#!/usr/bin/perl

use strict;
use warnings;
use Storable qw(nstore);
use HTML::Entities qw(decode_entities);

# --- Configuration ---
my $built_site_dir = '../_site/log'; 
my $output_file    = '../_site/cgi-bin/search_index.dat';
my %index;

print "Building search index from $built_site_dir...\n";

foreach my $path (glob("$built_site_dir/*/index.html")) {
    next unless open(my $fh, '<:utf8', $path);
    my $html = do { local $/; <$fh> };
    close($fh);

    # Extract Title and Main Content
    my ($title) = $html =~ m|<title>(.*?)</title>|is;
    my ($main)  = $html =~ m|<main>(.*?)</main>|is;
    $main //= '';

    # Strip HTML and clean prose
    $main =~ s|<pre[^>]*>.*?</pre>| |gs;
    $main =~ s|<code[^>]*>.*?</code>| |gs;
    $main =~ s|<[^>]+>| |g;
    $main = decode_entities($main);
    $main =~ s|\s+| |g;
    $main =~ s/^\s+|\s+$//g;

    # Normalize path
    my $url = $path;
    $url =~ s|^\.\./_site/||; # Remove local build directory
    $url =~ s|^\.\./||;       # Remove any leading dots
    $url =~ s|^/+||;          # Remove leading slashes

    $index{$url} = {
        t => $title || "Untitled",
        c => $main
    };
}

nstore(\%index, $output_file);
printf("Index complete: %d files (%.2f KB)\n", scalar(keys %index), (-s $output_file) / 1024);