<?php
/**
 * running_days.php — resolve a train's weekly running pattern by
 * querying MULTIPLE sources, normalising each to a Mon..Sun 7-bit
 * pattern, and taking a consensus vote. Result is cached.
 *
 * Design notes (read these):
 *  - The COMPARISON is deterministic code. No LLM is involved in deciding.
 *  - Sources mostly share one upstream (NTES/CRIS), so agreement means
 *    "high confidence", NOT "proven". We expose the per-source breakdown
 *    and a confidence level so the UI never overstates certainty.
 *  - Each adapter is responsible for its own source's format and MUST
 *    return a 7-element array [Mon,Tue,Wed,Thu,Fri,Sat,Sun] of 0/1, or null.
 *  - Add a source = add one callable to rd_adapters(). Three+ is the goal.
 */

/* ---------- Normalise many textual formats to a 7-bit array ---------- */
/* Returns [Mon..Sun] of 0/1, or null if the text doesn't pin down days. */
function rd_normalize(?string $raw): ?array {
    if ($raw === null) return null;
    $s = strtolower(trim($raw));
    if ($s === '') return null;

    // Pure 7-char bitstring, e.g. "1111111" or "1010100" (Mon..Sun)
    if (preg_match('/^[01]{7}$/', $s)) return array_map('intval', str_split($s));

    $map = ['mon'=>0,'tue'=>1,'wed'=>2,'thu'=>3,'fri'=>4,'sat'=>5,'sun'=>6];

    // "... except Sunday" style → start full, remove named days
    if (strpos($s, 'except') !== false) {
        $d = [1,1,1,1,1,1,1];
        foreach ($map as $a=>$i) if (strpos($s,$a)!==false) $d[$i]=0;
        return $d;
    }
    // Explicit daily
    if (preg_match('/\b(daily|every ?day|all ?7|7 ?days)\b/', $s)) return [1,1,1,1,1,1,1];

    // Day-name list (order-independent; works for "Mon,Wed,Fri", full names, etc.)
    $d = [0,0,0,0,0,0,0]; $found = false;
    foreach ($map as $a=>$i) if (strpos($s,$a)!==false) { $d[$i]=1; $found=true; }
    if ($found) return $d;

    // "weekly"/"bi-weekly" with no day named → cannot determine pattern
    return null;
}

function rd_to_string(?array $d): ?string {
    return $d === null ? null : implode('', array_map(fn($x)=>$x?1:0, $d));
}

/* Pure consensus vote. Input: [sourceName => "1111111"|null].
 * Returns [pattern|null, confidence]. confidence: high|medium|low|none. */
function rd_vote(array $perSource): array {
    $counts = [];
    foreach ($perSource as $patt) {
        if ($patt !== null) $counts[$patt] = ($counts[$patt] ?? 0) + 1;
    }
    if (!$counts) return [null, 'none'];
    arsort($counts);
    $pattern  = array_key_first($counts);
    $usable   = array_sum($counts);
    $distinct = count($counts);
    if      ($distinct === 1 && $usable >= 3) $confidence = 'high';
    elseif  ($distinct === 1 && $usable == 2) $confidence = 'medium';
    elseif  ($distinct === 1 && $usable == 1) $confidence = 'low';   // single source only
    else                                      $confidence = 'low';   // conflict → plurality, flagged
    return [$pattern, $confidence];
}

/* Human label: "Daily", or "Mon, Wed, Fri", or "—" */
function rd_label(?string $pattern): string {
    if ($pattern === null || !preg_match('/^[01]{7}$/', $pattern)) return '—';
    if ($pattern === '1111111') return 'Daily';
    $names = ['Mon','Tue','Wed','Thu','Fri','Sat','Sun'];
    $out = [];
    for ($i=0;$i<7;$i++) if ($pattern[$i]==='1') $out[] = $names[$i];
    return $out ? implode(', ', $out) : '—';
}

/* Does this train run on a given date? (Y-m-d) — uses cached pattern. */
function rd_runs_on(?string $pattern, string $ymd): ?bool {
    if ($pattern === null || !preg_match('/^[01]{7}$/', $pattern)) return null; // unknown
    $dow = (int)date('N', strtotime($ymd));   // 1=Mon..7=Sun
    return $pattern[$dow-1] === '1';
}

/* ---------- Source adapters ----------
 * Each returns [Mon..Sun] 0/1 array, or null on failure/unknown.
 * Keep each one small and isolated so one failing source never breaks others.
 */

// (1) irctc1 on RapidAPI — real adapter, only active if RAPIDAPI_KEY is set.
function src_irctc1(string $trainNo): ?array {
    if (!defined('RAPIDAPI_KEY') || RAPIDAPI_KEY === '') return null;
    $url = 'https://irctc1.p.rapidapi.com/api/v1/getTrainSchedule?trainNo=' . urlencode($trainNo);
    $ch = curl_init($url);
    curl_setopt_array($ch, [
        CURLOPT_RETURNTRANSFER => true,
        CURLOPT_TIMEOUT => 12,
        CURLOPT_HTTPHEADER => [
            'x-rapidapi-host: irctc1.p.rapidapi.com',
            'x-rapidapi-key: ' . RAPIDAPI_KEY,
        ],
    ]);
    $body = curl_exec($ch);
    $code = curl_getinfo($ch, CURLINFO_HTTP_CODE);
    curl_close($ch);
    if ($body === false || $code >= 400) return null;
    $j = json_decode($body, true);
    // The field name varies by version; check the common ones, then fall back
    // to normalising whatever string we find. Verify with one real call first.
    $raw = $j['data']['train_base']['running_days']
        ?? $j['data']['running_days']
        ?? $j['train_base']['running_days']
        ?? null;
    return rd_normalize(is_string($raw) ? $raw : null);
}

// (2) Template for an additional web/API source. Returns null until you wire it.
// NOTE: if you point this at a public site, confirm its terms allow automated
// access and expect to maintain the parser when their HTML changes.
function src_secondary(string $trainNo): ?array {
    return null; // TODO: fetch + parse, then `return rd_normalize($daysTextFromPage);`
}

// (3) Another source slot — same contract.
function src_tertiary(string $trainNo): ?array {
    return null; // TODO
}

function rd_adapters(): array {
    return [
        'irctc1'    => 'src_irctc1',
        'secondary' => 'src_secondary',
        'tertiary'  => 'src_tertiary',
    ];
}

/* ---------- The resolver: consensus + confidence + cache ---------- */
function resolve_running_days(string $trainNo, int $maxAgeDays = 180): array {
    // 1) cache hit?
    $st = db()->prepare('SELECT * FROM train_running_days WHERE train_no = ? LIMIT 1');
    $st->execute([$trainNo]);
    $row = $st->fetch();
    if ($row && strtotime($row['checked_at']) > time() - $maxAgeDays*86400) {
        return [
            'pattern'    => $row['pattern'],
            'confidence' => $row['confidence'],
            'label'      => rd_label($row['pattern']),
            'sources'    => json_decode($row['sources_json'] ?? '[]', true) ?: [],
            'cached'     => true,
        ];
    }

    // 2) query each source
    $perSource = [];     // name => "1111111" | null
    foreach (rd_adapters() as $name => $fn) {
        $arr = null;
        try { $arr = $fn($trainNo); } catch (Throwable $e) { $arr = null; }
        $perSource[$name] = rd_to_string($arr);
    }

    // 3) vote
    [$pattern, $confidence] = rd_vote($perSource);

    // 4) cache + return
    $sources_json = json_encode($perSource);
    db()->prepare(
        'INSERT INTO train_running_days (train_no, pattern, confidence, sources_json)
         VALUES (?,?,?,?)
         ON DUPLICATE KEY UPDATE pattern=VALUES(pattern), confidence=VALUES(confidence),
                                 sources_json=VALUES(sources_json), checked_at=NOW()'
    )->execute([$trainNo, $pattern, $confidence, $sources_json]);

    return [
        'pattern'    => $pattern,
        'confidence' => $confidence,
        'label'      => rd_label($pattern),
        'sources'    => $perSource,
        'cached'     => false,
    ];
}