Merge branch 'release/v1.03'

README.md

@@ -15,7 +15,7 @@ Currently supported backends:
 
 With [Composer](http://getcomposer.org/):
 
-    $ composer require mekras/php-speller:~1.00
+    $ composer require mekras/php-speller:~1.03
 
 ## Usage
 
@@ -64,3 +64,25 @@ $speller = new Hunspell();
 $speller->setDictionaryPath('/my_app/spelling');
 $speller->setCustomDictionaries(['tech', 'titles']);
 ```
+
+## Sources
+
+Sources — is an abstraction layer allowing spellers receive text from different sources like strings
+or files.
+
+Supported sources:
+
+* [StringSource](src/Source/StringSource.php) — simple PHP string;
+* [FileSource](src/Source/FileSource.php) — generic file source;
+* [XliffSource](src/Source/XliffSource.php) —
+  [XLIFF](http://docs.oasis-open.org/xliff/xliff-core/v2.0/xliff-core-v2.0.html) files.
+
+## Filters
+
+Filters used internally to filter out all non text contents received from source. In order to save
+original word location (line and column numbers) all filters replaces non text content with spaces.
+
+Available filters:
+
+* [StripAllFilter](src/Source/Filter/StripAllFilter.php) — strips all input text;
+* [HtmlFilter](src/Source/Filter/HtmlFilter.php) — strips HTML tags.

VERSION

@@ -1 +1 @@
-1.02
+1.03

src/Source/Filter/HtmlFilter.php

@@ -0,0 +1,135 @@
+<?php
+/**
+ * PHP Speller
+ *
+ * @copyright 2015, Михаил Красильников <m.krasilnikov@yandex.ru>
+ * @author    Михаил Красильников <m.krasilnikov@yandex.ru>
+ * @license   http://opensource.org/licenses/MIT MIT
+ */
+namespace Mekras\Speller\Source\Filter;
+
+/**
+ * Filter replaces HTML tags with spaces
+ *
+ * @since 1.03
+ */
+class HtmlFilter implements Filter
+{
+    /**
+     * Attrs with text contents
+     *
+     * @var string[]
+     */
+    private $textAttrs = ['title'];
+
+    /**
+     * Filter string
+     *
+     * @param string $string string to be filtered
+     *
+     * @return string filtered string
+     *
+     * @since 1.03
+     */
+    public function filter($string)
+    {
+        $result = '';
+
+        // Current/last tag name
+        $tagName = null;
+        // Current/last attribute name
+        $attrName = null;
+        // Current context
+        $context = null;
+        // Expected context
+        $expecting = null;
+
+        $length = mb_strlen($string);
+        for ($i = 0; $i < $length; $i++) {
+            $ch = mb_substr($string, $i, 1);
+            switch ($ch) {
+
+                case '<':
+                    $context = 'tag_name';
+                    $tagName = null;
+                    $ch = ' ';
+                    break;
+
+                case '>':
+                    $context = null;
+                    $expecting = null;
+                    $ch = ' ';
+                    break;
+
+                case ' ':
+                case "\n":
+                case "\t":
+                    switch ($context) {
+
+                        case 'tag_name':
+                            $context = 'tag_attrs';
+                            break;
+
+                        case 'attr_name':
+                            $context = 'tag_attrs';
+                            break;
+                    }
+                    break;
+
+                case '=':
+                    if ('attr_name' === $context || 'tag_attrs' === $context) {
+                        $expecting = 'attr_value';
+                        $ch = ' ';
+                    }
+                    break;
+
+                case '"':
+                case "'":
+                    switch (true) {
+
+                        case 'attr_value' === $expecting:
+                            $context = 'attr_value';
+                            if (in_array(strtolower($attrName), $this->textAttrs, true)) {
+                                $context = 'attr_text';
+                            }
+                            $expecting = null;
+                            $ch = ' ';
+                            break;
+
+                        case 'attr_value' === $context:
+                        case 'attr_text' === $context:
+                            $context = 'tag_attrs';
+                            $ch = ' ';
+                            break;
+                    }
+                    break;
+
+                default:
+                    switch ($context) {
+
+                        case 'tag_name':
+                            $tagName .= $ch;
+                            $ch = ' ';
+                            break;
+
+                        /** @noinspection PhpMissingBreakStatementInspection */
+                        case 'tag_attrs':
+                            $context = 'attr_name';
+                            $attrName = null;
+                            // no break needed
+                        case 'attr_name':
+                            $attrName .= $ch;
+                            $ch = ' ';
+                            break;
+
+                        case 'attr_value':
+                            $ch = ' ';
+                            break;
+                    }
+            }
+            $result .= $ch;
+        }
+
+        return $result;
+    }
+}

src/Source/XliffSource.php

@@ -9,6 +9,7 @@
 namespace Mekras\Speller\Source;
 
 use Mekras\Speller\Source\Filter\Filter;
+use Mekras\Speller\Source\Filter\HtmlFilter;
 use Mekras\Speller\Source\Filter\StripAllFilter;
 
 /**
@@ -57,20 +58,14 @@ public function getAsString()
         $text = parent::getAsString();
 
         $stripAll = new StripAllFilter();
+        $htmlFilter = new HtmlFilter();
 
         /* Removing CDATA tags */
         $text = preg_replace_callback(
             '#<!\[CDATA\[(.*?)\]\]>#ums',
-            function ($match) use ($stripAll) {
-                $string = $match[1];
-                $string = preg_replace_callback(
-                    '#(<[^>]+>)([^<]*)(</[^>]+>)#',
-                    function ($match) use ($stripAll) {
-                        return $stripAll->filter($match[1]) . $match[2]
-                        . $stripAll->filter($match[3]);
-                    },
-                    $string
-                );
+            function ($match) use ($htmlFilter) {
+                $string = $htmlFilter->filter($match[1]);
+                //      <![CDATA[               ]]
                 return '         ' . $string . '  ';
             },
             $text

tests/Source/Filter/HtmlFilterTest.php

@@ -0,0 +1,31 @@
+<?php
+/**
+ * PHP Speller
+ *
+ * @copyright 2015, Михаил Красильников <m.krasilnikov@yandex.ru>
+ * @author Михаил Красильников <m.krasilnikov@yandex.ru>
+ * @license http://opensource.org/licenses/MIT MIT
+ */
+namespace Mekras\Speller\Tests\Source\Filter;
+
+use Mekras\Speller\Source\Filter\HtmlFilter;
+use PHPUnit_Framework_TestCase as TestCase;
+
+/**
+ * Tests for Mekras\Speller\Source\Filter\HtmlFilter
+ *
+ * @covers Mekras\Speller\Source\Filter\HtmlFilter
+ */
+class HtmlFilterTest extends TestCase
+{
+    /**
+     * Test basic functional
+     */
+    public function testBasics()
+    {
+        $filter = new HtmlFilter();
+        $html = "<br>foo <a\nhref = '#' title='bar'>\nbaz</a>";
+        $text = "    foo   \n                  bar  \nbaz    ";
+        static::assertEquals($text, $filter->filter($html));
+    }
+}