Merge pull request #43 from bwnek/fix/improve-character-encoding

fix: improve char encoding
YetiForceCompany · Mar 14, 2024 · 30932ed · 30932ed
2 parents b561502 + 2bdeb9e
commit 30932ed
Show file tree

Hide file tree

Showing 2 changed files with 23 additions and 10 deletions.
diff --git a/lib/Document.php b/lib/Document.php
@@ -14,6 +14,8 @@
 
 namespace YetiForcePDF;
 
+use Exception;
+use YetiForcePDF\Html\Parser;
 use YetiForcePDF\Layout\FooterBox;
 use YetiForcePDF\Layout\HeaderBox;
 use YetiForcePDF\Layout\WatermarkBox;
@@ -95,7 +97,7 @@ class Document
 	protected $objects = [];
 
 	/**
-	 * @var \YetiForcePDF\Html\Parser
+	 * @var Parser
 	 */
 	protected $htmlParser;
 
@@ -786,14 +788,19 @@ public function removeObject(PdfObject $object): self
 	 * Load html string.
 	 *
 	 * @param string $html
-	 * @param string $inputEncoding
+	 * @param string $fromEncoding
 	 *
 	 * @return $this
+	 * @throws Exception
 	 */
-	public function loadHtml(string $html, string $inputEncoding = 'UTF-8')
+	public function loadHtml(string $html, string $fromEncoding = 'UTF-8'): self
 	{
-		$this->htmlParser = (new \YetiForcePDF\Html\Parser())->setDocument($this)->init();
-		$this->htmlParser->loadHtml($html, $inputEncoding);
+		if ($fromEncoding === '') {
+			throw new Exception('Encoding can not be empty');
+		}
+
+		$this->htmlParser = (new Parser())->setDocument($this)->init();
+		$this->htmlParser->loadHtml($html, $fromEncoding);
 
 		return $this;
 	}

diff --git a/lib/Html/Parser.php b/lib/Html/Parser.php
@@ -14,12 +14,13 @@
 
 namespace YetiForcePDF\Html;
 
+use YetiForcePDF\Base;
 use YetiForcePDF\Layout\PageGroupBox;
 
 /**
  * Class Parser.
  */
-class Parser extends \YetiForcePDF\Base
+class Parser extends Base
 {
 	/**
 	 * @var \DOMDocument
@@ -61,13 +62,18 @@ protected function cleanUpHtml(string $html)
 	 * @param string $html
 	 * @param string $fromEncoding
 	 *
-	 * @return \YetiForcePDF\Html\Parser
+	 * @return Parser
 	 */
-	public function loadHtml(string $html, string $fromEncoding = ''): self
+	public function loadHtml(string $html, string $fromEncoding): self
 	{
-		$html = htmlspecialchars_decode($html, ENT_HTML5);
+		$this->html = htmlspecialchars_decode($html, ENT_HTML5);
 		$this->html = $this->cleanUpHtml($html);
-		$this->html = html_entity_decode($this->html, ENT_COMPAT, $fromEncoding);
+
+		// 0x80 - start of unicode range
+		// 0x10FFFF - end of unicode range
+		// 0 - do not ommit any unicode char
+		// ~0 - negated 0 - convert negation of nothing (so convert all)
+		$this->html = mb_encode_numericentity($this->html, [0x80, 0x10FFFF, 0, ~0], $fromEncoding);
 
 		return $this;
 	}