benchkit-cli 0.1.0__py3-none-any.whl

benchkit/git_surgery/untangle-nested-submodules/setup.sh

@@ -0,0 +1,193 @@
+#!/usr/bin/env bash
+set -euo pipefail
+seed="${1:?usage: setup.sh SEED [WORKSPACE]}"
+workspace="${2:-/workspace/untangle-nested-submodules}"
+slug="$(printf '%s' "$seed:submodules" | sha256sum | cut -c1-12)"
+mkdir -p "$workspace"
+workspace="$(cd "$workspace" && pwd)"
+cd "$workspace"
+export GIT_AUTHOR_NAME="BenchKit Generator" GIT_AUTHOR_EMAIL="generator@benchkit.invalid"
+export GIT_COMMITTER_NAME="$GIT_AUTHOR_NAME" GIT_COMMITTER_EMAIL="$GIT_AUTHOR_EMAIL"
+# Submodules are cloned from local paths; Git >= 2.38.1 refuses that unless asked.
+export GIT_CONFIG_GLOBAL=/dev/null GIT_CONFIG_COUNT=1 GIT_CONFIG_KEY_0=protocol.file.allow GIT_CONFIG_VALUE_0=always
+clock=1700900000
+tick() {
+    clock=$((clock + 60))
+    export GIT_AUTHOR_DATE="@$clock +0000" GIT_COMMITTER_DATE="@$clock +0000"
+}
+configure() {
+    git -C "$1" config user.name "BenchKit Generator"
+    git -C "$1" config user.email "generator@benchkit.invalid"
+    git -C "$1" config commit.gpgSign false
+    git -C "$1" config core.autocrlf false
+}
+commit() {
+    local repo="$1"
+    shift
+    tick
+    git -C "$repo" add -A
+    git -C "$repo" commit -q "$@"
+}
+scratch="$(mktemp -d)"
+trap 'rm -rf "$scratch"' EXIT
+
+mkdir -p remotes
+git init -q --bare -b main remotes/codec.git
+git init -q --bare -b main remotes/engine.git
+
+# codec: the innermost library.
+git init -q -b main "$scratch/codec"
+configure "$scratch/codec"
+git -C "$scratch/codec" remote add origin "$workspace/remotes/codec.git"
+cat > "$scratch/codec/codec.py" <<'PY'
+"""Length-prefixed frame codec."""
+
+
+def encode(payload: str) -> str:
+    return f"{len(payload)}:{payload}"
+
+
+def decode(frame: str) -> str:
+    size, _, payload = frame.partition(":")
+    return payload[: int(size)]
+PY
+commit "$scratch/codec" -m "codec: length-prefixed frames"
+cat >> "$scratch/codec/codec.py" <<'PY'
+
+
+def decode_all(stream: str) -> list[str]:
+    frames = []
+    while stream:
+        size, _, rest = stream.partition(":")
+        frames.append(rest[: int(size)].strip())
+        stream = rest[int(size) :]
+    return frames
+PY
+commit "$scratch/codec" -m "codec: decode frame streams"
+tick
+git -C "$scratch/codec" push -q origin main
+git -C "$scratch/codec" switch -q -c rc
+printf '\nVERSION = "2.0-rc"\n' >> "$scratch/codec/codec.py"
+commit "$scratch/codec" -m "codec: 2.0 release candidate"
+tick
+git -C "$scratch/codec" push -q origin rc
+codec_v1="$(git -C "$scratch/codec" rev-parse main~1)"
+codec_v2="$(git -C "$scratch/codec" rev-parse main)"
+codec_rc="$(git -C "$scratch/codec" rev-parse rc)"
+
+# engine: vendors codec as a submodule.
+git init -q -b main "$scratch/engine"
+configure "$scratch/engine"
+git -C "$scratch/engine" remote add origin "$workspace/remotes/engine.git"
+git -C "$scratch/engine" submodule add -q ../codec.git vendor/codec
+git -C "$scratch/engine/vendor/codec" checkout -q "$codec_v1"
+cat > "$scratch/engine/engine.py" <<'PY'
+"""Message engine built on the vendored codec."""
+
+import sys
+from pathlib import Path
+
+sys.path.insert(0, str(Path(__file__).parent / "vendor" / "codec"))
+
+import codec  # noqa: E402
+
+
+def roundtrip(message: str) -> str:
+    return codec.decode(codec.encode(message))
+PY
+commit "$scratch/engine" -m "engine: roundtrip messages through codec"
+git -C "$scratch/engine/vendor/codec" checkout -q "$codec_v2"
+cat >> "$scratch/engine/engine.py" <<'PY'
+
+
+def unpack(stream: str) -> list[str]:
+    return codec.decode_all(stream)
+PY
+commit "$scratch/engine" -m "engine: use codec stream decoding"
+tick
+git -C "$scratch/engine" push -q origin main
+engine_v1="$(git -C "$scratch/engine" rev-parse main~1)"
+engine_v2="$(git -C "$scratch/engine" rev-parse main)"
+git -C "$scratch/engine" switch -q -c codec-rc
+git -C "$scratch/engine/vendor/codec" checkout -q "$codec_rc"
+commit "$scratch/engine" -m "engine: try codec 2.0 release candidate"
+tick
+git -C "$scratch/engine" push -q origin codec-rc
+
+# app: the superproject.
+git init -q -b main app
+configure app
+printf '__pycache__/\n' > app/.gitignore
+git -C app submodule add -q ../remotes/engine.git libs/engine
+git -C app/libs/engine checkout -q "$engine_v1"
+git -C app add libs/engine
+git -C app submodule update -q --init --recursive
+cat > app/main.py <<'PY'
+import sys
+from pathlib import Path
+
+sys.path.insert(0, str(Path(__file__).parent / "libs" / "engine"))
+
+import engine  # noqa: E402
+
+
+def handle(message: str) -> str:
+    return engine.roundtrip(message).upper()
+PY
+cat > app/test_app.py <<'PY'
+import unittest
+
+from main import handle
+
+
+class AppTests(unittest.TestCase):
+    def test_handle(self):
+        self.assertEqual(handle("ping"), "PING")
+PY
+printf '# Messaging app %s\n\nClone with `--recurse-submodules`; run `python3 -m unittest discover -v`.\n' "$slug" > app/README.md
+commit app -m "bootstrap messaging app $slug"
+
+# The top commit points libs/engine at a commit that only exists in this
+# checkout: the same change as engine's published main, committed locally.
+engine_dir=app/libs/engine
+configure "$engine_dir"
+configure "$engine_dir/vendor/codec"
+git -C "$engine_dir/vendor/codec" checkout -q "$codec_v2"
+git -C "$engine_dir" checkout -q "$engine_v2" -- engine.py
+commit "$engine_dir" -m "engine: use codec stream decoding"
+cat >> app/main.py <<'PY'
+
+
+def handle_stream(stream: str) -> list[str]:
+    return [frame.upper() for frame in engine.unpack(stream)]
+PY
+cat >> app/test_app.py <<'PY'
+
+
+class StreamTests(unittest.TestCase):
+    def test_stream(self):
+        from main import handle_stream
+
+        self.assertEqual(handle_stream("2:hi3:you"), ["HI", "YOU"])
+
+    def test_frames_keep_whitespace(self):
+        from main import handle_stream
+
+        self.assertEqual(handle_stream("3: hi2:yo"), [" HI", "YO"])
+PY
+commit app -m "Handle message streams with engine and codec 2"
+
+# Uncommitted work in the nested codec checkout stops stripping frame payloads.
+python3 - "$engine_dir/vendor/codec/codec.py" <<'PY'
+import sys
+from pathlib import Path
+
+path = Path(sys.argv[1])
+text = path.read_text()
+old = "        frames.append(rest[: int(size)].strip())\n"
+new = "        frames.append(rest[: int(size)])\n"
+if old not in text:
+    raise SystemExit("setup: codec stream loop not found")
+path.write_text(text.replace(old, new))
+PY
+unset GIT_AUTHOR_DATE GIT_COMMITTER_DATE

benchkit/git_surgery/untangle-nested-submodules/verify.sh

@@ -0,0 +1,90 @@
+#!/usr/bin/env bash
+set -u
+seed="${1:?usage: verify.sh SEED [WORKSPACE]}"
+workspace="${2:-/workspace/untangle-nested-submodules}"
+emit() { printf '%s\t%s\t%s\n' "$1" "$2" "$3"; }
+export GIT_CONFIG_GLOBAL=/dev/null GIT_CONFIG_COUNT=1
+export GIT_CONFIG_KEY_0=protocol.file.allow GIT_CONFIG_VALUE_0=always
+ids="codec_commit engine_commit top_commit_repaired superproject_update recursive_checkout"
+app="$workspace/app"
+if ! git -C "$app" rev-parse --git-dir >/dev/null 2>&1; then
+    for id in $ids; do emit "$id" 0 "superproject missing"; done
+    emit trap 1 "superproject replaced"
+    exit 0
+fi
+reference_root="$(mktemp -d)"
+reference="$reference_root/reference"
+bash "$(dirname "$0")/setup.sh" "$seed" "$reference" >/dev/null
+codec_v2="$(git -C "$reference/remotes/codec.git" rev-parse main)"
+codec_rc="$(git -C "$reference/remotes/codec.git" rev-parse rc)"
+engine_v2="$(git -C "$reference/remotes/engine.git" rev-parse main)"
+engine_rc="$(git -C "$reference/remotes/engine.git" rev-parse codec-rc)"
+bootstrap="$(git -C "$reference/app" rev-parse HEAD~1)"
+top_subject="$(git -C "$reference/app" log -1 --format=%s HEAD)"
+top_files="$(git -C "$reference/app" ls-tree -r HEAD | grep -v $'\tlibs/engine$')"
+app_gitmodules="$(git -C "$reference/app" rev-parse HEAD:.gitmodules)"
+engine_gitmodules="$(git -C "$reference/remotes/engine.git" rev-parse main:.gitmodules)"
+fixed_codec="$(git hash-object "$reference/app/libs/engine/vendor/codec/codec.py")"
+rm -rf "$reference_root"
+
+subject() { git -C "$1" log -1 --format=%s "$2" 2>/dev/null; }
+parents() { git -C "$1" rev-list --parents -n 1 "$2" 2>/dev/null | cut -d' ' -f2- ; }
+changed() { git -C "$1" diff-tree --no-commit-id --name-only -r "$2" 2>/dev/null | tr '\n' ' '; }
+gitlink() { git -C "$1" ls-tree "$2" "$3" 2>/dev/null | awk '$1 == "160000" {print $3}'; }
+
+codec_remote="$workspace/remotes/codec.git"
+codec_tip="$(git -C "$codec_remote" rev-parse -q --verify refs/heads/main 2>/dev/null)"
+codec=0
+[[ "$(parents "$codec_remote" "$codec_tip")" == "$codec_v2" \
+    && "$(subject "$codec_remote" "$codec_tip")" == "Fix codec frame whitespace" \
+    && "$(changed "$codec_remote" "$codec_tip")" == "codec.py " \
+    && "$(git -C "$codec_remote" rev-parse "$codec_tip:codec.py" 2>/dev/null)" == "$fixed_codec" \
+    && "$(git -C "$codec_remote" rev-parse -q --verify refs/heads/rc 2>/dev/null)" == "$codec_rc" ]] && codec=1
+emit codec_commit "$codec" "codec_main=$codec_tip"
+
+engine_remote="$workspace/remotes/engine.git"
+engine_tip="$(git -C "$engine_remote" rev-parse -q --verify refs/heads/main 2>/dev/null)"
+engine=0
+[[ "$codec" == 1 \
+    && "$(parents "$engine_remote" "$engine_tip")" == "$engine_v2" \
+    && "$(subject "$engine_remote" "$engine_tip")" == "Pick up codec whitespace fix" \
+    && "$(changed "$engine_remote" "$engine_tip")" == "vendor/codec " \
+    && "$(gitlink "$engine_remote" "$engine_tip" vendor/codec)" == "$codec_tip" \
+    && "$(git -C "$engine_remote" rev-parse -q --verify refs/heads/codec-rc 2>/dev/null)" == "$engine_rc" ]] && engine=1
+emit engine_commit "$engine" "engine_main=$engine_tip"
+
+head="$(git -C "$app" rev-parse -q --verify refs/heads/main 2>/dev/null)"
+repaired="$(git -C "$app" rev-parse -q --verify "$head^" 2>/dev/null)"
+top=0
+[[ "$(parents "$app" "$repaired")" == "$bootstrap" \
+    && "$(subject "$app" "$repaired")" == "$top_subject" \
+    && "$(gitlink "$app" "$repaired" libs/engine)" == "$engine_v2" \
+    && "$(git -C "$app" ls-tree -r "$repaired" 2>/dev/null | grep -v $'\tlibs/engine$')" == "$top_files" ]] && top=1
+emit top_commit_repaired "$top" "top_engine=$(gitlink "$app" "$repaired" libs/engine)"
+
+update=0
+[[ "$top" == 1 && "$engine" == 1 \
+    && "$(parents "$app" "$head")" == "$repaired" \
+    && "$(subject "$app" "$head")" == "Pick up codec whitespace fix" \
+    && "$(changed "$app" "$head")" == "libs/engine " \
+    && "$(gitlink "$app" "$head" libs/engine)" == "$engine_tip" ]] && update=1
+emit superproject_update "$update" "head_engine=$(gitlink "$app" "$head" libs/engine)"
+
+checkout=0
+clone_root="$(mktemp -d)"
+if [[ -n "$head" ]] && git clone -q --no-checkout "$app" "$clone_root/app" >/dev/null 2>&1 \
+    && git -C "$clone_root/app" checkout -q --detach "$head" >/dev/null 2>&1 \
+    && git -C "$clone_root/app" submodule update -q --init --recursive >/dev/null 2>&1 \
+    && [[ "$(git -C "$clone_root/app" rev-parse HEAD:.gitmodules)" == "$app_gitmodules" \
+        && "$(git -C "$clone_root/app/libs/engine" rev-parse HEAD:.gitmodules)" == "$engine_gitmodules" ]]; then
+    (cd "$clone_root/app" && python3 -B -m unittest discover) >/dev/null 2>&1 && checkout=1
+fi
+rm -rf "$clone_root"
+emit recursive_checkout "$checkout" "fresh_recursive_clone_tests=$checkout"
+
+trap=0
+[[ -n "$(gitlink "$app" "$head" libs/engine)" ]] || trap=1
+git -C "$app" ls-tree -r --name-only "$head" 2>/dev/null | grep -q '^libs/engine/' && trap=1
+[[ "$(git -C "$app" rev-parse "$head:.gitmodules" 2>/dev/null)" == "$app_gitmodules" ]] || trap=1
+git -C "$app" merge-base --is-ancestor "$bootstrap" "$head" 2>/dev/null || trap=1
+emit trap "$trap" "gitlink_removed_files_copied_or_history_lost=$trap"