+        url="https://authorized_only?X-API-QUERY-KEY=my_provided_api_key", method="GET"
+    )
+
+    async with httpx.AsyncClient() as client:
+        await client.get("https://authorized_only", auth=auth)