Hack to fix a broken url using the GitHub API

fix_repos.sh

#!/usr/bin/env bash

source token.sh # should export the ACCESS_TOKEN and USER variables

PAGE=1 # Run once per page (manually for now)
PER_PAGE=100 # Max 100

function list_repos {
    echo "Listing repos" > /dev/stderr
    curl -s -H "Authorization: token $ACCESS_TOKEN" "https://api.github.com/search/code?page=$PAGE&per_page=$PER_PAGE&q=http://www.eecs.harvard.edu/~kirsch/pubs/bbbf/esa06.pdf" | \
        jq --raw-output '.items[]["repository"] | select(.fork == false) | .full_name' | sort -u
}

function fork {
    repo=$1
    echo "Forking $1"
    curl -s -H "Authorization: token $ACCESS_TOKEN" -XPOST "https://api.github.com/repos/$repo/forks" >> fork.log
}

function repo_name {
    repo=$1
    echo "$repo" | cut -f2 -d/
}

function clone {
    repo=$1
    fork=$USER/$(repo_name "$repo")
    github="[email protected]:${fork}.git"
    mkdir -p repos
    (
        cd repos
        for retry in $(seq 6); do
            if git clone --depth=1 "$github"; then
                break
            fi
            # There is a window after a fork where clones fail, just wait it out
            echo "Clone failed, retrying in 10 seconds"
            sleep 10;
        done
    )
}

function fix_repo {
    repo=$1
    (
        cd repos/$(repo_name "$repo") &&
            git --no-pager grep -l --full-name http://www.eecs.harvard.edu/\~kirsch/pubs/bbbf/esa06.pdf | \
                while read -r f; do
                    sed -i "s#http://www\.eecs\.harvard\.edu/~kirsch/pubs/bbbf/esa06\.pdf#https://www.eecs.harvard.edu/~michaelm/postscripts/tr-02-05.pdf#" "$f"
                done &&
            git add -Av &&
            git commit -m "Fix broken url to \"Building a Better Bloom Filter\"" &&
            git push -f
    )
}

function reset_repo {
    repo=$1
    path=repos/$(repo_name "$repo")
    echo "Resetting $1"
    (
        cd "$path" &&
            git reset --hard
    )
}

function undo_commit {
    repo=$1
    path=repos/$(repo_name "$repo")
    echo "Reverting to origin: $path"
    (
        cd "$path" &&
            git reset --hard "origin/$(git rev-parse --symbolic-full-name --abbrev-ref HEAD)"
    )
}

function submit_pr {
    repo=$1
    path=repos/$(repo_name "$repo")
    (
        cd "$path"
        branch=$(git rev-parse --symbolic-full-name --abbrev-ref HEAD)
        echo "Using branch $branch"
        curl -s -H "Authorization: token $ACCESS_TOKEN" -XPOST "https://api.github.com/repos/$repo/pulls" --data-binary \
             '{"title": "Fix broken url to \"Better Bloom Filter\" paper", "body": "I noticed that this link was broken so I replaced it with a link to a copy provided by the other author", "head": "'$USER':'"$branch"'", "base": "'"$branch"'"}' >> prs.log
    )
    echo "$repo" >> done.txt
}

function wait_for_fork {
    # Fork API is async
    repo=$1
    repo_name=$(repo_name "$repo")
    while true; do
        if curl -s -H "Authorization: token $ACCESS_TOKEN" "https://api.github.com/repos/${USER}/${repo_name}" | jq --raw-output '.' | grep "$repo_name" >/dev/null ; then
            echo "$repo is forked"
            break
        else
            echo "$repo not forked yet"
            sleep 3;
        fi
    done
}

function delete_repo {
    repo=$1
    repo_name=$(repo_name "$repo")
    fork=${USER}/${repo_name}
    curl -s -H "Authorization: token ${ACCESS_TOKEN}" -XDELETE "https://api.github.com/repos/${fork}"
    (
        cd repos &&
            rm -rf "$repo_name"
    )
}

for repo in $(list_repos); do
    if grep "$repo" done.txt >/dev/null; then
        echo "Already done, skipping $repo"
        continue
    fi
    echo "Doing $repo"
    fork "$repo"
    wait_for_fork "$repo"
    clone "$repo"
    fix_repo "$repo"
    submit_pr "$repo"
    delete_repo "$repo"
    # Used for testing:
    ## reset_repo "$repo"
    ## undo_commit "$repo"
done