From 28f9cfa1f505cdc5d44c63e3fdd8335218caf397 Mon Sep 17 00:00:00 2001
From: Alberto Restifo <alberto@restifo.dev>
Date: Wed, 22 Apr 2020 09:33:46 +0200
Subject: [PATCH] Add support for string streaming

---
 spec/rfc_spec.cr              |  2 +-
 src/cbor.cr                   |  3 +-
 src/cbor/diagnostic.cr        |  2 ++
 src/cbor/lexer.cr             | 58 +++++++++++++++++++++++++++++++----
 src/cbor/token.cr             |  2 +-
 src/cbor/type/string_array.cr | 13 ++++++++
 6 files changed, 70 insertions(+), 10 deletions(-)
 create mode 100644 src/cbor/type/string_array.cr

diff --git a/spec/rfc_spec.cr b/spec/rfc_spec.cr
index 3dccb55..0755e24 100644
--- a/spec/rfc_spec.cr
+++ b/spec/rfc_spec.cr
@@ -73,7 +73,7 @@ tests = [
   # { %(["a", {"b": "c"}]), "82 61 61 a1 61 62 61 63" },
   # { %({"a": "A", "b": "B", "c": "C", "d": "D", "e": "E"}), "a5 61 61 61 41 61 62 61 42 61 63 61 43 61 64 61 44 61 65 61 45" },
   { %((_ h'0102', h'030405')), "5f 42 01 02 43 03 04 05 ff" },
-  # { %((_ "strea", "ming")), "7f 65 73 74 72 65 61 64 6d 69 6e 67 ff" },
+  { %((_ "strea", "ming")), "7f 65 73 74 72 65 61 64 6d 69 6e 67 ff" },
   # { %([_ ]), "9f ff" },
   # { %([_ 1, [2, 3], [_ 4, 5]]), "9f 01 82 02 03 9f 04 05 ff ff" },
   # { %([_ 1, [2, 3], [4, 5]]), "9f 01 82 02 03 82 04 05 ff" },
diff --git a/src/cbor.cr b/src/cbor.cr
index d6bc043..8418855 100644
--- a/src/cbor.cr
+++ b/src/cbor.cr
@@ -19,6 +19,5 @@ module CBOR
                UInt32 |
                Int64 |
                UInt64 |
-               Int128 |
-               BytesArray
+               Int128
 end
diff --git a/src/cbor/diagnostic.cr b/src/cbor/diagnostic.cr
index 9774374..63eb6ea 100644
--- a/src/cbor/diagnostic.cr
+++ b/src/cbor/diagnostic.cr
@@ -34,6 +34,8 @@ class CBOR::Diagnostic
       "h'#{token[:value].as(Bytes).hexstring}'"
     when Kind::BytesArray
       token[:value].as(BytesArray).to_diagnostic
+    when Kind::StringArray
+      token[:value].as(StringArray).to_diagnostic
     else
       token[:kind].to_s
     end
diff --git a/src/cbor/lexer.cr b/src/cbor/lexer.cr
index 850cc0d..2b69efe 100644
--- a/src/cbor/lexer.cr
+++ b/src/cbor/lexer.cr
@@ -1,6 +1,25 @@
 require "./token"
 
 class CBOR::Lexer
+  # Types returned by the lexer
+  alias Type = Nil |
+               Bool |
+               String |
+               Bytes |
+               Array(Type) |
+               Hash(Type, Type) |
+               Int8 |
+               UInt8 |
+               Int16 |
+               UInt16 |
+               Int32 |
+               UInt32 |
+               Int64 |
+               UInt64 |
+               Int128 |
+               BytesArray |
+               StringArray
+
   def self.new(string : String)
     new IO::Memory.new(string)
   end
@@ -47,6 +66,8 @@ class CBOR::Lexer
       {kind: token[:kind], value: nil}
     when Kind::BytesArray
       {kind: token[:kind], value: read_bytes_array}
+    when Kind::StringArray
+      {kind: token[:kind], value: read_string_array}
     end
   end
 
@@ -54,17 +75,24 @@ class CBOR::Lexer
   def read_bytes_array : CBOR::BytesArray
     bytes = BytesArray.new
 
-    loop do
-      token = next_token
-      raise ParseError.new("Unexpected EOF while reading bytes array") unless token
-      break if token[:kind] == Kind::BytesArrayEnd
-      raise ParseError.new("Illegal token #{token.class} while reading bytes array") unless token[:kind] == Kind::Bytes
-      bytes << token[:value].as(Bytes)
+    read_until(Kind::BytesArrayEnd, only: Kind::Bytes) do |chunk|
+      bytes << chunk.as(Bytes)
     end
 
     bytes
   end
 
+  # Reads until break for chunks of strings
+  def read_string_array : CBOR::StringArray
+    strings = StringArray.new
+
+    read_until(Kind::StringArrayEnd, only: Kind::String) do |chunk|
+      strings << chunk.as(String)
+    end
+
+    strings
+  end
+
   private def next_token : Token?
     return nil if @eof
 
@@ -118,6 +146,8 @@ class CBOR::Lexer
       consume_string(read(UInt32))
     when 0x7b
       consume_string(read(UInt16))
+    when 0x7f
+      {kind: open_token(Kind::StringArray), value: nil}
     when 0xff
       {kind: finish_token, value: nil}
     else
@@ -125,6 +155,22 @@ class CBOR::Lexer
     end
   end
 
+  # Reads tokens until it meets the stop kind.
+  # Optionally it can fail when the read token is not of the passed kind.
+  private def read_until(stop : Kind, only : Kind?, &block)
+    loop do
+      token = next_token
+      raise ParseError.new("Unexpected EOF") unless token
+      break if token[:kind] == stop
+
+      if only && token[:kind] != only
+        raise ParseError.new("Illegal token #{token[:kind].to_s} while reading #{only.to_s} array")
+      end
+
+      yield token[:value]
+    end
+  end
+
   private def next_byte : UInt8?
     byte = @io.read_byte
     if byte
diff --git a/src/cbor/token.cr b/src/cbor/token.cr
index 5a5b44d..38da57f 100644
--- a/src/cbor/token.cr
+++ b/src/cbor/token.cr
@@ -15,4 +15,4 @@ enum CBOR::Kind
   Map
 end
 
-alias CBOR::Token = NamedTuple(kind: Kind, value: Type)
+alias CBOR::Token = NamedTuple(kind: Kind, value: Lexer::Type)
diff --git a/src/cbor/type/string_array.cr b/src/cbor/type/string_array.cr
new file mode 100644
index 0000000..10d09dd
--- /dev/null
+++ b/src/cbor/type/string_array.cr
@@ -0,0 +1,13 @@
+class CBOR::StringArray < Array(String)
+  def to_s : String
+    join
+  end
+
+  def to_diagnostic : String
+    "(_ #{map { |s| quote(s) }.join(", ")})"
+  end
+
+  private def quote(chunk : String) : String
+    %("#{chunk}")
+  end
+end