Merge pull request #485 from wordpress-mobile/issue/482-format-crash

Issue/482 format crash

Author: khaykov

CHANGELOG.md

@@ -1,5 +1,9 @@
 # Changelog
 
+## [Unreleased]
+### Fixed
+- crash when HTML tags contained special regex characters ('$' or '\\')
+
 ## [1.0-beta.9](https://github.com/wordpress-mobile/AztecEditor-Android/releases/tag/v1.0-beta.9) - 2017-09-25
 ### Fixed
 - OOB crash when unmarshalling URL span

aztec/src/main/kotlin/org/wordpress/aztec/source/Format.kt

@@ -95,7 +95,7 @@ internal object Format {
             sb = StringBuffer()
             if (m.find()) {
                 val result = replaceAll(content.substring(m.start(), m.end()), "<br([^>]*)>", "<wp-temp-br$1>")
-                m.appendReplacement(sb, replace(result, "[\\r\\n\\t]+", ""))
+                m.appendReplacement(sb, Matcher.quoteReplacement(replace(result, "[\\r\\n\\t]+", "")))
             }
             m.appendTail(sb)
             content = sb.toString()
@@ -143,7 +143,7 @@ internal object Format {
             m = p.matcher(content)
             sb = StringBuffer()
             if (m.find()) {
-                m.appendReplacement(sb, replace(content.substring(m.start(), m.end()), "[\\r\\n]+", ""))
+                m.appendReplacement(sb, Matcher.quoteReplacement(replace(content.substring(m.start(), m.end()), "[\\r\\n]+", "")))
             }
             m.appendTail(sb)
             content = sb.toString()
@@ -198,7 +198,7 @@ internal object Format {
             m = p.matcher(html)
             sb = StringBuffer()
             while (m.find()) {
-                m.appendReplacement(sb, replaceAll(html.substring(m.start(), m.end()), "[\\r\\n]+", ""))
+                m.appendReplacement(sb, Matcher.quoteReplacement(replaceAll(html.substring(m.start(), m.end()), "[\\r\\n]+", "")))
             }
             m.appendTail(sb)
             html = sb.toString()
@@ -208,7 +208,7 @@ internal object Format {
         m = p.matcher(html)
         sb = StringBuffer()
         while (m.find()) {
-            m.appendReplacement(sb, replaceAll(html.substring(m.start(), m.end()), "[\\r\\n]+", ""))
+            m.appendReplacement(sb, Matcher.quoteReplacement(replaceAll(html.substring(m.start(), m.end()), "[\\r\\n]+", "")))
         }
         m.appendTail(sb)
         html = sb.toString()
@@ -231,21 +231,21 @@ internal object Format {
             sb = StringBuffer()
             while (m.find()) {
                 // keep existing <br>
-                m.appendReplacement(sb, replaceAll(html.substring(m.start(), m.end()), "<br([^>]*)>", "<wp-temp-br$1>"))
+                m.appendReplacement(sb, Matcher.quoteReplacement(replaceAll(html.substring(m.start(), m.end()), "<br([^>]*)>", "<wp-temp-br$1>")))
 
                 // no line breaks inside HTML tags
                 val p2 = Pattern.compile("<[a-zA-Z0-9]+( [^<>]+)?>")
                 val content = html.substring(m.start(), m.end())
                 val m2 = p2.matcher(content)
                 val sb2 = StringBuffer()
                 while (m2.find()) {
-                    m2.appendReplacement(sb2, replace(content.substring(m2.start(), m2.end()), "[\\r\\n\\t]+", " "))
+                    m2.appendReplacement(sb2, Matcher.quoteReplacement(replace(content.substring(m2.start(), m2.end()), "[\\r\\n\\t]+", " ")))
                 }
                 m2.appendTail(sb2)
-                m.appendReplacement(sb, sb2.toString())
+                m.appendReplacement(sb, Matcher.quoteReplacement(sb2.toString()))
 
                 // convert remaining line breaks to <br>
-                m.appendReplacement(sb, replaceAll(html.substring(m.start(), m.end()), "\\s*\\n\\s*", "<wp-temp-br />"))
+                m.appendReplacement(sb, Matcher.quoteReplacement(replaceAll(html.substring(m.start(), m.end()), "\\s*\\n\\s*", "<wp-temp-br />")))
             }
             m.appendTail(sb)
             html = sb.toString()

aztec/src/test/kotlin/org/wordpress/aztec/CalypsoFormattingTest.kt

@@ -76,6 +76,19 @@ class CalypsoFormattingTest : AndroidTestCase() {
                     "<div class=\"fourth\"><u>Under</u>line</div>\n" +
                     "<div class=\"fifth\"></div>\n</div>\n</div>"
 
+    private val HTML_MIXED_REGEX =
+            "\n\n<span><i>Italic</i></span>\n\n<b>Bold</b><br>" +
+                    "\t<div class=\"\$\$\$first\">" +
+                    "<a href=\"https://github.com/wordpress-mobile/Word\\Press-Aztec-Android\">Link</a>" +
+                    "    \t<div class=\"sec $8 ond\"></div></div>"
+
+    private val HTML_MIXED_REGEX_CALYPSO =
+            "<span><i>Italic</i></span>\n\n<b>Bold</b>\n" +
+                    "<div class=\"\$\$\$first\">" +
+                    "<a href=\"https://github.com/wordpress-mobile/Word\\Press-Aztec-Android\">Link</a>\n" +
+                    "<div class=\"sec $8 ond\"></div>\n" +
+                    "</div>"
+
     /**
      * Initialize variables.
      */
@@ -125,4 +138,18 @@ class CalypsoFormattingTest : AndroidTestCase() {
         val output = Format.addSourceEditorFormatting(parser.toHtml(span), true)
         Assert.assertEquals(HTML_MIXED_WITH_NEWLINES_CALYPSO, output)
     }
+
+    /**
+     * Test the conversion of HTML containing special regex characters
+     *
+     * @throws Exception
+     */
+    @Test
+    @Throws(Exception::class)
+    fun formatRegexSpecialCharactersCalypso() {
+        val input = Format.removeSourceEditorFormatting(HTML_MIXED_REGEX, true)
+        val span = SpannableString(parser.fromHtml(input, context))
+        val output = Format.addSourceEditorFormatting(parser.toHtml(span), true)
+        Assert.assertEquals(HTML_MIXED_REGEX_CALYPSO, output)
+    }
 }