Add date comparison

src/imf_reader/sdr/read_announcements.py

@@ -25,7 +25,7 @@ def read_tsv(url: str) -> pd.DataFrame:
         return pd.read_csv(url, delimiter="/t", engine="python")
 
     except pd.errors.ParserError:
-        raise ValueError("SDR _data not available for this date")
+        raise ValueError("SDR data not available for this date")
 
 
 def clean_df(df: pd.DataFrame) -> pd.DataFrame:
@@ -73,14 +73,16 @@ def get_holdings_and_allocations_data(
 
 
 @lru_cache
-def get_latest_allocations_holdings_date() -> tuple[int, int]:
+def get_latest_allocations_holdings_date(log_info=True) -> tuple[int, int]:
     """
     Get the latest available SDR allocation holdings date.
-
+    Args:
+        log_info (bool, optional): If True, log info about the download. Defaults to True.
     Returns:
         tuple[int, int]: A tuple containing the year and month of the latest SDR data.
     """
-    logger.info("Fetching latest date")
+    if log_info:
+        logger.info("Fetching latest date")
 
     response = make_request(MAIN_PAGE_URL)
     soup = BeautifulSoup(response.content, "html.parser")
@@ -94,16 +96,27 @@ def get_latest_allocations_holdings_date() -> tuple[int, int]:
 
 
 def fetch_allocations_holdings(date: tuple[int, int] | None = None) -> pd.DataFrame:
-    """Fetch SDR holdings and allocations data for a given date
+    """
+    Fetch SDR holdings and allocations data for a given date. If date is not specified or exceeds the latest available
+    date, it fetches data for the latest date
 
     Args:
-        date: The year and month to get allocations and holdings data for. e.g. (2024, 11) for November 2024. If None, the latest announcements released are fetched
+        date: The year and month to get allocations and holdings data for. e.g. (2024, 11) for November 2024.
+        If None, the latest announcements released are fetched
 
     returns:
         A dataframe with the SDR allocations and holdings data
     """
 
     if date is None:
         date = get_latest_allocations_holdings_date()
+    else:
+        latest_date = get_latest_allocations_holdings_date(log_info=False)
+        date_obj = datetime(date[0], date[1], 1)
+        latest_date_obj = datetime(latest_date[0], latest_date[1], 1)
+        if date_obj > latest_date_obj:
+            raise ValueError(
+                f"SDR data unavailable for: ({date[0]}, {date[1]}).\nLatest available: ({latest_date[0]}, {latest_date[1]})"
+            )
 
     return get_holdings_and_allocations_data(*date)

tests/test_sdr/test_read_announcements.py

@@ -1,6 +1,7 @@
 from unittest.mock import patch, Mock
 import pytest
 import pandas as pd
+import re
 from imf_reader import sdr
 from imf_reader.sdr.read_announcements import (
     read_tsv,
@@ -50,7 +51,7 @@ def test_read_tsv_success(self, mock_read_csv):
     def test_read_tsv_failure(self, mock_read_csv):
         """Test read_tsv raises ValueError on malformed data."""
         mock_read_csv.side_effect = pd.errors.ParserError
-        with pytest.raises(ValueError, match="SDR _data not available for this date"):
+        with pytest.raises(ValueError, match="SDR data not available for this date"):
             read_tsv("mock_url")
 
     def test_clean_df_correct_format(self, input_df):
@@ -128,7 +129,7 @@ def test_get_holdings_and_allocations_data_success(
     def test_get_holdings_and_allocations_data_failure(self, mock_read_tsv):
         """Test get_holdings_and_allocations_data raises ValueError when read_tsv fails."""
         with pytest.raises(ValueError, match="Data not available"):
-            get_holdings_and_allocations_data(2024, 2)
+            get_holdings_and_allocations_data(1800, 1)
 
     @patch("imf_reader.sdr.read_announcements.make_request")
     def test_get_latest_date_success(self, mock_make_request):
@@ -192,7 +193,7 @@ def test_fetch_allocations_holdings_default_date(
     ):
         """Test fetch_allocations_holdings when no date is provided."""
         # Mock get_latest_allocations_holdings_date to return a specific date
-        mock_get_latest_date.return_value = (2, 2024)
+        mock_get_latest_date.return_value = (2024, 2)
 
         # Mock read_tsv
         mock_read_tsv.return_value = input_df
@@ -217,24 +218,26 @@ def test_fetch_allocations_holdings_default_date(
         # Assertions
         mock_get_latest_date.assert_called_once()  # Ensure get_latest_allocations_holdings_date was called
         mock_get_holdings_and_allocations_data.assert_called_once_with(
-            2, 2024
+            2024, 2
         )  # Ensure the correct call
         pd.testing.assert_frame_equal(
             result, cleaned_df
         )  # Ensure the result matches the cleaned data
 
-    @patch("imf_reader.sdr.read_announcements.get_holdings_and_allocations_data")
-    @patch("imf_reader.sdr.read_announcements.get_latest_allocations_holdings_date")
-    def test_fetch_allocations_holdings_failure(
-        self, mock_get_latest_date, mock_get_holdings_data
-    ):
-        """Test fetch_allocations_holdings raises ValueError when data fetching fails."""
-        # Mock get_latest_allocations_holdings_date to return a date
-        mock_get_latest_date.return_value = (2, 2024)
-
-        # Simulate a failure in get_holdings_and_allocations_data
-        mock_get_holdings_data.side_effect = ValueError("Data not available")
 
-        # Assertions
-        with pytest.raises(ValueError, match="Data not available"):
-            fetch_allocations_holdings()
+@patch("imf_reader.sdr.read_announcements.get_holdings_and_allocations_data")
+@patch(
+    "imf_reader.sdr.read_announcements.get_latest_allocations_holdings_date",
+    return_value=(2024, 1),
+)
+def test_fetch_allocations_holdings_future_date(
+    mock_get_latest_date, mock_get_holdings_and_allocations_data
+):
+    """Test fetch_allocations_holdings when unavailable future date is provided."""
+    with pytest.raises(
+        ValueError,
+        match=re.escape(
+            "SDR data unavailable for: (2025, 1).\nLatest available: (2024, 1)"
+        ),
+    ):
+        fetch_allocations_holdings((2025, 1))