feat: html entities handling on iOS

Author: szydlovsky

ios/inputParser/InputParser.mm

@@ -3,6 +3,7 @@
 #import "StyleHeaders.h"
 #import "UIView+React.h"
 #import "TextInsertionUtils.h"
+#import "StringExtension.h"
 
 @implementation InputParser {
   EnrichedTextInputView *_input;
@@ -204,8 +205,8 @@ - (NSString *)parseToHtmlFromRange:(NSRange)range {
         [result appendString: [NSString stringWithFormat:@"<%@>", tagContent]];
       }
 
-      // append the letter
-      [result appendString:currentCharacterStr];
+      // append the letter and escape it if needed
+      [result appendString: [NSString stringByEscapingHtml:currentCharacterStr]];
 
       // save current styles for next character's checks
       previousActiveStyles = currentActiveStyles;
@@ -490,6 +491,7 @@ - (NSArray *)getTextAndStylesFromHtml:(NSString *)fixedHtml {
   BOOL closingTag = NO;
   NSMutableString *currentTagName = [[NSMutableString alloc] initWithString:@""];
   NSMutableString *currentTagParams = [[NSMutableString alloc] initWithString:@""];
+  NSDictionary *htmlEntitiesDict = [NSString getEscapedCharactersInfoFrom:fixedHtml];
 
   // firstly, extract text and initially processed tags
   for(int i = 0; i < fixedHtml.length; i++) {
@@ -511,7 +513,7 @@ - (NSArray *)getTextAndStylesFromHtml:(NSString *)fixedHtml {
       } else if(!closingTag) {
         // we finish opening tag - get its location and optionally params and put them under tag name key in ongoingTags
         NSMutableArray *tagArr = [[NSMutableArray alloc] init];
-        [tagArr addObject:[NSNumber numberWithInt:plainText.length]];
+        [tagArr addObject:[NSNumber numberWithInteger:plainText.length]];
         if(currentTagParams.length > 0) {
           [tagArr addObject:[currentTagParams copy]];
         }
@@ -550,8 +552,19 @@ - (NSArray *)getTextAndStylesFromHtml:(NSString *)fixedHtml {
       currentTagParams = [[NSMutableString alloc] initWithString:@""];
     } else {
       if(!insideTag) {
-        // no tags logic - just append text
-        [plainText appendString:currentCharacterStr];
+        // no tags logic - just append the right text
+        
+        // html entity on the index; use unescaped character and forward iterator accordingly
+        NSArray *entityInfo = htmlEntitiesDict[@(i)];
+        if(entityInfo != nullptr) {
+          NSString *escaped = entityInfo[0];
+          NSString *unescaped = entityInfo[1];
+          [plainText appendString:unescaped];
+          // the iterator will forward by 1 itself
+          i += escaped.length - 1;
+        } else {
+          [plainText appendString:currentCharacterStr];
+        }
       } else {
         if(gettingTagName) {
           if(currentCharacterChar == ' ') {

ios/inputTextView/InputTextView.mm

@@ -14,7 +14,7 @@ - (void)copy:(id)sender {
   NSString *plainText = [typedInput->textView.textStorage.string substringWithRange:typedInput->textView.selectedRange];
   NSString *fixedPlainText = [plainText stringByReplacingOccurrencesOfString:@"\u200B" withString:@""];
 
-  NSString *escapedHtml = [NSString stringByEscapingHtml:[typedInput->parser parseToHtmlFromRange:typedInput->textView.selectedRange]];
+  NSString *parsedHtml = [typedInput->parser parseToHtmlFromRange:typedInput->textView.selectedRange];
 
   NSMutableAttributedString *attrStr = [[typedInput->textView.textStorage attributedSubstringFromRange:typedInput->textView.selectedRange] mutableCopy];
   NSRange fullAttrStrRange = NSMakeRange(0, attrStr.length);
@@ -28,7 +28,7 @@ - (void)copy:(id)sender {
   UIPasteboard *pasteboard = [UIPasteboard generalPasteboard];
   [pasteboard setItems:@[@{
     UTTypeUTF8PlainText.identifier : fixedPlainText,
-    UTTypeHTML.identifier : escapedHtml,
+    UTTypeHTML.identifier : parsedHtml,
     UTTypeRTF.identifier : rtfData
   }]];
 }
@@ -53,9 +53,7 @@ - (void)paste:(id)sender {
       htmlString = htmlValue;
     }
 
-    // unescape the html
-    htmlString = [NSString stringByUnescapingHtml:htmlString];
-    // validate it
+    // validate the html
     NSString *initiallyProcessedHtml = [typedInput->parser initiallyProcessHtml:htmlString];
 
     if(initiallyProcessedHtml != nullptr) {

ios/utils/StringExtension.h

@@ -6,7 +6,7 @@
 - (std::string)toCppString;
 + (NSString *)fromCppString:(std::string)string;
 + (NSString *)stringByEscapingHtml:(NSString *)html;
-+ (NSString *)stringByUnescapingHtml:(NSString *)html;
++ (NSDictionary *)getEscapedCharactersInfoFrom:(NSString *)text;
 @end
 
 @interface NSMutableString (StringExtension)

ios/utils/StringExtension.mm

@@ -17,7 +17,8 @@ + (NSString *)stringByEscapingHtml:(NSString *)html {
     @"<": @"&lt;",
     @">": @"&gt;",
     @"\"": @"&quot;",
-    @"'": @"&apos;"
+    @"'": @"&apos;",
+    @"/": @"&sol;",
   };
 
   for(NSString *key in escapeMap) {
@@ -26,20 +27,35 @@ + (NSString *)stringByEscapingHtml:(NSString *)html {
   return escaped;
 }
 
-+ (NSString *)stringByUnescapingHtml:(NSString *)html {
-  NSMutableString *unescaped = [html mutableCopy];
++ (NSDictionary *)getEscapedCharactersInfoFrom:(NSString *)text {
   NSDictionary *unescapeMap = @{
     @"&amp;": @"&",
     @"&lt;": @"<",
     @"&gt;": @">",
     @"&quot;": @"\"",
     @"&apos;": @"'",
+    @"&sol;": @"/",
   };
 
+  NSMutableDictionary *results = [[NSMutableDictionary alloc] init];
+  
   for(NSString *key in unescapeMap) {
-    [unescaped replaceOccurrencesOfString:key withString:unescapeMap[key] options:NSLiteralSearch range:NSMakeRange(0, unescaped.length)];
+    NSRange searchRange = NSMakeRange(0, text.length);
+    NSRange foundRange;
+
+    while(searchRange.location < text.length) {
+      foundRange = [text rangeOfString:key options:0 range:searchRange];
+      if(foundRange.location != NSNotFound) {
+        results[@(foundRange.location)] = @[key, unescapeMap[key]];
+        searchRange.location = foundRange.location + foundRange.length;
+        searchRange.length = text.length - searchRange.location;
+      } else {
+        break;
+      }
+    }
   }
-  return unescaped;
+  
+  return results;
 }
 
 @end