Bracket-Matching.md

@@ -116,6 +116,44 @@ after any value or Comma.
    errAfterLBrak = ScanTo(Advance(arrayStart, 1), anyToken) & ~ (nested | valueToken | RBrak)
 ```
 
+### Validating objects
+
+Assume we have two bit streams for tokens: `valueToken` (explained above) and `anyToken`, a bit stream marking the end position of any legal or illegal token. Assume also that `LBrace`, `RBrace`, `DQuote`, `Colon` and `Comma` are streams marking the position of JSON `{`, `}`, `"`, `:` and `,` tokens respectively.
+
+```
+   str = valueToken & DQuote
+   valueTokenMinusStr = valueToken ^ str
+
+   atDepth = bnc.EQ(ND, d)
+   nested = bnc.UGT(ND, d)
+   objStart = atDepth & LBrace
+   objEnd = ScanThru(objStart, nested | (atDepth & ~ (RBrak | RBrace)))
+   objAtEnd = objEnd & RBrak
+   objSpan = ExclusiveSpan(objStart, objEnd)
+   // Now validate that every value or nested item is followed 
+   // either by a comma or a the end RBrace.
+   afterNested = Advance(nested & objSpan, 1) & atDepth
+
+   // process all values that are not strings
+   afterTokenMinusStr = Advance(valueTokenMinusStr & objSpan, 1)
+   tokenNextMinusStr = ScanThru(afterNested | afterTokenMinusStr, whitespace)
+   errAfterValueMinusStr = tokenNextMinusStr &~(Comma | RBrace)
+  
+   // process strings as both key and value
+   afterTokenStr = Advance(str & objSpan, 1)
+   tokenNextStr = ScanThru(afterNested | afterTokenStr, whitespace)
+   errAfterValueStr = tokenNextStr &~(Comma | RBrace | Colon)
+
+   // join errors
+   errAfterValue = errAfterValueMinusStr | errAfterValueStr
+
+   //
+   // Every Comma must be followed by a value
+   errAfterComma = ScanTo(Advance(Comma & arraySpan, 1), anyToken) & ~ (nested | valueToken)
+   // After the LBrak we must have either a value or an RBrak.
+   errAfterLBrak = ScanTo(Advance(arrayStart, 1), anyToken) & ~ (nested | valueToken | RBrak)
+```
+
 # CFG Kernel (_note: this doc isn't ready_)
 
 The CFG kernel is a Parabix kernel (to be implemented) that can compute a context-free grammars given a set of rules with its own syntax (close to EBNF), a BixNum representing the nesting depth of its tokens (see NestingDepth kernel), and a StreamSet representing all valid markers.