840d7509e1
Flash is the default model; --m3 falls back to M3, --effort overrides. Defaults: max (review-minimax, review-loop, mm-worktree-task), medium (explore), low (ci-triage). ask-minimax-run applies MINIMAX_EFFORT to Flash models when --effort is not passed. Bump stale fixture date.
158 lines
6.6 KiB
Bash
Executable File
158 lines
6.6 KiB
Bash
Executable File
#!/usr/bin/env bash
|
|
# test-review-minimax-run.sh — Verify review-minimax-run against its spec.
|
|
#
|
|
# Strategy: a throwaway git repo with a scratch branch provides real diffs;
|
|
# REVIEW_MINIMAX_FAKE_JSON drives the sub-session paths so exits 0/2/3 and
|
|
# the result gate are all exercised offline. No live MiniMax calls.
|
|
|
|
set -uo pipefail
|
|
|
|
SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
|
|
REPO_ROOT="$(cd "$SCRIPT_DIR/.." && pwd)"
|
|
SCRIPT="$REPO_ROOT/scripts/review-minimax-run"
|
|
|
|
GREEN='\033[0;32m'
|
|
RED='\033[0;31m'
|
|
NC='\033[0m'
|
|
PASS=0
|
|
FAIL=0
|
|
|
|
assert_contains() {
|
|
local label=$1 needle=$2 haystack=$3
|
|
if [[ "$haystack" == *"$needle"* ]]; then
|
|
printf " ${GREEN}PASS${NC} %s\n" "$label"; PASS=$((PASS + 1))
|
|
else
|
|
printf " ${RED}FAIL${NC} %s\n" "$label"
|
|
printf " expected to contain: %s\n" "$needle"; FAIL=$((FAIL + 1))
|
|
fi
|
|
}
|
|
|
|
assert_not_contains() {
|
|
local label=$1 needle=$2 haystack=$3
|
|
if [[ "$haystack" != *"$needle"* ]]; then
|
|
printf " ${GREEN}PASS${NC} %s\n" "$label"; PASS=$((PASS + 1))
|
|
else
|
|
printf " ${RED}FAIL${NC} %s\n" "$label"
|
|
printf " expected NOT to contain: %s\n" "$needle"; FAIL=$((FAIL + 1))
|
|
fi
|
|
}
|
|
|
|
assert_eq() {
|
|
local label=$1 expected=$2 actual=$3
|
|
if [[ "$expected" == "$actual" ]]; then
|
|
printf " ${GREEN}PASS${NC} %s\n" "$label"; PASS=$((PASS + 1))
|
|
else
|
|
printf " ${RED}FAIL${NC} %s\n" "$label"
|
|
printf " expected: %s actual: %s\n" "$expected" "$actual"; FAIL=$((FAIL + 1))
|
|
fi
|
|
}
|
|
|
|
# --- fixture repo -------------------------------------------------------------
|
|
TMP=$(mktemp -d /tmp/test-review-minimax.XXXXXX)
|
|
trap 'rm -rf "$TMP"' EXIT
|
|
REPO="$TMP/repo"
|
|
mkdir -p "$REPO"
|
|
git -C "$REPO" -c init.defaultBranch=main init -q
|
|
git -C "$REPO" -c user.email=t@t -c user.name=t commit -q --allow-empty -m init
|
|
echo "safe line" > "$REPO/app.sh"
|
|
git -C "$REPO" add -A
|
|
git -C "$REPO" -c user.email=t@t -c user.name=t commit -qm base
|
|
git -C "$REPO" checkout -qb feature
|
|
echo 'eval "$user_input"' >> "$REPO/app.sh"
|
|
git -C "$REPO" add -A
|
|
git -C "$REPO" -c user.email=t@t -c user.name=t commit -qm change
|
|
|
|
FAKE_OK_JSON="$TMP/ok.json"
|
|
echo '{"is_error":false,"result":"done","usage":{"input_tokens":1,"output_tokens":1}}' > "$FAKE_OK_JSON"
|
|
|
|
echo "== review-minimax-run: --help =="
|
|
HELP_OUT="$("$SCRIPT" --help 2>&1)"
|
|
assert_contains "help shows usage" "--worktree" "$HELP_OUT"
|
|
|
|
echo "== usage errors (exit 1) =="
|
|
"$SCRIPT" >/dev/null 2>&1; assert_eq "no args exits 1" 1 $?
|
|
"$SCRIPT" --worktree /nonexistent >/dev/null 2>&1; assert_eq "missing worktree exits 1" 1 $?
|
|
"$SCRIPT" --worktree "$TMP" >/dev/null 2>&1; assert_eq "non-git worktree exits 1" 1 $?
|
|
|
|
echo "== dryrun =="
|
|
DRY_OUT="$("$SCRIPT" --worktree "$REPO" --dryrun 2>&1)"
|
|
assert_eq "dryrun exits 0" 0 $?
|
|
assert_contains "dryrun shows branch" "branch: feature" "$DRY_OUT"
|
|
assert_contains "dryrun defaults to Flash" "MiniMax-M3.1-Flash-Preview" "$DRY_OUT"
|
|
assert_contains "dryrun default effort max" "--effort max" "$DRY_OUT"
|
|
M3_OUT="$("$SCRIPT" --worktree "$REPO" --m3 --dryrun 2>&1)"
|
|
assert_eq "m3 dryrun exits 0" 0 $?
|
|
assert_contains "m3 dryrun shows M3 model" "MiniMax-M3[1m]" "$M3_OUT"
|
|
assert_not_contains "m3 dryrun sends no effort" "--effort" "$M3_OUT"
|
|
EF_OUT="$("$SCRIPT" --worktree "$REPO" --effort high --dryrun 2>&1)"
|
|
assert_contains "explicit effort overrides default" "--effort high" "$EF_OUT"
|
|
"$SCRIPT" --worktree "$REPO" --effort bogus --dryrun >/dev/null 2>&1; assert_eq "bad effort exits 1" 1 $?
|
|
assert_contains "dryrun shows deny list" "Bash,WebFetch,WebSearch" "$DRY_OUT"
|
|
assert_contains "dryrun prompt has findings format" "NO FINDINGS" "$DRY_OUT"
|
|
assert_contains "dryrun shows checklist" "shell=True" "$DRY_OUT"
|
|
|
|
echo "== profile =="
|
|
PROF_OUT="$("$SCRIPT" --worktree "$REPO" --profile agent-runtimes --dryrun 2>&1)"
|
|
assert_contains "agent-runtimes profile adds test-bug checklist" "/v1/ prefix" "$PROF_OUT"
|
|
NOPROF_OUT="$("$SCRIPT" --worktree "$REPO" --dryrun 2>&1)"
|
|
if [[ "$NOPROF_OUT" != *"/v1/ prefix"* ]]; then
|
|
printf " ${GREEN}PASS${NC} no profile omits repo-specific checklist\n"; PASS=$((PASS + 1))
|
|
else
|
|
printf " ${RED}FAIL${NC} no profile omits repo-specific checklist\n"; FAIL=$((FAIL + 1))
|
|
fi
|
|
|
|
echo "== diff gates (exit 1) =="
|
|
"$SCRIPT" --worktree "$REPO" --max-diff-lines 1 --dryrun >/dev/null 2>&1
|
|
assert_eq "oversize diff exits 1" 1 $?
|
|
git -C "$REPO" checkout -q main
|
|
git -C "$REPO" checkout -qb empty-branch
|
|
"$SCRIPT" --worktree "$REPO" --dryrun >/dev/null 2>&1
|
|
assert_eq "empty diff exits 1" 1 $?
|
|
git -C "$REPO" checkout -q feature
|
|
|
|
echo "== credential-shaped diff (exit 1) =="
|
|
git -C "$REPO" checkout -qb cred-branch main
|
|
echo 'url = https://x-access-token:tok@example.com/r.git' > "$REPO/cfg"
|
|
git -C "$REPO" add -A
|
|
git -C "$REPO" -c user.email=t@t -c user.name=t commit -qm cred
|
|
CRED_OUT="$("$SCRIPT" --worktree "$REPO" --dryrun 2>&1)"
|
|
assert_eq "credential-shaped diff exits 1" 1 $?
|
|
assert_contains "credential message" "credential-shaped" "$CRED_OUT"
|
|
git -C "$REPO" checkout -q feature
|
|
|
|
echo "== fake seam: no findings file -> exit 3 after retry =="
|
|
OUT3="$TMP/findings3.md"
|
|
REVIEW_MINIMAX_FAKE_JSON="$FAKE_OK_JSON" "$SCRIPT" --worktree "$REPO" --out "$OUT3" >/dev/null 2>&1
|
|
assert_eq "missing findings file exits 3" 3 $?
|
|
|
|
echo "== fake seam: uncited findings -> exit 3 =="
|
|
FAKE_BAD="$TMP/fake-bad-findings.md"
|
|
printf '### [NIT] vague worry\n- Location: made/up/file.py:12\n- Confidence: suspected\n' > "$FAKE_BAD"
|
|
REVIEW_MINIMAX_FAKE_JSON="$FAKE_OK_JSON" REVIEW_MINIMAX_FAKE_FINDINGS="$FAKE_BAD" \
|
|
"$SCRIPT" --worktree "$REPO" --out "$TMP/findings4.md" >/dev/null 2>&1
|
|
assert_eq "uncited findings exits 3" 3 $?
|
|
|
|
echo "== fake seam: cited findings -> exit 0 =="
|
|
FAKE_GOOD="$TMP/fake-good-findings.md"
|
|
printf '### [BLOCKER] eval on user input\n- Location: app.sh:2\n- Category: Injection\n- Confidence: confirmed\n' > "$FAKE_GOOD"
|
|
REVIEW_MINIMAX_FAKE_JSON="$FAKE_OK_JSON" REVIEW_MINIMAX_FAKE_FINDINGS="$FAKE_GOOD" \
|
|
"$SCRIPT" --worktree "$REPO" --out "$TMP/findings0.md" >/dev/null 2>&1
|
|
assert_eq "cited findings exits 0" 0 $?
|
|
|
|
echo "== fake seam: NO FINDINGS -> exit 0 =="
|
|
FAKE_CLEAN="$TMP/fake-clean.md"
|
|
echo 'NO FINDINGS' > "$FAKE_CLEAN"
|
|
REVIEW_MINIMAX_FAKE_JSON="$FAKE_OK_JSON" REVIEW_MINIMAX_FAKE_FINDINGS="$FAKE_CLEAN" \
|
|
"$SCRIPT" --worktree "$REPO" --out "$TMP/findings-clean.md" >/dev/null 2>&1
|
|
assert_eq "NO FINDINGS exits 0" 0 $?
|
|
|
|
echo "== fake seam: is_error envelope -> exit 2 =="
|
|
FAKE_ERR_JSON="$TMP/err.json"
|
|
echo '{"is_error":true,"result":"boom"}' > "$FAKE_ERR_JSON"
|
|
REVIEW_MINIMAX_FAKE_JSON="$FAKE_ERR_JSON" "$SCRIPT" --worktree "$REPO" --out "$TMP/f5.md" >/dev/null 2>&1
|
|
assert_eq "is_error envelope exits 2" 2 $?
|
|
|
|
echo ""
|
|
echo "review-minimax-run: $PASS passed, $FAIL failed"
|
|
[[ "$FAIL" -eq 0 ]]
|