Revise ui::DomKey to unify character and non-character codes.

ui/events/keycodes/dom/dom_key.h

Index: ui/events/keycodes/dom/dom_key.h
diff --git a/ui/events/keycodes/dom/dom_key.h b/ui/events/keycodes/dom/dom_key.h
index 25ea09209b522e649fae8c8f5fe6e6a3438974d8..e692733ee8a1389dfc3355c14aafb088b3db8e78 100644
--- a/ui/events/keycodes/dom/dom_key.h
+++ b/ui/events/keycodes/dom/dom_key.h
@@ -5,13 +5,154 @@
 #ifndef UI_EVENTS_KEYCODES_DOM3_DOM_KEY_H_
 #define UI_EVENTS_KEYCODES_DOM3_DOM_KEY_H_
 
+#include <stdint.h>
+
+#include "base/logging.h"
+
 namespace ui {
 
+// Integer representation of UI Events KeyboardEvent.key value.
+//
+// The semantics follow the web string form[1]: the value is either a
+// Unicode character or one of a defined set of additional values[2].
+// There is one notable difference from the UI Events string key: for
+// the 'Dead' key, this type provides a whole range of values that also
+// encode the associated combining character. (They are not quite the
+// same thing: a dead key is a non-printing operator that modifies a
+// subsequent printing character, whereas a Unicode combining character
+// is a printable character in its own right that attaches to a preceding
+// character in a string.) This allows the interpretation of any keystroke
+// to be carried as a single integer value.
+//
+// DomKey::NONE is a sentinel used to indicate an error or undefined value.
+// It is not the same as Unicode code point 0 (ASCII NUL) or the valid DOM
+// key 'Unidentified'.

[Wez, 2015/08/26 23:38:33]

nit: It's confusing to mention "undefined value" above, versus "Unidentified"
here - is Unidentified specific to keys we can't identify, where otherwise DOM
key works for other keys..?

[kpschoedel, 2015/08/27 15:47:54]

I'm not sure how to write this. DomKey::UNIDENTIFIED is 'Unidentified'
https://w3c.github.io/DOM-Level-3-Events-key/#keys-special and as far as code
using DomKey is concerned it's just another standard-defined |key|.


DomKey::NONE should not ever be exposed to the web. Currently, in ui::KeyEvent,
|key_ == DomKey::NONE| indicates that keyboard layout hasn't been done yet. If
layout fails (e.g. if |code| is not mapped), then key_ is set to
DomKey::UNDEFINED.

+//
+// References:
+// [1] http://www.w3.org/TR/uievents/#widl-KeyboardEvent-key
+// [2] http://www.w3.org/TR/DOM-Level-3-Events-key/
+//
+class DomKey {
+ public:
+  using Base = int32_t;
+
+ private:
+  // Integer representation of DomKey. This is arranged so that DomKey encoded
+  // values are distinct from Unicode code points, so that we can dynamically
+  // verify that they are not accidentally conflated.
+  //
+  // 31             24              16              8               0
+  // |       |       |       |       |       |       |       |       |
+  // | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
+  // |        z        |c|s|                    v                    |
+  //
+  //  From low to high:
+  //  - |v| is a value whose interpretation depends on the kind of key:
+  //    - for a Unicode value, it is the code point (0 <= v <= 0x10FFFF);
+  //    - for a dead key, the code point of the associated combining character;
+  //    - for others, an arbitrary distinct value.
+  //  - |s| is set for a valid symbolic key (i.e. not a Unicode character).
+  //  - |c| is set if |v| holds a code point (for either a Unicode character
+  //        directly, or a dead-key combining character).
+  //  - |z| is reserved and always zero.
+  //
+  //  As consequences of this representation,
+  //  - all valid DomKey encodings have at least one of |c| or |s| set, so
+  //    they can't be confused with raw Unicode characters (where both are 0).
+  //  - integer 0 is not a valid encoding, and can be used for DomKey::NONE.
+  //
+  enum { VALUE_BITS = 21 };
+  enum Type : Base {
+    VALUE_MASK = (1L << VALUE_BITS) - 1,
+    TF_SYMBOLIC = (1L << VALUE_BITS),
+    TF_CODEPOINT = (1L << (VALUE_BITS + 1)),
+    TYPE_MASK = TF_CODEPOINT | TF_SYMBOLIC,
+    TYPE_UNICODE = TF_CODEPOINT,
+    TYPE_NON_UNICODE = TF_SYMBOLIC,
+    TYPE_DEAD = TF_CODEPOINT | TF_SYMBOLIC,
+  };
+  static_assert(TYPE_UNICODE != 0 && TYPE_NON_UNICODE != 0 && TYPE_DEAD != 0,
+                "suspicious representation change");

[Wez, 2015/08/26 23:38:33]

:D

+
+ public:
+  enum InvalidKey : Base { NONE = 0 };
+// |dom_key_data.inc| describes the non-printable DomKey values, and is
+// included here to create constants for them in the DomKey:: scope.
+#define DOM_KEY_MAP_DECLARATION enum Key : Base
+#define DOM_KEY_UNI(key, id, value) id = (TYPE_UNICODE | (value))
+#define DOM_KEY_MAP_BEGIN FIRST_NON_UNICODE = TYPE_NON_UNICODE,
 #define DOM_KEY_MAP(key, id) id
-#define DOM_KEY_MAP_DECLARATION enum class DomKey
+#define DOM_KEY_MAP_END LAST_NON_UNICODE
 #include "ui/events/keycodes/dom/dom_key_data.inc"
-#undef DOM_KEY_MAP
 #undef DOM_KEY_MAP_DECLARATION
+#undef DOM_KEY_MAP_BEGIN
+#undef DOM_KEY_MAP
+#undef DOM_KEY_MAP_END
+#undef DOM_KEY_UNI
+
+  // Create a DomKey, with the undefined-value sentinel DomKey::NONE.
+  DomKey() : value_(NONE) {}
+
+  // Create a DomKey from an encoded integer value. This is implicit so
+  // that DomKey::NAME constants don't need to be explicitly converted
+  // to DomKey.
+  DomKey(Base value) : value_(value) {
+    DCHECK(value == 0 || IsValid()) << value;
+  }
+
+  // Obtain the encoded integer representation of the DomKey.
+  operator Base() const { return value_; }
+
+  // True if the value is a valid DomKey (which excludes DomKey::NONE and
+  // integers not following the DomKey format).
+  bool IsValid() const { return (value_ & TYPE_MASK) != 0; }
+
+  // True if the value is a Unicode code point.
+  bool IsCharacter() const { return (value_ & TYPE_MASK) == TYPE_UNICODE; }
+
+  // True if the value is a dead key.
+  bool IsDead() const { return (value_ & TYPE_MASK) == TYPE_DEAD; }

[Wez, 2015/08/26 23:38:33]

nit: Name this and ToDeadKeyCharacter, below, consistently.

+
+  // Returns the Unicode code point for a Unicode key.
+  // It is incorrect to call this for other kinds of key.
+  int32_t ToCharacter() const {
+    DCHECK(IsCharacter()) << value_;
+    return value_ & VALUE_MASK;
+  }
+
+  // Returns the associated combining code point for a dead key.
+  // It is incorrect to call this for other kinds of key.
+  int32_t ToDeadKeyCharacter() const {
+    DCHECK(IsDead()) << value_;
+    return value_ & VALUE_MASK;
+  }
+
+  // Returns a DomKey for the given Unicode character.
+  static DomKey FromCharacter(int32_t character) {
+    DCHECK(character >= 0 && character <= 0x10FFFF);
+    return DomKey(TYPE_UNICODE | character);
+  }
+
+  // Returns a dead-key DomKey for the given combining character.
+  static DomKey Dead(int32_t combining_character) {

[Wez, 2015/08/26 23:38:33]

nit: Name this consistently w/ FromCharacter and IsDead, ToDeadKeyCharacter,
e.g. FromDeadKeyCharacter?

[kpschoedel, 2015/08/27 15:47:54]

For dead keys, the combining character is an added attribute (not present in the
web KeyboardEvent) rather than its identity (as for a Unicode key). How about
IsDeadKey(), ToDeadKeyCombiningCharacter(), DeadKeyFromCombiningCharacter()?

+    DCHECK(combining_character >= 0 && combining_character <= 0x10FFFF);
+    return DomKey(TYPE_DEAD | combining_character);
+  }
+
+  // Provide means to generate constant DomKey::Base values, primarily to
+  // allow conversion tables to be constant, without startup construction.
+  // In the future (cue the theremin) this can be replaced with constexpr
+  // functions.
+  template<Base C> struct Constant {
+    enum : Base {
+      Character = TYPE_UNICODE | C,
+      Dead = TYPE_DEAD | C,
+    };
+  };
+
+ private:
+  Base value_;
+};
 
 }  // namespace ui