From a188d21e41c07b7a25bb15be3597dc497cdef40b Mon Sep 17 00:00:00 2001
From: Matthew Phillips <matt@mattp.name>
Date: Thu, 16 Apr 2020 16:18:53 +0930
Subject: [PATCH] Add '<>:#' as valid characters in symbols

---
 MPEdn/MPEdnParser.m           | 22 +++++++++++++++-------
 MPEdnTests/MPEdnParserTests.m |  4 +++-
 2 files changed, 18 insertions(+), 8 deletions(-)

diff --git a/MPEdn/MPEdnParser.m b/MPEdn/MPEdnParser.m
index 6ad0149..8953fea 100644
--- a/MPEdn/MPEdnParser.m
+++ b/MPEdn/MPEdnParser.m
@@ -179,14 +179,22 @@ - (void) addTagReader: (id<MPEdnTaggedValueReader>) reader
 
 #pragma mark - Tokeniser
 
-static BOOL is_sym_punct (unichar ch)
+/// is this a character that may begin a symbol?
+static BOOL isStartSymPunct (unichar ch)
 {
   // NB: '/' is added to the specified EDN set of punctuation allowed in
   // symbol names. The check for its valid use will be done by the parser.
   return (ch >= '!' && ch <= '_' &&
           (ch == '!' || ch == '$' || ch == '%' || ch == '&' || ch == '*' ||
            ch == '+' || ch == '-' || ch == '.' || ch == '=' || ch == '?' ||
-           ch == '_' || ch == '/' || ch == ':' || ch == '#'));
+           ch == '_' || ch == '/' || ch == ':' || ch == '#' || ch == '<' ||
+           ch == '>'));
+}
+
+/// is this a character that may be in a symbol?
+static BOOL isSymPunct (unichar ch)
+{
+  return isStartSymPunct (ch) || ch == ':' || ch == '#';
 }
 
 - (unichar) currentEndIdxChar
@@ -326,7 +334,7 @@ - (void) nextToken
       {
         [self readTagName];
       }
-    } else if (isalpha (ch) || is_sym_punct (ch))
+    } else if (isalpha (ch) || isStartSymPunct (ch))
     {
       [self readNameToken];
     } else
@@ -507,7 +515,7 @@ - (void) readNameToken
   do
   {
     ch = [self advanceEndIdx];
-  } while (isalnum (ch) || is_sym_punct (ch));
+  } while (isalnum (ch) || isSymPunct (ch));
   
   token = TOKEN_NAME;
   tokenValue =
@@ -521,7 +529,7 @@ - (void) readKeywordToken
   do
   {
     ch = [self advanceEndIdx];
-  } while (isalnum (ch) || is_sym_punct (ch));
+  } while (isalnum (ch) || isSymPunct (ch));
   
   NSInteger length = endIdx - startIdx - 1;
   
@@ -627,8 +635,8 @@ - (void) readTagName
   do
   {
     ch = [self advanceEndIdx];
-  } while (isalnum (ch) || is_sym_punct (ch));
-  
+  } while (isalnum (ch) || isStartSymPunct (ch));
+
   NSInteger tagLen = endIdx - startIdx - 1;
   
   if (tagLen > 0)
diff --git a/MPEdnTests/MPEdnParserTests.m b/MPEdnTests/MPEdnParserTests.m
index f2e033e..3c61401 100644
--- a/MPEdnTests/MPEdnParserTests.m
+++ b/MPEdnTests/MPEdnParserTests.m
@@ -121,11 +121,13 @@ - (void) testSymbols
   
   MPAssertParseOK (@"+", [MPEdnSymbol symbolWithName: @"+"], @"Symbol");
   MPAssertParseOK (@"-", [MPEdnSymbol symbolWithName: @"-"], @"Symbol");
+  MPAssertParseOK (@"<", [MPEdnSymbol symbolWithName: @"<"], @"Symbol");
+  MPAssertParseOK (@"<:>", [MPEdnSymbol symbolWithName: @"<:>"], @"Symbol");
   MPAssertParseOK (@"+a", [MPEdnSymbol symbolWithName: @"+a"], @"Symbol");
   MPAssertParseOK (@"?", [MPEdnSymbol symbolWithName: @"?"], @"Symbol");
   MPAssertParseOK (@"?a", [MPEdnSymbol symbolWithName: @"?a"], @"Symbol");
   MPAssertParseOK (@"a?", [MPEdnSymbol symbolWithName: @"a?"], @"Symbol");
-  
+
   MPAssertParseError (@"}", @"Not a symbol");
   MPAssertParseError (@"]", @"Not a symbol");
   MPAssertParseError (@")", @"Not a symbol");