arkindex-client 1.1.1__tar.gz → 1.1.3__tar.gz

{arkindex-client-1.1.1 → arkindex-client-1.1.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: arkindex-client
-Version: 1.1.1
+Version: 1.1.3
 Summary: API client for the Arkindex project
 Home-page: https://gitlab.teklia.com/arkindex/api-client
 Author: Teklia <contact@teklia.com>

arkindex-client-1.1.3/VERSION

@@ -0,0 +1 @@
+1.1.3

{arkindex-client-1.1.1 → arkindex-client-1.1.3}/arkindex/client/client.py

@@ -2,14 +2,16 @@
 """
 Arkindex API Client
 """
+import json
 import logging
 import os
 import warnings
+from importlib.metadata import version
 from time import sleep
-from urllib.parse import urljoin, urlsplit, urlunsplit
+from urllib.parse import quote, urljoin, urlparse, urlsplit
 
 import requests
-import yaml
+import typesystem
 from tenacity import (
     before_sleep_log,
     retry,
@@ -19,20 +21,14 @@ from tenacity import (
 )
 
 from arkindex.auth import TokenSessionAuthentication
-from arkindex.client.base import BaseClient
-from arkindex.exceptions import ErrorResponse, SchemaError
+from arkindex.client import decoders
+from arkindex.exceptions import ClientError, ErrorMessage, ErrorResponse, SchemaError
 from arkindex.pagination import ResponsePaginator
+from arkindex.schema.validator import validate
 
 logger = logging.getLogger(__name__)
 
-try:
-    from yaml import CSafeLoader as SafeLoader
-
-    logger.debug("Using LibYAML-based parser")
-except ImportError:
-    from yaml import SafeLoader
-
-    logger.debug("Using default PyYAML parser")
+REQUEST_TIMEOUT = (30, 60)
 
 BASE_DIR = os.path.dirname(os.path.abspath(__file__))
 
@@ -82,22 +78,7 @@ def options_from_env():
     return options
 
 
-def _find_operation(schema, operation_id):
-    for path_object in schema["paths"].values():
-        for operation in path_object.values():
-            if operation["operationId"] == operation_id:
-                return operation
-    raise KeyError("Operation '{}' not found".format(operation_id))
-
-
-def _find_param(operation, param_name):
-    for parameter in operation.get("parameters", []):
-        if parameter["name"] == param_name:
-            return parameter
-    raise KeyError("Parameter '{}' not found".format(param_name))
-
-
-class ArkindexClient(BaseClient):
+class ArkindexClient:
     """
     An Arkindex API client.
     """
@@ -111,7 +92,6 @@ class ArkindexClient(BaseClient):
         csrf_cookie=None,
         sleep=0,
         verify=True,
-        **kwargs,
     ):
         r"""
         :param token: An API token to use. If omitted, access is restricted to public endpoints.
@@ -125,55 +105,49 @@ class ArkindexClient(BaseClient):
         :type csrf_cookie: str or None
         :param float sleep: Number of seconds to wait before sending each API request,
            as a simple means of throttling.
-        :param \**kwargs: Keyword arguments to send to ``arkindex.client.base.BaseClient``.
+        :param bool verify: Whether to verify the SSL certificate on each request. Enabled by default.
         """
+        self.decoders = [
+            decoders.JSONDecoder(),
+            decoders.TextDecoder(),
+            decoders.DownloadDecoder(),
+        ]
+
+        self.session = requests.Session()
+        self.session.verify = verify
+        client_version = version("arkindex-client")
+        self.session.headers.update(
+            {
+                "accept": ", ".join([decoder.media_type for decoder in self.decoders]),
+                "user-agent": f"arkindex-client/{client_version}",
+            }
+        )
+
         if not schema_url:
             schema_url = urljoin(base_url, SCHEMA_ENDPOINT)
 
-        self.verify = verify
         try:
             split = urlsplit(schema_url)
             if split.scheme == "file" or not (split.scheme or split.netloc):
                 # This is a local path
                 with open(schema_url) as f:
-                    schema = yaml.load(f, Loader=SafeLoader)
+                    schema = json.load(f)
             else:
-                resp = requests.get(schema_url, verify=self.verify)
+                resp = self.session.get(
+                    schema_url,
+                    headers={
+                        # Explicitly request an OpenAPI schema in JSON and not YAML
+                        "Accept": "application/vnd.oai.openapi+json, application/json",
+                    },
+                )
                 resp.raise_for_status()
-                schema = yaml.load(resp.content, Loader=SafeLoader)
+                schema = resp.json()
         except Exception as e:
             raise SchemaError(
                 f"Could not retrieve a proper OpenAPI schema from {schema_url}"
             ) from e
 
-        super().__init__(schema, **kwargs)
-
-        # An OpenAPI schema is considered valid even when there are no endpoints, making the client completely useless.
-        if not len(self.document.walk_links()):
-            raise SchemaError(
-                f"The OpenAPI schema from {base_url} has no defined endpoints"
-            )
-
-        # Post-processing of the parsed schema
-        for link_info in self.document.walk_links():
-            # Look for deprecated links
-            # https://github.com/encode/apistar/issues/664
-            operation = _find_operation(schema, link_info.link.name)
-            link_info.link.deprecated = operation.get("deprecated", False)
-            for item in link_info.link.get_query_fields():
-                parameter = _find_param(operation, item.name)
-                item.deprecated = parameter.get("deprecated", False)
-
-            # Detect paginated links
-            if "x-paginated" in operation:
-                link_info.link._paginated = operation["x-paginated"]
-
-            # Remove domains from each endpoint; allows to properly handle our base URL
-            # https://github.com/encode/apistar/issues/657
-            original_url = urlsplit(link_info.link.url)
-            # Removes the scheme and netloc
-            new_url = ("", "", *original_url[2:])
-            link_info.link.url = urlunsplit(new_url)
+        self.document = validate(schema)
 
         # Try to autodetect the CSRF cookie:
         # - Try to find a matching server for this base URL and look for the x-csrf-cookie extension
@@ -228,7 +202,7 @@ class ArkindexClient(BaseClient):
         """
         if not csrf_cookie:
             csrf_cookie = "arkindex.csrf"
-        self.transport.session.auth = TokenSessionAuthentication(
+        self.session.auth = TokenSessionAuthentication(
             token,
             csrf_cookie_name=csrf_cookie,
             scheme=auth_scheme,
@@ -242,7 +216,15 @@ class ArkindexClient(BaseClient):
             self.document.url = base_url
 
         # Add the Referer header to allow Django CSRF to function
-        self.transport.headers.setdefault("Referer", self.document.url)
+        self.session.headers.setdefault("Referer", self.document.url)
+
+    def lookup_operation(self, operation_id: str):
+        if operation_id in self.document.links:
+            return self.document.links[operation_id]
+
+        text = 'Operation ID "%s" not found in schema.' % operation_id
+        message = ErrorMessage(text=text, code="invalid-operation")
+        raise ClientError(messages=[message])
 
     def paginate(self, operation_id, *args, **kwargs):
         """
@@ -251,9 +233,10 @@ class ArkindexClient(BaseClient):
         :return: An iterator for a paginated endpoint.
         :rtype: Union[arkindex.pagination.ResponsePaginator, dict, list]
         """
+
         link = self.lookup_operation(operation_id)
         # If there was no x-paginated, trust the caller and assume the endpoint is paginated
-        if getattr(link, "_paginated", True):
+        if link.paginated is not False:
             return ResponsePaginator(self, operation_id, *args, **kwargs)
         return self.request(operation_id, *args, **kwargs)
 
@@ -264,15 +247,77 @@ class ArkindexClient(BaseClient):
         """
         resp = self.request("Login", body={"email": email, "password": password})
         if "auth_token" in resp:
-            self.transport.session.auth.scheme = "Token"
-            self.transport.session.auth.token = resp["auth_token"]
+            self.session.auth.scheme = "Token"
+            self.session.auth.token = resp["auth_token"]
         return resp
 
-    def single_request(self, operation_id, *args, **kwargs):
+    def get_query_params(self, link, params):
+        return {
+            field.name: params[field.name]
+            for field in link.get_query_fields()
+            if field.name in params
+        }
+
+    def get_url(self, link, params):
+        url = urljoin(self.document.url, link.url)
+
+        scheme = urlparse(url).scheme.lower()
+
+        if not scheme:
+            text = "URL missing scheme '%s'." % url
+            message = ErrorMessage(text=text, code="invalid-url")
+            raise ClientError(messages=[message])
+
+        if scheme not in ("http", "https"):
+            text = "Unsupported URL scheme '%s'." % scheme
+            message = ErrorMessage(text=text, code="invalid-url")
+            raise ClientError(messages=[message])
+
+        for field in link.get_path_fields():
+            value = str(params[field.name])
+            if "{%s}" % field.name in url:
+                url = url.replace("{%s}" % field.name, quote(value, safe=""))
+            elif "{+%s}" % field.name in url:
+                url = url.replace("{+%s}" % field.name, quote(value, safe="/"))
+
+        return url
+
+    def get_content(self, link, params):
+        body_field = link.get_body_field()
+        if body_field and body_field.name in params:
+            assert (
+                link.encoding == "application/json"
+            ), "Only JSON request bodies are supported"
+            return params[body_field.name]
+
+    def get_decoder(self, content_type=None):
+        """
+        Given the value of a 'Content-Type' header, return the appropriate
+        decoder for handling the response content.
+        """
+        if content_type is None:
+            return self.decoders[0]
+
+        content_type = content_type.split(";")[0].strip().lower()
+        main_type = content_type.split("/")[0] + "/*"
+        wildcard_type = "*/*"
+
+        for codec in self.decoders:
+            if codec.media_type in (content_type, main_type, wildcard_type):
+                return codec
+
+        text = (
+            "Unsupported encoding '%s' in response Content-Type header." % content_type
+        )
+        message = ErrorMessage(text=text, code="cannot-decode-response")
+        raise ClientError(messages=[message])
+
+    def single_request(self, operation_id, **parameters):
         """
         Perform an API request.
-        :param args: Arguments passed to the BaseClient.
-        :param kwargs: Keyword arguments passed to the BaseClient.
+
+        :param str operation_id: Name of the API endpoint.
+        :param path_parameters: Path parameters for this endpoint.
         """
         link = self.lookup_operation(operation_id)
         if link.deprecated:
@@ -282,8 +327,23 @@ class ArkindexClient(BaseClient):
                 stacklevel=2,
             )
 
-        query_params = self.get_query_params(link, kwargs)
+        validator = typesystem.Object(
+            properties={field.name: typesystem.Any() for field in link.fields},
+            required=[field.name for field in link.fields if field.required],
+            additional_properties=False,
+        )
+        try:
+            validator.validate(parameters)
+        except typesystem.ValidationError as exc:
+            raise ClientError(messages=exc.messages()) from None
+
+        method = link.method
+        url = self.get_url(link, parameters)
+
+        content = self.get_content(link, parameters)
+        query_params = self.get_query_params(link, parameters)
         fields = link.get_query_fields()
+
         for field in fields:
             if field.deprecated and field.name in query_params:
                 warnings.warn(
@@ -291,12 +351,41 @@ class ArkindexClient(BaseClient):
                     DeprecationWarning,
                     stacklevel=2,
                 )
+
         if self.sleep_duration:
             logger.debug(
                 "Delaying request by {:f} seconds...".format(self.sleep_duration)
             )
             sleep(self.sleep_duration)
-        return super().request(operation_id, *args, **kwargs)
+
+        return self._send_request(
+            method, url, query_params=query_params, content=content
+        )
+
+    def _send_request(self, method, url, query_params=None, content=None):
+        options = {
+            "params": query_params,
+            "timeout": REQUEST_TIMEOUT,
+        }
+        if content is not None:
+            options["json"] = content
+
+        response = self.session.request(method, url, **options)
+
+        # Given an HTTP response, return the decoded data.
+        result = None
+        if response.content:
+            content_type = response.headers.get("content-type")
+            decoder = self.get_decoder(content_type)
+            result = decoder.decode(response)
+
+        if 400 <= response.status_code <= 599:
+            title = "%d %s" % (response.status_code, response.reason)
+            raise ErrorResponse(
+                title=title, status_code=response.status_code, content=result
+            )
+
+        return result
 
     @retry(
         retry=retry_if_exception(_is_500_error),
@@ -305,15 +394,34 @@ class ArkindexClient(BaseClient):
         stop=stop_after_attempt(5),
         before_sleep=before_sleep_log(logger, logging.INFO),
     )
-    def request(self, operation_id, *args, **kwargs):
+    def request(self, operation_id, **parameters):
         """
-        Proxy all Arkindex API requests with a retry mechanism in case of 50X errors.
-        The same API call will be retried 5 times, with an exponential sleep time
-        going through 3, 4, 8 and 16 seconds of wait between call.
+        Perform an API request with an automatic retry mechanism in case of 50X errors.
+        A failing API call will be retried 5 times, with an exponential sleep time going
+        through 3, 4, 8 and 16 seconds of wait between call.
         If the 5th call still gives a 50x, the exception is re-raised and the caller should catch it.
         Log messages are displayed before sleeping (when at least one exception occurred).
 
-        :param args: Arguments passed to the BaseClient.
-        :param kwargs: Keyword arguments passed to the BaseClient.
+        :param str operation_id: Name of the API endpoint.
+        :param parameters: Body, Path or Query parameters passed as kwargs.
+            Body parameters must be passed using the `body` keyword argument, others can be set directly.
+
+        Example usage for POST and unpaginated GET requests:
+
+        >>> request(
+        ...     "CreateMetaDataBulk",
+        ...     id="8f8f196f-49bc-444e-9cfe-c705c3cd01ae",
+        ...     body={
+        ...         "worker_run_id": "50e1f2d4-2087-41ed-a862-d17576bae480",
+        ...         "metadata_list": [
+        ...             …
+        ...         ],
+        ...     },
+        ... )
+        >>> request(
+        ...     "ListElements",
+        ...     corpus="7358ab03-cc36-4160-86ce-98f70e993a0f",
+        ...     top_level=True,
+        ... )
         """
-        return self.single_request(operation_id, *args, **kwargs)
+        return self.single_request(operation_id, **parameters)

{arkindex-client-1.1.1 → arkindex-client-1.1.3}/arkindex/client/decoders.py

@@ -1,10 +1,10 @@
 # -*- coding: utf-8 -*-
-import cgi
 import json
 import os
 import posixpath
 import shutil
 import tempfile
+from email.message import EmailMessage
 from urllib.parse import urlparse
 
 from arkindex.compat import DownloadedFile
@@ -213,10 +213,11 @@ def _get_filename_from_content_disposition(content_disposition):
     """
     Determine an output filename based on the `Content-Disposition` header.
     """
-    params = value, params = cgi.parse_header(content_disposition)
+    message = EmailMessage()
+    message["content-disposition"] = content_disposition
+    filename = message["content-disposition"].params.get("filename")
 
-    if "filename" in params:
-        filename = params["filename"]
+    if filename:
         return _safe_filename(filename)
 
     return None

arkindex-client-1.1.3/arkindex/document.py

@@ -0,0 +1,128 @@
+# -*- coding: utf-8 -*-
+import re
+import typing
+
+from arkindex.exceptions import SchemaError
+
+
+class Document:
+    def __init__(
+        self,
+        links: typing.Sequence["Link"],
+        url: str = "",
+    ):
+        if not len(links):
+            raise SchemaError(
+                "An OpenAPI document must contain at least one valid operation."
+            )
+
+        links_by_name = {}
+
+        # Ensure all names within a document are unique.
+        for link in links:
+            assert (
+                link.name not in links_by_name
+            ), f'Link "{link.name}" in Document must have a unique name.'
+            links_by_name[link.name] = link
+
+        self.links = links_by_name
+        self.url = url
+
+
+class Link:
+    """
+    Links represent the actions that a client may perform.
+    """
+
+    def __init__(
+        self,
+        url: str,
+        method: str,
+        handler: typing.Callable = None,
+        name: str = "",
+        encoding: str = "",
+        fields: typing.Sequence["Field"] = None,
+        deprecated: bool = False,
+        paginated: typing.Optional[bool] = None,
+    ):
+        method = method.upper()
+        fields = [] if (fields is None) else list(fields)
+
+        url_path_names = set(
+            [item.strip("{}").lstrip("+") for item in re.findall("{[^}]*}", url)]
+        )
+        path_fields = [field for field in fields if field.location == "path"]
+        body_fields = [field for field in fields if field.location == "body"]
+
+        assert method in (
+            "GET",
+            "POST",
+            "PUT",
+            "PATCH",
+            "DELETE",
+            "OPTIONS",
+            "HEAD",
+            "TRACE",
+        )
+        assert len(body_fields) < 2
+        if body_fields:
+            assert encoding
+        for field in path_fields:
+            assert field.name in url_path_names
+
+        # Add in path fields for any "{param}" items that don't already have
+        # a corresponding path field.
+        for path_name in url_path_names:
+            if path_name not in [field.name for field in path_fields]:
+                fields += [Field(name=path_name, location="path", required=True)]
+
+        self.url = url
+        self.method = method
+        self.handler = handler
+        self.name = name if name else handler.__name__
+        self.encoding = encoding
+        self.fields = fields
+        self.deprecated = deprecated
+        self.paginated = paginated
+
+    def get_path_fields(self):
+        return [field for field in self.fields if field.location == "path"]
+
+    def get_query_fields(self):
+        return [field for field in self.fields if field.location == "query"]
+
+    def get_body_field(self):
+        for field in self.fields:
+            if field.location == "body":
+                return field
+        return None
+
+    def get_expanded_body(self):
+        field = self.get_body_field()
+        if field is None or not hasattr(field.schema, "properties"):
+            return None
+        return field.schema.properties
+
+
+class Field:
+    def __init__(
+        self,
+        name: str,
+        location: str,
+        required: bool = None,
+        schema: typing.Any = None,
+        example: typing.Any = None,
+        deprecated: bool = False,
+    ):
+        assert location in ("path", "query", "body", "cookie", "header", "formData")
+        if required is None:
+            required = True if location in ("path", "body") else False
+        if location == "path":
+            assert required, "May not set 'required=False' on path fields."
+
+        self.name = name
+        self.location = location
+        self.required = required
+        self.schema = schema
+        self.example = example
+        self.deprecated = deprecated