fix: Excel implementation from polars

README.md

@@ -51,45 +51,27 @@ Clone the [repository](https://github.com/oasci/vaxstats):
 git clone https://github.com/oasci/vaxstats.git
 ```
 
-### Conda environment
-
-Move into `vaxstats` directory (`cd vaxstats`) and install the development conda environment using [GNU Make](https://www.gnu.org/software/make/) (which could be installed by default on your system).
-
-```bash
-make environment
-```
-
-Now you can activate the new conda environment `vaxstats-dev` and use `vaxstats` commands.
-
-```sh
-conda activate vaxstats-dev
-```
-
-### Manual install
-
-Alternatively, you can manually install `vaxstats` using `pip` after moving into the directory.
+Install `vaxstats` using `pip` after moving into the directory.
 
 ```sh
 pip install .
 ```
 
 This will install all dependencies and `vaxstats` into your current Python environment.
 
-## Deploying
+## Development
 
-We use [bump-my-version](https://github.com/callowayproject/bump-my-version) to release a new version.
-This will create a git tag used by [poetry-dynamic-version](https://github.com/mtkennerly/poetry-dynamic-versioning) to generate version strings and update `CHANGELOG.md`.
-
-For example, you would run the following command to bump the `minor` version.
+We use [pixi](https://pixi.sh/latest/) to manage Python environments and simplify the developer workflow.
+Once you have [pixi](https://pixi.sh/latest/) installed, move into `vaxstats` directory (e.g., `cd vaxstats`) and install the  environment using the command
 
 ```bash
-poetry run bump-my-version bump minor
+pixi install
 ```
 
-After releasing a new version, you must push and include all tags.
+Now you can activate the new virtual environment using
 
-```bash
-git push --follow-tags
+```sh
+pixi shell
 ```
 
 ## License

pyproject.toml

@@ -34,13 +34,15 @@ black = { cmd = ["black", "--config", "pyproject.toml", "./content"] }
 format = { depends-on = ["mdlint", "isort", "black"] }
 
 [tool.pixi.dependencies]
-python = ">=3.11.0,<3.13"
+python = ">=3.12.0,<3.13"
 polars = ">=1.12.0,<2"
 loguru = ">=0.7.2,<0.8"
 statsforecast = ">=1.7.8,<2"
 xlsx2csv = ">=0.8.3,<0.9"
 pandas = ">=2.2.3,<3"
 matplotlib = ">=3.9.2,<4"
+pyyaml = ">=6.0.2,<7"
+fastexcel = ">=0.12.0,<0.13"
 
 [tool.pixi.feature.dev.dependencies]
 ruff = ">=0.7.2,<0.8"

tests/test_df_prep.py

@@ -18,7 +18,7 @@ def test_clean_df(path_example_excel):
 def test_prep_df(path_example_excel):
     df = load_file(path_example_excel)
     df = clean_df(df)
-    df = prep_forecast_df(df, 0, 1, 6)
+    df = prep_forecast_df(df, 0, 0, 6)
     assert df.columns == ["unique_id", "ds", "y"]
     assert df.shape == (2_721, 3)
 

vaxstats/io.py

@@ -74,6 +74,10 @@ def prep_forecast_df(
         DataFrame's columns.
     ValueError: If the date and time strings do not match the specified formats.
 
+    Notes:
+        If `date_idx` and `time_idx` are the same, we combine `input_date_fmt` and
+        `input_time_fmt` and load from the specified column.
+
     Examples:
     >>> import polars as pl
     >>> data = {'date': ["01-01-23", "01-02-23"], 'time': ["01:00:00 PM", "02:00:00 PM"], 'y': [10, 20]}
@@ -96,35 +100,30 @@ def prep_forecast_df(
         raise IndexError("One or more column indices are out of range")
 
     # Select only the required columns using indices
-    df = df.select(df.columns[date_idx], df.columns[time_idx], df.columns[y_idx])
-
-    logger.debug("Combining date and time columns")
-    df = df.with_columns(
-        [pl.concat_str([df.columns[0], df.columns[1]], separator=" ").alias("ds")]
-    )
-    logger.debug(f"Example row: {df[0]}")
-
-    logger.debug(
+    if date_idx == time_idx:
+        df = df.select(df.columns[date_idx], df.columns[y_idx])
+        df = df.rename({df.columns[0]: "ds"})
+    else:
+        df = df.select(df.columns[date_idx], df.columns[time_idx], df.columns[y_idx])
+        logger.debug("Combining date and time columns")
+        df = df.with_columns(
+            [pl.concat_str([df.columns[0], df.columns[1]], separator=" ").alias("ds")]
+        )
+        logger.debug(
         f"Parsing datetimes with date format '{input_date_fmt}' and time format '{input_time_fmt}'"
-    )
-    df = df.with_columns(
-        [
-            pl.col("ds")
-            .str.strptime(pl.Datetime, format=f"{input_date_fmt} {input_time_fmt}")
-            .alias("parsed_datetime")
-        ]
-    )
-    logger.debug(f"Example row: {df[0]}")
+        )
+        df = df.with_columns(
+            [
+                pl.col("ds")
+                .str.strptime(pl.Datetime, format=f"{input_date_fmt} {input_time_fmt}", strict=False)
+                .alias("parsed_datetime")
+            ]
+        )
 
-    logger.debug(f"Writing datetimes in '{output_fmt}'")
-    df = df.with_columns(
-        [pl.col("parsed_datetime").dt.strftime(output_fmt).alias("ds")]
-    )
-
-    df = df.drop("parsed_datetime")
+    logger.debug(f"Example row: {df[0]}")
 
     # Rename the y column
-    df = df.rename({df.columns[2]: "y"})
+    df = df.rename({df.columns[1]: "y"})
 
     logger.debug("Adding unique_id column")
     df = df.with_columns(pl.lit(0).alias("unique_id"))