Differential D16085 Diff 38701 src/lint/linter/ArcanistTextLinter.php

Changeset View

Standalone View

src/lint/linter/ArcanistTextLinter.php

Show All 10 Lines	final class ArcanistTextLinter extends ArcanistLinter {
const LINT_EOF_NEWLINE = 4;		const LINT_EOF_NEWLINE = 4;
const LINT_BAD_CHARSET = 5;		const LINT_BAD_CHARSET = 5;
const LINT_TRAILING_WHITESPACE = 6;		const LINT_TRAILING_WHITESPACE = 6;
const LINT_BOF_WHITESPACE = 8;		const LINT_BOF_WHITESPACE = 8;
const LINT_EOF_WHITESPACE = 9;		const LINT_EOF_WHITESPACE = 9;
const LINT_EMPTY_FILE = 10;		const LINT_EMPTY_FILE = 10;

private $maxLineLength = 80;		private $maxLineLength = 80;
		private $allowUTF8 = false;

public function getInfoName() {		public function getInfoName() {
return pht('Basic Text Linter');		return pht('Basic Text Linter');
}		}

public function getInfoDescription() {		public function getInfoDescription() {
return pht(		return pht(
'Enforces basic text rules like line length, character encoding, '.		'Enforces basic text rules like line length, character encoding, '.
'and trailing whitespace.');		'and trailing whitespace.');
}		}

public function getLinterPriority() {		public function getLinterPriority() {
return 0.5;		return 0.5;
}		}

public function getLinterConfigurationOptions() {		public function getLinterConfigurationOptions() {
$options = array(		$options = array(
'text.max-line-length' => array(		'text.max-line-length' => array(
'type' => 'optional int',		'type' => 'optional int',
'help' => pht(		'help' => pht(
'Adjust the maximum line length before a warning is raised. By '.		'Adjust the maximum line length before a warning is raised. By '.
'default, a warning is raised on lines exceeding 80 characters.'),		'default, a warning is raised on lines exceeding 80 characters.'),
),		),
		'text.allow-utf8' => array(
		'type' => 'optional bool',
		'help' => pht(
		'Allow any valid UTF-8 character. By default, only ASCII bytes '.
		'with ordinal decimal values between 32 and 126 inclusive, plus '.
		'linefeed are allowed.'),
		),
);		);

return $options + parent::getLinterConfigurationOptions();		return $options + parent::getLinterConfigurationOptions();
}		}

public function setMaxLineLength($new_length) {		public function setMaxLineLength($new_length) {
$this->maxLineLength = $new_length;		$this->maxLineLength = $new_length;
return $this;		return $this;
}		}

public function setLinterConfigurationValue($key, $value) {		public function setLinterConfigurationValue($key, $value) {
switch ($key) {		switch ($key) {
case 'text.max-line-length':		case 'text.max-line-length':
$this->setMaxLineLength($value);		$this->setMaxLineLength($value);
return;		return;
		case 'text.allow-utf8':
		$this->allowUTF8 = $value;
		return;
}		}

return parent::setLinterConfigurationValue($key, $value);		return parent::setLinterConfigurationValue($key, $value);
}		}

public function getLinterName() {		public function getLinterName() {
return 'TXT';		return 'TXT';
}		}
▲ Show 20 Lines • Show All 79 Lines • ▼ Show 20 Lines	if (preg_match('/^\s*$/', $data)) {
self::LINT_EMPTY_FILE,		self::LINT_EMPTY_FILE,
pht("Empty files usually don't serve any useful purpose."));		pht("Empty files usually don't serve any useful purpose."));
$this->stopAllLinters();		$this->stopAllLinters();
}		}
}		}

protected function lintNewlines($path) {		protected function lintNewlines($path) {
$data = $this->getData($path);		$data = $this->getData($path);
		$allowUTF8 = $this->allowUTF8;
		Lint: Naming Conventions Inline Actions Follow naming conventions: variables should be named using `lowercase_with_underscores`. Lint: Naming Conventions: Follow naming conventions: variables should be named using `lowercase_with_underscores`.
$pos = strpos($this->getData($path), "\r");		$pos = strpos($this->getData($path), "\r");

if ($pos !== false) {		if ($pos !== false) {
$this->raiseLintAtOffset(		$this->raiseLintAtOffset(
0,		0,
self::LINT_DOS_NEWLINE,		self::LINT_DOS_NEWLINE,
pht('You must use ONLY Unix linebreaks ("%s") in source code.', '\n'),		pht('You must use ONLY Unix linebreaks ("%s") in source code.', '\n'),
$data,		$data,
▲ Show 20 Lines • Show All 45 Lines • ▼ Show 20 Lines	if (!strlen($data) \|\| $data[strlen($data) - 1] != "\n") {
pht('Files must end in a newline.'),		pht('Files must end in a newline.'),
'',		'',
"\n");		"\n");
}		}
}		}

protected function lintCharset($path) {		protected function lintCharset($path) {
$data = $this->getData($path);		$data = $this->getData($path);
		$allowUTF8 = $this->allowUTF8;
		Lint: Naming Conventions Inline Actions Follow naming conventions: variables should be named using `lowercase_with_underscores`. Lint: Naming Conventions: Follow naming conventions: variables should be named using `lowercase_with_underscores`.

$matches = null;		$matches = null;
$bad = '[^\x09\x0A\x20-\x7E]';		// Allow newline, tab, 32 to 126, Letters, Marks, Numbers,
		// Punctuation and Symbols
		$badUTF8 = '[^\x09\x0A\x20-\x7E\p{L}\p{M}\p{N}\p{P}\p{S}]';
		Lint: Naming Conventions Inline Actions Follow naming conventions: variables should be named using `lowercase_with_underscores`. Lint: Naming Conventions: Follow naming conventions: variables should be named using `lowercase_with_underscores`.
		$badUTF8Regex = "/{$badUTF8}(.*{$badUTF8})?/u";
		Lint: Naming Conventions Inline Actions Follow naming conventions: variables should be named using `lowercase_with_underscores`. Lint: Naming Conventions: Follow naming conventions: variables should be named using `lowercase_with_underscores`.

		$badAscii = '[^\x09\x0A\x20-\x7E]';
		Lint: Naming Conventions Inline Actions Follow naming conventions: variables should be named using `lowercase_with_underscores`. Lint: Naming Conventions: Follow naming conventions: variables should be named using `lowercase_with_underscores`.
		$badAsciiRegex = "/{$badAscii}(.*{$badAscii})?/";
		Lint: Naming Conventions Inline Actions Follow naming conventions: variables should be named using `lowercase_with_underscores`. Lint: Naming Conventions: Follow naming conventions: variables should be named using `lowercase_with_underscores`.

		$badRegex = ($allowUTF8 ? $badUTF8Regex : $badAsciiRegex);
		Lint: Naming Conventions Inline Actions Follow naming conventions: variables should be named using `lowercase_with_underscores`. Lint: Naming Conventions: Follow naming conventions: variables should be named using `lowercase_with_underscores`.
		Lint: Naming Conventions Inline Actions Follow naming conventions: variables should be named using `lowercase_with_underscores`. Lint: Naming Conventions: Follow naming conventions: variables should be named using `lowercase_with_underscores`.
		Lint: Naming Conventions Inline Actions Follow naming conventions: variables should be named using `lowercase_with_underscores`. Lint: Naming Conventions: Follow naming conventions: variables should be named using `lowercase_with_underscores`.
		Lint: Naming Conventions Inline Actions Follow naming conventions: variables should be named using `lowercase_with_underscores`. Lint: Naming Conventions: Follow naming conventions: variables should be named using `lowercase_with_underscores`.

$preg = preg_match_all(		$preg = preg_match_all(
"/{$bad}(.*{$bad})?/",		$badRegex,
		Lint: Naming Conventions Inline Actions Follow naming conventions: variables should be named using `lowercase_with_underscores`. Lint: Naming Conventions: Follow naming conventions: variables should be named using `lowercase_with_underscores`.
$data,		$data,
$matches,		$matches,
PREG_OFFSET_CAPTURE);		PREG_OFFSET_CAPTURE);

if (!$preg) {		if (!$preg) {
return;		return;
}		}

		$badAsciiHelpText = pht('Source code should contain only ASCII bytes with ordinal '.
		Lint: Naming Conventions Inline Actions Follow naming conventions: variables should be named using `lowercase_with_underscores`. Lint: Naming Conventions: Follow naming conventions: variables should be named using `lowercase_with_underscores`.
		Lint: Line Too Long Inline Actions This line is 88 characters long, but the convention is 80 characters. Lint: Line Too Long: This line is 88 characters long, but the convention is 80 characters.
		'decimal values between 32 and 126 inclusive, plus linefeed. '.
		'Do not use UTF-8 or other multibyte charsets.');

		$badUTF8HelpText = pht('Source code should contain only printable UTF-8 characters.');
		Lint: Naming Conventions Inline Actions Follow naming conventions: variables should be named using `lowercase_with_underscores`. Lint: Naming Conventions: Follow naming conventions: variables should be named using `lowercase_with_underscores`.
		Lint: Line Too Long Inline Actions This line is 90 characters long, but the convention is 80 characters. Lint: Line Too Long: This line is 90 characters long, but the convention is 80 characters.

		$helpText = ($allowUTF8 ? $badUTF8HelpText : $badAsciiHelpText);
		Lint: Naming Conventions Inline Actions Follow naming conventions: variables should be named using `lowercase_with_underscores`. Lint: Naming Conventions: Follow naming conventions: variables should be named using `lowercase_with_underscores`.
		Lint: Naming Conventions Inline Actions Follow naming conventions: variables should be named using `lowercase_with_underscores`. Lint: Naming Conventions: Follow naming conventions: variables should be named using `lowercase_with_underscores`.
		Lint: Naming Conventions Inline Actions Follow naming conventions: variables should be named using `lowercase_with_underscores`. Lint: Naming Conventions: Follow naming conventions: variables should be named using `lowercase_with_underscores`.
		Lint: Naming Conventions Inline Actions Follow naming conventions: variables should be named using `lowercase_with_underscores`. Lint: Naming Conventions: Follow naming conventions: variables should be named using `lowercase_with_underscores`.

foreach ($matches[0] as $match) {		foreach ($matches[0] as $match) {
list($string, $offset) = $match;		list($string, $offset) = $match;
$this->raiseLintAtOffset(		$this->raiseLintAtOffset(
$offset,		$offset,
self::LINT_BAD_CHARSET,		self::LINT_BAD_CHARSET,
pht(		$helpText,
		Lint: Naming Conventions Inline Actions Follow naming conventions: variables should be named using `lowercase_with_underscores`. Lint: Naming Conventions: Follow naming conventions: variables should be named using `lowercase_with_underscores`.
'Source code should contain only ASCII bytes with ordinal '.
'decimal values between 32 and 126 inclusive, plus linefeed. '.
'Do not use UTF-8 or other multibyte charsets.'),
$string);		$string);
}		}

if ($this->isMessageEnabled(self::LINT_BAD_CHARSET)) {		if ($this->isMessageEnabled(self::LINT_BAD_CHARSET)) {
$this->stopAllLinters();		$this->stopAllLinters();
}		}
}		}

▲ Show 20 Lines • Show All 79 Lines • Show Last 20 Lines