#!/bin/bash
# FreshRSS Smart Digest
# Pulls unread articles, ranks them by relevance, delivers categorized summary
# Runs daily at 7:00 AM (before AI newsletter digest)

set -e

SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
source "$SCRIPT_DIR/../../.env" 2>/dev/null || true

# Config
FRESHRSS_URL="${FRESHRSS_URL:-http://freshrss.kangaroo-eel.ts.net}"
FRESHRSS_USER="${FRESHRSS_USER:-anthony}"
TELEGRAM_CHAT="${TELEGRAM_CHAT:-1793951355}"
NTFY_URL="${NTFY_URL:-}"
NTFY_TOPIC="${NTFY_TOPIC:-}"
NTFY_MIN_PRIORITY="${NTFY_MIN_PRIORITY:-4}"

# Your interest keywords for relevance ranking
INTERESTS=(
    "AI" "artificial intelligence" "machine learning" "LLM"
    "renewable energy" "solar" "wind" "battery" "EV" "electric vehicle"
    "politics" "Labor" "election" "government"
    "LGBTQ" "LGBT" "queer" "transgender"
    "Perth" "Western Australia" "WA"
    "climate" "environment"
    "mental health" "depression" "therapy"
)

# Priority sources (always include these)
PRIORITY_SOURCES=(
    "CNN"
    "MSNBC"
    "Al Jazeera"
    "ABC News"
    "The Guardian"
    "BlueSky"
)

log() {
    echo "[$(date '+%Y-%m-%d %H:%M:%S')] $1"
}

fetch_unread() {
    log "Fetching unread articles from FreshRSS..."
    
    local raw_output
    raw_output=$(/home/openclaw/.openclaw/workspace/skills/freshrss-reader/scripts/freshrss.sh headlines \
        --unread --count 50 2>/dev/null) || {
        log "ERROR: Failed to fetch from FreshRSS"
        return 1
    }
    
    echo "$raw_output"
}

# Score an article based on interest keywords
score_article() {
    local title="$1"
    local source="$2"
    local categories="$3"
    local score=0
    local text="${title,,} ${categories,,}"
    
    # Check interest keywords
    for interest in "${INTERESTS[@]}"; do
        if [[ "$text" == *"${interest,,}"* ]]; then
            ((score+=2))
        fi
    done
    
    # Priority sources get a boost
    for priority in "${PRIORITY_SOURCES[@]}"; do
        if [[ "$source" == *"$priority"* ]]; then
            ((score+=3))
        fi
    done
    
    echo "$score"
}

# Parse articles and build digest
build_digest() {
    local raw="$1"
    local output=""
    local must_read=""
    local skimmable=""
    local total_count=0
    
    # Parse the output - each article is 3 lines: date/source/title, URL, categories
    local current_date=""
    local current_source=""
    local current_title=""
    local current_url=""
    local current_cats=""
    
    while IFS= read -r line; do
        # Skip empty lines
        [[ -z "$line" ]] && continue
        
        # Check if this is a title line (starts with [)
        if [[ "$line" =~ ^\[.+Z\].*\]\ (.*)$ ]]; then
            # Save previous article if exists
            if [[ -n "$current_title" ]]; then
                local score
                score=$(score_article "$current_title" "$current_source" "$current_cats")
                
                local article_formatted="• [${current_title}](${current_url}) \n   _${current_source}_\n"
                
                if ((score >= 5)); then
                    must_read+="$article_formatted\n"
                elif ((score >= 2)); then
                    skimmable+="$article_formatted\n"
                fi
                
                ((total_count++))
            fi
            
            # Parse new article - format: [timestamp] [source] title
            current_date=$(echo "$line" | sed 's/^\[\([^]]*\)\].*/\1/')
            current_source=$(echo "$line" | sed -E 's/^\[.+Z\] \[([^]]+)\].*/\1/')
            current_title=$(echo "$line" | sed -E 's/^\[.+Z\] \[[^]]+\] //')
            current_url=""
            current_cats=""
            
        # Check if this is a URL line
        elif [[ "$line" =~ ^[[:space:]]*https?:// ]]; then
            current_url="$(echo "$line" | xargs)"
            
        # Check if this is categories line
        elif [[ "$line" =~ ^Categories:\  ]]; then
            current_cats=$(echo "$line" | sed 's/Categories: //')
        fi
    done <<< "$raw"
    
    # Don't forget the last article
    if [[ -n "$current_title" ]]; then
        local score
        score=$(score_article "$current_title" "$current_source" "$current_cats")
        
        local article_formatted="• [${current_title}](${current_url}) \n   _${current_source}_\n"
        
        if ((score >= 5)); then
            must_read+="$article_formatted\n"
        elif ((score >= 2)); then
            skimmable+="$article_formatted\n"
        fi
        
        ((total_count++))
    fi
    
    # Limit items per section to keep message under Telegram's 4096 char limit
    local must_read_limited="$(echo -e "$must_read" | head -n 16)"  # ~8 items (2 lines each)
    local skimmable_limited="$(echo -e "$skimmable" | head -n 10)"  # ~5 items (2 lines each)
    
    # Build final message
    output="📰 *FreshRSS Daily Digest*\n\n"
    output+="Found *$total_count* unread articles.\n\n"
    
    if [[ -n "$must_read" ]]; then
        local count_must=$(($(echo -e "$must_read" | grep -c '^•') + 0))
        output+="🔥 *Must Read* ($count_must)\n$must_read_limited"
        [[ "$must_read" != "$must_read_limited" ]] && output+="\n   _...and more_\n\n" || output+="\n"
    fi
    
    if [[ -n "$skimmable" ]]; then
        local count_skim=$(($(echo -e "$skimmable" | grep -c '^•') + 0))
        output+="📎 *Skimmable* ($count_skim)\n$skimmable_limited"
        [[ "$skimmable" != "$skimmable_limited" ]] && output+="\n   _...and more_\n\n" || output+="\n"
    fi
    
    if [[ -z "$must_read" && -z "$skimmable" ]]; then
        output+="No high-priority articles today. You're caught up! 🎉\n"
    fi
    
    output+="\n📖 [Open FreshRSS]($FRESHRSS_URL)"
    
    # Truncate if still too long (Telegram limit is 4096)
    if [[ ${#output} -gt 4000 ]]; then
        output="${output:0:3950}...\n\n_(truncated — more articles in FreshRSS)_\n\n📖 [Open FreshRSS]($FRESHRSS_URL)"
    fi
    
    echo -e "$output"
}

send_telegram() {
    local message="$1"
    
    log "Sending to Telegram..."
    
    if [[ -z "${TELEGRAM_BOT_TOKEN:-}" ]]; then
        log "ERROR: TELEGRAM_BOT_TOKEN is not set"
        return 1
    fi

    local tg_response
    tg_response=$(curl -s -X POST "https://api.telegram.org/bot${TELEGRAM_BOT_TOKEN}/sendMessage" \
        -H "Content-Type: application/json" \
        -d "{
            \"chat_id\": \"$TELEGRAM_CHAT\",
            \"text\": \"$message\",
            \"parse_mode\": \"Markdown\",
            \"disable_web_page_preview\": true
        }") || {
        log "ERROR: Failed to send Telegram message"
        return 1
    }

    if ! echo "$tg_response" | jq -e '.ok == true' > /dev/null 2>&1; then
        log "ERROR: Telegram API rejected message: $tg_response"
        return 1
    fi
    
    log "Sent to Telegram successfully"
}

send_ntfy() {
    local title="$1"
    local message="$2"
    local priority="${3:-4}"
    local sound="${4:-default}"

    [[ -z "$NTFY_URL" ]] && return 0
    [[ -z "$NTFY_TOPIC" ]] && return 0

    # Enforce minimum priority (default 4)
    if [[ "$priority" =~ ^[0-9]+$ ]] && [[ "$NTFY_MIN_PRIORITY" =~ ^[0-9]+$ ]]; then
        if (( priority < NTFY_MIN_PRIORITY )); then
            priority="$NTFY_MIN_PRIORITY"
        fi
    fi

    log "Sending to ntfy..."
    curl -s -X POST "${NTFY_URL%/}/${NTFY_TOPIC}" \
        -H "Title: $title" \
        -H "Priority: $priority" \
        -H "Sound: $sound" \
        -d "$message" > /dev/null || {
        log "ERROR: Failed to send ntfy message"
        return 1
    }
    log "Sent to ntfy successfully"
}

main() {
    log "Starting FreshRSS digest..."
    
    # Fetch articles
    local raw_articles
    raw_articles=$(fetch_unread) || exit 1
    
    # Build digest
    local digest
    digest=$(build_digest "$raw_articles")
    
    # Send to both channels
    send_telegram "$digest"
    
    # Plain text version for ntfy (strip markdown)
    local plain_digest
    plain_digest=$(echo -e "$digest" | sed 's/\*//g' | sed 's/\[\([^]]*\)\]([^)]*)/\1/g')
    send_ntfy "FreshRSS Digest" "$plain_digest" 4 default
    
    log "Digest complete!"
}

main "$@"