#!/bin/bash

set -e

KEYWORD_LIST="./30_keyword-list"
SOURCE_LIST="./20_source-list-raw"
MATCHES_ALL="./30_matches-all"
MATCHES_NEW="./30_matches-new"

if [ ! -f "$KEYWORD_LIST" ]; then
    echo "Error: $KEYWORD_LIST not found"
    exit 1
fi

if [ ! -f "$SOURCE_LIST" ]; then
    echo "Error: $SOURCE_LIST not found"
    exit 1
fi

# Create matches-all if it doesn't exist
touch "$MATCHES_ALL"

# Clear matches-new for this run
> "$MATCHES_NEW"

echo "Searching for keywords in $SOURCE_LIST..."

new_matches=0
total_matches=0

while IFS= read -r keyword; do
    keyword=$(echo "$keyword" | tr -d '\r\n' | sed 's/[[:space:]]*$//')

    if [ -z "$keyword" ]; then
        continue
    fi

    echo "  Searching for keyword: $keyword"

    # Find entries containing the keyword
    matches=$(grep -i "$keyword" "$SOURCE_LIST" || true)

    if [ -n "$matches" ]; then
        match_count=$(echo "$matches" | wc -l)
        echo "    Found $match_count matches"
        total_matches=$((total_matches + match_count))

        # Check each match against matches-all
        echo "$matches" | while IFS= read -r entry; do
            if ! grep -Fxq "$entry" "$MATCHES_ALL" 2>/dev/null; then
                echo "$entry" >> "$MATCHES_ALL"
                echo "$entry" >> "$MATCHES_NEW"
                new_matches=$((new_matches + 1))
            fi
        done
    else
        echo "    No matches found"
    fi

done < "$KEYWORD_LIST"

# Count new matches
new_count=$(wc -l < "$MATCHES_NEW" 2>/dev/null || echo "0")
all_count=$(wc -l < "$MATCHES_ALL" 2>/dev/null || echo "0")

echo ""
echo "Summary:"
echo "  Total matches found: $total_matches"
echo "  New matches: $new_count"
echo "  Total matches (all): $all_count"