Bugfix/15 fix column dtype conversion

tcmb/auth.py

@@ -33,7 +33,9 @@ def check_status(response):
             # Therefore the traceback is printed as well as the most
             # probable cause of the HTTPError: ApiKeyError
             print(err.with_traceback(None))
-            raise ApiKeyError("API key is invalid.") from err
+            raise ApiKeyError(
+                "API key is invalid or wrong key. See error message for details."
+            ) from err
     else:
         try:
             response.json()

tcmb/core.py

@@ -122,7 +122,9 @@ def read(
     # convert list to str seperated by "-"
     # as descibed in the api reference
     if isinstance(series, list):
-        series = "-".join(series)
+        series_str = "-".join(series)
+    else:
+        series_str = series
 
     # convert list to str seperated by "-"
     if isinstance(agg, list):
@@ -135,7 +137,7 @@ def read(
         end = utils.standardize_date(end)
 
     params = {
-        "series": series,
+        "series": series_str,
         "startDate": start or "01-01-1970",
         "endDate": end or date.today().strftime("%d-%m-%Y"),
         "type": "json",  # csv, xml, json
@@ -156,7 +158,7 @@ def read(
 
     # convert response JSON to DataFrame
     #   time series data is in the "items"
-    data = utils.to_dataframe(res.json()["items"])
+    data = utils.to_dataframe(res.json()["items"], series=series)
 
     return data
 

tcmb/utils.py

@@ -10,6 +10,8 @@
 import numpy as np
 import pandas as pd
 
+from tcmb._data import fetch_dg_series_codes
+
 
 def standardize_date(date_str: str) -> str:
     """Standardize date string format to output DD-MM-YYYY.
@@ -45,14 +47,22 @@ def standardize_date(date_str: str) -> str:
     return datetime.strptime(date_str, date_format).strftime("%d-%m-%Y")
 
 
-def to_dataframe(data: dict) -> pd.DataFrame:
+def to_dataframe(data: dict, series: str | list) -> pd.DataFrame:
     """Convert data from the json response to pandas DataFrame."""
+    if isinstance(series, str):
+        series = [series]
+
+    # generate column names from series keys by replacing "." with "_"
+    # these columns will be selected as the value columns
+    # the columns other than date (Tarih) and value columns will be dropped
+    value_cols = [col.replace(".", "_") for col in series]
+
     df = pd.DataFrame(data)
     # drop unused column
-    df = df.drop("UNIXTIME", axis=1)
+    df = df[["Tarih", *value_cols]]
     # set date as index
-    # TODO: check if "Tarih" always the first column
-    df = df.set_index(df.columns[0])
+    # TODO: check if "Tarih" always the date column
+    df = df.set_index("Tarih")
 
     # detect date format
     if re.match(r"\d+-\d+-\d{4}", df.index[0]):
@@ -63,6 +73,7 @@ def to_dataframe(data: dict) -> pd.DataFrame:
         date_format = "%Y-%m"
     elif re.match(r"\d{4}", df.index[0]):
         date_format = "%Y"
+
     # convert date strings to datetime
     df.index = pd.to_datetime(df.index, format=date_format)
 
@@ -128,8 +139,6 @@ def wildcard_search(
                 dg_series = json.load(file)
 
         else:
-            from tcmb._data import fetch_dg_series_codes
-
             dg_series = fetch_dg_series_codes()
 
         # merge series of all datagroups into one list

tests/test_utils.py

@@ -0,0 +1,31 @@
+from tcmb import utils
+
+import pytest
+
+
+test_date_data = [
+    ("01-01-2024", "01-01-2024"),
+    ("17.12.2009", "17-12-2009"),
+    ("2011-06-18", "18-06-2011"),
+    ("2018.04.10", "10-04-2018"),
+]
+
+
+@pytest.mark.parametrize("date_str, expected", test_date_data)
+def test_standardize_date(date_str, expected):
+    result = utils.standardize_date(date_str)
+    print(result)
+    assert result == expected
+
+
+def test_wildcard_search():
+    items_expected = [
+        "TP.API.REP.TL.A12",
+        "TP.API.REP.TL.A23",
+        "TP.API.REP.TL.G1",
+        "TP.API.REP.TL.G1530",
+        "TP.API.REP.TL.G214",
+    ]
+    items = utils.wildcard_search("TP.API.REP.TL.*")
+
+    assert all(item in items_expected for item in items)