obikmer/scripts/smoke_test_index.sh

#!/usr/bin/env bash
# smoke_test_index.sh — end-to-end smoke test of `obikmer index` + `obikmer query`
#
# The unit/integration test suite never exercises obikmer's real,
# file-driven CLI path (every test builds indexes by calling library APIs
# directly — see DevDocMD/implementation/partition_layer_cache.md, "(3)
# done"/"(6) done": a green `cargo test --workspace` has twice missed a
# real bug in this exact path). This script is the fast, repeatable
# substitute for hand-rolling that check each time.
#
# What it does:
#   1. builds `obikmer` (debug, via `cargo run`)
#   2. generates a small deterministic random FASTA
#   3. runs `obikmer index` on it
#   4. picks a real k-mer from the source sequence, avoiding low-complexity
#      substrings (a homopolymer run tripped up an earlier manual run of
#      this check — it gets rejected by query's own entropy filter, which
#      looks like a bug but isn't one)
#   5. runs `obikmer query` and checks the k-mer round-trips
#   6. prints total-kmers-indexed and a clear PASS/FAIL, exit code matches
#
# Usage:
#   scripts/smoke_test_index.sh [-k KMER_SIZE] [-m MINIMIZER_SIZE] [-p PARTITIONS] [--keep]
#
# --keep leaves the temp directory in place (path printed) instead of
# deleting it on exit, for manual inspection of a failure.

set -euo pipefail

K=11
M=5
PARTITIONS=4
KEEP=0

while [ $# -gt 0 ]; do
    case "$1" in
        -k) K="$2"; shift 2 ;;
        -m) M="$2"; shift 2 ;;
        -p) PARTITIONS="$2"; shift 2 ;;
        --keep) KEEP=1; shift ;;
        *) echo "unknown argument: $1" >&2; exit 2 ;;
    esac
done

REPO_ROOT="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
WORK="$(mktemp -d -t obikmer_smoke.XXXXXX)"

cleanup() {
    if [ "$KEEP" -eq 1 ]; then
        echo "kept: $WORK"
    else
        rm -rf "$WORK"
    fi
}
trap cleanup EXIT

fail() {
    echo "FAIL: $1" >&2
    exit 1
}

# ── 1. generate a small deterministic random FASTA ─────────────────────────
python3 - "$WORK/test.fasta" "$K" <<'EOF'
import random, sys
path, k = sys.argv[1], int(sys.argv[2])
random.seed(1234)
bases = "ACGT"
with open(path, "w") as f:
    for i in range(3):
        seq = "".join(random.choice(bases) for _ in range(300))
        f.write(f">seq{i}\n{seq}\n")
EOF

# ── 2. build + run index ────────────────────────────────────────────────────
cd "$REPO_ROOT/src"
INDEX_LOG="$WORK/index.log"
if ! cargo run -q -p obikmer --bin obikmer -- \
        index -k "$K" -m "$M" --theta 0 -p "$PARTITIONS" \
        -o "$WORK/out.idx" "$WORK/test.fasta" > "$INDEX_LOG" 2>&1
then
    cat "$INDEX_LOG" >&2
    fail "obikmer index exited non-zero"
fi

N_KMERS="$(grep -o '[0-9]* total kmers indexed' "$INDEX_LOG" | grep -o '^[0-9]*' || true)"
[ -n "$N_KMERS" ] || { cat "$INDEX_LOG" >&2; fail "could not find 'N total kmers indexed' in index log"; }
[ "$N_KMERS" -gt 0 ] || fail "index reports 0 kmers indexed"

# ── 3+4. try candidate k-mers spread across the source sequence until one
#         round-trips. `query` applies its own entropy filter to the query
#         sequence (not exposed as a CLI flag, undocumented threshold) — a
#         window that fails it reports kmer_count:0 even though the k-mer
#         is genuinely in the index (not a bug, just a bad test fixture).
#         Rather than reverse-engineer the filter's formula, just ask the
#         real binary and move to the next candidate on a miss.
readarray -t CANDIDATES < <(python3 - "$WORK/test.fasta" "$K" <<'EOF'
import sys
path, k = sys.argv[1], int(sys.argv[2])
with open(path) as f:
    seq = "".join(l.strip() for l in f if not l.startswith(">"))
for start in range(0, len(seq) - k, 17):
    print(seq[start:start+k])
EOF
)
[ "${#CANDIDATES[@]}" -gt 0 ] || fail "could not extract any candidate k-mer from the source FASTA"

QUERY_LOG="$WORK/query.log"
FOUND=0
for QUERY_KMER in "${CANDIDATES[@]}"; do
    printf ">q1\n%s\n" "$QUERY_KMER" > "$WORK/query.fasta"
    if ! cargo run -q -p obikmer --bin obikmer -- \
            query "$WORK/out.idx" "$WORK/query.fasta" > "$QUERY_LOG" 2>&1
    then
        cat "$QUERY_LOG" >&2
        fail "obikmer query exited non-zero"
    fi
    if grep -q '"kmer_count":1' "$QUERY_LOG"; then
        FOUND=1
        break
    fi
done

if [ "$FOUND" -ne 1 ]; then
    cat "$QUERY_LOG" >&2
    fail "no candidate k-mer round-tripped (tried ${#CANDIDATES[@]}) — likely a real regression, not a low-complexity fixture"
fi

echo "PASS: index+query round-trip OK — $N_KMERS kmers indexed, query k-mer '$QUERY_KMER' found"