stage2.ipython

from tqdm import tqdm 
 
%cd /home/vadim 
!rm -rf boilerplate 
!mkdir -p boilerplate 
!ln -s ../hercules.bin boilerplate/hercules.bin 
with open("tasks1_tail.txt") as fin: 
    %cd boilerplate 
    rid = "" 
    lines = fin.readlines() 
    for seq, line in tqdm(enumerate(lines), total=len(lines)): 
        commit, repo = line.strip().split(" ") 
        if "llvm" in repo or "linux" in repo: 
            continue 
        new_rid = repo.replace("/", "@") 
        if new_rid != rid: 
            !rm -rf $rid 
        rid = new_rid 
        !GIT_TERMINAL_PROMPT=0 git clone --bare https://github.com/$repo $rid 
        if not os.path.exists(rid): 
            print("failed to clone") 
            continue 
        %cd $rid 
        try: 
            parents = !git rev-list --parents -n 1 $commit 
            parent = parents[0].split(" ")[1] 
        except: 
            continue 
        finally: 
            %cd .. 
        if len(parent) != 40: 
            print("failed to get the parent from", parents) 
            continue 
        with open("commits.txt", "w") as fout: 
            print(commit, file=fout) 
            print(parent, file=fout) 
        result = !../hercules --typos-dataset --feature uast --commits commits.txt $rid 
        print(result[:10]) 
        if result and result[-1] != "TyposDataset:" and "hercules:" in result: 
            with open("%s_%d.yaml" % (rid, seq), "w") as fout: 
                ri = result.index("hercules:") 
                print("\n".join(result[ri:]), file=fout) 
    !rm -rf $rid 
    %cd ..