fix: "wordGlues" works wrongly under some circumstances (#25)

If we have a diffed circumstance like the following,

old: ... a-<del>B</del>-<del>C</del> ...
new: ... a-<ins>b</ins>- ...

we want to glue them into

old: ... a-<del>B-C</del> ...
new: ... a-<ins>b-</ins> ...

But instead, because of this bug, we have incorrect

old: ... a-<del>B-C</del> ...
new: ... a-<ins>b</ins>- ...

We can fix this by adding some dummy closures before glueing like

old: ... a-<del>B</del>-<del>C</del> ...
new: ... a-<ins>b</ins>-<ins></ins> ...
                        ^^^^^^^^^^^ dummy closure

so that now "old" and "new" have the same amounts of closures.

#25

Signed-off-by: Jack Cherng <jfcherng@gmail.com>

Author: jfcherng

src/Renderer/Html/AbstractHtml.php

@@ -119,7 +119,9 @@ public function getChanges(Differ $differ): array
      */
     protected function renderWorker(Differ $differ): string
     {
-        return $this->redererChanges($this->getChanges($differ));
+        $rendered = $this->redererChanges($this->getChanges($differ));
+
+        return $this->cleanUpDummyHtmlClosures($rendered);
     }
 
     /**
@@ -129,7 +131,9 @@ protected function renderArrayWorker(array $differArray): string
     {
         $this->ensureChangesUseIntTag($differArray);
 
-        return $this->redererChanges($differArray);
+        $rendered = $this->redererChanges($differArray);
+
+        return $this->cleanUpDummyHtmlClosures($rendered);
     }
 
     /**
@@ -317,4 +321,21 @@ protected function ensureChangesUseIntTag(array &$changes): void
             }
         }
     }
+
+    /**
+     * Clean up empty HTML closures in the given string.
+     *
+     * @param string $string the string
+     */
+    protected function cleanUpDummyHtmlClosures(string $string): string
+    {
+        return \str_replace(
+            [
+                RendererConstant::HTML_CLOSURES_DEL[0] . RendererConstant::HTML_CLOSURES_DEL[1],
+                RendererConstant::HTML_CLOSURES_INS[0] . RendererConstant::HTML_CLOSURES_INS[1],
+            ],
+            '',
+            $string
+        );
+    }
 }

src/Renderer/Html/LineRenderer/Word.php

@@ -20,6 +20,7 @@ final class Word extends AbstractLineRenderer
     public function render(MbString $mbOld, MbString $mbNew): LineRendererInterface
     {
         static $splitRegex = '/([' . RendererConstant::PUNCTUATIONS_RANGE . '])/uS';
+        static $dummyHtmlClosure = RendererConstant::HTML_CLOSURES[0] . RendererConstant::HTML_CLOSURES[1];
 
         // using PREG_SPLIT_NO_EMPTY will make "wordGlues" work wrongly under some rare cases
         // failure case:
@@ -33,18 +34,37 @@ public function render(MbString $mbOld, MbString $mbNew): LineRendererInterface
         $hunk = $this->getChangedExtentSegments($oldWords, $newWords);
 
         // reversely iterate hunk
+        $dummyDelIdxes = $dummyInsIdxes = [];
         foreach (ReverseIterator::fromArray($hunk) as [$op, $i1, $i2, $j1, $j2]) {
             if ($op & (SequenceMatcher::OP_REP | SequenceMatcher::OP_DEL)) {
                 $oldWords[$i1] = RendererConstant::HTML_CLOSURES[0] . $oldWords[$i1];
                 $oldWords[$i2 - 1] .= RendererConstant::HTML_CLOSURES[1];
+
+                if ($op === SequenceMatcher::OP_DEL) {
+                    $dummyInsIdxes[] = $j1;
+                }
             }
 
             if ($op & (SequenceMatcher::OP_REP | SequenceMatcher::OP_INS)) {
                 $newWords[$j1] = RendererConstant::HTML_CLOSURES[0] . $newWords[$j1];
                 $newWords[$j2 - 1] .= RendererConstant::HTML_CLOSURES[1];
+
+                if ($op === SequenceMatcher::OP_INS) {
+                    $dummyDelIdxes[] = $i1;
+                }
             }
         }
 
+        // insert dummy HTML closure to make sure there are always
+        // the same amounts of HTML closures in $oldWords and $newWords
+        // thus, this should ensure that "wordGlues" works correctly
+        foreach (ReverseIterator::fromArray($dummyDelIdxes) as $idx) {
+            \array_splice($oldWords, $idx, 0, [$dummyHtmlClosure]);
+        }
+        foreach (ReverseIterator::fromArray($dummyInsIdxes) as $idx) {
+            \array_splice($newWords, $idx, 0, [$dummyHtmlClosure]);
+        }
+
         if (!empty($hunk) && !empty($this->rendererOptions['wordGlues'])) {
             $regexGlues = \array_map(
                 function (string $glue): string {