SchemaExtractor::parseFields() - Code Metrics - Sysbot-org/tgscraper - Measure and Improve Code Quality continuously with Scrutinizer

SchemaExtractor::parseFields() A
last analyzed 2024-07-02 10:38 UTC

↳ Parent: SchemaExtractor

Complexity

Conditions	4
Paths	4

Size

Total Lines	24
Code Lines	17

Duplication

Lines	0
Ratio	0 %

Importance

Changes	2
Bugs	0	Features	0

Metric	Value
eloc	17
c	2
b	0
f	0
dl	0
loc	24
rs	9.7
cc	4
nc	4
nop	2

<?php

namespace TgScraper\Common;

use Composer\InstalledVersions;
use Exception;
use GuzzleHttp\Client;
use GuzzleHttp\Exception\GuzzleException;
use InvalidArgumentException;
use OutOfBoundsException;
use Psr\Log\LoggerInterface;
use RuntimeException;
use TgScraper\Parsers\Field;
use TgScraper\Parsers\ObjectDescription;
use TgScraper\Constants\Versions;
use voku\helper\HtmlDomParser;
use voku\helper\SimpleHtmlDomInterface;
use voku\helper\SimpleHtmlDomNode;
use voku\helper\SimpleHtmlDomNodeInterface;

/**
 * Class SchemaExtractor
 * @package TgScraper\Common
 */
class SchemaExtractor
{
    /**
     * @var string
     */
    private string $version;

    /**
     * SchemaExtractor constructor.
     * @param LoggerInterface $logger
     * @param HtmlDomParser $dom
     */
    public function __construct(private LoggerInterface $logger, private HtmlDomParser $dom)
    {
        $this->version = $this->parseVersion();
        $this->logger->info('Bot API version: ' . $this->version);
    }


    /**
     * @param LoggerInterface $logger
     * @param string $version
     * @return SchemaExtractor
     * @throws OutOfBoundsException
     * @throws Exception
     * @throws GuzzleException
     */
    public static function fromVersion(LoggerInterface $logger, string $version = Versions::LATEST): SchemaExtractor
    {
        if (InstalledVersions::isInstalled('sysbot/tgscraper-cache') and class_exists('\TgScraper\Cache\CacheLoader')) {
            $logger->info('Cache package detected, searching for a cached version.');
            try {
                /** @noinspection PhpFullyQualifiedNameUsageInspection */
                /** @noinspection PhpUndefinedNamespaceInspection */
                /** @psalm-suppress UndefinedClass */
                $path = \TgScraper\Cache\CacheLoader::getCachedVersion($version);
                $logger->info('Cached version found.');
                return self::fromFile($logger, $path);
            } catch (OutOfBoundsException) {
                $logger->info('Cached version not found, continuing with URL.');
            }
        }
        $url = Versions::getUrlFromText($version);
        $logger->info(sprintf('Using URL: %s', $url));
        return self::fromUrl($logger, $url);
    }

    /**
     * @param LoggerInterface $logger
     * @param string $path
     * @return SchemaExtractor
     * @throws InvalidArgumentException
     * @throws RuntimeException
     */
    public static function fromFile(LoggerInterface $logger, string $path): SchemaExtractor
    {
        if (!file_exists($path) or is_dir($path)) {
            throw new InvalidArgumentException('File not found');
        }
        $path = realpath($path);
        try {
            $logger->info(sprintf('Loading data from file "%s".', $path));
            $dom = HtmlDomParser::file_get_html($path);
            $logger->info('Data loaded.');
        } catch (RuntimeException $e) {
            $logger->critical(sprintf('Unable to load data from "%s": %s', $path, $e->getMessage()));
            throw $e;
        }
        return new self($logger, $dom);
    }

    /**
     * @param LoggerInterface $logger
     * @param string $url
     * @return SchemaExtractor
     * @throws GuzzleException
     */
    public static function fromUrl(LoggerInterface $logger, string $url): SchemaExtractor
    {
        $client = new Client();
        try {
            $html = $client->get($url)->getBody();
            $dom = HtmlDomParser::str_get_html((string)$html);
        } catch (GuzzleException $e) {
            $logger->critical(sprintf('Unable to load data from URL "%s": %s', $url, $e->getMessage()));
            throw $e;
        }
        $logger->info(sprintf('Data loaded from "%s".', $url));
        return new self($logger, $dom);
    }

    /**
     * @param SimpleHtmlDomInterface $node
     * @return array{description: string, table: ?SimpleHtmlDomNodeInterface, extended_by: string[]}
     */
    private static function parseNode(SimpleHtmlDomInterface $node): array
    {
        $description = '';
        $table = null;
        $extendedBy = [];
        $tag = '';
        $sibling = $node;
        while (!str_starts_with($tag ?? '', 'h')) {
            $sibling = $sibling?->nextSibling();
            $tag = $sibling?->tag;
            if (empty($node->text()) or empty($tag) or $tag == 'text' or empty($sibling)) {
                continue;
            }
            switch ($tag) {
                case 'p':
                    $description .= PHP_EOL . $sibling->innerHtml();
                    break;
                case 'ul':
                    $items = $sibling->findMulti('li');
                    foreach ($items as $item) {
                        $extendedBy[] = $item->text();
                    }
                    break 2;
                case 'table':
                    /** @var SimpleHtmlDomNodeInterface $table */
                    $table = $sibling->findOne('tbody')->findMulti('tr');
                    break 2;
            }
        }
        return ['description' => $description, 'table' => $table, 'extended_by' => $extendedBy];
    }

    /**
     * @return string
     */
    private function parseVersion(): string
    {
        $element = $this->dom->findOne('h3');
        $tag = '';
        while ($tag != 'p' and !empty($element)) {
            $element = $element->nextSibling();
            $tag = $element?->tag;
        }
        if (empty($element)) {
            return '1.0.0';
        }
        $versionNumbers = explode('.', str_replace('Bot API ', '', $element->text()));
        return sprintf(
            '%s.%s.%s',
            $versionNumbers[0] ?? '1',
            $versionNumbers[1] ?? '0',
            $versionNumbers[2] ?? '0'
        );
    }

    /**
     * @return string
     */
    public function getVersion(): string
    {
        return $this->version;
    }

    /**
     * @return array{version: string, methods: array, types: array}
     * @throws Exception
     */
    public function extract(): array
    {
        $elements = $this->dom->findMultiOrFalse('h4');
        if (false === $elements) {
            throw new Exception('Unable to fetch required DOM nodes');
        }
        $data = ['version' => $this->version, 'methods' => [], 'types' => []];
        foreach ($elements as $element) {
            if (!str_contains($name = $element->text(), ' ')) {
                $isMethod = lcfirst($name) == $name;
                $path = $isMethod ? 'methods' : 'types';
                ['description' => $description, 'table' => $table, 'extended_by' => $extendedBy] = self::parseNode(
                    $element
                );
                $data[$path][] = self::generateElement(
                    $name,
                    trim($description),
                    $table,
                    $extendedBy,
                    $isMethod
                );
            }
        }
        return $data;
    }

    /**
     * @param string $name
     * @param string $description
     * @param SimpleHtmlDomNodeInterface|null $unparsedFields
     * @param array $extendedBy
     * @param bool $isMethod
     * @return array
     */
    private static function generateElement(
        string $name,
        string $description,
        ?SimpleHtmlDomNodeInterface $unparsedFields,
        array $extendedBy,
        bool $isMethod
    ): array {
        $fields = self::parseFields($unparsedFields, $isMethod);
        $result = [
            'name' => $name,
            'description' => htmlspecialchars_decode(strip_tags($description), ENT_QUOTES),
            'fields' => $fields
        ];
        if ($isMethod) {
            $description = new ObjectDescription($description);
            $returnTypes = $description->getTypes();
            $result['return_types'] = $returnTypes;
            return $result;
        }
        $result['extended_by'] = $extendedBy;
        return $result;
    }

    /**
     * @param SimpleHtmlDomNodeInterface|null $fields
     * @param bool $isMethod
     * @return array
     */
    private static function parseFields(?SimpleHtmlDomNodeInterface $fields, bool $isMethod): array
    {
        $parsedFields = [];
        $fields ??= [];
        /** @var SimpleHtmlDomInterface $field */
        foreach ($fields as $field) {
            /** @var SimpleHtmlDomNode $fieldData */
            $fieldData = $field->findMulti('td');
            $name = $fieldData[0]->text();
            if (empty($name)) {
                continue;
            }
            $types = $fieldData[1]->text();
            if ($isMethod) {
                $optional = $fieldData[2]->text() != 'Yes';
                $description = $fieldData[3]->innerHtml();
            } else {
                $description = $fieldData[2]->innerHtml();
                $optional = str_starts_with($fieldData[2]->text(), 'Optional.');
            }
            $field = new Field($name, $types, $optional, $description);
            $parsedFields[] = $field->toArray();
        }
        return $parsedFields;
    }
}


1			<?php
2
3			namespace TgScraper\Common;
4
5			use Composer\InstalledVersions;
6			use Exception;
7			use GuzzleHttp\Client;
8			use GuzzleHttp\Exception\GuzzleException;
9			use InvalidArgumentException;
10			use OutOfBoundsException;
11			use Psr\Log\LoggerInterface;
12			use RuntimeException;
13			use TgScraper\Parsers\Field;
14			use TgScraper\Parsers\ObjectDescription;
15			use TgScraper\Constants\Versions;
16			use voku\helper\HtmlDomParser;
17			use voku\helper\SimpleHtmlDomInterface;
18			use voku\helper\SimpleHtmlDomNode;
19			use voku\helper\SimpleHtmlDomNodeInterface;
20
21			/**
22			* Class SchemaExtractor
23			* @package TgScraper\Common
24			*/
25			class SchemaExtractor
26			{
27			/**
28			* @var string
29			*/
30			private string $version;
31
32			/**
33			* SchemaExtractor constructor.
34			* @param LoggerInterface $logger
35			* @param HtmlDomParser $dom
36			*/
37			public function __construct(private LoggerInterface $logger, private HtmlDomParser $dom)
38			{
39			$this->version = $this->parseVersion();
40			$this->logger->info('Bot API version: ' . $this->version);
41			}
42
43
44			/**
45			* @param LoggerInterface $logger
46			* @param string $version
47			* @return SchemaExtractor
48			* @throws OutOfBoundsException
49			* @throws Exception
50			* @throws GuzzleException
51			*/
52			public static function fromVersion(LoggerInterface $logger, string $version = Versions::LATEST): SchemaExtractor
53			{
54			if (InstalledVersions::isInstalled('sysbot/tgscraper-cache') and class_exists('\TgScraper\Cache\CacheLoader')) {
55			$logger->info('Cache package detected, searching for a cached version.');
56			try {
57			/** @noinspection PhpFullyQualifiedNameUsageInspection */
58			/** @noinspection PhpUndefinedNamespaceInspection */
59			/** @psalm-suppress UndefinedClass */
60			$path = \TgScraper\Cache\CacheLoader::getCachedVersion($version);
61			$logger->info('Cached version found.');
62			return self::fromFile($logger, $path);
63			} catch (OutOfBoundsException) {
64			$logger->info('Cached version not found, continuing with URL.');
65			}
66			}
67			$url = Versions::getUrlFromText($version);
68			$logger->info(sprintf('Using URL: %s', $url));
69			return self::fromUrl($logger, $url);
70			}
71
72			/**
73			* @param LoggerInterface $logger
74			* @param string $path
75			* @return SchemaExtractor
76			* @throws InvalidArgumentException
77			* @throws RuntimeException
78			*/
79			public static function fromFile(LoggerInterface $logger, string $path): SchemaExtractor
80			{
81			if (!file_exists($path) or is_dir($path)) {
82			throw new InvalidArgumentException('File not found');
83			}
84			$path = realpath($path);
85			try {
86			$logger->info(sprintf('Loading data from file "%s".', $path));
87			$dom = HtmlDomParser::file_get_html($path);
88			$logger->info('Data loaded.');
89			} catch (RuntimeException $e) {
90			$logger->critical(sprintf('Unable to load data from "%s": %s', $path, $e->getMessage()));
91			throw $e;
92			}
93			return new self($logger, $dom);
94			}
95
96			/**
97			* @param LoggerInterface $logger
98			* @param string $url
99			* @return SchemaExtractor
100			* @throws GuzzleException
101			*/
102			public static function fromUrl(LoggerInterface $logger, string $url): SchemaExtractor
103			{
104			$client = new Client();
105			try {
106			$html = $client->get($url)->getBody();
107			$dom = HtmlDomParser::str_get_html((string)$html);
108			} catch (GuzzleException $e) {
109			$logger->critical(sprintf('Unable to load data from URL "%s": %s', $url, $e->getMessage()));
110			throw $e;
111			}
112			$logger->info(sprintf('Data loaded from "%s".', $url));
113			return new self($logger, $dom);
114			}
115
116			/**
117			* @param SimpleHtmlDomInterface $node
118			* @return array{description: string, table: ?SimpleHtmlDomNodeInterface, extended_by: string[]}
119			*/
120			private static function parseNode(SimpleHtmlDomInterface $node): array
121			{
122			$description = '';
123			$table = null;
124			$extendedBy = [];
125			$tag = '';
126			$sibling = $node;
127			while (!str_starts_with($tag ?? '', 'h')) {
128			$sibling = $sibling?->nextSibling();
129			$tag = $sibling?->tag;
130			if (empty($node->text()) or empty($tag) or $tag == 'text' or empty($sibling)) {
131			continue;
132			}
133			switch ($tag) {
134			case 'p':
135			$description .= PHP_EOL . $sibling->innerHtml();
136			break;
137			case 'ul':
138			$items = $sibling->findMulti('li');
139			foreach ($items as $item) {
140			$extendedBy[] = $item->text();
141			}
142			break 2;
143			case 'table':
144			/** @var SimpleHtmlDomNodeInterface $table */
145			$table = $sibling->findOne('tbody')->findMulti('tr');
146			break 2;
147			}
148			}
149			return ['description' => $description, 'table' => $table, 'extended_by' => $extendedBy];
150			}
151
152			/**
153			* @return string
154			*/
155			private function parseVersion(): string
156			{
157			$element = $this->dom->findOne('h3');
158			$tag = '';
159			while ($tag != 'p' and !empty($element)) {
160			$element = $element->nextSibling();
161			$tag = $element?->tag;
162			}
163			if (empty($element)) {
164			return '1.0.0';
165			}
166			$versionNumbers = explode('.', str_replace('Bot API ', '', $element->text()));
167			return sprintf(
168			'%s.%s.%s',
169			$versionNumbers[0] ?? '1',
170			$versionNumbers[1] ?? '0',
171			$versionNumbers[2] ?? '0'
172			);
173			}
174
175			/**
176			* @return string
177			*/
178			public function getVersion(): string
179			{
180			return $this->version;
181			}
182
183			/**
184			* @return array{version: string, methods: array, types: array}
185			* @throws Exception
186			*/
187			public function extract(): array
188			{
189			$elements = $this->dom->findMultiOrFalse('h4');
190			if (false === $elements) {
191			throw new Exception('Unable to fetch required DOM nodes');
192			}
193			$data = ['version' => $this->version, 'methods' => [], 'types' => []];
194			foreach ($elements as $element) {
195			if (!str_contains($name = $element->text(), ' ')) {
196			$isMethod = lcfirst($name) == $name;
197			$path = $isMethod ? 'methods' : 'types';
198			['description' => $description, 'table' => $table, 'extended_by' => $extendedBy] = self::parseNode(
199			$element
200			);
201			$data[$path][] = self::generateElement(
202			$name,
203			trim($description),
204			$table,
205			$extendedBy,
206			$isMethod
207			);
208			}
209			}
210			return $data;
211			}
212
213			/**
214			* @param string $name
215			* @param string $description
216			* @param SimpleHtmlDomNodeInterface\|null $unparsedFields
217			* @param array $extendedBy
218			* @param bool $isMethod
219			* @return array
220			*/
221			private static function generateElement(
222			string $name,
223			string $description,
224			?SimpleHtmlDomNodeInterface $unparsedFields,
225			array $extendedBy,
226			bool $isMethod
227			): array {
228			$fields = self::parseFields($unparsedFields, $isMethod);
229			$result = [
230			'name' => $name,
231			'description' => htmlspecialchars_decode(strip_tags($description), ENT_QUOTES),
232			'fields' => $fields
233			];
234			if ($isMethod) {
235			$description = new ObjectDescription($description);
236			$returnTypes = $description->getTypes();
237			$result['return_types'] = $returnTypes;
238			return $result;
239			}
240			$result['extended_by'] = $extendedBy;
241			return $result;
242			}
243
244			/**
245			* @param SimpleHtmlDomNodeInterface\|null $fields
246			* @param bool $isMethod
247			* @return array
248			*/
249			private static function parseFields(?SimpleHtmlDomNodeInterface $fields, bool $isMethod): array
250			{
251			$parsedFields = [];
252			$fields ??= [];
253			/** @var SimpleHtmlDomInterface $field */
254			foreach ($fields as $field) {
255			/** @var SimpleHtmlDomNode $fieldData */
256			$fieldData = $field->findMulti('td');
257			$name = $fieldData[0]->text();
258			if (empty($name)) {
259			continue;
260			}
261			$types = $fieldData[1]->text();
262			if ($isMethod) {
263			$optional = $fieldData[2]->text() != 'Yes';
264			$description = $fieldData[3]->innerHtml();
265			} else {
266			$description = $fieldData[2]->innerHtml();
267			$optional = str_starts_with($fieldData[2]->text(), 'Optional.');
268			}
269			$field = new Field($name, $types, $optional, $description);
270			$parsedFields[] = $field->toArray();
271			}
272			return $parsedFields;
273			}
274			}
275

Sysbot-org / tgscraper

SchemaExtractor::parseFields() A last analyzed 2024-07-02 10:38 UTC

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like

SchemaExtractor::parseFields() A
last analyzed 2024-07-02 10:38 UTC