docset-viewer/lib/docset_repository.php

<?php

class DocsetRepository
{
	private string $root;

	public function __construct(?string $root = null)
	{
		$this->root = realpath((string)($root ?: cfg('docsets/path'))) ?: (string)($root ?: cfg('docsets/path'));
	}

	public function listDocsets(): array
	{
		if(!is_dir($this->root))
			return array();
		$docsets = array();
		foreach(scandir($this->root) ?: array() as $entry)
		{
			if($entry === '.' || $entry === '..')
				continue;
			if(!str_ends_with($entry, '.docset'))
				continue;
			try
			{
				$docsets[] = $this->readDocset($entry);
			}
			catch(Throwable $e)
			{
				Log::text('docsets', 'Skipping '.$entry.': '.$e->getMessage());
			}
		}
		usort($docsets, fn($a, $b) => strcasecmp($a['name'], $b['name']));
		return $docsets;
	}

	public function getDocset(string $id): ?array
	{
		$id = $this->cleanId($id);
		if($id === '')
			return null;
		$path = $this->root.'/'.$id.'.docset';
		if(!is_dir($path))
			return null;
		return $this->readDocset($id.'.docset');
	}

	public function search(string $query, string $docsetId = '', int $limit = 80): array
	{
		$query = trim($query);
		if($query === '')
			return array();
		$docsets = $docsetId !== '' ? array_filter(array($this->getDocset($docsetId))) : $this->listDocsets();
		$all = array();
		foreach($docsets as $docset)
		{
			$all = array_merge($all, $this->searchOne($docset, $query, $limit));
		}
		usort($all, function($a, $b) use ($query) {
			$qa = strtolower($query);
			$an = strtolower($a['name']);
			$bn = strtolower($b['name']);
			$scoreA = $an === $qa ? 0 : (str_starts_with($an, $qa) ? 1 : 2);
			$scoreB = $bn === $qa ? 0 : (str_starts_with($bn, $qa) ? 1 : 2);
			return $scoreA <=> $scoreB ?: strcasecmp($a['name'], $b['name']);
		});
		return array_slice($all, 0, $limit);
	}

	private function searchOne(array $docset, string $query, int $limit): array
	{
		if(!class_exists('SQLite3'))
			throw new RuntimeException('PHP SQLite3 extension is not installed.');
		$db = new SQLite3($docset['index_path'], SQLITE3_OPEN_READONLY);
		try
		{
			$sql = 'select name, type, path from searchIndex where name like :prefix escape "\\" order by length(name), lower(name) limit :limit';
			$stmt = $db->prepare($sql);
			$stmt->bindValue(':prefix', $this->escapeLike($query).'%', SQLITE3_TEXT);
			$stmt->bindValue(':limit', max(1, min($limit, 200)), SQLITE3_INTEGER);
			$rows = $this->fetchRows($stmt);
			if(count($rows) < $limit)
			{
				$sql = 'select name, type, path from searchIndex where name like :contains escape "\\" and name not like :prefix escape "\\" order by lower(name) limit :limit';
				$stmt = $db->prepare($sql);
				$stmt->bindValue(':contains', '%'.$this->escapeLike($query).'%', SQLITE3_TEXT);
				$stmt->bindValue(':prefix', $this->escapeLike($query).'%', SQLITE3_TEXT);
				$stmt->bindValue(':limit', max(1, min($limit - count($rows), 200)), SQLITE3_INTEGER);
				$rows = array_merge($rows, $this->fetchRows($stmt));
			}
			return array_map(function($row) use ($docset) {
				$row['docset'] = $docset['id'];
				$row['docset_name'] = $docset['name'];
				$row['href'] = URL::Link('view', array('docset' => $docset['id'], 'path' => $row['path']));
				return $row;
			}, $rows);
		}
		finally
		{
			$db->close();
		}
	}

	private function fetchRows(SQLite3Stmt $stmt): array
	{
		$result = $stmt->execute();
		$rows = array();
		if($result)
		{
			while($row = $result->fetchArray(SQLITE3_ASSOC))
				$rows[] = $row;
			$result->finalize();
		}
		$stmt->close();
		return $rows;
	}

	public function resolveDocumentPath(string $docsetId, string $relativePath): ?array
	{
		$docset = $this->getDocset($docsetId);
		if(!$docset)
			return null;
		$relativePath = ltrim(str_replace('\\', '/', $relativePath), '/');
		$anchorParts = explode('#', $relativePath, 2);
		$queryParts = explode('?', $anchorParts[0], 2);
		$filePath = $queryParts[0];
		$anchor = $anchorParts[1] ?? '';
		$real = realpath($docset['documents_path'].'/'.$filePath);
		$documentsRoot = realpath($docset['documents_path']);
		if(!$real || !$documentsRoot || !$this->isWithin($real, $documentsRoot) || !is_file($real))
			return null;
		return array('docset' => $docset, 'file' => $real, 'relative' => $filePath, 'anchor' => $anchor);
	}

	public function assetUrl(string $docsetId, string $relativePath): string
	{
		return URL::Link('asset', array('docset' => $docsetId, 'path' => ltrim($relativePath, '/')));
	}

	private function readDocset(string $entry): array
	{
		$id = substr($entry, 0, -strlen('.docset'));
		$base = $this->root.'/'.$entry;
		$resources = $base.'/Contents/Resources';
		$index = $resources.'/docSet.dsidx';
		$documents = $resources.'/Documents';
		if(!is_file($index))
			throw new RuntimeException('missing docSet.dsidx');
		if(!is_dir($documents))
			throw new RuntimeException('missing Documents directory');
		$plist = $this->readInfoPlist($base.'/Contents/Info.plist');
		$count = $this->countEntries($index);
		return array(
			'id' => $id,
			'name' => $plist['CFBundleName'] ?? $id,
			'family' => $plist['DocSetPlatformFamily'] ?? '',
			'index_file' => $plist['dashIndexFilePath'] ?? 'index.html',
			'path' => $base,
			'index_path' => $index,
			'documents_path' => $documents,
			'entry_count' => $count,
		);
	}

	private function countEntries(string $index): int
	{
		if(!class_exists('SQLite3'))
			return 0;
		$db = new SQLite3($index, SQLITE3_OPEN_READONLY);
		try
		{
			$result = $db->querySingle('select count(*) from searchIndex');
			return (int)$result;
		}
		catch(Throwable $e)
		{
			return 0;
		}
		finally
		{
			$db->close();
		}
	}

	private function readInfoPlist(string $path): array
	{
		if(!is_file($path))
			return array();
		$xml = @simplexml_load_file($path);
		if(!$xml || !isset($xml->dict))
			return array();
		$dict = array();
		$key = null;
		foreach($xml->dict->children() as $node)
		{
			if($node->getName() === 'key')
			{
				$key = (string)$node;
				continue;
			}
			if($key !== null)
			{
				$dict[$key] = (string)$node;
				$key = null;
			}
		}
		return $dict;
	}

	private function cleanId(string $id): string
	{
		$id = trim($id);
		$out = '';
		foreach(str_split($id) as $char)
		{
			if(ctype_alnum($char) || $char === '-' || $char === '_' || $char === '.')
				$out .= $char;
		}
		return trim($out, '.');
	}

	private function escapeLike(string $value): string
	{
		$value = str_replace('\\', '\\\\', $value);
		$value = str_replace('%', '\\%', $value);
		return str_replace('_', '\\_', $value);
	}

	private function isWithin(string $path, string $root): bool
	{
		$root = rtrim($root, DIRECTORY_SEPARATOR).DIRECTORY_SEPARATOR;
		return str_starts_with($path, $root) || $path === rtrim($root, DIRECTORY_SEPARATOR);
	}
}