Merge pull request #176 from mariob0y/feat/detect_format_reader

feat: added custom reader support for `detect_format`

MANIFEST.in

@@ -7,3 +7,4 @@ recursive-include tests *.json
 recursive-include tests *.jsonl
 recursive-include tests *.py
 recursive-include tests *.yaml
+recursive-include tests *.gz

docs/changelog.rst

@@ -1,6 +1,14 @@
 Changelog
 =========
 
+1.0.1 (Unreleased)
+------------------
+
+Added
+~~~~~
+
+-  :meth:`ocdskit.util.detect_format` accepts a ``reader`` keyword argument: for example, ``gzip.open`` instead of ``open``.
+
 1.0.0 (2021-05-19)
 ------------------
 
@@ -32,7 +40,7 @@ Fixed
 Fixed
 ~~~~~
 
--   :ref:`schema-strict`: Don't add ``"uniqueItems": true`` to coordinates fields.
+-  :ref:`schema-strict`: Don't add ``"uniqueItems": true`` to coordinates fields.
 
 0.2.21 (2021-04-10)
 -------------------

ocdskit/util.py

@@ -188,7 +188,7 @@ def _empty_package(uri, publisher, published_date, version):
     }
 
 
-def detect_format(path, root_path=''):
+def detect_format(path, root_path='', reader=open):
     """
     Returns the format of OCDS data, and whether the OCDS data is concatenated or in an array.
 
@@ -200,7 +200,7 @@ def detect_format(path, root_path=''):
     :rtype: tuple
     :raises UnknownFormatError: if the format cannot be detected
     """
-    with open(path, 'rb') as f:
+    with reader(path, 'rb') as f:
         events = iter(ijson.parse(f, multiple_values=True))
 
         while True:

tests/test_util.py

@@ -1,3 +1,4 @@
+import gzip
 import json
 
 import pytest
@@ -142,3 +143,12 @@ def test_detect_format(filename, expected):
     result = detect_format(path(filename))
 
     assert result == expected
+
+
+@pytest.mark.parametrize('filename,expected', [
+    ('ocds-sample-data.json.gz', ('release package', False, False))
+])
+def test_detect_format_gz(filename, expected):
+    result = detect_format(path(filename), reader=gzip.open)
+
+    assert result == expected