Sorting future FV (#188)

* updated statement to sort documents for future FV

* Added tests

dev_requirements/requirements-tests.in

@@ -4,3 +4,4 @@ pytest-datafiles
 pytest-mock
 pytest-asyncio==0.22.0 #pinned for bugs/breaking changes in 0.23.x
 aioresponses
+freezegun

dev_requirements/requirements-tests.txt

@@ -12,6 +12,10 @@ aiosignal==1.3.1
     # via aiohttp
 attrs==23.2.0
     # via aiohttp
+colorama==0.4.6
+    # via pytest
+freezegun==1.5.1
+    # via -r dev_requirements/requirements-tests.in
 frozenlist==1.4.1
     # via
     #   aiohttp
@@ -40,5 +44,9 @@ pytest-datafiles==3.0.0
     # via -r dev_requirements/requirements-tests.in
 pytest-mock==3.14.0
     # via -r dev_requirements/requirements-tests.in
+python-dateutil==2.9.0.post0
+    # via freezegun
+six==1.16.0
+    # via python-dateutil
 yarl==1.9.4
     # via aiohttp

requirements.txt

@@ -4,8 +4,6 @@
 #
 #    pip-compile pyproject.toml
 #
-aiodns==3.2.0
-    # via aiohttp
 aiohttp[speedups]==3.9.5
     # via
     #   aiohttp-requests
@@ -22,8 +20,6 @@ beautifulsoup4==4.12.3
     # via edi_energy_scraper (pyproject.toml)
 brotli==1.1.0
     # via aiohttp
-cffi==1.16.0
-    # via pycares
 coworker==2.0.1
     # via aiohttp-requests
 frozenlist==1.4.1
@@ -44,10 +40,6 @@ multidict==6.0.5
     #   yarl
 packaging==24.0
     # via marshmallow
-pycares==4.4.0
-    # via aiodns
-pycparser==2.22
-    # via cffi
 pypdf==4.2.0
     # via edi_energy_scraper (pyproject.toml)
 pytz==2024.1

src/edi_energy_scraper/__init__.py

@@ -459,7 +459,12 @@ def get_edifact_format_version_range_from_filename(path: Path) -> List[EdifactFo
 
     is_format_version_end_greater_than_next_format_version = format_version_end > next_format_version
     if is_format_version_end_greater_than_next_format_version:
-        format_version_end = next_format_version
+        # files which have been published earlier and not been replaced yet are used until the next format version
+        if format_version_start < next_format_version:
+            format_version_end = next_format_version
+        # new documents which are not valdi yet are used for the future format version
+        else:
+            format_version_end = format_version_start
 
     format_versions = [efv.value for efv in EdifactFormatVersion]
 

unittests/test_edienergyscraper.py

@@ -4,6 +4,7 @@
 import pytest
 from aioresponses import aioresponses
 from bs4 import BeautifulSoup
+from freezegun import freeze_time
 from maus.edifact import EdifactFormatVersion
 
 from edi_energy_scraper import (
@@ -505,6 +506,7 @@ def test_get_edifact_version_and_formats(self, input_filename: str, expected_res
 
         assert actual == expected_result
 
+    @freeze_time("2024-06-25")
     @pytest.mark.parametrize(
         "input_filename, expected_result",
         [
@@ -527,6 +529,16 @@ def test_get_edifact_version_and_formats(self, input_filename: str, expected_res
                 [EdifactFormatVersion.FV2310, EdifactFormatVersion.FV2404],
                 id="valid for two format versions",
             ),
+            pytest.param(
+                "IFTSTAMIG2.0e_20280930_20231001.pdf",
+                [EdifactFormatVersion.FV2310, EdifactFormatVersion.FV2404, EdifactFormatVersion.FV2410],
+                id="valid in future, starting in past",
+            ),
+            pytest.param(
+                "IFTSTAMIG2.0e_20280930_20250404.pdf",
+                [EdifactFormatVersion.FV2504],
+                id="starting in future",
+            ),
         ],
     )
     def test_get_edifact_format_version_range(self, input_filename: str, expected_result: List[EdifactFormatVersion]):