| name | csv-data-loader |
| description | Load and parse CSV files into structured data frames with configurable encoding, delimiter, and date parsing options. |
| version | 0.3.1 |
CSV Data Loader
Skill Summary
Reads CSV files from disk and parses them into pandas DataFrames. Supports configurable delimiters (comma, tab, semicolon), file encodings (UTF-8, Latin-1, GBK), and automatic date column detection. Handles common CSV quirks like BOM markers, inconsistent quoting, and mixed line endings.
Inputs
--input / -i: Path to CSV file (required)
--delimiter / -d: Column delimiter (default: ,)
--encoding / -e: File encoding (default: utf-8)
--date-columns: Comma-separated list of columns to parse as dates
--output / -o: Output directory for parsed result (Parquet format)
Processing Steps
- Detect file encoding if not specified (chardet fallback)
- Read CSV with pandas
read_csv, applying delimiter and encoding
- Strip leading/trailing whitespace from string columns
- Auto-detect and parse date columns (ISO 8601, US, EU formats)
- Report column types and row count
Output
- Parsed DataFrame saved as Parquet file in output directory
- Console summary: row count, column names and dtypes, detected date columns
Implementation
- Command:
python3 ./skills/csv_data_loader/loader.py -i <file> -o <output_dir>
- Dependencies:
pandas, chardet