Merge pull request #59 from PHPCSStandards/feature/new-orthography-class

New Utils\Orthography class

Author: jrfnl

PHPCSUtils/Utils/Orthography.php

@@ -0,0 +1,121 @@
+<?php
+/**
+ * PHPCSUtils, utility functions and classes for PHP_CodeSniffer sniff developers.
+ *
+ * @package   PHPCSUtils
+ * @copyright 2019 PHPCSUtils Contributors
+ * @license   https://opensource.org/licenses/LGPL-3.0 LGPL3
+ * @link      https://github.com/PHPCSStandards/PHPCSUtils
+ */
+
+namespace PHPCSUtils\Utils;
+
+use PHPCSUtils\BackCompat\Helper;
+
+/**
+ * Utility functions for checking the orthography of arbitrary text strings.
+ *
+ * > An orthography is a set of conventions for writing a language. It includes norms of spelling,
+ * > hyphenation, capitalization, word breaks, emphasis, and punctuation.
+ * > Source: https://en.wikipedia.org/wiki/Orthography
+ *
+ * @since 1.0.0
+ */
+class Orthography
+{
+
+    /**
+     * Characters which are considered terminal points for a sentence.
+     *
+     * @link https://www.thepunctuationguide.com/terminal-points.html
+     *
+     * @since 1.0.0
+     *
+     * @var string
+     */
+    const TERMINAL_POINTS = '.?!';
+
+    /**
+     * Check if the first character of an arbitrary text string is a capital letter.
+     *
+     * Letter characters which do not have a concept of lower/uppercase will
+     * be accepted as correctly capitalized.
+     *
+     * @since 1.0.0
+     *
+     * @param string $string The text string to examine.
+     *                       This can be the contents of a text string token,
+     *                       but also, for instance, a comment text.
+     *                       Potential text delimiter quotes should be stripped
+     *                       off a text string before passing it to this method.
+     *
+     * @return bool True when the first character is a capital letter or a letter
+     *              which doesn't have a concept of capitalization.
+     *              False otherwise, including for non-letter characters.
+     */
+    public static function isFirstCharCapitalized($string)
+    {
+        $string = \ltrim($string);
+        return (\preg_match('`^[\p{Lu}\p{Lt}\p{Lo}]`u', $string) > 0);
+    }
+
+    /**
+     * Check if the first character of an arbitrary text string is a lowercase letter.
+     *
+     * @since 1.0.0
+     *
+     * @param string $string The text string to examine.
+     *                       This can be the contents of a text string token,
+     *                       but also, for instance, a comment text.
+     *                       Potential text delimiter quotes should be stripped
+     *                       off a text string before passing it to this method.
+     *
+     * @return bool True when the first character is a lowercase letter.
+     *              False otherwise, including for letters which don't have a concept of
+     *              capitalization and for non-letter characters.
+     */
+    public static function isFirstCharLowercase($string)
+    {
+        $string = \ltrim($string);
+        return (\preg_match('`^\p{Ll}`u', $string) > 0);
+    }
+
+    /**
+     * Check if the last character of an arbitrary text string is a valid punctuation character.
+     *
+     * @since 1.0.0
+     *
+     * @param string $string       The text string to examine.
+     *                             This can be the contents of a text string token,
+     *                             but also, for instance, a comment text.
+     *                             Potential text delimiter quotes should be stripped
+     *                             off a text string before passing it to this method.
+     * @param string $allowedChars Characters which are considered valid punctuation
+     *                             to end the text string.
+     *                             Defaults to '.?!', i.e. a full stop, question mark
+     *                             or exclamation mark.
+     *
+     * @return bool
+     */
+    public static function isLastCharPunctuation($string, $allowedChars = self::TERMINAL_POINTS)
+    {
+        static $encoding;
+
+        if (isset($encoding) === false) {
+            $encoding = Helper::getConfigData('encoding');
+        }
+
+        $string = \rtrim($string);
+        if (\function_exists('iconv_substr') === true) {
+            $lastChar = \iconv_substr($string, -1, 1, $encoding);
+        } else {
+            $lastChar = \substr($string, -1);
+        }
+
+        if (\function_exists('iconv_strpos') === true) {
+            return (\iconv_strpos($allowedChars, $lastChar, 0, $encoding) !== false);
+        } else {
+            return (\strpos($allowedChars, $lastChar) !== false);
+        }
+    }
+}