diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml index 78a2790..490cc05 100644 --- a/.github/workflows/ci.yml +++ b/.github/workflows/ci.yml @@ -19,10 +19,24 @@ jobs: experimental-features = nix-command flakes - name: build dependencies run: nix build path://$PWD/ + + - name: python3 --version + run: PATH="$PWD/result/bin" python3 --version + - name: awk --version + run: PATH="$PWD/result/bin" awk --version + - name: sh --version + run: PATH="$PWD/result/bin" sh --version + - name: sha256sum --version + run: PATH="$PWD/result/bin" sha256sum --version + - name: sqlite3 --version + run: PATH="$PWD/result/bin" sqlite3 --version + - name: git --version + run: PATH="$PWD/result/bin" git --version + - name: Run tests - run: PATH="$PWD/result/bin:$PATH" ./test.py + run: PATH="$PWD/result/bin" python3 ./test.py - name: Run tests - run: PATH="$PWD/result/bin" ./test.py + run: PATH="$PWD/result/bin" python3 ./test.py - uses: actions/upload-artifact@v2 with: name: debug-git.tar.gz diff --git a/hash-files.py b/hash-files.py index 93a20ac..f5e14c3 100755 --- a/hash-files.py +++ b/hash-files.py @@ -5,12 +5,14 @@ import sys import subprocess def hashFile(filename): - #print("hashFile " + filename) - return subprocess.check_output(['sha256sum', '--binary', '--zero', filename])[0:64] + result = subprocess.check_output(['sha256sum', '--binary', '--zero', filename])[0:64] + print("hashFile("+x+") = "+str(result), file=sys.stderr) + return result def hash1(bytes): - #print("hash1 " + bytes) - return subprocess.check_output(['sha256sum', '--binary', '--zero'], input=bytes)[0:64] + result = subprocess.check_output(['sha256sum', '--binary', '--zero'], input=bytes)[0:64] + print("hash1("+x+") = "+str(result), file=sys.stderr) + return result # # TODO: use this to get the hashes and names for all roots of the DAG (commits that are reachable only through one (or several) direct branch names, but not transitively as ancestors of other commits) @@ -35,10 +37,14 @@ def ref_exists(path, ref): def hashGit(path): FETCH_HEAD = "FETCH_HEAD" if ref_exists(path, "FETCH_HEAD") else '' HEAD = "HEAD" if ref_exists(path, "HEAD") else '' - return subprocess.check_output(['sh', '-c', git_command.format(HEAD=HEAD, FETCH_HEAD=FETCH_HEAD)], cwd=path) + result = subprocess.check_output(['sh', '-c', git_command.format(HEAD=HEAD, FETCH_HEAD=FETCH_HEAD)], cwd=path) + print("hashGit("+x+") = "+str(result), file=sys.stderr) + return result def hashSqlite3(path): - return subprocess.check_output(['sh', '-c', 'sqlite3 "$1" .dump | sort | sha256sum --binary --zero', '--', os.path.abspath(path)]) + result= subprocess.check_output(['sh', '-c', 'sqlite3 "$1" .dump | sort | sha256sum --binary --zero', '--', os.path.abspath(path)]) + print("hashSqlite3("+x+") = "+str(result), file=sys.stderr) + return result def ignore_exitcode(cmd, **kwargs): try: @@ -52,24 +58,28 @@ def is_git(x): ignore_exitcode("git rev-parse --is-inside-work-tree 2>/dev/null", cwd=x, shell=True).strip() == b'true') # TODO: if a file which is inside a larger git dir is passed on the CLI, this still returns True :-( -def recur(x): +def recur(depth, x): #print(x) # initial list of paths if isinstance(x, list): - return b'root\0' + b''.join(recur(os.path.abspath(path)) + b' ' + path.encode('utf-8') + b'\0' for path in sorted(x)) + print("ROOT " + str(depth) + ' ' + x, file=sys.stderr) + return b'root\0' + b''.join(recur(depth + 1, os.path.abspath(path)) + b' ' + path.encode('utf-8') + b'\0' for path in sorted(x)) # GIT repo elif is_git(x): - #print("GIT DIR " + x, file=sys.stderr) + print("GIT DIR " + str(depth) + ' ' + x, file=sys.stderr) return hash1(b'git-versioned folder\0' + hashGit(x)) # directory elif os.path.isdir(x): - return hash1(b'directory\0' + b''.join(recur(os.path.join(x, entry)) + b' ' + entry.encode('utf-8') + b'\0' for entry in os.listdir(x))) + print("DIR " + str(depth) + ' ' + x, file=sys.stderr) + return hash1(b'directory\0' + b''.join(recur(depth + 1, os.path.join(x, entry)) + b' ' + entry.encode('utf-8') + b'\0' for entry in os.listdir(x))) elif b'SQLite 3.x database' in subprocess.check_output(["file", x]): + print("SQLITE3 " + str(depth) + ' ' + x, file=sys.stderr) return hashSqlite3(x) # Just a file elif os.path.isfile(x): + print("PLAIN FILE " + str(depth) + ' ' + x, file=sys.stderr) return hashFile(x) else: sys.exit("unknown file type for %s" % os.path.abspath(x)) -print(hash1(recur(sys.argv[1:])).decode('utf-8')) +print(hash1(recur(0, sys.argv[1:])).decode('utf-8'))