#!/usr/bin/env bash

# This script outputs statistics for the current git repository. This script
# is used by the search-core team to understand the size and shape of a
# repository. In particular we use this when understanding the scale of a
# monorepo to help us guide our work.

set -e

# Do everything from the gitdir. Makes ls-tree below also not work on a
# subtree.
cd "$(git rev-parse --git-dir)"

# The size of the git store (not the working copy).
echo "$(du -sh .)" gitdir

# The number of commits reachable from HEAD
echo "$(git rev-list --count HEAD)" commits


# Some awk which extracts statistics on the files in the latest commit.
echo
echo HEAD statistics
git ls-tree -r --long HEAD | awk '
BEGIN {
    base = 10
    logbase = log(base)
}
$4 != "-" {
    if ($4 == 0) {
        hist[0]++
    } else {
        hist[int(log($4) / logbase) + 1]++
    }
    total += $4
    count++
}
END {
    printf("%.3fGiB\n%d files\n", total / 1024 / 1024 / 1024, count)
    printf("histogram:\n")
    for (x in hist) {
        printf("%d^%d\t%d\n", base, x, hist[x])
    }
}
'