<?php
/**
 * Warrant Search scraper → CSV exporter (browser + CLI, with path + permission checks)
 */

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

$isCli = (php_sapi_name() === 'cli');

// === CONFIG: where to put the CSV ===
// It will save IN THE SAME FOLDER as this script.
$baseDir    = __DIR__;
$timestamp  = date('Ymd_His');
$outputCsv  = $baseDir . "/warrants_export_$timestamp.csv";
$debugText0 = $baseDir . "/debug_page0.txt";

$baseUrl       = "https://doc.wa.gov/records/incarcerated-data-search/warrant-search";
$outputCsvUrl  = basename($outputCsv); // URL relative to this script
$userAgent     = "Mozilla/5.0 (compatible; WCFS-scraper/1.0; +you@example.com)";

// -------------- Simple logger -------------------

function logmsg($msg)
{
    global $isCli;
    if ($isCli) {
        echo $msg . "\n";
    } else {
        echo htmlspecialchars($msg, ENT_QUOTES, 'UTF-8') . "\n";
    }
}

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

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

    $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 -------------------

function parseWarrantPage($html, $debugFile = null, $pageNum = 0)
{
    // Strip HTML tags and decode entities
    $text = html_entity_decode(strip_tags($html));

    // For debugging: write raw text of first page once
    if ($debugFile && $pageNum === 0 && !file_exists($debugFile)) {
        file_put_contents($debugFile, $text);
    }

    // Normalize all whitespace to single spaces
    $normalized = preg_replace('/\s+/', ' ', $text);
    $normalized = trim($normalized);

    $records = [];

    $pattern = '/(\d{2}\/\d{2}\/\d{4})\s+([^0-9]+?)\s+([A-Z0-9 ,\'\-\/]+?)\s+([A-Za-z ]+?)\s+Name:/';

    $matches = [];
    preg_match_all($pattern, $normalized, $matches, PREG_SET_ORDER | PREG_OFFSET_CAPTURE);

    if (!$matches) {
        return [];
    }

    $numMatches = count($matches);
    $textLen    = strlen($normalized);

    for ($i = 0; $i < $numMatches; $i++) {
        $m = $matches[$i];

        $start = $m[0][1];
        $end   = ($i + 1 < $numMatches) ? $matches[$i + 1][0][1] : $textLen;

        $chunk = substr($normalized, $start, $end - $start);

        $record = [
            'warrant_date_header' => $m[1][0],
            'name_header'         => trim($m[2][0]),
            'crime_header'        => trim($m[3][0]),
            'county_header'       => trim($m[4][0]),
            'name'                => '',
            'doc_number'          => '',
            'supervision_county'  => '',
            'crime_type'          => '',
            'warrant_issue_date'  => '',
            'age'                 => '',
            'birthdate'           => '',
            'height'              => '',
            'weight'              => '',
            'eye_color'           => '',
            'hair_color'          => '',
            'race'                => '',
            'hispanic'            => '',
        ];

        if (preg_match('/Name:\s*(.+?)\s*DOC Number:/', $chunk, $m2)) {
            $record['name'] = trim($m2[1]);
        }
        if (preg_match('/DOC Number:\s*(.+?)\s*Supervision County:/', $chunk, $m2)) {
            $record['doc_number'] = trim($m2[1]);
        }
        if (preg_match('/Supervision County:\s*(.+?)\s*Crime Type:/', $chunk, $m2)) {
            $record['supervision_county'] = trim($m2[1]);
        }
        if (preg_match('/Crime Type:\s*(.+?)\s*Secretary\'?s Warrant/', $chunk, $m2)) {
            $record['crime_type'] = trim($m2[1]);
        }
        if (preg_match('/Date Issued:\s*(.+?)\s+Submit a Tip/', $chunk, $m2)) {
            $record['warrant_issue_date'] = trim($m2[1]);
        }

        if (preg_match('/Age:\s*([^ ]+)\s+Birthdate:/', $chunk, $m2)) {
            $record['age'] = trim($m2[1]);
        }
        if (preg_match('/Birthdate:\s*(.+?)\s+Height:/', $chunk, $m2)) {
            $record['birthdate'] = trim($m2[1]);
        }
        if (preg_match('/Height:\s*(.+?)\s+Weight:/', $chunk, $m2)) {
            $record['height'] = trim($m2[1]);
        }
        if (preg_match('/Weight:\s*(.+?)\s+Eye Color:/', $chunk, $m2)) {
            $record['weight'] = trim($m2[1]);
        }
        if (preg_match('/Eye Color:\s*(.+?)\s+Hair Color:/', $chunk, $m2)) {
            $record['eye_color'] = trim($m2[1]);
        }
        if (preg_match('/Hair Color:\s*(.+?)\s+Race:/', $chunk, $m2)) {
            $record['hair_color'] = trim($m2[1]);
        }
        if (preg_match('/Race:\s*(.+?)\s+Hispanic:/', $chunk, $m2)) {
            $record['race'] = trim($m2[1]);
        }
        if (preg_match('/Hispanic:\s*(Yes|No)/i', $chunk, $m2)) {
            $record['hispanic'] = trim($m2[1]);
        }

        $records[] = $record;
    }

    return $records;
}

// -------------- MAIN -------------------

if (!$isCli) {
    header('Content-Type: text/html; charset=utf-8');
    echo "<!DOCTYPE html><html><head><title>Warrant Export</title></head><body>";
    echo "<h1>Warrant Export</h1>";
    echo "<p><strong>Script directory (where files are written):</strong><br><code>" .
        htmlspecialchars($baseDir, ENT_QUOTES, 'UTF-8') . "</code></p>";
    echo "<pre>";
}

logmsg("Starting scrape...");
logmsg("Output CSV will be: $outputCsv");

$page       = 0;
$allRecords = [];
$maxPages   = 500;

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

    $pageRecords = parseWarrantPage($html, $debugText0, $page);
    $countPage   = count($pageRecords);

    logmsg("  Parsed $countPage records on this page.");

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

    $allRecords = array_merge($allRecords, $pageRecords);
    $page++;
    sleep(1);
}

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

if (empty($allRecords)) {
    logmsg("No records collected. Nothing to write.");
    if (!$isCli) {
        echo "</pre><p>No CSV was generated.</p></body></html>";
    }
    exit(0);
}

// Try to open file
$fp = @fopen($outputCsv, 'w');
if (!$fp) {
    $dirPerms = substr(sprintf('%o', fileperms($baseDir)), -4);
    logmsg("ERROR: Unable to open CSV for writing: $outputCsv");
    logmsg("Directory permissions for $baseDir are: $dirPerms");
    logmsg("The web server user probably does NOT have write permission here.");
    if (!$isCli) {
        echo "</pre><p><strong>ERROR:</strong> Cannot write CSV. Check folder permissions.</p></body></html>";
    }
    exit(1);
}

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

// Header row
fputcsv($fp, $headers);

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

fclose($fp);

// Verify existence
clearstatcache();
if (file_exists($outputCsv)) {
    $real = realpath($outputCsv);
    logmsg("Done. Wrote " . count($allRecords) . " records.");
    logmsg("CSV file exists at: $real");
} else {
    logmsg("ERROR: After writing, CSV file DOES NOT EXIST. Something is wrong with filesystem/permissions.");
}

if ($isCli) {
    exit(0);
} else {
    echo "</pre>";
    if (file_exists($outputCsv)) {
        echo "<p><strong>CSV ready:</strong> ";
        echo "<a href=\"" . htmlspecialchars($outputCsvUrl, ENT_QUOTES, 'UTF-8') . "\">Download " .
             htmlspecialchars($outputCsvUrl, ENT_QUOTES, 'UTF-8') . "</a></p>";
    } else {
        echo "<p><strong>ERROR:</strong> CSV file was not found on disk. Check permissions on:<br><code>" .
             htmlspecialchars($baseDir, ENT_QUOTES, 'UTF-8') . "</code></p>";
    }
    echo "</body></html>";
}