codegen-propose

#!/usr/bin/env bash
# codegen-propose — turn codegen-analyze turn-waste clusters into
# structured, human-gated proposed-change records via one codegen-call per
# qualifying cluster.
#
# Usage: codegen-propose --since YYYY-MM-DD [--max N] [--min-wasted-turns T]
#                        [--model M]
#
# Read-only against the repo except for its own output sink:
#   codegen/analysis-proposals/<from>_<to>.md (ephemeral, gitignored).
# Never writes source files. Never proposes changes itself — every cluster's
# proposal is drafted by an LLM call and requires human review before any
# fix is applied.
#
# Exit: 0 = batch completed (even with some errored/declined clusters).
#       2 = usage error.
set -euo pipefail

# Resolve the codegen root so both direct invocations (from repo) and
# installed copies (~/.local/bin/) can locate harnesses/ and analysis/.
CODEGEN_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd -P)"
if [[ ! -d "$CODEGEN_DIR/harnesses" ]] && [[ -n "${OCG_CODEGEN_DIR:-}" ]]; then
    CODEGEN_DIR="$OCG_CODEGEN_DIR"
fi

DEFAULT_MODEL="opus"
DEFAULT_EFFORT="high"
DEFAULT_MAX=5
DEFAULT_MIN_WASTED_TURNS=4

SINCE=""
MAX="$DEFAULT_MAX"
MIN_WASTED_TURNS="$DEFAULT_MIN_WASTED_TURNS"
MODEL="$DEFAULT_MODEL"
EFFORT="$DEFAULT_EFFORT"

usage() {
    printf 'Usage: codegen-propose --since YYYY-MM-DD [--max N] [--min-wasted-turns T] [--model M]\n' >&2
    exit 2
}

while [[ $# -gt 0 ]]; do
    case "$1" in
    --since=*)
        SINCE="${1#--since=}"
        shift
        ;;
    --since)
        SINCE="$2"
        shift 2
        ;;
    --max=*)
        MAX="${1#--max=}"
        shift
        ;;
    --max)
        MAX="$2"
        shift 2
        ;;
    --min-wasted-turns=*)
        MIN_WASTED_TURNS="${1#--min-wasted-turns=}"
        shift
        ;;
    --min-wasted-turns)
        MIN_WASTED_TURNS="$2"
        shift 2
        ;;
    --model=*)
        MODEL="${1#--model=}"
        shift
        ;;
    --model)
        MODEL="$2"
        shift 2
        ;;
    *)
        printf 'codegen-propose: unknown argument %q\n' "$1" >&2
        usage
        ;;
    esac
done

if [[ -z "$SINCE" ]]; then
    printf 'codegen-propose: --since YYYY-MM-DD is required\n' >&2
    usage
fi

FROM_STAMP="$(printf '%s' "$SINCE" | tr -d '-')"
TO_STAMP="$(date -u +%Y%m%d)"

PROPOSALS_DIR="$CODEGEN_DIR/codegen/analysis-proposals"
PROPOSALS_FILE="$PROPOSALS_DIR/${FROM_STAMP}_${TO_STAMP}.md"
mkdir -p "$PROPOSALS_DIR"

SYSTEM_PROMPT_FILE="$CODEGEN_DIR/harnesses/claude/propose-system-prompt.md"
SCHEMA_FILE="$CODEGEN_DIR/harnesses/claude/proposal.schema.json"

# Prefer codegen-analyze on PATH (allows test stubbing); fall back to direct
# module invocation.
if command -v codegen-analyze >/dev/null 2>&1; then
    ANALYZE_CMD=(codegen-analyze)
else
    ANALYZE_CMD=(env PYTHONPATH="$CODEGEN_DIR" python3 -m analysis)
fi

SELECT_CMD=(env PYTHONPATH="$CODEGEN_DIR" python3 -m analysis.proposer.select
    --min-wasted-turns "$MIN_WASTED_TURNS" --max "$MAX")

SELECTED_JSONL="$("${ANALYZE_CMD[@]}" --json --since "$SINCE" | "${SELECT_CMD[@]}")"

if [[ -z "$SELECTED_JSONL" ]]; then
    NOTE="no proposals above threshold ${MIN_WASTED_TURNS} wasted turns (since ${SINCE})"
    printf '%s\n' "$NOTE" >"$PROPOSALS_FILE"
    printf '%s\n' "$NOTE"
    exit 0
fi

PROPOSED_COUNT=0
DECLINED_COUNT=0
ERRORED_COUNT=0
RECORDS_TMP="$(mktemp)"
trap 'rm -f "$RECORDS_TMP"' EXIT

while IFS= read -r cluster_line; do
    [[ -z "$cluster_line" ]] && continue

    counter="$(printf '%s' "$cluster_line" | jq -r '.counter')"
    pattern_key="$(printf '%s' "$cluster_line" | jq -r '.pattern_key')"
    wasted_turns="$(printf '%s' "$cluster_line" | jq -r '.wasted_turns')"
    sessions="$(printf '%s' "$cluster_line" | jq -r '.sessions')"
    top_evidence="$(printf '%s' "$cluster_line" | jq -r '.top_evidence')"

    prompt="Cluster to draft a proposed-change record for:
counter: ${counter}
pattern_key: ${pattern_key}
wasted_turns: ${wasted_turns}
sessions: ${sessions}
top_evidence: ${top_evidence}"

    set +e
    call_out=$(codegen-call --harness=claude_code --model="$MODEL" --effort="$EFFORT" \
        --system-prompt "@$SYSTEM_PROMPT_FILE" \
        --json-schema "@$SCHEMA_FILE" \
        -- "$prompt" 2>/dev/null)
    call_rc=$?
    set -e

    if [[ $call_rc -ne 0 ]]; then
        ERRORED_COUNT=$((ERRORED_COUNT + 1))
        {
            printf '## Proposals\n\n'
            printf '### %s: %s (ERRORED)\n\n' "$counter" "$pattern_key"
            printf -- '- wasted_turns: %s\n' "$wasted_turns"
            printf -- '- error: codegen-call exited %s\n\n' "$call_rc"
        } >>"$RECORDS_TMP"
        continue
    fi

    status="$(printf '%s' "$call_out" | jq -r '.result.status // empty')"
    value="$(printf '%s' "$call_out" | jq -c '.result.value // empty')"

    if [[ "$status" != "success" || -z "$value" || "$value" == "null" ]]; then
        ERRORED_COUNT=$((ERRORED_COUNT + 1))
        reason="$(printf '%s' "$call_out" | jq -r '.result.reason // "no reason given"')"
        {
            printf '## Proposals\n\n'
            printf '### %s: %s (ERRORED)\n\n' "$counter" "$pattern_key"
            printf -- '- wasted_turns: %s\n' "$wasted_turns"
            printf -- '- error: result.status=%s, reason=%s\n\n' "$status" "$reason"
        } >>"$RECORDS_TMP"
        continue
    fi

    confidence="$(printf '%s' "$value" | jq -r '.confidence // "low"')"
    change="$(printf '%s' "$value" | jq -c '.change // empty')"
    target_file="$(printf '%s' "$value" | jq -r '.target_file // "unknown"')"
    anchor="$(printf '%s' "$value" | jq -r '.anchor // ""')"
    rationale="$(printf '%s' "$value" | jq -r '.rationale // ""')"

    if [[ -z "$change" || "$change" == "null" ]]; then
        DECLINED_COUNT=$((DECLINED_COUNT + 1))
        {
            printf '## Proposals\n\n'
            printf '### %s: %s (DECLINED-LOW-CONFIDENCE)\n\n' "$counter" "$pattern_key"
            printf -- '- wasted_turns: %s\n' "$wasted_turns"
            printf -- '- confidence: %s\n' "$confidence"
            printf -- '- rationale: %s\n\n' "$rationale"
        } >>"$RECORDS_TMP"
        continue
    fi

    PROPOSED_COUNT=$((PROPOSED_COUNT + 1))
    {
        printf '## Proposals\n\n'
        printf '### %s: %s (PROPOSED)\n\n' "$counter" "$pattern_key"
        printf -- '- wasted_turns: %s\n' "$wasted_turns"
        printf -- '- confidence: %s\n' "$confidence"
        printf -- '- target_file: %s\n' "$target_file"
        printf -- '- anchor: %s\n' "$anchor"
        printf -- '- rationale: %s\n' "$rationale"
        printf -- '- change:\n\n```json\n%s\n```\n\n' "$change"
    } >>"$RECORDS_TMP"
done <<<"$SELECTED_JSONL"

{
    printf 'Summary: %s proposed / %s declined-low-confidence / %s errored\n\n' \
        "$PROPOSED_COUNT" "$DECLINED_COUNT" "$ERRORED_COUNT"
    cat "$RECORDS_TMP"
} >"$PROPOSALS_FILE"

printf 'Summary: %s proposed / %s declined-low-confidence / %s errored\n' \
    "$PROPOSED_COUNT" "$DECLINED_COUNT" "$ERRORED_COUNT"
printf 'Proposals written: %s\n' "$PROPOSALS_FILE"
exit 0