pgload

load: Faster Loading for CSV & JSONL Data into PostgreSQL

Loading large CSV or JSONL files into PostgreSQL can often be a slow process. load is a command-line tool designed to help speed this up. It leverages PostgreSQL's efficient COPY command and parallel processing to handle large datasets, including multi-gigabyte files with millions of rows, more effectively than standard single-file insertion methods.

This tool can be particularly helpful when you need to regularly import large data files and want to automate parts of the process, like table creation.

Features

• Faster CSV Imports: Uses PostgreSQL's native COPY command for efficient bulk loading of CSV data.
• JSONL File Support: Can process JSONL files (where each line is a JSON object). It converts the data to CSV format on the fly and then uses the COPY command to load. While the conversion adds some overhead compared to direct CSV loading, it's still designed to handle large JSONL files effectively.
• Handles Large Files: Tested with multi-gigabyte files containing millions of rows (see examples below).
• Concurrent File Loading: Speeds up loading multiple files by processing them concurrently using 8 internal workers.
• File Pattern Matching: Accepts multiple file paths and supports glob patterns (e.g., data/*.csv) for easily selecting files.
• Automatic Table Creation: Analyzes the input file(s) to infer a schema and create the target PostgreSQL table if it doesn't exist.
  • Supported data types for auto-schema: TEXT, NUMERIC, JSON. (This covers common cases but may need manual adjustment for more complex types).
• Schema Inference Tuning: Provides an option (lookup size) to adjust how many rows are sampled for determining data types, allowing a trade-off between speed and accuracy.
• Compressed File Handling: Reads .gz compressed CSV and JSONL files directly, avoiding a separate decompression step.

Installation

Ensure you have Go installed.

go install github.com/anvesh9652/pgload/cmd/load@latest

(Make sure your Go binary path ($GOPATH/bin or $HOME/go/bin) is included in your system's PATH environment variable.)

View Help and Options:

load -h

Command-Line Flags

Use these flags to customize the loading process:

Example Commands:

# Load multiple CSV files (including a compressed one).
# Assumes default format ('csv') and connection settings.
load file1.csv file2.csv file3.csv.gz

# Load multiple JSON/JSONL files (including compressed).
# Explicitly specifies the format using -f jsonl.
load -f jsonl file1.json file2.jsonl file3.json.gz

# Load a CSV file, specifying a non-default PostgreSQL port (54321).
load -p 54321 data.csv

# Load a mix of CSV and JSONL files (-f both).
# Specifies a non-default port and uses a glob pattern (*) to include files.
load -f both -p 54321 data.csv data.json all_files/*

# Load CSV files matching patterns, using specific connection parameters:
# User 'test', Password '123', Database 'temp', Schema 'testing',
# and connects to PostgreSQL at 'localhost:123'.
load -U test -P 123 -d temp -s testing -u "localhost:123" file_2*.csv test1.csv dummy/*/*.csv

(Note: Table names are inferred from filenames.)

Loading Speed Stats

These examples show load's performance loading large files on specific hardware (MacBook Pro 15-inch, M1 Pro, 10 cores, 16GB RAM). Your results may vary based on your hardware, database configuration, and network.

License

This project is licensed under the MIT License - see the LICENSE file for details.