VerseRef.TrySetVerseUnicode: Improve handling of non-decimal numerals…

… and surrogate pair numerals (sillsdev#1000)
elisunger · Jan 4, 2022 · 3a829c5 · 3a829c5
1 parent 9823d1d
commit 3a829c5
Show file tree

Hide file tree

Showing 3 changed files with 57 additions and 13 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -29,6 +29,7 @@ and this project adheres to [Semantic Versioning](http://semver.org/).
 - [SIL.WritingSystems] Update langtags.json to the latest
 - [SIL.Scripture] Made VerseRef class implement new IScrVerseRef interface
 - [SIL.Forms.Scripture] Changed VerseControl to use IScrVerseRef and not depend directly on ScrVers
+- [SIL.Scripture] VerseRef.TrySetVerseUnicode: Improve handling of non-decimal numerals and surrogate pair numerals (#1000)
 
 ### Fixed
 

diff --git a/SIL.Scripture.Tests/VerseRefTests.cs b/SIL.Scripture.Tests/VerseRefTests.cs
@@ -2076,9 +2076,11 @@ public void UnBridge()
  [TestCase("ะ", ExpectedResult = -1, TestName = "Thai non-numeral")]
  [TestCase("᠔-᠔", ExpectedResult = 4, TestName = "Mongolian complex verse")]
  [TestCase("᠔ᠠ", ExpectedResult = 4, TestName = "Mongolian complex verse - lettered")]
- [TestCase("二十", ExpectedResult = 20, TestName = "Japanese numeral", IgnoreReason = "Non-decimal numeral systems not yet implemented. (See issue #1000.)")]
- [TestCase("יא", ExpectedResult = 11, TestName = "Hebrew numeral", IgnoreReason = "Non-decimal numeral systems not yet implemented. (See issue #1000.)")]
- [TestCase("\U0001113A\U00011138", ExpectedResult = 42, TestName = "Chakma numeral", IgnoreReason = "Surrogate pair handling not yet implemented. (See issue #1000.)")]
+ [TestCase("二十", ExpectedResult = 20, TestName = "Japanese numeral", IgnoreReason = "Non-decimal numeral systems not yet implemented. See issue #1000")]
+ [TestCase("יא", ExpectedResult = 11, TestName = "Hebrew numeral", IgnoreReason = "Non-decimal numeral systems not yet implemented. See issue #1000")]
+ [TestCase("\U0001113A\U00011138", ExpectedResult = 42, TestName = "Chakma numeral")]
+ [TestCase("Ⅻ", ExpectedResult = 12, TestName = "Roman numeral LetterNumber")]
+ [TestCase("፹", ExpectedResult = 80, TestName = "Ethiopic numeral OtherNumber")]
  public int TrySetVerseUnicode_InterpretNumerals(string verseStr)
  {
  VerseRef vref = new VerseRef("EXO 6:1");

diff --git a/SIL.Scripture/VerseRef.cs b/SIL.Scripture/VerseRef.cs
@@ -5,6 +5,7 @@
 using System.Diagnostics;
 using System.Text;
 using JetBrains.Annotations;
+using System.Globalization;
 
 namespace SIL.Scripture
 {
@@ -316,19 +317,59 @@ static bool TryGetVerseNum(string verseStr, bool romanOnly , out short vNum)
  for (int i = 0; i < verseStr.Length; i++)
  {
  char ch = verseStr[i];
- if (romanOnly ? (ch < '0' || ch > '9') : !char.IsDigit(ch))
+ if (!char.IsSurrogate(ch))
  {
- if (i == 0)
- vNum = -1;
- return false;
- }
+ switch (char.GetUnicodeCategory(ch))
+ {
+ case UnicodeCategory.DecimalDigitNumber:
+ if (romanOnly ? (ch < '0' || ch > '9') : !char.IsDigit(ch))
+ {
+ if (i == 0)
+ vNum = -1;
+ return false;
+ }
 
- vNum = (short)(vNum * 10 + (romanOnly ? ch - '0' : char.GetNumericValue(ch)));
- if (vNum > bcvMaxValue)
+ vNum = (short)(vNum * 10 + (romanOnly ? ch - '0' : char.GetNumericValue(ch)));
+ if (vNum > bcvMaxValue)
+ {
+ // whoops, we got too big!
+ vNum = -1;
+ return false;
+ }
+
+ break;
+ case UnicodeCategory.Surrogate:
+ vNum = (short)(vNum * 10 + (char.GetNumericValue(verseStr, i)));
+
+ if (vNum > bcvMaxValue)
+ {
+ // whoops, we got too big!
+ vNum = -1;
+ return false;
+ }
+ break;
+ case UnicodeCategory.LetterNumber:
+ case UnicodeCategory.OtherNumber:
+ vNum += (short)char.GetNumericValue(verseStr, i);
+ break;
+ default:
+ if (i == 0)
+ vNum = -1;
+ return false;
+ }
+
+ }
+ else if (i + 1 < verseStr.Length && char.IsSurrogatePair(verseStr[i], verseStr[i + 1]))
  {
- // whoops, we got too big!
- vNum = -1;
- return false;
+ vNum = (short)(vNum * 10 + (char.GetNumericValue(verseStr, i)));
+
+ if (vNum > bcvMaxValue)
+ {
+ // whoops, we got too big!
+ vNum = -1;
+ return false;
+ }
+
  }
  }
  return true;