<?php
/**
 * Geoplot — XMP reader for image and video files.
 *
 * Container-level XMP packet extraction uses two extractors:
 *
 *   - extractFromIsoBmffUuid() handles ISO BMFF containers (MP4, MOV,
 *     M4V, ...) by reading the top-level uuid box with the Adobe XMP
 *     UUID.
 *
 *   - extractViaGetID3() handles everything else (PNG, JPEG, TIFF,
 *     MKV, and ISO BMFF files whose XMP lives inside moov/udta/meta
 *     rather than the top-level uuid box).
 *
 * XMP packet parsing is done here with DOMDocument / DOMXPath against
 * the standard XMP namespaces.
 *
 * (c) Wizardry and Steamworks 2026
 * SPDX-License-Identifier: MIT
 */

if (!defined('DOKU_INC')) die();

require_once __DIR__ . '/vendor/autoload.php';

class Geoplot_XMP {

    const NS_DC    = 'http://purl.org/dc/elements/1.1/';
    const NS_EXIF  = 'http://ns.adobe.com/exif/1.0/';
    const NS_XMPDM = 'http://ns.adobe.com/xmp/1.0/DynamicMedia/';

    const ADOBE_XMP_UUID = "be7acfcb97a942e89c71999491e3afac";

    private static $ISO_BMFF_EXT = array(
        'mp4', 'm4v', 'mov', 'm4a', '3gp', '3g2',
    );

    public static function read($path) {
        if (!is_file($path)) return null;

        $ext = strtolower(pathinfo($path, PATHINFO_EXTENSION));

        $packet = null;

        if (in_array($ext, self::$ISO_BMFF_EXT)) {
            $packet = self::extractFromIsoBmffUuid($path);
        }

        if ($packet === null) {
            $packet = self::extractViaGetID3($path);
        }

        if ($packet === null) return null;

        return self::parse($packet);
    }

    private static function extractViaGetID3($path) {
        if (!class_exists('\\getID3')) return null;

        $getID3 = new \getID3();
        $info   = $getID3->analyze($path);

        if (!empty($info['error'])) return null;

        $candidates = array(
            array('tags', 'png',       'XML:com.adobe.xmp'),
            array('tags', 'jpeg',      'XMP'),
            array('tags', 'jpeg',      'XML:com.adobe.xmp'),
            array('tags', 'quicktime', 'XMP'),
            array('tags', 'quicktime', 'XML:com.adobe.xmp'),
            array('tags', 'matroska',  'XMP'),
            array('tags', 'matroska',  'XML:com.adobe.xmp'),
            array('tags', 'id3v2',     'XMP'),
            array('xmp',  'data'),
        );

        foreach ($candidates as $keys) {
            $node = $info;
            foreach ($keys as $k) {
                if (!is_array($node) || !isset($node[$k])) {
                    $node = null;
                    break;
                }
                $node = $node[$k];
            }
            $packet = self::coerceToString($node);
            if ($packet !== null && self::isWellFormedXmpPacket($packet)) {
                return $packet;
            }
        }

        return self::deepFindXmp($info, 0);
    }

    private static function coerceToString($node) {
        if (is_string($node) && strlen($node) > 32) return $node;
        if (is_array($node)) {
            foreach ($node as $v) {
                if (is_string($v) && strlen($v) > 32) return $v;
            }
        }
        return null;
    }

    private static function deepFindXmp($node, $depth) {
        if ($depth > 10) return null;

        if (is_string($node)) {
            if (self::isWellFormedXmpPacket($node)) {
                return $node;
            }
            return null;
        }
        if (!is_array($node)) return null;
        foreach ($node as $v) {
            $found = self::deepFindXmp($v, $depth + 1);
            if ($found !== null) return $found;
        }
        return null;
    }

    private static function isWellFormedXmpPacket($s) {
        if (!is_string($s) || strlen($s) < 64) return false;

        $head = $s;
        if (substr($head, 0, 3) === "\xEF\xBB\xBF") {
            $head = substr($head, 3);
        }
        $head = ltrim($head, " \t\r\n");

        if (strpos($head, '<?xpacket') !== 0
            && strpos($head, '<?xml') !== 0
            && strpos($head, '<x:xmpmeta') !== 0) {
            return false;
        }

        return (strpos($s, '<x:xmpmeta') !== false
             && strpos($s, '</x:xmpmeta>') !== false);
    }

    private static function extractFromIsoBmffUuid($path) {
        $fh = @fopen($path, 'rb');
        if (!$fh) return null;

        $result = null;
        $uuidBytes = hex2bin(self::ADOBE_XMP_UUID);

        while (!feof($fh)) {
            $hdr = fread($fh, 8);
            if (strlen($hdr) < 8) break;

            $unpacked = unpack('Nsize/a4type', $hdr);
            $size = $unpacked['size'];
            $type = $unpacked['type'];
            $header = 8;

            if ($size === 1) {
                $big = fread($fh, 8);
                if (strlen($big) < 8) break;
                $hi = unpack('N', substr($big, 0, 4))[1];
                $lo = unpack('N', substr($big, 4, 4))[1];
                $size = ($hi << 32) | $lo;
                $header = 16;
            } elseif ($size === 0) {
                $size = PHP_INT_MAX;
            }

            if ($size < $header) break;

            $payloadLength = $size - $header;

            if ($type === 'uuid' && $payloadLength >= 16) {
                $uuid = fread($fh, 16);
                if (strlen($uuid) < 16) break;

                if ($uuid === $uuidBytes) {
                    $lenBytes = fread($fh, 4);
                    if (strlen($lenBytes) === 4) {
                        $packetLen = unpack('N', $lenBytes)[1];
                        $packet = fread($fh, $packetLen);
                        if (strlen($packet) === $packetLen) {
                            $result = $packet;
                        }
                    }
                    break;
                }

                fseek($fh, $payloadLength - 16, SEEK_CUR);
                continue;
            }

            fseek($fh, $payloadLength, SEEK_CUR);
        }

        fclose($fh);
        return $result;
    }

    public static function parse($packet) {
        if ($packet === null || $packet === '') return null;

        $start = strpos($packet, '<?xpacket');
        if ($start !== false) {
            $head = substr($packet, 0, $start);
            $head = str_replace("\xEF\xBB\xBF", '', $head);
            if (trim($head) === '') {
                $end = strpos($packet, '?>', $start);
                if ($end !== false) {
                    $packet = substr($packet, $end + 2);
                }
            }
        }
        $packet = ltrim($packet, "\xEF\xBB\xBF \t\r\n");

        $doc = new DOMDocument('1.0', 'UTF-8');
        $prev = libxml_use_internal_errors(true);
        $loaded = $doc->loadXML($packet, LIBXML_NONET | LIBXML_NOCDATA);
        libxml_clear_errors();
        libxml_use_internal_errors($prev);
        if (!$loaded) return null;

        $xpath = new DOMXPath($doc);
        $xpath->registerNamespace('dc',    self::NS_DC);
        $xpath->registerNamespace('exif',  self::NS_EXIF);
        $xpath->registerNamespace('xmpDM', self::NS_XMPDM);

        $latRaw = self::firstText($xpath, '//exif:GPSLatitude');
        $lonRaw = self::firstText($xpath, '//exif:GPSLongitude');
        if ($latRaw === null || $lonRaw === null) return null;

        $lat = self::parseGps($latRaw);
        $lon = self::parseGps($lonRaw);
        if ($lat === null || $lon === null) return null;

        $shotRaw = self::firstText($xpath, '//xmpDM:ShotDate');
        if ($shotRaw === null) {
            $shotRaw = self::firstText($xpath, '//xmpDM:StartDate');
        }
        if ($shotRaw === null) return null;

        $ts = strtotime($shotRaw);
        if ($ts === false) return null;

        $timecode = self::firstText($xpath, '//xmpDM:StartTimecode');
        if ($timecode === null) {
            $timecode = self::firstText($xpath, '//xmpDM:startTimecode');
        }
        if ($timecode === null) $timecode = '';

        $creator = '';
        $creatorNodes = $xpath->query('//dc:creator//rdf:li');
        if ($creatorNodes !== false) {
            $parts = array();
            foreach ($creatorNodes as $n) {
                $parts[] = trim($n->textContent);
            }
            $creator = implode(', ', $parts);
        }

        return array(
            'lat'      => $lat,
            'lon'      => $lon,
            'date'     => (int) $ts,
            'date_raw' => (string) $shotRaw,
            'timecode' => (string) $timecode,
            'creator'  => $creator,
        );
    }

    private static function firstText(DOMXPath $xpath, $query) {
        $nodes = $xpath->query($query);
        if ($nodes === false || $nodes->length === 0) return null;
        $text = trim($nodes->item(0)->textContent);
        return $text === '' ? null : $text;
    }

    private static function parseGps($s) {
        $s = trim($s);
        if (!preg_match('/^([0-9.]+),([0-9.]+)([NSEWnsew])$/', $s, $m)) {
            if (is_numeric($s)) return (float) $s;
            return null;
        }
        $v = (float) $m[1] + (float) $m[2] / 60.0;
        $hemi = strtoupper($m[3]);
        if ($hemi === 'S' || $hemi === 'W') $v = -$v;
        return $v;
    }
}

// END_OF_GEOPLOT_XMP