extract-sample-cache-data

extract-sample-cache-data

Extracts a sample of author_login_cache pairs from the claude-leaderboard database for testing purposes.

Purpose

This tool reads from the claude-leaderboard's author_login_cache table and outputs a CSV sample containing between 10-100 pairs. It ensures both valid logins and NULL logins are included (if available), and preserves the original ISO 8601 timestamp format with microsecond precision.

Building

cd /home/coding/commitgraph
go build -o extract-sample-cache-data ./cmd/extract-sample-cache-data/

Usage

Basic usage:

./extract-sample-cache-data -output sample.csv

With custom sample size:

./extract-sample-cache-data -output sample.csv -count 25

With custom database path:

./extract-sample-cache-data -output sample.csv -db /path/to/hot.db

Options

• -db: Path to claude-leaderboard SQLite database (default: ~/backups/claude-leaderboard/hot.db)
• -output: Output CSV file path (required)
• -count: Number of pairs to extract, between 10-100 inclusive (default: 50)

Output Format

The output CSV has three columns:

• author_email: Email address of the author
• github_login: GitHub username/login (literal string "NULL" for unresolved identities)
• resolved_at: ISO 8601 timestamp with microsecond precision

Example:

author_email,github_login,resolved_at
bot@quantifieduncertainty.org,quri-bot,2026-03-14T21:20:01.065651Z
unknown.user@example.com,NULL,2026-08-06T10:00:00.000000Z

Database Schema

The tool reads from the author_login_cache table in the claude-leaderboard SQLite database:

CREATE TABLE author_login_cache (
    author_email TEXT PRIMARY KEY,
    github_login TEXT NOT NULL,
    resolved_at TIMESTAMP NOT NULL
);

Behavior

• Sample size: Automatically clamps to range [10, 100]
• NULL logins: Includes at least 20% NULL logins if available; otherwise logs a warning
• Timestamp format: Preserves original ISO 8601 format with microsecond precision
• Error handling: Validates inputs and reports errors clearly

Related Documentation

See notes/cg-4p4nu.md for detailed author_login_cache data structure documentation.

Task

This tool was created for bead cg-5l4fq: "Write script to extract sample data from cache"