bug #35 Handle multiple codes in sequences (jacklul)

fabpot · fabpot · commit f9a7bea58483 · 2025-03-15T09:53:30.000-07:00
This PR was squashed before being merged into the master branch. Discussion ---------- Handle multiple codes in sequences This PR adds support for handling multiple sequences of ansi codes - preserving the styling until reset code is found. This addresses #17 ### Current: ![sequences current](https://github.com/user-attachments/assets/9ba757d8-cc92-4e44-8a68-c5dfff22469e) ### With this PR: ![sequences pr](https://github.com/user-attachments/assets/fdc5fb5e-1d92-4683-8543-0673441b360b) Like in my other PR, PHPUnit passes once the test class is fixed. ## TODO: ~~ANSI codes 22-29 should disable appropriate styling, currently it will not work because of `in_array` checks on 1-9, might need to remove values directly in `tokenize` or rewrite the styling code into something else **I have an idea how to approach this, will update on that soon**~~ Done Commits ------- de5ed7a Handle style reset sequences 62572d1 Handle multiple codes in sequences
diff --git a/SensioLabs/AnsiConverter/AnsiToHtmlConverter.php b/SensioLabs/AnsiConverter/AnsiToHtmlConverter.php
@@ -101,7 +101,7 @@ protected function convertAnsiToColor($ansi)
         if ('0' != $ansi && '' != $ansi) {
             $options = explode(';', $ansi);
 
-            foreach ($options as $option) {
+            foreach ($options as $key => $option) {
                 if ($option >= 30 && $option < 38) {
                     $fg = $option - 30;
                 } elseif ($option >= 40 && $option < 48) {
@@ -115,6 +115,22 @@ protected function convertAnsiToColor($ansi)
                     $fg = 7;
                 } elseif (49 == $option) {
                     $bg = 0;
+                } elseif ($option >= 22 && $option < 30) { // 21 has varying effects, best to ignored it
+                    $unset = $option - 20;
+
+                    foreach ($options as $i => $v) {
+                        if ($v == $unset) {
+                            unset($options[$i]);
+                        }
+
+                        if (2 == $unset && 1 == $v) { // 22 also unsets bold
+                            unset($options[$i]);
+                        }
+
+                        if ($i >= $key) { // do not unset options after current position in parent loop
+                            break;
+                        }
+                    }
                 }
             }
 
@@ -162,14 +178,30 @@ protected function tokenize($text)
         $tokens = array();
         preg_match_all("/(?:\e\[(.*?)m|(\x08))/", $text, $matches, PREG_OFFSET_CAPTURE);
 
+        $codes = array();
         $offset = 0;
         foreach ($matches[0] as $i => $match) {
             if ($match[1] - $offset > 0) {
                 $tokens[] = array('text', substr($text, $offset, $match[1] - $offset));
             }
-            $tokens[] = array("\x08" == $match[0] ? 'backspace' : 'color', $matches[1][$i][0]);
+
+            foreach (explode(';', $matches[1][$i][0]) as $code) {
+                if ('0' == $code || '' == $code) {
+                    $codes = array();
+                } else {
+                    // remove existing occurrence to avoid processing duplicate styles
+                    if (in_array($code, $codes) && ($key = array_search($code, $codes)) !== false) {
+                        unset($codes[$key]);
+                    }
+                }
+
+                $codes[] = $code;
+            }
+
+            $tokens[] = array("\x08" == $match[0] ? 'backspace' : 'color', implode(';', $codes));
             $offset = $match[1] + strlen($match[0]);
         }
+
         if ($offset < strlen($text)) {
             $tokens[] = array('text', substr($text, $offset));
         }
diff --git a/SensioLabs/AnsiConverter/Tests/AnsiToHtmlConverterTest.php b/SensioLabs/AnsiConverter/Tests/AnsiToHtmlConverterTest.php
@@ -75,6 +75,9 @@ public function getConvertData()
 
             // non valid unicode codepoints substitution (only available with PHP >= 5.4)
             PHP_VERSION_ID < 50400 ?: array('<span style="background-color: black; color: white">foo '."\xEF\xBF\xBD".'</span>', "foo \xF4\xFF\xFF\xFF"),
+
+            // codes in sequence - remember enabled styling like bold, italic, etc. (until we hit a reset)
+            array('<span style="background-color: black; color: lightgreen">foo</span><span style="background-color: black; color: lightgreen">bar</span><span style="background-color: black; color: white">foo</span>', "\e[1;32mfoo\e[32mbar\e[mfoo"),
         );
     }
 }

Original file line number	Diff line number	Diff line change
`@@ -75,6 +75,9 @@ public function getConvertData()`
`75`	`75`
`76`	`76`	`// non valid unicode codepoints substitution (only available with PHP >= 5.4)`
`77`	`77`	`PHP_VERSION_ID < 50400 ?: array('<span style="background-color: black; color: white">foo '."\xEF\xBF\xBD".'</span>', "foo \xF4\xFF\xFF\xFF"),`
	`78`	`+`
	`79`	`+ // codes in sequence - remember enabled styling like bold, italic, etc. (until we hit a reset)`
	`80`	`+ array('<span style="background-color: black; color: lightgreen">foo</span><span style="background-color: black; color: lightgreen">bar</span><span style="background-color: black; color: white">foo</span>', "\e[1;32mfoo\e[32mbar\e[mfoo"),`
`78`	`81`	`);`
`79`	`82`	`}`
`80`	`83`	`}`