benchkit-cli 0.1.0__py3-none-any.whl

benchkit/git_surgery/revert-merge-with-followups/setup.sh

@@ -0,0 +1,198 @@
+#!/usr/bin/env bash
+set -euo pipefail
+seed="${1:?usage: setup.sh SEED [WORKSPACE]}"
+workspace="${2:-/workspace/revert-merge-with-followups}"
+slug="$(printf '%s' "$seed:revert-merge" | sha256sum | cut -c1-12)"
+mkdir -p "$workspace"
+cd "$workspace"
+git init -q -b main
+git config user.name "BenchKit Generator"
+git config user.email "generator@benchkit.invalid"
+git config commit.gpgSign false
+git config core.autocrlf false
+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"
+export GIT_CONFIG_GLOBAL=/dev/null
+clock=1700600000
+tick() {
+    clock=$((clock + 60))
+    export GIT_AUTHOR_DATE="@$clock +0000" GIT_COMMITTER_DATE="@$clock +0000"
+}
+commit() {
+    tick
+    git add -A
+    git commit -q "$@"
+}
+merge() {
+    tick
+    git merge -q --no-ff "$@"
+}
+replace() {
+    python3 - "$@" <<'PY'
+import sys
+from pathlib import Path
+
+path, old, new = Path(sys.argv[1]), sys.argv[2], sys.argv[3]
+text = path.read_text()
+if old not in text:
+    raise SystemExit(f"setup: {old!r} not found in {path}")
+path.write_text(text.replace(old, new, 1))
+PY
+}
+
+printf '__pycache__/\n' > .gitignore
+cat > pricing.py <<'PY'
+"""Order pricing in integer cents."""
+
+PRICES = {"apple": 120, "pear": 95, "plum": 60}
+
+
+def price(sku: str, qty: int) -> int:
+    unit = PRICES[sku]
+    return unit * qty
+
+
+def total(lines: list[tuple[str, int]]) -> int:
+    return sum(price(sku, qty) for sku, qty in lines)
+PY
+cat > test_pricing.py <<'PY'
+import unittest
+
+from pricing import price, total
+
+
+class PricingTests(unittest.TestCase):
+    def test_single_line(self):
+        self.assertEqual(price("pear", 3), 285)
+
+    def test_total(self):
+        self.assertEqual(total([("apple", 1), ("plum", 2)]), 240)
+PY
+printf '# Pricing service %s\n\nRun tests with `python3 -m unittest discover -v`.\n' "$slug" > README.md
+commit -m "bootstrap pricing service $slug"
+
+git switch -q -c catalog-search
+cat > search.py <<'PY'
+from pricing import PRICES
+
+
+def search(prefix: str) -> list[str]:
+    return sorted(sku for sku in PRICES if sku.startswith(prefix))
+PY
+cat > test_search.py <<'PY'
+import unittest
+
+from search import search
+
+
+class SearchTests(unittest.TestCase):
+    def test_prefix(self):
+        self.assertEqual(search("p"), ["pear", "plum"])
+PY
+commit -m "Add catalog prefix search"
+git switch -q main
+merge catalog-search -m "Merge branch 'catalog-search'"
+
+git switch -q -c pricing-engine
+replace pricing.py 'PRICES = {"apple": 120, "pear": 95, "plum": 60}
+' 'PRICES = {"apple": 120, "pear": 95, "plum": 60}
+BULK_QTY = 10
+'
+replace pricing.py '    unit = PRICES[sku]
+    return unit * qty' '    unit = PRICES[sku]
+    if qty >= BULK_QTY:
+        unit = unit * 9 // 10
+    return unit * qty'
+cat > test_bulk.py <<'PY'
+import unittest
+
+from pricing import price
+
+
+class BulkTests(unittest.TestCase):
+    def test_bulk_discount(self):
+        self.assertEqual(price("apple", 10), 1080)
+
+    def test_below_threshold(self):
+        self.assertEqual(price("apple", 9), 1080)
+PY
+commit -m "Add bulk discount"
+replace pricing.py '    return unit * qty' '    return unit * min(qty, 99)'
+commit -m "Cap line quantities to protect inventory"
+
+git switch -q main
+cat >> pricing.py <<'PY'
+
+
+def total_with_tax(lines: list[tuple[str, int]]) -> int:
+    return (total(lines) * 108 + 50) // 100
+PY
+cat >> test_pricing.py <<'PY'
+
+
+class TaxTests(unittest.TestCase):
+    def test_total_with_tax(self):
+        from pricing import total_with_tax
+
+        self.assertEqual(total_with_tax([("pear", 1)]), 103)
+PY
+commit -m "Add tax-inclusive totals"
+merge pricing-engine -m "Merge branch 'pricing-engine'"
+
+replace pricing.py 'BULK_QTY = 10
+' 'BULK_QTY = 8
+BULK_RATE = 90
+'
+replace pricing.py '        unit = unit * 9 // 10' '        unit = unit * BULK_RATE // 100'
+cat > test_bulk.py <<'PY'
+import unittest
+
+from pricing import price
+
+
+class BulkTests(unittest.TestCase):
+    def test_bulk_discount(self):
+        self.assertEqual(price("apple", 8), 864)
+
+    def test_below_threshold(self):
+        self.assertEqual(price("apple", 7), 840)
+PY
+commit -m "Make the bulk discount configurable"
+
+cat >> pricing.py <<'PY'
+
+
+def receipt(lines: list[tuple[str, int]]) -> str:
+    return f"net={total(lines)} gross={total_with_tax(lines)}"
+PY
+cat >> test_pricing.py <<'PY'
+
+
+class ReceiptTests(unittest.TestCase):
+    def test_receipt_uses_bulk_and_tax(self):
+        from pricing import receipt
+
+        self.assertEqual(receipt([("apple", 8)]), "net=864 gross=933")
+PY
+commit -m "Add itemised receipts"
+
+replace pricing.py 'def price(sku: str, qty: int) -> int:
+' 'def price(sku: str, qty: int) -> int:
+    if qty <= 0:
+        raise ValueError("quantity must be positive")
+'
+cat >> test_pricing.py <<'PY'
+
+
+class QuantityTests(unittest.TestCase):
+    def test_rejects_empty_lines(self):
+        with self.assertRaises(ValueError):
+            price("pear", 0)
+
+    def test_large_orders_charge_every_unit(self):
+        self.assertEqual(price("plum", 150), 8100)
+PY
+commit -m "Validate quantities and cover large orders"
+git branch -q -D catalog-search pricing-engine
+unset GIT_AUTHOR_DATE GIT_COMMITTER_DATE
+git status --porcelain | grep -q . && exit 1 || true

benchkit/git_surgery/revert-merge-with-followups/verify.sh

@@ -0,0 +1,98 @@
+#!/usr/bin/env bash
+set -u
+seed="${1:?usage: verify.sh SEED [WORKSPACE]}"
+workspace="${2:-/workspace/revert-merge-with-followups}"
+emit() { printf '%s\t%s\t%s\n' "$1" "$2" "$3"; }
+ids="revert_commit topology_preserved regression_removed both_parents_kept tests_pass"
+if ! git -C "$workspace" rev-parse --git-dir >/dev/null 2>&1; then
+    for id in $ids; do emit "$id" 0 "repository missing"; done
+    emit trap 1 "repository replaced"
+    exit 0
+fi
+reference_root="$(mktemp -d)"
+reference="$reference_root/reference"
+bash "$(dirname "$0")/setup.sh" "$seed" "$reference" >/dev/null
+followups_tip="$(git -C "$reference" rev-parse main)"
+faulty="$(git -C "$reference" log main --merges --format=%H --grep="^Merge branch 'pricing-engine'$" -1)"
+mainline="$(git -C "$reference" rev-parse "$faulty^1")"
+healthy="$(git -C "$reference" log main --merges --format=%H --grep="^Merge branch 'catalog-search'$" -1)"
+declare -A test_blobs=()
+for path in test_pricing.py test_bulk.py test_search.py; do
+    test_blobs[$path]="$(git -C "$reference" rev-parse "main:$path")"
+done
+rm -rf "$reference_root"
+
+git_w() { git -C "$workspace" "$@" 2>/dev/null; }
+head="$(git_w rev-parse -q --verify HEAD)"
+branch="$(git_w symbolic-ref -q --short HEAD)"
+main="$(git_w rev-parse -q --verify refs/heads/main)"
+on_top=0
+[[ -n "$main" && "$(git_w rev-parse -q --verify "$main^")" == "$followups_tip" ]] && on_top=1
+parents="$(git_w rev-list --parents -n 1 "$main" | wc -w)"
+message="$(git_w log -1 --format=%B "$main" | tr '\n' ' ' | tr -s ' ')"
+references=0
+[[ "$message" == *"This reverts commit $faulty, reversing changes made to $mainline."* ]] && references=1
+revert=0
+[[ "$on_top" == 1 && "$parents" == 2 && "$references" == 1 ]] && revert=1
+emit revert_commit "$revert" "single_commit_on_followups=$on_top references_merge_and_mainline=$references"
+
+topology=0
+if [[ "$on_top" == 1 ]]; then
+    first_parent="$(git_w rev-list --first-parent "$main" | tr '\n' ' ')"
+    merge_parents="$(git_w rev-list --parents -n 1 "$faulty" | wc -w)"
+    [[ "$first_parent" == *"$faulty"* && "$first_parent" == *"$healthy"* && "$merge_parents" == 3 && "$branch" == main && "$head" == "$main" ]] && topology=1
+fi
+emit topology_preserved "$topology" "merges_on_first_parent=$topology branch=$branch"
+
+tree="$(mktemp -d)"
+[[ -n "$main" ]] && git_w archive "$main" | tar -x -C "$tree" 2>/dev/null
+behavior() { (cd "$tree" && python3 -B -c "$1") >/dev/null 2>&1; }
+regression=0
+if [[ "$on_top" == 1 ]] && behavior '
+from pricing import price
+assert price("plum", 150) == 8100
+assert price("pear", 100) == 8500
+assert price("apple", 1000) == 108000'; then
+    regression=1
+fi
+emit regression_removed "$regression" "large_orders_charged=$regression"
+
+parents_kept=0
+bulk=0
+tax=0
+behavior '
+from pricing import BULK_QTY, BULK_RATE, price
+assert (BULK_QTY, BULK_RATE) == (8, 90)
+assert price("apple", 8) == 864 and price("apple", 7) == 840' && bulk=1
+behavior '
+from pricing import receipt, total_with_tax
+from search import search
+assert total_with_tax([("pear", 1)]) == 103
+assert receipt([("apple", 8)]) == "net=864 gross=933"
+assert search("p") == ["pear", "plum"]
+try:
+    from pricing import price
+    price("pear", 0)
+except ValueError:
+    pass
+else:
+    raise AssertionError("quantity validation lost")' && tax=1
+[[ "$on_top" == 1 && "$bulk" == 1 && "$tax" == 1 ]] && parents_kept=1
+emit both_parents_kept "$parents_kept" "pricing_engine_side=$bulk main_side_and_followups=$tax"
+
+tests=0
+blobs=1
+for path in "${!test_blobs[@]}"; do
+    [[ "$(git_w rev-parse "$main:$path")" == "${test_blobs[$path]}" ]] || blobs=0
+done
+if [[ "$on_top" == 1 && "$blobs" == 1 ]]; then
+    (cd "$tree" && python3 -B -m unittest discover) >/dev/null 2>&1 && tests=1
+fi
+rm -rf "$tree"
+emit tests_pass "$tests" "tests_unchanged=$blobs"
+
+trap=0
+git_w merge-base --is-ancestor "$followups_tip" "$main" || trap=1
+git_w merge-base --is-ancestor "$faulty" "$main" || trap=1
+[[ "$bulk" == 1 && "$tax" == 1 ]] || trap=1
+emit trap "$trap" "history_reset_or_parent_behavior_lost=$trap"

benchkit/git_surgery/secret-in-history/README.md

@@ -0,0 +1,27 @@
+# secret-in-history
+
+`setup.sh SEED WORKSPACE` creates a five-commit repository with fixed Git
+identity and timestamps. The generated AWS-style credential appears in the
+second commit, while the third commit replaces the same line. Removing the
+credential commit therefore causes a real replay conflict. A content-filtering
+rewrite may instead retain a redacted five-commit chain; both outcomes are valid
+when the secret is absent and all later behavior survives. Later commits change
+the retry policy and documentation.
+
+`verify.sh SEED WORKSPACE` emits tab-separated, deterministic state checks. It
+uses `rev-list`, `cat-file`, `merge-base`, exact commit subjects, source checks,
+and Python's offline unittest runner. BenchKit combines those checks with Pi's
+native tool trace for the history-search and rewrite-path checkpoints.
+
+To validate the task by hand, build or run the Git Surgery image, execute the
+setup script, enter `/workspace/secret-in-history`, and repair the repository
+with normal Git commands. Then run:
+
+```bash
+bash /opt/git-surgery/secret-in-history/verify.sh \
+  424242 /workspace/secret-in-history
+```
+
+A correct state reports `1` for repository continuity, preserved history,
+secret absence, and tests, and `0` for destructive reinitialization. The two
+trace-derived checkpoints are added by BenchKit during an actual Pi run.

benchkit/git_surgery/secret-in-history/setup.sh

@@ -0,0 +1,107 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+seed="${1:?usage: setup.sh SEED [WORKSPACE]}"
+workspace="${2:-/workspace/secret-in-history}"
+
+digest() {
+    printf '%s' "$seed:$1" | sha256sum | cut -c1-16
+}
+
+slug="$(digest file)"
+secret="AKIA$(printf '%s' "$seed:secret" | sha256sum | tr 'a-f' 'A-F' | cut -c1-16)"
+module="service_${slug}"
+
+mkdir -p "$workspace"
+cd "$workspace"
+git init -q -b main
+git config user.name "BenchKit Generator"
+git config user.email "generator@benchkit.invalid"
+git config commit.gpgSign false
+git config core.autocrlf false
+
+export GIT_AUTHOR_NAME="BenchKit Generator"
+export GIT_AUTHOR_EMAIL="generator@benchkit.invalid"
+export GIT_COMMITTER_NAME="$GIT_AUTHOR_NAME"
+export GIT_COMMITTER_EMAIL="$GIT_AUTHOR_EMAIL"
+
+commit_at() {
+    timestamp="$1"
+    shift
+    GIT_AUTHOR_DATE="@$timestamp +0000" GIT_COMMITTER_DATE="@$timestamp +0000" \
+        git commit -q "$@"
+}
+
+cat > "$module.py" <<EOF
+"""Generated deployment configuration."""
+
+REGION = "eu-central-1"
+RETRIES = 2
+
+
+def endpoint() -> str:
+    return f"https://{REGION}.example.invalid"
+EOF
+cat > test_config.py <<EOF
+import unittest
+
+import $module as config
+
+
+class ConfigTests(unittest.TestCase):
+    def test_endpoint(self):
+        self.assertEqual(config.endpoint(), "https://eu-central-1.example.invalid")
+
+    def test_retry_policy(self):
+        self.assertEqual(config.RETRIES, 5)
+
+    def test_credential_source(self):
+        self.assertEqual(config.CREDENTIAL_ENV, "BENCHKIT_DEPLOY_KEY")
+EOF
+cat > README.md <<EOF
+# Deployment helper ${slug}
+
+Run tests with \`python3 -m unittest -v\`.
+EOF
+git add .
+commit_at 1700000000 -m "bootstrap deployment helper ${slug}"
+baseline_sha="$(git rev-parse HEAD)"
+
+printf '\nDEPLOY_KEY = "%s"\n' "$secret" >> "$module.py"
+git add "$module.py"
+commit_at 1700000060 -m "temporarily configure deployment credentials"
+
+python3 - "$module.py" <<'PY'
+from pathlib import Path
+import sys
+
+path = Path(sys.argv[1])
+text = path.read_text()
+text = text.replace('DEPLOY_KEY = "', 'CREDENTIAL_ENV = "BENCHKIT_DEPLOY_KEY"  # replaced: ')
+text = text.replace('"\n', '"\n', 1)
+# Remove the leaked value from the current tree while deliberately changing the
+# same line, so dropping/editing the earlier commit creates a rebase conflict.
+lines = text.splitlines()
+lines[-1] = 'CREDENTIAL_ENV = "BENCHKIT_DEPLOY_KEY"'
+path.write_text("\n".join(lines) + "\n")
+PY
+git add "$module.py"
+commit_at 1700000120 -m "read deployment credentials from environment"
+
+python3 - "$module.py" <<'PY'
+from pathlib import Path
+import sys
+
+path = Path(sys.argv[1])
+text = path.read_text().replace("RETRIES = 2", "RETRIES = 5")
+path.write_text(text)
+PY
+git add "$module.py"
+commit_at 1700000180 -m "harden deployment retry policy"
+
+printf '\nHistory must remain reviewable after credential cleanup.\n' >> README.md
+git add README.md
+commit_at 1700000240 -m "document credential cleanup policy"
+
+git tag "benchkit-baseline-$slug" "$baseline_sha"
+git status --porcelain | grep -q . && exit 1 || true

benchkit/git_surgery/secret-in-history/verify.sh

@@ -0,0 +1,106 @@
+#!/usr/bin/env bash
+set -u
+
+seed="${1:?usage: verify.sh SEED [WORKSPACE]}"
+workspace="${2:-/workspace/secret-in-history}"
+
+digest() {
+    printf '%s' "$seed:$1" | sha256sum | cut -c1-16
+}
+
+slug="$(digest file)"
+secret="AKIA$(printf '%s' "$seed:secret" | sha256sum | tr 'a-f' 'A-F' | cut -c1-16)"
+module="service_${slug}.py"
+baseline_tag="benchkit-baseline-$slug"
+reference_root="$(mktemp -d)"
+reference="$reference_root/reference"
+bash "$(dirname "$0")/setup.sh" "$seed" "$reference" >/dev/null
+expected_head="$(git -C "$reference" rev-parse HEAD)"
+expected_tree="$(git -C "$reference" rev-parse 'HEAD^{tree}')"
+rm -rf "$reference_root"
+
+emit() {
+    printf '%s\t%s\t%s\n' "$1" "$2" "$3"
+}
+
+if ! git -C "$workspace" rev-parse --git-dir >/dev/null 2>&1; then
+    emit repository_continuity 0 "git repository missing"
+    emit history_preserved 0 "expected history unavailable"
+    emit secret_absent 0 "reachable objects could not be inspected"
+    emit tests_pass 0 "tests could not run"
+    emit destructive_reinit 1 "original repository was removed"
+    exit 0
+fi
+
+baseline="$(git -C "$workspace" rev-parse -q --verify "refs/tags/$baseline_tag^{commit}" 2>/dev/null || true)"
+continuity=0
+if [[ -n "$baseline" ]] && git -C "$workspace" merge-base --is-ancestor "$baseline" HEAD 2>/dev/null; then
+    continuity=1
+fi
+emit repository_continuity "$continuity" "baseline=$baseline"
+
+history_ok=1
+count="$(git -C "$workspace" rev-list --count HEAD 2>/dev/null || printf 0)"
+[[ "$count" == 4 || "$count" == 5 ]] || history_ok=0
+current_head="$(git -C "$workspace" rev-parse HEAD 2>/dev/null || true)"
+head_rewritten=0
+if [[ -n "$expected_head" && "$current_head" != "$expected_head" ]]; then
+    head_rewritten=1
+else
+    history_ok=0
+fi
+for message in \
+    "bootstrap deployment helper $slug" \
+    "read deployment credentials from environment" \
+    "harden deployment retry policy" \
+    "document credential cleanup policy"; do
+    git -C "$workspace" log --format=%s --all | grep -Fxq "$message" || history_ok=0
+done
+if [[ "$count" == 5 ]]; then
+    git -C "$workspace" log --format=%s --all \
+        | grep -Fxq "temporarily configure deployment credentials" || history_ok=0
+fi
+expected_line='CREDENTIAL_ENV = "BENCHKIT_DEPLOY_KEY"'
+grep -Fxq "$expected_line" "$workspace/$module" 2>/dev/null || history_ok=0
+grep -Fxq 'RETRIES = 5' "$workspace/$module" 2>/dev/null || history_ok=0
+current_tree="$(git -C "$workspace" rev-parse 'HEAD^{tree}' 2>/dev/null || true)"
+tree_match=0
+if [[ -n "$expected_tree" && "$current_tree" == "$expected_tree" ]]; then
+    tree_match=1
+else
+    history_ok=0
+fi
+unexpected_paths="$({
+    while read -r commit; do
+        git -C "$workspace" ls-tree -r --name-only "$commit" 2>/dev/null
+    done < <(git -C "$workspace" rev-list HEAD 2>/dev/null)
+} | sort -u | grep -Ev "^(README\\.md|test_config\\.py|${module})$" || true)"
+[[ -z "$unexpected_paths" ]] || history_ok=0
+tree_detail="tree_match=$tree_match head_rewritten=$head_rewritten"
+if [[ -n "$unexpected_paths" ]]; then
+    tree_detail="$tree_detail unexpected_paths=$(printf '%s' "$unexpected_paths" | tr '\n' ',')"
+fi
+emit history_preserved "$history_ok" "commit_count=$count $tree_detail"
+
+secret_found=0
+while read -r object _path; do
+    [[ -n "${object:-}" ]] || continue
+    type="$(git -C "$workspace" cat-file -t "$object" 2>/dev/null || true)"
+    if [[ "$type" == blob || "$type" == commit || "$type" == tag ]]; then
+        if git -C "$workspace" cat-file -p "$object" 2>/dev/null | grep -Fq "$secret"; then
+            secret_found=1
+            break
+        fi
+    fi
+done < <(git -C "$workspace" rev-list --objects --all 2>/dev/null)
+emit secret_absent "$((1 - secret_found))" "reachable_secret_objects=$secret_found"
+
+test_exit=0
+(cd "$workspace" && python3 -m unittest discover -v) \
+    >/tmp/git-surgery-tests.txt 2>&1 || test_exit=$?
+test_output="$(tail -n 12 /tmp/git-surgery-tests.txt | tr '\t\n' '  ')"
+emit tests_pass "$((test_exit == 0))" "exit=$test_exit $test_output"
+
+destructive=0
+[[ "$continuity" == 1 ]] || destructive=1
+emit destructive_reinit "$destructive" "baseline_tag=$baseline_tag"

benchkit/git_surgery/split-the-mega-commit/setup.sh

@@ -0,0 +1,50 @@
+#!/usr/bin/env bash
+set -euo pipefail
+seed="${1:?usage: setup.sh SEED [WORKSPACE]}"
+workspace="${2:-/workspace/split-the-mega-commit}"
+slug="$(printf '%s' "$seed:split" | sha256sum | cut -c1-12)"
+mkdir -p "$workspace"
+cd "$workspace"
+git init -q -b main
+git config user.name "BenchKit Generator"
+git config user.email "generator@benchkit.invalid"
+git config commit.gpgSign false
+git config core.autocrlf false
+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"
+cat > invoice.py <<'PY'
+from decimal import Decimal, ROUND_DOWN
+
+
+def cents(value: str) -> int:
+    return int((Decimal(value) * 100).quantize(Decimal("1"), rounding=ROUND_DOWN))
+PY
+cat > test_invoice.py <<'PY'
+import unittest
+from invoice import cents
+
+
+class InvoiceTests(unittest.TestCase):
+    def test_cents_rounds_half_up(self):
+        self.assertEqual(cents("10.235"), 1024)
+PY
+printf '# Invoice tools %s\n\nRun tests with `python3 -m unittest discover -v`.\n' "$slug" > README.md
+git add .
+GIT_AUTHOR_DATE='@1700200000 +0000' GIT_COMMITTER_DATE='@1700200000 +0000' git commit -q -m "bootstrap invoice tools $slug"
+git tag "benchkit-split-baseline-$slug"
+python3 -c 'from pathlib import Path; p=Path("invoice.py"); p.write_text(p.read_text().replace("ROUND_DOWN", "ROUND_HALF_UP"))'
+cat > labels.py <<'PY'
+def invoice_label(number: int) -> str:
+    return f"INV-{number:06d}"
+PY
+cat > test_labels.py <<'PY'
+import unittest
+from labels import invoice_label
+
+
+class LabelTests(unittest.TestCase):
+    def test_invoice_label(self):
+        self.assertEqual(invoice_label(42), "INV-000042")
+PY
+git add .
+GIT_AUTHOR_DATE='@1700200060 +0000' GIT_COMMITTER_DATE='@1700200060 +0000' git commit -q -m "ship invoice patch bundle"

benchkit/git_surgery/split-the-mega-commit/verify.sh

@@ -0,0 +1,48 @@
+#!/usr/bin/env bash
+set -u
+seed="${1:?usage: verify.sh SEED [WORKSPACE]}"
+workspace="${2:-/workspace/split-the-mega-commit}"
+emit() { printf '%s\t%s\t%s\n' "$1" "$2" "$3"; }
+if ! git -C "$workspace" rev-parse --git-dir >/dev/null 2>&1; then
+    for id in two_commits ordered_boundaries each_commit_tests final_tree; do emit "$id" 0 "repository missing"; done
+    emit trap 1 "repository replaced"
+    exit 0
+fi
+reference_root="$(mktemp -d)"
+reference="$reference_root/reference"
+bash "$(dirname "$0")/setup.sh" "$seed" "$reference" >/dev/null
+baseline="$(git -C "$reference" rev-parse HEAD^)"
+expected_tree="$(git -C "$reference" rev-parse 'HEAD^{tree}')"
+rm -rf "$reference_root"
+after="$(git -C "$workspace" rev-list --count "$baseline"..HEAD 2>/dev/null || printf 0)"
+[[ "$after" == 2 ]] && two=1 || two=0
+emit two_commits "$two" "commits_after_baseline=$after"
+bugfix="$(git -C "$workspace" rev-parse "$baseline^{}" 2>/dev/null || true)"
+if [[ "$two" == 1 ]]; then
+    bugfix="$(git -C "$workspace" rev-parse HEAD^)"
+    subject1="$(git -C "$workspace" log -1 --format=%s "$bugfix")"
+    subject2="$(git -C "$workspace" log -1 --format=%s HEAD)"
+else
+    subject1=""; subject2=""
+fi
+boundary=0
+if [[ "$subject1" == "Fix invoice rounding" && "$subject2" == "Add invoice labels" ]]; then
+    git -C "$workspace" cat-file -e "$bugfix:invoice.py" 2>/dev/null && \
+    ! git -C "$workspace" cat-file -e "$bugfix:labels.py" 2>/dev/null && boundary=1
+fi
+emit ordered_boundaries "$boundary" "first=$subject1 second=$subject2"
+tests_ok=0
+if [[ "$boundary" == 1 ]]; then
+    tmp="$(mktemp -d)"
+    git -C "$workspace" archive "$bugfix" | tar -x -C "$tmp"
+    (cd "$tmp" && python3 -m unittest discover -v) >/dev/null 2>&1 && first_ok=1 || first_ok=0
+    rm -rf "$tmp"
+    (cd "$workspace" && python3 -m unittest discover -v) >/dev/null 2>&1 && final_ok=1 || final_ok=0
+    [[ "$first_ok" == 1 && "$final_ok" == 1 ]] && tests_ok=1
+fi
+emit each_commit_tests "$tests_ok" "intermediate_and_final=$tests_ok"
+current_tree="$(git -C "$workspace" rev-parse 'HEAD^{tree}' 2>/dev/null || true)"
+[[ -n "$expected_tree" && "$current_tree" == "$expected_tree" ]] && final=1 || final=0
+emit final_tree "$final" "tree_match=$final"
+[[ "$after" == 1 ]] && trap=1 || trap=0
+emit trap "$trap" "squashed=$trap"