Fault-tolerant Date parsing for dashes

Summary: This is addressing a gap between Hermes and V8 regarding date parsing, raised in facebook#796 . There is whitespace allowed between M D Y. However, we should also support allowing a dash to be present as the first character in those gaps.

Reviewed By: jpporto

Differential Revision: D40368889

fbshipit-source-id: 5a9529300a221609ebb5c59dc44c4cf97a153700

lib/VM/JSLib/DateUtil.cpp

@@ -901,7 +901,7 @@ static double parseESDate(StringView str) {
 
   /// Reads the next \p len characters into `tok`,
   /// but instead of consuming \p len chars, it consumes a single word
-  /// whatever how long it is (i.e. until a space is encountered).
+  /// whatever how long it is (i.e. until a space or dash is encountered).
   /// e.g.
   ///     &str ="Garbage G MayG"
   ///     scanStrAndSkipWord(3); consumeSpaces();  // &str="G MayG", &tok="Gar"
@@ -913,7 +913,7 @@ static double parseESDate(StringView str) {
     if (it + len > str.end())
       return false;
     tok = str.slice(it, it + len);
-    while (it != str.end() && !std::isspace(*it))
+    while (it != str.end() && !std::isspace(*it) && *it != '-')
       it++;
     return true;
   };
@@ -931,6 +931,21 @@ static double parseESDate(StringView str) {
       ++it;
   };
 
+  /// Only one dash is ever allowed, and the dash must come first.
+  /// There is no limit to the number of spaces.
+  /// This is in line with V8's behavior.
+  auto consumeSpacesOrDash = [&]() {
+    auto first = true;
+    while (it != str.end()) {
+      if (std::isspace(*it) || (first && *it == '-')) {
+        ++it;
+      } else {
+        return;
+      }
+      first = false;
+    }
+  };
+
   // Weekday
   if (!scanStr(3))
     return nan;
@@ -964,7 +979,7 @@ static double parseESDate(StringView str) {
       // Day
       scanInt(it, end, d);
       // Month
-      consumeSpaces();
+      consumeSpacesOrDash();
       // e.g. `Janwhatever` will get read as `Jan`
       if (!scanStrAndSkipWord(3))
         return nan;
@@ -980,7 +995,7 @@ static double parseESDate(StringView str) {
       // `tok` is now set to the Month candidate.
       if (tokIsMonth()) {
         // Day
-        consumeSpaces();
+        consumeSpacesOrDash();
         if (!scanInt(it, end, d))
           return nan;
         break;
@@ -993,12 +1008,12 @@ static double parseESDate(StringView str) {
   }
 
   // Year
-  consumeSpaces();
+  consumeSpacesOrDash();
   if (!scanInt(it, end, y))
     return nan;
 
   // Hour:minute:second.
-  consumeSpaces();
+  consumeSpacesOrDash();
 
   if (it != end) {
     if (!scanInt(it, end, h))

test/hermes/date-constructor.js

@@ -313,6 +313,25 @@ print(Date.parse('Tue May 05 2020 00:00:00'.padEnd(60)));  // trailing spaces.
 print(Date.parse('Tue May 05 2020'.padEnd(60)));
 // CHECK-NEXT: 1588662000000
 
+// Fault tolerance on dashes
+print(Date.parse("Friday, 02 Aug 2019 07:14:03 UTC"));
+// CHECK-NEXT: 1564730043000
+print(Date.parse("Friday, 02-Aug-2019-07:14:03 UTC"));
+// CHECK-NEXT: 1564730043000
+print(Date.parse("Friday, 02-  Aug-2019-07:14:03 UTC"));
+// CHECK-NEXT: 1564730043000
+print(Date.parse("Friday, 02-  Aug-   2019- 07:14:03 UTC"));
+// CHECK-NEXT: 1564730043000
+// Invalid use of dashes
+print(Date.parse("Friday, 02 -  Aug-2019-07:14:03 UTC"));
+// CHECK-NEXT: NaN
+print(Date.parse("Friday, 02-Aug -2019-07:14:03 UTC"));
+// CHECK-NEXT: NaN
+print(Date.parse("Friday, 02-Aug-2019-  -07:14:03 UTC"));
+// CHECK-NEXT: NaN
+print(Date.parse("Friday, 02-Aug-2019 -07:14:03 UTC"));
+// CHECK-NEXT: NaN
+
 // Quick check that getters work; internal functions are unit tested instead.
 print('getters');
 // CHECK-LABEL: getters