Fix data access and redirects

src/pyDataverse/api.py

@@ -122,7 +122,7 @@ def get_request(self, url, params=None, auth=False):
 
         try:
             url = urljoin(self.base_url_api, url)
-            resp = httpx.get(url, params=params)
+            resp = httpx.get(url, params=params, follow_redirects=True)
             if resp.status_code == 401:
                 error_msg = resp.json()["message"]
                 raise ApiAuthorizationError(
@@ -175,7 +175,7 @@ def post_request(self, url, data=None, auth=False, params=None, files=None):
             params["key"] = self.api_token
 
         try:
-            resp = httpx.post(url, data=data, params=params, files=files)
+            resp = httpx.post(url, data=data, params=params, files=files, follow_redirects=True)
             if resp.status_code == 401:
                 error_msg = resp.json()["message"]
                 raise ApiAuthorizationError(
@@ -216,7 +216,7 @@ def put_request(self, url, data=None, auth=False, params=None):
             params["key"] = self.api_token
 
         try:
-            resp = httpx.put(url, data=data, params=params)
+            resp = httpx.put(url, data=data, params=params, follow_redirects=True)
             if resp.status_code == 401:
                 error_msg = resp.json()["message"]
                 raise ApiAuthorizationError(
@@ -255,7 +255,7 @@ def delete_request(self, url, auth=False, params=None):
             params["key"] = self.api_token
 
         try:
-            return httpx.delete(url, params=params)
+            return httpx.delete(url, params=params, follow_redirects=True)
         except ConnectError:
             raise ConnectError(
                 "ERROR: DELETE could not establish connection to api {}.".format(url)
@@ -338,13 +338,13 @@ def get_datafile(
         """
         is_first_param = True
         if is_pid:
-            url = "{0}/datafile/{1}".format(self.base_url_api_data_access, identifier)
-            if data_format or no_var_header or image_thumb:
-                url += "?"
-        else:
             url = "{0}/datafile/:persistentId/?persistentId={1}".format(
                 self.base_url_api_data_access, identifier
             )
+        else:
+            url = "{0}/datafile/{1}".format(self.base_url_api_data_access, identifier)
+            if data_format or no_var_header or image_thumb:
+                url += "?"
         if data_format:
             url += "format={0}".format(data_format)
             is_first_param = False

tests/api/test_access.py

@@ -0,0 +1,146 @@
+import os
+import json
+import httpx
+
+from pyDataverse.api import DataAccessApi, NativeApi
+
+class TestDataAccess:
+
+    def test_get_data_by_id(self):
+        """Tests getting data file by id."""
+
+        # Arrange
+        BASE_URL = os.getenv("BASE_URL").rstrip("/")
+        API_TOKEN = os.getenv("API_TOKEN")
+
+        assert BASE_URL is not None, "BASE_URL is not set"
+        assert API_TOKEN is not None, "API_TOKEN is not set"
+
+        # Create dataset
+        metadata = json.load(open("tests/data/file_upload_ds_minimum.json"))
+        pid = self._create_dataset(BASE_URL, API_TOKEN, metadata)
+        api = DataAccessApi(BASE_URL, API_TOKEN)
+
+        # Upload a file
+        self._upload_datafile(BASE_URL, API_TOKEN, pid)
+
+        # Retrieve the file ID
+        file_id = self._get_file_id(BASE_URL, API_TOKEN, pid)
+
+        # Act
+        response = api.get_datafile(file_id, is_pid=False)
+        response.raise_for_status()
+        content = response.content.decode("utf-8")
+
+        # Assert
+        expected = open("tests/data/datafile.txt").read()
+        assert content == expected, "Data retrieval failed."
+
+    def test_get_data_by_pid(self):
+        """Tests getting data file by id.
+
+        Test runs with a PID instead of a file ID from Harvard.
+        No PID given if used within local containers
+
+        TODO - Check if possible with containers
+        """
+
+        # Arrange
+        BASE_URL = "https://dataverse.harvard.edu"
+        pid = "doi:10.7910/DVN/26093/IGA4JD"
+        api = DataAccessApi(BASE_URL)
+
+        # Act
+        response = api.get_datafile(pid, is_pid=True)
+        response.raise_for_status()
+        content = response.content
+
+        # Assert
+        expected = self._get_file_content(BASE_URL, pid)
+        assert content == expected, "Data retrieval failed."
+
+    @staticmethod
+    def _create_dataset(
+        BASE_URL: str,
+        API_TOKEN: str,
+        metadata: dict,
+    ):
+        """
+        Create a dataset in the Dataverse.
+
+        Args:
+            BASE_URL (str): The base URL of the Dataverse instance.
+            API_TOKEN (str): The API token for authentication.
+            metadata (dict): The metadata for the dataset.
+
+        Returns:
+            str: The persistent identifier (PID) of the created dataset.
+        """
+        url = f"{BASE_URL}/api/dataverses/root/datasets"
+        response = httpx.post(
+            url=url,
+            json=metadata,
+            headers={
+                "X-Dataverse-key": API_TOKEN,
+                "Content-Type": "application/json",
+            },
+        )
+
+        response.raise_for_status()
+
+        return response.json()["data"]["persistentId"]
+
+    @staticmethod
+    def _get_file_id(
+        BASE_URL: str,
+        API_TOKEN: str,
+        pid: str,
+    ):
+        """Retrieves a file ID for a given persistent identifier (PID) in Dataverse."""
+
+        response = httpx.get(
+            url=f"{BASE_URL}/api/datasets/:persistentId/?persistentId={pid}",
+            headers={
+                "X-Dataverse-key": API_TOKEN,
+                "Content-Type": "application/json",
+            }
+        )
+
+        response.raise_for_status()
+
+        return response.json()["data"]["latestVersion"]["files"][0]["dataFile"]["id"]
+
+    @staticmethod
+    def _upload_datafile(
+        BASE_URL: str,
+        API_TOKEN: str,
+        pid: str,
+    ):
+        """Uploads a file to Dataverse"""
+
+        url = f"{BASE_URL}/api/datasets/:persistentId/add?persistentId={pid}"
+        response = httpx.post(
+            url=url,
+            files={"file": open("tests/data/datafile.txt", "rb")},
+            headers={
+                "X-Dataverse-key": API_TOKEN,
+            },
+        )
+
+        response.raise_for_status()
+
+    @staticmethod
+    def _get_file_content(
+        BASE_URL: str,
+        pid: str,
+    ):
+        """Retrieves the file content for testing purposes."""
+
+        response = httpx.get(
+            url=f"{BASE_URL}/api/access/datafile/:persistentId/?persistentId={pid}",
+            follow_redirects=True,
+        )
+
+        response.raise_for_status()
+
+        return response.content