<?php
/** @license MIT
 * Copyright 2018 J. King et al.
 * See LICENSE and AUTHORS files for details */

declare(strict_types=1);
namespace MensBeam\Lax\Parser\XML;

use MensBeam\Lax\Category\Category;
use MensBeam\Lax\Category\Collection as CategoryCollection;
use MensBeam\Lax\Person\Person;
use MensBeam\Lax\Person\Collection as PersonCollection;
use MensBeam\Lax\Text;
use MensBeam\Lax\Date;
use MensBeam\Lax\MimeType;
use MensBeam\Lax\Url;

abstract class Construct {
    use \MensBeam\Lax\Parser\Construct;

    protected const TEXT_LOOSE = "loose";
    protected const TEXT_PLAIN = "plain";
    protected const TEXT_HTML = "html";

    protected const DATE_ANY = 0;
    protected const DATE_LATEST = 1;
    protected const DATE_EARLIEST = 2;
    protected const DATE_ALL = 3;

    protected const QUERY_AMBIGUOUS_DATES = "rss2:pubDate|rss2:lastBuildDate|dc:date|dc:available|dc:dateAccepted|dc:dateCopyrighted|dc:dateSubmitted|dc:issued|dc:modified|dc:valid|dct:date|dct:available|dct:dateAccepted|dct:dateCopyrighted|dct:dateSubmitted|dct:issued|dct:modified|dct:valid";
    protected const QUERY_RSS_PERMALINK = "rss2:guid[not(@isPermaLink) or @isPermaLink='true']";

    /** @var \DOMDocument */
    protected $document;
    /** @var \DOMXPath */
    protected $xpath;
    /** @var \DOMElement */
    protected $subject;

    /** Retrieves an element node based on an XPath query */
    protected function fetchElement(string $query, \DOMNode $context = null): ?\DOMElement {
        $node = @$this->xpath->query("(".$query.")[1]", $context ?? $this->subject);
        if ($node === false) {
            throw new \Exception("Invalid XPath query: $query"); // @codeCoverageIgnore
        }
        return ($node->length) ? $node->item(0) : null;
    }

    /** Retrieves the trimmed text content of one or more DOM elements based on an XPath query, optionally matching a pattern
     *
     * Returns null if no suitable nodes were found
     *
     * @param string $query The XPath query of the nodes to return
     * @param string|null $pattern The pattern to optionally filter matches with. The pattern should not include delimiters or anchors and is always case-insensitive
     * @param bool|null $multi Whether to return multiple results as an array (true) or one result as a string (false, default)
     * @param \DOMNode $context The context node for the XPath query
     * @return string|array|null
     */
    protected function fetchString(string $query, ?string $pattern = null, ?bool $multi = null, ?\DOMNode $context = null) {
        $out = [];
        $pattern = strlen($pattern ?? "") ? "/^(?:".str_replace("/", "\\/", $pattern).")$/i" : "";
        $multi = $multi ?? false;
        $nodes = $this->xpath->query($query, $context ?? $this->subject);
        foreach ($nodes as $node) {
            $t = $this->trimText($node->textContent);
            if (!$pattern || preg_match($pattern, $t)) {
                if (!$multi) {
                    return $t;
                } else {
                    $out[] = $t;
                }
            }
        }
        return ($out) ? $out : null;
    }

    /** Retrieves and parses a date from one or more DOM elements based on an XPath query
     *
     * Returns null if no suitable nodes were found
     *
     * @param string $query The XPath query of the nodes to return
     * @param bool|null $mode Whether to return the first valid date found (DATE_ANY), the earliest chronologically (DATE_EARLIEST), latest chronologically (DATE_LATEST), or all valid dates (DAATE_ALL) in a sorted array
     * @param \DOMNode $context The context node for the XPath query
     * @return \MensBeam\Lax\Date|array|null
     */
    protected function fetchDate(string $query, int $mode, \DOMNode $context = null) {
        $out = [];
        $tz = new \DateTimeZone("UTC");
        assert(in_array($mode, [self::DATE_ANY, self::DATE_ALL, self::DATE_EARLIEST, self::DATE_LATEST]));
        foreach ((array) $this->fetchString($query, null, true, $context) as $d) {
            if ($d = Date::createFromString($d ?? "")) {
                if ($mode === self::DATE_ANY) {
                    return $d;
                } else {
                    // add the date to the output only if it is a unique moment in time so far
                    $ts = $d->setTimezone($tz)->format("Y-m-d\TH:i:s.u\Z");
                    if (!isset($out[$ts])) {
                        $out[$ts] = $d;
                    }
                }
            }
        }
        // sort the dates earliest to latest and produce an indexed array
        ksort($out);
        $out = array_values($out);
        // return based on requested mode
        switch ($mode) {
            case self::DATE_ALL:
                return $out;
            case self::DATE_EARLIEST:
                return $out ? $out[0] : null;
            case self::DATE_LATEST:
                return $out ? array_pop($out) : null;
        }
    }

    /** Returns the first valid URL matching an XPath query. Relative URLs are resolved when possible
     *
     * @param string $query The XPath query of the node to return
     * @param \DOMNode $context The context node for the XPath query
     */
    protected function fetchUrl(string $query, \DOMNode $context = null): ?Url {
        foreach ($this->xpath->query($query, $context ?? $this->subject) as $node) {
            $url = trim($node->textContent);
            if (strlen($url)) {
                try {
                    return new Url($url, $node->baseURI);
                } catch (\InvalidArgumentException $e) {
                    // don't return a result that doesn't evaluate to a valid URL of some sort
                }
            }
        }
        return null;
    }

    protected function fetchText(string $query, string $format, \DOMNode $context = null): ?Text {
        foreach ($this->xpath->query($query, $context ?? $this->subject) as $node) {
            $data = trim($node->textContent);
            if (strlen($data)) {
                $out = new Text;
                if ($format === "plain") {
                    $data = $this->trimText($data);
                } elseif ($format === "html" || $format === "loose") {
                    $out->htmlBase = strlen($node->baseURI) ? $node->baseURI : null;
                }
                $out->$format = $data;
                return $out;
            }
        }
        return null;
    }

    /** Finds and parses RSS person-texts and returns a collection of person objects
     *
     * Each can have a name, e-mail address, or both
     *
     * The following forms will yield both a name and address:
     *
     * - user@example.com (Full Name)
     * - Full Name <user@example.com>
     * - Full Name <mailto:user@example.com>
     */
    protected function fetchPeople(string $query, string $role, ?\DOMNode $context = null): ?PersonCollection {
        $out = new PersonCollection;
        foreach ($this->fetchString($query, ".+", true, $context) ?? [] as $person) {
            $p = new Person;
            if (preg_match("/^([^@\s]+@\S+) \((.+?)\)$/", $person, $match)) { // tests "user@example.com (Full Name)" form
                if ($this->validateMail($match[1])) {
                    $p->name = trim($match[2]);
                    $p->mail = $match[1];
                } else {
                    $p->name = $person;
                }
            } elseif (preg_match("/^((?:\S|\s(?!<))+) <(?:mailto:)?([^>]+)>$/", $person, $match)) { // tests "Full Name <user@example.com>" form
                if ($this->validateMail($match[2])) {
                    $p->name = trim($match[1]);
                    $p->mail = $match[2];
                } else {
                    $p->name = $person;
                }
            } elseif ($this->validateMail($person)) {
                $p->name = $person;
                $p->mail = $person;
            } else {
                $p->name = $person;
            }
            $p->role = $role;
            $out[] = $p;
        }
        return count($out) ? $out : null;
    }

    /** Returns a node-list of Atom link elements with the desired relation or equivalents.
     *
     * Links without an href attribute are excluded.
     *
     * @see https://tools.ietf.org/html/rfc4287#section-4.2.7.2
     */
    protected function fetchAtomRelations(string $rel = "", \DOMNode $context = null): array {
        // normalize the relation
        $custom = false;
        $rel = trim($rel);
        if ($rel === "") {
            $rel = "alternate"; // @codeCoverageIgnore
        } elseif (strpos(strtolower($rel), "http://www.iana.org/assignments/relation/") === 0) {
            $rel = substr($rel, 41); // @codeCoverageIgnore
        } elseif (preg_match("<^[a-z\.\-]+$>i", $rel)) {
            $rel = strtolower($rel);
        } else {
            $custom = true;
            $url = (string) new Url($rel);
        }
        // look at all the links that have a non-empty href attribute
        $out = [];
        foreach ($this->xpath->query("atom:link[normalize-space(@href)]", $context ?? $this->subject) as $l) {
            try {
                new Url($l->getAttribute("href"));
            } catch (\InvalidArgumentException $e) {
                // reject any links which do not have valid URLs
                continue;
            }
            $r = trim($l->getAttribute("rel"));
            if ($custom) {
                if ($url === (string) new Url($r)) {
                    $out[] = $l;
                }
            } else {
                $r = trim(strtolower(rawurldecode($r)));
                $r = (strpos($r, "http://www.iana.org/assignments/relation/") === 0) ? substr($r, 41) : $r;
                $r = !strlen($r) ? "alternate" : $r;
                if ($r === $rel) {
                    $out[] = $l;
                }
            }
        }
        return $out;
    }

    /** Returns the first Atom link URL which matches the desired relation, with nearest desired media type */
    protected function fetchAtomRelation(string $rel = "", array $mediaTypes = [], \DOMNode $context = null): ?Url {
        // tidy ther list of media types; this orders them worst (0)to best (highest index) and then creates a hashtable
        $mediaTypes = array_flip(array_reverse(array_values(array_unique(array_map(function(string $t) {
            return strtolower(trim($t));
        }, $mediaTypes)))));
        $rels = $this->fetchAtomRelations($rel, $context);
        if ($rels && !$mediaTypes) {
            return new Url($rels[0]->getAttribute("href"), $rels[0]->baseURI);
        }
        $result = array_reduce($rels, function($best, $cur) use ($mediaTypes) {
            $t = trim($cur->getAttribute("type"));
            // absence of media type is acceptable if no better match yet exists
            if (!strlen($t) && (!$best || $best[1] < -1)) {
                return [$cur, -1]; // any preferred type will rank higher than -1
            }
            $t = MimeType::parse($t);
            if ($t) {
                $rank = $mediaTypes[$t->essence] ?? -2; // even no type will rank higher than a non-preferred type
                if (!$best || $rank > $best[1]) {
                    // if there is currently no candidate or the candidate ranks lower, use the current link
                    return [$cur, $rank];
                }
            }
            return $best;
        });
        return $result ? new Url($result[0]->getAttribute("href"), $result[0]->baseURI) : null;
    }

    protected function fetchAtomText(string $query, \DOMNode $context = null): ?Text {
        $out = new Text;
        $populated = false;
        foreach ($this->xpath->query($query, $context ?? $this->subject) as $node) {
            if ($node->hasAttribute("src")) {
                // ignore any external content
                continue;
            }
            // get the content type; assume "text" if not provided
            switch (MimeType::parseAtom(trim($node->getAttribute("type")))->essence) {
                case "text/plain":
                    if (is_null($out->plain)) {
                        $plain = $this->trimText($node->textContent);
                        if (strlen($plain)) {
                            $out->plain = $plain;
                        }
                    }
                    break;
                case "text/html":
                    if (is_null($out->html)) {
                        $html = trim($node->textContent);
                        if (strlen($html)) {
                            $out->html = $html;
                            $out->htmlBase = strlen($node->baseURI) ? $node->baseURI : null;
                        }
                    }
                    break;
                case "application/xhtml+xml":
                    if (is_null($out->xhtml) && ($xhtml = $this->fetchElement("html:div", $node))) {
                        $out->xhtml = $xhtml->ownerDocument->saveXML($xhtml);
                        $out->xhtmlBase = strlen($xhtml->baseURI) ? $xhtml->baseURI : null;
                    }
                    break;
            }
        }
        return (!$this->empty($out)) ? $out : null;
    }

    /** Finds and parses Atom person-constructs, and returns a collection of Person objects */
    protected function fetchAtomPeople(string $query, string $role, \DOMNode $context = null): ?PersonCollection {
        $nodes = $this->xpath->query($query, $context ?? $this->subject);
        $out = new PersonCollection;
        foreach ($nodes as $node) {
            $p = new Person;
            $mail = $this->fetchString("atom:email", null, null, $node) ?? "";
            $p->mail = $this->validateMail($mail) ? $mail : null;
            $p->name = $this->fetchString("atom:name", ".+", null, $node);
            $p->url = $this->fetchUrl("atom:uri", $node);
            $p->role = $role;
            if (!is_null($p->name)) {
                $out[] = $p;
            }
        }
        return count($out) ? $out : null;
    }

    /** Primitive to fetch an Atom feed/entry identifier */
    protected function getIdAtom(): ?string {
        return $this->fetchString("atom:id", ".+");
    }

    /** Primitive to fetch an RSS feed/entry identifier
     *
     * Using RSS' <guid> for feed identifiers is non-standard, but harmless
     */
    protected function getIdRss2(): ?string {
        return $this->fetchString("rss2:guid", ".+");
    }

    /** Primitive to fetch a Dublin Core feed/entry identifier */
    protected function getIdDC(): ?string {
        return $this->fetchString("dc:identifier|dct:identifier", ".+");
    }

    protected function getLangXML(): ?string {
        // walk up the tree looking for the nearest language tag
        $el = $this->subject;
        do {
            $out = $this->fetchString("@xml:lang", ".+", false, $el);
            $el = $el->parentNode;
        } while (is_null($out) && $el);
        return $out;
    }

    protected function getLangDC(): ?string {
        return $this->fetchString("dc:language|dct:language", ".+");
    }

    protected function getLangRss2(): ?string {
        return $this->fetchString("rss2:language", ".+");
    }

    protected function getLinkAtom(): ?Url {
        return $this->fetchAtomRelation("alternate", ["text/html", "application/xhtml+xml"]);
    }

    protected function getLinkRss2(): ?Url {
        return $this->fetchUrl("rss2:link") ?? $this->fetchUrl(self::QUERY_RSS_PERMALINK);
    }

    protected function getLinkRss1(): ?Url {
        return $this->fetchUrl("rss1:link") ?? $this->fetchUrl("rss0:link");
    }

    protected function getTitleAtom(): ?Text {
        return $this->fetchAtomText("atom:title");
    }

    protected function getTitleRss1(): ?Text {
        return $this->fetchText("rss1:title", self::TEXT_LOOSE) ?? $this->fetchText("rss0:title", self::TEXT_LOOSE);
    }

    protected function getTitleRss2(): ?Text {
        return $this->fetchText("rss2:title", self::TEXT_LOOSE);
    }

    protected function getTitleDC(): ?Text {
        return $this->fetchText("dc:title|dct:title", self::TEXT_PLAIN);
    }

    protected function getTitlePod(): ?Text {
        return $this->fetchText("apple:title", self::TEXT_PLAIN);
    }

    protected function getCategoriesFromNode(\DOMNode $context): ?CategoryCollection {
        return $this->getCategoriesAtom($context)
            ?? $this->getCategoriesRss2($context)
            ?? $this->getCategoriesGPlay($context)
            ?? $this->getCategoriesTunes($context)
            ?? $this->getCategoriesDC($context);
    }

    protected function getCategoriesAtom(\DOMNode $context): ?CategoryCollection {
        $out = new CategoryCollection;
        foreach ($this->xpath->query("atom:category[@term]", $context) as $node) {
            $c = new Category;
            $c->domain = $this->trimText($node->getAttribute("scheme"));
            $c->label = $this->trimText($node->getAttribute("label"));
            $c->name = $this->trimText($node->getAttribute("term"));
            if (strlen($c->name)) {
                $out[] = $c;
            }
        }
        return count($out) ? $out : null;
    }

    protected function getCategoriesRss2(\DOMNode $context): ?CategoryCollection {
        $out = new CategoryCollection;
        foreach ($this->xpath->query("rss2:category", $context) as $node) {
            $c = new Category;
            $c->domain = $this->trimText($node->getAttribute("domain"));
            $c->name = $this->trimText($node->textContent);
            if (strlen($c->name)) {
                $out[] = $c;
            }
        }
        return count($out) ? $out : null;
    }

    /** Dublin Core doesn't have an obvious category type, so we use 'subject' as a nearest approximation */
    protected function getCategoriesDC(\DOMNode $context): ?CategoryCollection {
        $out = new CategoryCollection;
        foreach ($this->fetchString("dc:subject|dct:subject", null, true, $context) ?? [] as $text) {
            if (strlen($text)) {
                $c = new Category;
                $c->name = $text;
                $out[] = $c;
            }
        }
        return count($out) ? $out : null;
    }

    protected function getCategoriesTunes(\DOMNode $context): ?CategoryCollection {
        $out = new CategoryCollection;
        foreach ($this->xpath->query("apple:category", $context) as $node) {
            $c = new Category;
            $c->name = $this->trimText($node->getAttribute("text"));
            if (strlen($c->name)) {
                foreach ($this->xpath->query("apple:category", $node) as $sub) {
                    $sname = $this->trimText($sub->getAttribute("text"));
                    if (strlen($sname)) {
                        $c->subcategory = $sname;
                        break;
                    }
                }
                $out[] = $c;
            }
        }
        return count($out) ? $out : null;
    }

    protected function getCategoriesGPlay(\DOMNode $context): ?CategoryCollection {
        $out = new CategoryCollection;
        foreach ($this->xpath->query("gplay:category", $context) as $node) {
            $c = new Category;
            $c->name = $this->trimText($node->getAttribute("text"));
            if (strlen($c->name)) {
                $out[] = $c;
            }
        }
        return count($out) ? $out : null;
    }
}