#!/bin/bash
# expand-categories.sh — Expand skill categories into full manifest
#
# Uses claude --print to expand each category in categories.json
# into 15-25 individual skill titles, producing manifest.json
#
# Usage:
#   ./expand-categories.sh                    # Expand all categories
#   ./expand-categories.sh azure-ai-services  # Expand single category
#   ./expand-categories.sh --wave 1           # Expand wave 1 only
#
# Prerequisites:
#   - claude CLI installed and authenticated
#   - jq installed
#   - categories.json in same directory

set -euo pipefail

SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
PLUGIN_DIR="$(cd "$SCRIPT_DIR/../.." && pwd)"
CATEGORIES_FILE="$SCRIPT_DIR/categories.json"
MANIFEST_FILE="$SCRIPT_DIR/manifest.json"
LOG_DIR="$SCRIPT_DIR/logs"

# Model for expansion (haiku is sufficient for generating titles)
MODEL="${MODEL:-haiku}"

# Colors
RED='\033[0;31m'
GREEN='\033[0;32m'
YELLOW='\033[1;33m'
BLUE='\033[0;34m'
NC='\033[0m'

log() { echo -e "${BLUE}[expand]${NC} $1" >&2; }
success() { echo -e "${GREEN}[expand]${NC} $1" >&2; }
warn() { echo -e "${YELLOW}[expand]${NC} $1" >&2; }
error() { echo -e "${RED}[expand]${NC} $1" >&2; }

# Check prerequisites
check_prereqs() {
    if ! command -v claude &>/dev/null; then
        error "claude CLI not found. Install: npm install -g @anthropic-ai/claude-code"
        exit 1
    fi
    if ! command -v jq &>/dev/null; then
        error "jq not found. Install: brew install jq"
        exit 1
    fi
    if [[ ! -f "$CATEGORIES_FILE" ]]; then
        error "categories.json not found at $CATEGORIES_FILE"
        exit 1
    fi
}

# Get list of existing reference files for context
get_existing_refs() {
    local category_dir="$1"
    local refs_dir="$PLUGIN_DIR/skills/ms-ai-engineering/references"

    # List all existing reference files
    find "$refs_dir" -name "*.md" -type f 2>/dev/null | while read -r f; do
        basename "$f" .md
    done | sort | tr '\n' ', '
}

# Expand a single category
expand_category() {
    local category_key="$1"

    local name description estimated examples existing_overlap notes
    name=$(jq -r ".categories[\"$category_key\"].name" "$CATEGORIES_FILE")
    description=$(jq -r ".categories[\"$category_key\"].description" "$CATEGORIES_FILE")
    estimated=$(jq -r ".categories[\"$category_key\"].estimated_skills" "$CATEGORIES_FILE")
    examples=$(jq -r ".categories[\"$category_key\"].examples | join(\", \")" "$CATEGORIES_FILE")
    existing_overlap=$(jq -r ".categories[\"$category_key\"].existing_overlap | join(\", \")" "$CATEGORIES_FILE")
    notes=$(jq -r ".categories[\"$category_key\"].notes" "$CATEGORIES_FILE")

    local existing_refs
    existing_refs=$(get_existing_refs "$category_key")

    log "Expanding: $name ($estimated skills estimated)"

    local prompt
    prompt="Du er en Microsoft AI Solution Architect som planlegger en kunnskapsbase.

Kategorien **${name}** trenger individuelle kunnskapsfiler (skills).

## Kategori-beskrivelse
${description}

## Eksisterende filer i kunnskapsbasen (unngå duplikering)
${existing_refs}

## Eksisterende overlapp å ta hensyn til
${existing_overlap}

## Eksempel-titler (for inspirasjon, ikke begrens deg til disse)
${examples}

## Notater
${notes}

## Oppgave

Generer en JSON-array med NØYAKTIG ${estimated} skills for denne kategorien.

Hver skill skal ha:
- \`id\`: kebab-case filnavn (uten .md)
- \`title\`: Engelsk tittel (kortfattet, beskrivende)
- \`description\`: 1-2 setninger på norsk om hva filen dekker
- \`subtopics\`: 3-5 viktige undertemaer som array

Regler:
1. Ikke dupliser emner som allerede finnes i eksisterende filer
2. Sørg for bred dekning uten overlapp mellom skills
3. Titler skal være spesifikke (\"Azure AI Vision OCR and Document Processing\", ikke bare \"Vision\")
4. Prioriter mest nyttige emner for enterprise AI-arkitekter i norsk offentlig sektor
5. Returner KUN valid JSON-array, ingen annen tekst

Eksempel-format:
[
  {
    \"id\": \"example-skill-name\",
    \"title\": \"Example Skill - Full Descriptive Title\",
    \"description\": \"Beskrivelse av hva denne kunnskapsfilen dekker.\",
    \"subtopics\": [\"subtopic-1\", \"subtopic-2\", \"subtopic-3\"]
  }
]"

    local output
    output=$(claude --print --model "$MODEL" "$prompt" 2>"$LOG_DIR/expand-${category_key}.err")

    # Extract JSON array from response (handles markdown code blocks, plain JSON, etc.)
    local json_output
    json_output=$(python3 -c "
import sys, json, re
text = sys.stdin.read()
# Try to find JSON array in code blocks first
m = re.search(r'\`\`\`(?:json)?\s*(\[[\s\S]*?\])\s*\`\`\`', text)
if m:
    arr = json.loads(m.group(1))
    print(json.dumps(arr))
    sys.exit(0)
# Try to find bare JSON array
m = re.search(r'(\[[\s\S]*\])', text)
if m:
    try:
        arr = json.loads(m.group(1))
        print(json.dumps(arr))
        sys.exit(0)
    except: pass
# Nothing found
sys.exit(1)
" <<< "$output" 2>/dev/null)

    # Validate JSON
    if ! echo "$json_output" | jq . &>/dev/null; then
        error "Invalid JSON for $category_key. Raw output saved to $LOG_DIR/expand-${category_key}.raw"
        echo "$output" > "$LOG_DIR/expand-${category_key}.raw"
        return 1
    fi

    local count
    count=$(echo "$json_output" | jq 'length')
    success "$name: $count skills generated"

    # Return the JSON
    echo "$json_output"
}

# Build or update manifest
build_manifest() {
    local categories_to_expand=("$@")

    # Initialize manifest if it doesn't exist
    if [[ ! -f "$MANIFEST_FILE" ]]; then
        echo '{"version":"1.0","created":"'"$(date +%Y-%m-%d)"'","categories":{}}' | jq . > "$MANIFEST_FILE"
    fi

    local total=0
    local failed=0

    for category_key in "${categories_to_expand[@]}"; do
        local skills_json
        if skills_json=$(expand_category "$category_key"); then
            # Add to manifest
            local dir
            dir=$(jq -r ".categories[\"$category_key\"].dir" "$CATEGORIES_FILE")
            local name
            name=$(jq -r ".categories[\"$category_key\"].name" "$CATEGORIES_FILE")
            local priority
            priority=$(jq -r ".categories[\"$category_key\"].priority" "$CATEGORIES_FILE")

            local category_obj
            category_obj=$(jq -n \
                --arg name "$name" \
                --arg dir "$dir" \
                --arg priority "$priority" \
                --argjson skills "$skills_json" \
                '{name: $name, dir: $dir, priority: $priority, skills: $skills}')

            # Merge into manifest
            jq --arg key "$category_key" --argjson cat "$category_obj" \
                '.categories[$key] = $cat' "$MANIFEST_FILE" > "$MANIFEST_FILE.tmp" \
                && mv "$MANIFEST_FILE.tmp" "$MANIFEST_FILE"

            local count
            count=$(echo "$skills_json" | jq 'length')
            total=$((total + count))
        else
            failed=$((failed + 1))
        fi

        # Rate limiting: pause between API calls
        sleep 2
    done

    echo "" >&2
    log "═══════════════════════════════════════"
    success "Total skills in manifest: $total"
    [[ $failed -gt 0 ]] && error "Failed categories: $failed"
    log "Manifest: $MANIFEST_FILE"
    log "═══════════════════════════════════════"
}

# Parse arguments
parse_args() {
    local wave=""
    local single_category=""

    while [[ $# -gt 0 ]]; do
        case "$1" in
            --wave)
                wave="$2"
                shift 2
                ;;
            --model)
                MODEL="$2"
                shift 2
                ;;
            --help|-h)
                echo "Usage: $0 [category-key] [--wave N] [--model MODEL]"
                echo ""
                echo "Options:"
                echo "  category-key    Expand single category"
                echo "  --wave N        Expand all categories in wave N (1 or 2)"
                echo "  --model MODEL   Claude model to use (default: haiku)"
                echo ""
                echo "Examples:"
                echo "  $0                          # Expand all categories"
                echo "  $0 azure-ai-services        # Expand single category"
                echo "  $0 --wave 1                 # Expand HIGH priority only"
                exit 0
                ;;
            *)
                single_category="$1"
                shift
                ;;
        esac
    done

    # Determine which categories to expand
    if [[ -n "$single_category" ]]; then
        echo "$single_category"
    elif [[ -n "$wave" ]]; then
        jq -r ".waves[] | select(.wave == $wave) | .categories[]" "$CATEGORIES_FILE"
    else
        jq -r '.categories | keys[]' "$CATEGORIES_FILE"
    fi
}

# Main
main() {
    check_prereqs
    mkdir -p "$LOG_DIR"

    log "Skill Category Expansion"
    log "Model: $MODEL | Categories file: $CATEGORIES_FILE"
    echo "" >&2

    local categories=()
    while IFS= read -r line; do
        categories+=("$line")
    done < <(parse_args "$@")

    if [[ ${#categories[@]} -eq 0 ]]; then
        error "No categories to expand"
        exit 1
    fi

    log "Categories to expand: ${#categories[@]}"
    for cat in "${categories[@]}"; do
        log "  - $cat"
    done
    echo "" >&2

    build_manifest "${categories[@]}"
}

main "$@"