Node::checkSyntax() - Code Metrics - Inspection of "Fixed some bugs .." - johnnymast/rivescript-php - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — develop ( ec92a4...023af3 )

by Johnny

created 2022-01-15 15:19 UTC

Node::checkSyntax() D

↳ Parent: Node

Complexity

Conditions	39
Paths	30

Size

Total Lines	107
Code Lines	57

Duplication

Lines	0
Ratio	0 %

Importance

Changes	2
Bugs	0	Features	0

Metric	Value
cc	39
eloc	57
nc	30
nop	0
dl	0
loc	107
rs	4.1666
c	2
b	0
f	0

How to fix Long Method Complexity

<?php
/*
 * This file is part of Rivescript-php
 *
 * (c) Shea Lewis <[email protected]>
 *
 * For the full copyright and license information, please view the LICENSE
 * file that was distributed with this source code.
 */

namespace Axiom\Rivescript\Cortex;

/**
 * Node class
 *
 * The Node class stores information about a parsed
 * line from the script.
 *
 * PHP version 7.4 and higher.
 *
 * @category Core
 * @package  Cortext
 * @author   Shea Lewis <[email protected]>
 * @license  https://opensource.org/licenses/MIT MIT
 * @link     https://github.com/axiom-labs/rivescript-php
 * @since    0.3.0
 */
class Node
{
    /**
     * The source string for the node.
     *
     * @var string
     */
    protected string $source = '';

    /**
     * The line number of the node source.
     *
     * @var int
     */
    protected int $number = 0;

    /**
     * The command symbol.
     *
     * @var string
     */
    protected string $command = '';

    /**
     * The source without the command symbol.
     *
     * @var string
     */
    protected string $value = '';

    /**
     * Is this line part of a docblock.
     *
     * @var bool
     */
    protected bool $isInterrupted = false;

    /**
     * Is this node a comment.
     *
     * @var bool
     */
    protected bool $isComment = false;


    /**
     * Is this an empty line.
     *
     * @var bool
     */
    public bool $isEmpty = false;

    /**
     * Is UTF8 modes enabled.
     *
     * @var bool
     */
    protected bool $allowUtf8 = false;

    /**
     * Create a new Source instance.
     *
     * @param string $source The source line.
     * @param int    $number The line number in the script.
     */
    public function __construct(string $source, int $number)
    {
        $this->source = remove_whitespace($source);
        $this->number = $number;

        $this->determineEmpty();
        $this->determineComment();
        $this->determineCommand();
        $this->determineValue();
    }

    /**
     * Returns the node's command trigger.
     *
     * @return string
     */
    public function command(): string
    {
        return $this->command;
    }

    /**
     * Returns the node's value.
     *
     * @return string
     */
    public function value(): string
    {
        return $this->value;
    }

    /**
     * Returns the node's source.
     *
     * @return string
     */
    public function source(): string
    {
        return $this->source;
    }

    /**
     * Returns the node's line number.
     *
     * @return int
     */
    public function number(): int
    {
        return $this->number;
    }

    public function isEmpty(): bool
    {
        return $this->isEmpty;
    }

    /**
     * Returns true if node is a comment.
     *
     * @return bool
     */
    public function isComment(): bool
    {
        return $this->isComment;
    }

    /**
     * Returns true is node has been interrupted.
     *
     * @return bool
     */
    public function isInterrupted(): bool
    {
        return $this->isInterrupted;
    }

    /**
     * Determine the command type of the node.
     *
     * @return void
     */
    protected function determineCommand(): void
    {
        if ($this->source === '') {
            $this->isInterrupted = true;

            return;
        }

        $this->command = mb_substr($this->source, 0, 1);
    }

    protected function determineEmpty(): void
    {
        $this->isEmpty = empty(trim($this->source()));
    }

    /**
     * Determine if the current node source is a comment.
     *
     * @return void
     */
    protected function determineComment(): void
    {
        if (starts_with($this->source, '//')) {
            $this->isInterrupted = true;
        } elseif (starts_with($this->source, '#')) {
            log_warning('Using the # symbol for comments is deprecated');
            $this->isInterrupted = true;
        } elseif (starts_with($this->source, '/*')) {
            if (ends_with($this->source, '*/')) {
                return;
            }
            $this->isComment = true;
        } elseif (ends_with($this->source, '*/')) {
            $this->isComment = false;
        }
    }

    /**
     * Determine the value of the node.
     *
     * @return void
     */
    protected function determineValue(): void
    {
        $this->value = trim(mb_substr($this->source, 1));
    }

    /**
     * Enable the UTF8 mode.
     *
     * @param bool $allowUtf8 True of false.
     */
    public function setAllowUtf8(bool $allowUtf8): void
    {
        $this->allowUtf8 = $allowUtf8;
    }

    /**
     * Check the syntax
     *
     * @return string
     */
    public function checkSyntax(): string
    {
        if (starts_with($this->source, '!')) {
            # ! Definition
            #   - Must be formatted like this:
            #     ! type name = value
            #     OR
            #     ! type = value
            #   - Type options are NOT enforceable, for future compatibility; if RiveScript
            #     encounters a new type that it can't handle, it can safely warn and skip it.
            if ($this->matchesPattern("/^.+(?:\s+.+|)\s*=\s*.+?$/", $this->source) === false) {
                return "Invalid format for !Definition line: must be '! type name = value' OR '! type = value'";
            }
        } elseif (starts_with($this->source, '>')) {
            # > Label
            #   - The "begin" label must have only one argument ("begin")
            #   - "topic" labels must be lowercase but can inherit other topics ([A-Za-z0-9_\s])
            #   - "object" labels follow the same rules as "topic" labels, but don't need be lowercase
            if ($this->matchesPattern("/^begin/", $this->value) === true
                && $this->matchesPattern("/^begin$/", $this->value) === false) {
                return "The 'begin' label takes no additional arguments, should be verbatim '> begin'";
            } elseif ($this->matchesPattern("/^topic/", $this->value) === true
                && $this->matchesPattern("/[^a-z0-9_\-\s]/", $this->value) === true) {
                return "Topics should be lowercased and contain only numbers and letters!";
            } elseif ($this->matchesPattern("/^object/", $this->value) === true
                && $this->matchesPattern("/[^a-z0-9_\-\s]/", $this->value) === true) {
                return "Objects can only contain numbers and lowercase letters!";
            }
        } elseif (starts_with($this->source, '+') || starts_with($this->source, '%')
            || starts_with($this->source, '@')) {
            # + Trigger, % Previous, @ Redirect
            #   This one is strict. The triggers are to be run through Perl's regular expression
            #   engine. Therefore, it should be acceptable by the regexp engine.
            #   - Entirely lowercase
            #   - No symbols except: ( | ) [ ] * _ # @ { } < > =
            #   - All brackets should be matched

            if ($this->allowUtf8 === true) {
                if ($this->matchesPattern("/[A-Z\\.]/", $this->value) === true) {
                    return "Triggers can't contain uppercase letters, backslashes or dots in UTF-8 mode.";
                }
            } elseif ($this->matchesPattern("/[^a-z0-9(\|)\[\]*_#\@{}<>=\s]/", $this->value) === true) {
                return "Triggers may only contain lowercase letters, numbers, and these symbols: ( | ) [ ] * _ # @ { } < > =";
            }

            $parens = 0; # Open parenthesis
            $square = 0; # Open square brackets
            $curly = 0; # Open curly brackets
            $chevron = 0; # Open angled brackets
            $len = strlen($this->value);

            for ($i = 0; $i < $len; $i++) {
                $chr = $this->value[$i];

                # Count brackets.
                if ($chr === '(') {
                    $parens++;
                }
                if ($chr === ')') {
                    $parens--;
                }
                if ($chr === '[') {
                    $square++;
                }
                if ($chr === ']') {
                    $square--;
                }
                if ($chr === '{') {
                    $curly++;
                }
                if ($chr === '}') {
                    $curly--;
                }
                if ($chr === '<') {
                    $chevron++;
                }
                if ($chr === '>') {
                    $chevron--;
                }
            }

            if ($parens) {
                return "Unmatched " . ($parens > 0 ? "left" : "right") . " parenthesis bracket ()";
            }
            if ($square) {
                return "Unmatched " . ($square > 0 ? "left" : "right") . " square bracket []";
            }
            if ($curly) {
                return "Unmatched " . ($curly > 0 ? "left" : "right") . " curly bracket {}";
            }
            if ($chevron) {
                return "Unmatched " . ($chevron > 0 ? "left" : "right") . " angled bracket <>";
            }
        } elseif (starts_with($this->source, '-') || starts_with($this->source, '^')
            || starts_with($this->source, '/')) {
            # - Trigger, ^ Continue, / Comment
            # These commands take verbatim arguments, so their syntax is loose.
        } elseif (starts_with($this->source, '*') === true && $this->isComment() === false) {
            # * Condition
            #   Syntax for a conditional is as follows:
            #   * value symbol value => response
            if ($this->matchesPattern("/.+?\s(==|eq|!=|ne|<>|<|<=|>|>=)\s.+?=>.+?$/", $this->value) === false) {
                return "Invalid format for !Condition: should be like `* value symbol value => response`";
            }
        }

        return "";
    }

    /**
     * Check for patterns in a given string.
     *
     * @param string $regex  The pattern to detect.
     * @param string $string The string that could contain the pattern.
     *
     * @return bool
     */
    private function matchesPattern(string $regex = '', string $string = ''): bool
    {
        preg_match_all($regex, $string, $matches);

        return isset($matches[0][0]);
    }
}


1			<?php
2			/*
3			* This file is part of Rivescript-php
4			*
5			* (c) Shea Lewis <[email protected]>
6			*
7			* For the full copyright and license information, please view the LICENSE
8			* file that was distributed with this source code.
9			*/
10
11			namespace Axiom\Rivescript\Cortex;
12
13			/**
14			* Node class
15			*
16			* The Node class stores information about a parsed
17			* line from the script.
18			*
19			* PHP version 7.4 and higher.
20			*
21			* @category Core
22			* @package Cortext
23			* @author Shea Lewis <[email protected]>
24			* @license https://opensource.org/licenses/MIT MIT
25			* @link https://github.com/axiom-labs/rivescript-php
26			* @since 0.3.0
27			*/
28			class Node
29			{
30			/**
31			* The source string for the node.
32			*
33			* @var string
34			*/
35			protected string $source = '';
36
37			/**
38			* The line number of the node source.
39			*
40			* @var int
41			*/
42			protected int $number = 0;
43
44			/**
45			* The command symbol.
46			*
47			* @var string
48			*/
49			protected string $command = '';
50
51			/**
52			* The source without the command symbol.
53			*
54			* @var string
55			*/
56			protected string $value = '';
57
58			/**
59			* Is this line part of a docblock.
60			*
61			* @var bool
62			*/
63			protected bool $isInterrupted = false;
64
65			/**
66			* Is this node a comment.
67			*
68			* @var bool
69			*/
70			protected bool $isComment = false;
71
72
73			/**
74			* Is this an empty line.
75			*
76			* @var bool
77			*/
78			public bool $isEmpty = false;
79
80			/**
81			* Is UTF8 modes enabled.
82			*
83			* @var bool
84			*/
85			protected bool $allowUtf8 = false;
86
87			/**
88			* Create a new Source instance.
89			*
90			* @param string $source The source line.
91			* @param int $number The line number in the script.
92			*/
93			public function __construct(string $source, int $number)
94			{
95			$this->source = remove_whitespace($source);
96			$this->number = $number;
97
98			$this->determineEmpty();
99			$this->determineComment();
100			$this->determineCommand();
101			$this->determineValue();
102			}
103
104			/**
105			* Returns the node's command trigger.
106			*
107			* @return string
108			*/
109			public function command(): string
110			{
111			return $this->command;
112			}
113
114			/**
115			* Returns the node's value.
116			*
117			* @return string
118			*/
119			public function value(): string
120			{
121			return $this->value;
122			}
123
124			/**
125			* Returns the node's source.
126			*
127			* @return string
128			*/
129			public function source(): string
130			{
131			return $this->source;
132			}
133
134			/**
135			* Returns the node's line number.
136			*
137			* @return int
138			*/
139			public function number(): int
140			{
141			return $this->number;
142			}
143
144			public function isEmpty(): bool
145			{
146			return $this->isEmpty;
147			}
148
149			/**
150			* Returns true if node is a comment.
151			*
152			* @return bool
153			*/
154			public function isComment(): bool
155			{
156			return $this->isComment;
157			}
158
159			/**
160			* Returns true is node has been interrupted.
161			*
162			* @return bool
163			*/
164			public function isInterrupted(): bool
165			{
166			return $this->isInterrupted;
167			}
168
169			/**
170			* Determine the command type of the node.
171			*
172			* @return void
173			*/
174			protected function determineCommand(): void
175			{
176			if ($this->source === '') {
177			$this->isInterrupted = true;
178
179			return;
180			}
181
182			$this->command = mb_substr($this->source, 0, 1);
183			}
184
185			protected function determineEmpty(): void
186			{
187			$this->isEmpty = empty(trim($this->source()));
188			}
189
190			/**
191			* Determine if the current node source is a comment.
192			*
193			* @return void
194			*/
195			protected function determineComment(): void
196			{
197			if (starts_with($this->source, '//')) {
198			$this->isInterrupted = true;
199			} elseif (starts_with($this->source, '#')) {
200			log_warning('Using the # symbol for comments is deprecated');
201			$this->isInterrupted = true;
202			} elseif (starts_with($this->source, '/*')) {
203			if (ends_with($this->source, '*/')) {
204			return;
205			}
206			$this->isComment = true;
207			} elseif (ends_with($this->source, '*/')) {
208			$this->isComment = false;
209			}
210			}
211
212			/**
213			* Determine the value of the node.
214			*
215			* @return void
216			*/
217			protected function determineValue(): void
218			{
219			$this->value = trim(mb_substr($this->source, 1));
220			}
221
222			/**
223			* Enable the UTF8 mode.
224			*
225			* @param bool $allowUtf8 True of false.
226			*/
227			public function setAllowUtf8(bool $allowUtf8): void
228			{
229			$this->allowUtf8 = $allowUtf8;
230			}
231
232			/**
233			* Check the syntax
234			*
235			* @return string
236			*/
237			public function checkSyntax(): string
238			{
239			if (starts_with($this->source, '!')) {
240			# ! Definition
241			# - Must be formatted like this:
242			# ! type name = value
243			# OR
244			# ! type = value
245			# - Type options are NOT enforceable, for future compatibility; if RiveScript
246			# encounters a new type that it can't handle, it can safely warn and skip it.
247			if ($this->matchesPattern("/^.+(?:\s+.+\|)\s=\s.+?$/", $this->source) === false) {
248			return "Invalid format for !Definition line: must be '! type name = value' OR '! type = value'";
249			}
250			} elseif (starts_with($this->source, '>')) {
251			# > Label
252			# - The "begin" label must have only one argument ("begin")
253			# - "topic" labels must be lowercase but can inherit other topics ([A-Za-z0-9_\s])
254			# - "object" labels follow the same rules as "topic" labels, but don't need be lowercase
255			if ($this->matchesPattern("/^begin/", $this->value) === true
256			&& $this->matchesPattern("/^begin$/", $this->value) === false) {
257			return "The 'begin' label takes no additional arguments, should be verbatim '> begin'";
258			} elseif ($this->matchesPattern("/^topic/", $this->value) === true
259			&& $this->matchesPattern("/[^a-z0-9_\-\s]/", $this->value) === true) {
260			return "Topics should be lowercased and contain only numbers and letters!";
261			} elseif ($this->matchesPattern("/^object/", $this->value) === true
262			&& $this->matchesPattern("/[^a-z0-9_\-\s]/", $this->value) === true) {
263			return "Objects can only contain numbers and lowercase letters!";
264			}
265			} elseif (starts_with($this->source, '+') \|\| starts_with($this->source, '%')
266			\|\| starts_with($this->source, '@')) {
267			# + Trigger, % Previous, @ Redirect
268			# This one is strict. The triggers are to be run through Perl's regular expression
269			# engine. Therefore, it should be acceptable by the regexp engine.
270			# - Entirely lowercase
271			# - No symbols except: ( \| ) [ ] * _ # @ { } < > =
272			# - All brackets should be matched
273
274			if ($this->allowUtf8 === true) {
275			if ($this->matchesPattern("/[A-Z\\.]/", $this->value) === true) {
276			return "Triggers can't contain uppercase letters, backslashes or dots in UTF-8 mode.";
277			}
278			} elseif ($this->matchesPattern("/[^a-z0-9(\\|)\[\]*_#\@{}<>=\s]/", $this->value) === true) {
279			return "Triggers may only contain lowercase letters, numbers, and these symbols: ( \| ) [ ] * _ # @ { } < > =";
280			}
281
282			$parens = 0; # Open parenthesis
283			$square = 0; # Open square brackets
284			$curly = 0; # Open curly brackets
285			$chevron = 0; # Open angled brackets
286			$len = strlen($this->value);
287
288			for ($i = 0; $i < $len; $i++) {
289			$chr = $this->value[$i];
290
291			# Count brackets.
292			if ($chr === '(') {
293			$parens++;
294			}
295			if ($chr === ')') {
296			$parens--;
297			}
298			if ($chr === '[') {
299			$square++;
300			}
301			if ($chr === ']') {
302			$square--;
303			}
304			if ($chr === '{') {
305			$curly++;
306			}
307			if ($chr === '}') {
308			$curly--;
309			}
310			if ($chr === '<') {
311			$chevron++;
312			}
313			if ($chr === '>') {
314			$chevron--;
315			}
316			}
317
318			if ($parens) {
319			return "Unmatched " . ($parens > 0 ? "left" : "right") . " parenthesis bracket ()";
320			}
321			if ($square) {
322			return "Unmatched " . ($square > 0 ? "left" : "right") . " square bracket []";
323			}
324			if ($curly) {
325			return "Unmatched " . ($curly > 0 ? "left" : "right") . " curly bracket {}";
326			}
327			if ($chevron) {
328			return "Unmatched " . ($chevron > 0 ? "left" : "right") . " angled bracket <>";
329			}
330			} elseif (starts_with($this->source, '-') \|\| starts_with($this->source, '^')
331			\|\| starts_with($this->source, '/')) {
332			# - Trigger, ^ Continue, / Comment
333			# These commands take verbatim arguments, so their syntax is loose.
334			} elseif (starts_with($this->source, '*') === true && $this->isComment() === false) {
335			# * Condition
336			# Syntax for a conditional is as follows:
337			# * value symbol value => response
338			if ($this->matchesPattern("/.+?\s(==\|eq\|!=\|ne\|<>\|<\|<=\|>\|>=)\s.+?=>.+?$/", $this->value) === false) {
339			return "Invalid format for !Condition: should be like `* value symbol value => response`";
340			}
341			}
342
343			return "";
344			}
345
346			/**
347			* Check for patterns in a given string.
348			*
349			* @param string $regex The pattern to detect.
350			* @param string $string The string that could contain the pattern.
351			*
352			* @return bool
353			*/
354			private function matchesPattern(string $regex = '', string $string = ''): bool
355			{
356			preg_match_all($regex, $string, $matches);
357
358			return isset($matches[0][0]);
359			}
360			}
361

johnnymast / rivescript-php

Push — develop ( ec92a4...023af3 )

Node::checkSyntax() D

Complexity

Size

Duplication

Importance

How to fix Long Method Complexity

Long Method

Duplication Side-by-Side

Filter issues like