Merge pull request #790 from CitrineInformatics/feature/raw_download

adding a read method to read from S3

Author: btreecat

src/citrine/__version__.py

@@ -1 +1 @@
-__version__ = '1.44.0'
+__version__ = '1.44.1'

src/citrine/resources/file_link.py

@@ -687,6 +687,37 @@ def download(self, *, file_link: Union[str, UUID, FileLink], local_path: Union[s
         download_response = requests.get(final_url)
         write_file_locally(download_response.content, final_path)
 
+    def read(self, *, file_link: Union[str, UUID, FileLink]):
+        """
+        Read the file associated with a given FileLink.
+
+        Parameters
+        ----------
+        file_link: FileLink, str, UUID
+            Resource referencing the file.
+
+        Returns
+        -------
+        I/O stream
+            The contents of the file.
+
+        """
+        file_link = self._resolve_file_link(file_link)
+
+        if self._is_external_url(file_link.url):  # Pull it from where ever it lives
+            final_url = file_link.url
+        elif self._validate_local_url(file_link.url):
+            # The "/content-link" route returns a pre-signed url to download the file.
+            content_link = self._get_path_from_file_link(file_link, action='content-link')
+            content_link_response = self.session.get_resource(content_link)
+            pre_signed_url = content_link_response['pre_signed_read_link']
+            final_url = rewrite_s3_links_locally(pre_signed_url, self.session.s3_endpoint_url)
+        else:  # Unrecognized
+            raise ValueError(f"URL was malformed for a local file resource ({file_link.url}).")
+
+        download_response = requests.get(final_url)
+        return download_response.content
+
     def process(self, *, file_link: Union[FileLink, str, UUID],
                 processing_type: FileProcessingType,
                 wait_for_response: bool = True,

tests/resources/test_file_link.py

@@ -410,6 +410,63 @@ def _checked_write(path, content):
     with pytest.raises(ValueError, match="malformed"):
         collection.download(file_link=bad_file, local_path=target_dir)
 
+def test_read(collection: FileCollection, session):
+    """
+    Test that reading a file works as expected.
+
+    """
+    # Given
+    filename = 'diagram.pdf'
+    url = f"projects/{collection.project_id}/datasets/{collection.dataset_id}/files/{uuid4()}/versions/{uuid4()}"
+    file = FileLink.build(FileLinkDataFactory(url=url, filename=filename))
+    pre_signed_url = "http://files.citrine.io/secret-codes/jiifema987pjfsda"  # arbitrary
+    session.set_response({
+        'pre_signed_read_link': pre_signed_url,
+    })
+   
+    with requests_mock.mock() as mock_get:
+        mock_get.get(pre_signed_url, text="lorem ipsum")
+        # When
+        io = collection.read(file_link=file)
+        assert io.decode('UTF-8') == 'lorem ipsum'
+        # When
+        assert mock_get.call_count == 1
+        expected_call = FakeCall(
+            method='GET',
+            path=url + '/content-link'
+        )
+        assert expected_call == session.last_call
+
+
+
+    bad_url = f"bin/uuid3/versions/uuid4"
+    bad_file = FileLink.build(FileLinkDataFactory(url=bad_url, filename=filename))
+    with pytest.raises(ValueError, match="malformed"):
+        collection.read(file_link=bad_file)
+
+
+def test_external_file_read(collection: FileCollection, session):
+    """
+    Test that reading a file works as expected for external files.
+
+    """
+    # Given
+    filename = 'spreadsheet.xlsx'
+    url = "http://customer.com/data-lake/files/123/versions/456"
+    file = FileLink.build(FileLinkDataFactory(url=url, filename=filename))
+
+
+    with requests_mock.mock() as mock_get:
+        mock_get.get(url, text='010111011')
+
+        # When
+        io = collection.read(file_link=file)
+        assert io.decode('UTF-8') == '010111011'
+
+        # When
+        assert mock_get.call_count == 1
+
+    # assert local_path.read_text() == '010111011'
 
 def test_external_file_download(collection: FileCollection, session, tmpdir):
     """