<?php
/**
 * Warrant Search scraper → CSV exporter
 *
 * Usage (from CLI):
 *   php warrant_export.php
 *
 * This will create warrants_export.csv in the same directory.
 */

ini_set('memory_limit', '512M');
set_time_limit(0);

$baseUrl   = "https://doc.wa.gov/records/incarcerated-data-search/warrant-search";
$outputCsv = __DIR__ . "/warrants_export.csv";

$userAgent = "Mozilla/5.0 (compatible; WCFS-scraper/1.0; +you@example.com)";

// -------------- HTTP FETCH -------------------

function fetchPage($url, $pageNum, $userAgent)
{
    $ch = curl_init();

    // Only pass the page parameter – leave everything else to defaults.
    $params  = ['page' => $pageNum];
    $fullUrl = $url . '?' . http_build_query($params);

    curl_setopt_array($ch, [
        CURLOPT_URL            => $fullUrl,
        CURLOPT_RETURNTRANSFER => true,
        CURLOPT_FOLLOWLOCATION => true,
        CURLOPT_USERAGENT      => $userAgent,
        CURLOPT_TIMEOUT        => 30,
        CURLOPT_SSL_VERIFYPEER => true,
    ]);

    $html = curl_exec($ch);

    if ($html === false) {
        $err = curl_error($ch);
        curl_close($ch);
        throw new RuntimeException("cURL error fetching $fullUrl: $err");
    }

    $status = curl_getinfo($ch, CURLINFO_HTTP_CODE);
    curl_close($ch);

    if ($status >= 400) {
        throw new RuntimeException("HTTP $status fetching $fullUrl");
    }

    return $html;
}

// -------------- PARSER -------------------

/**
 * Parse a single HTML page into an array of records.
 *
 * Each record is an associative array with consistent keys.
 */
function parseWarrantPage($html)
{
    // Strip HTML tags, keep whitespace/newlines
    $text = html_entity_decode(strip_tags($html));
    $lines = preg_split('/\R+/', $text); // split on any line break
    $cleanLines = [];

    foreach ($lines as $line) {
        $line = trim($line);
        if ($line !== '') {
            $cleanLines[] = $line;
        }
    }

    $records = [];
    $i = 0;
    $count = count($cleanLines);

    while ($i < $count) {
        $line = $cleanLines[$i];

        // Header line pattern:
        // "MM/DD/YYYY  LASTNAME, FIRST  CRIME  County"
        if (preg_match('/^(\d{2}\/\d{2}\/\d{4})\s+(.+?)\s{2,}(.+?)\s{2,}(.+)$/', $line, $m)) {
            $record = [
                'warrant_date_header' => $m[1],
                'name_header'         => trim($m[2]),
                'crime_header'        => trim($m[3]),
                'county_header'       => trim($m[4]),
                'name'                => '',
                'doc_number'          => '',
                'supervision_county'  => '',
                'crime_type'          => '',
                'warrant_issue_date'  => '',
                'age'                 => '',
                'birthdate'           => '',
                'height'              => '',
                'weight'              => '',
                'eye_color'           => '',
                'hair_color'          => '',
                'race'                => '',
                'hispanic'            => '',
            ];

            $i++;
            // Read forward until next header or end
            while ($i < $count) {
                $line2 = $cleanLines[$i];

                // If we hit another header line, stop this record
                if (preg_match('/^(\d{2}\/\d{2}\/\d{4})\s+(.+?)\s{2,}(.+?)\s{2,}(.+)$/', $line2)) {
                    break;
                }

                // Map label/value pairs
                if (stripos($line2, 'Name:') === 0) {
                    $record['name'] = trim(substr($line2, strlen('Name:')));
                } elseif (stripos($line2, 'DOC Number:') === 0) {
                    $record['doc_number'] = trim(substr($line2, strlen('DOC Number:')));
                } elseif (stripos($line2, 'Supervision County:') === 0) {
                    $record['supervision_county'] = trim(substr($line2, strlen('Supervision County:')));
                } elseif (stripos($line2, 'Crime Type:') === 0) {
                    $record['crime_type'] = trim(substr($line2, strlen('Crime Type:')));
                } elseif (stripos($line2, 'Date Issued:') === 0) {
                    $record['warrant_issue_date'] = trim(substr($line2, strlen('Date Issued:')));
                } elseif (stripos($line2, 'Age:') === 0) {
                    $record['age'] = trim(substr($line2, strlen('Age:')));
                } elseif (stripos($line2, 'Birthdate:') === 0) {
                    $record['birthdate'] = trim(substr($line2, strlen('Birthdate:')));
                } elseif (stripos($line2, 'Height:') === 0) {
                    $record['height'] = trim(substr($line2, strlen('Height:')));
                } elseif (stripos($line2, 'Weight:') === 0) {
                    $record['weight'] = trim(substr($line2, strlen('Weight:')));
                } elseif (stripos($line2, 'Eye Color:') === 0) {
                    $record['eye_color'] = trim(substr($line2, strlen('Eye Color:')));
                } elseif (stripos($line2, 'Hair Color:') === 0) {
                    $record['hair_color'] = trim(substr($line2, strlen('Hair Color:')));
                } elseif (stripos($line2, 'Race:') === 0) {
                    $record['race'] = trim(substr($line2, strlen('Race:')));
                } elseif (stripos($line2, 'Hispanic:') === 0) {
                    $record['hispanic'] = trim(substr($line2, strlen('Hispanic:')));
                }

                $i++;
            }

            $records[] = $record;
            continue;
        }

        $i++;
    }

    return $records;
}

// -------------- MAIN SCRAPE LOOP -------------------

$page = 0;
$allRecords = [];

// Optional safety: stop after a high page number in case something goes weird
$maxPages = 500;

echo "Starting scrape...\n";

while ($page < $maxPages) {
    echo "Fetching page $page...\n";
    try {
        $html = fetchPage($baseUrl, $page, $userAgent);
    } catch (Exception $e) {
        echo "Error fetching page $page: " . $e->getMessage() . "\n";
        break;
    }

    $pageRecords = parseWarrantPage($html);
    $countPage = count($pageRecords);

    echo "  Parsed $countPage records on this page.\n";

    if ($countPage === 0) {
        echo "No more records found. Stopping.\n";
        break;
    }

    $allRecords = array_merge($allRecords, $pageRecords);

    $page++;
    // Be polite to their server.
    sleep(1);
}

// -------------- WRITE CSV -------------------

if (empty($allRecords)) {
    echo "No records collected. Nothing to write.\n";
    exit(0);
}

// Determine columns dynamically from keys
$headers = array_keys($allRecords[0]);

$fp = fopen($outputCsv, 'w');
if (!$fp) {
    throw new RuntimeException("Unable to open $outputCsv for writing.");
}

// Write header row
fputcsv($fp, $headers);

// Write data
foreach ($allRecords as $row) {
    $line = [];
    foreach ($headers as $h) {
        $line[] = isset($row[$h]) ? $row[$h] : '';
    }
    fputcsv($fp, $line);
}

fclose($fp);

echo "Done. Wrote " . count($allRecords) . " records to $outputCsv\n";