From 2b957702c7e046b6229eb395ae0bb8a7c53cab47 Mon Sep 17 00:00:00 2001
From: Jonathan Kew <jfkthame@gmail.com>
Date: Mon, 20 Feb 2012 20:39:59 +0000
Subject: [PATCH] bug 728866 - pt 1 - expose low-level APIs for canonical
 composition and decomposition of a single Unicode character. r=smontagu

--HG--
rename : intl/unicharutil/src/nsUnicodeNormalizer.h => intl/unicharutil/public/nsUnicodeNormalizer.h
---
 intl/unicharutil/public/Makefile.in           |  1 +
 .../{src => public}/nsUnicodeNormalizer.h     |  4 +++
 intl/unicharutil/src/nsUnicodeNormalizer.cpp  | 27 +++++++++++++++++++
 3 files changed, 32 insertions(+)
 rename intl/unicharutil/{src => public}/nsUnicodeNormalizer.h (91%)

diff --git a/intl/unicharutil/public/Makefile.in b/intl/unicharutil/public/Makefile.in
index 9bf08cd718c9..6ebd062b3952 100644
--- a/intl/unicharutil/public/Makefile.in
+++ b/intl/unicharutil/public/Makefile.in
@@ -48,6 +48,7 @@ EXPORTS		= \
 		nsICaseConversion.h \
 		nsIUGenCategory.h \
 		nsUnicharUtilCIID.h \
+		nsUnicodeNormalizer.h \
 		$(NULL)
 
 include $(topsrcdir)/config/rules.mk
diff --git a/intl/unicharutil/src/nsUnicodeNormalizer.h b/intl/unicharutil/public/nsUnicodeNormalizer.h
similarity index 91%
rename from intl/unicharutil/src/nsUnicodeNormalizer.h
rename to intl/unicharutil/public/nsUnicodeNormalizer.h
index 242f26e8e93f..6d500edcfb85 100644
--- a/intl/unicharutil/src/nsUnicodeNormalizer.h
+++ b/intl/unicharutil/public/nsUnicodeNormalizer.h
@@ -57,6 +57,10 @@ public:
    NS_IMETHOD NormalizeUnicodeNFC( const nsAString& aSrc, nsAString& aDest);
    NS_IMETHOD NormalizeUnicodeNFKD( const nsAString& aSrc, nsAString& aDest);
    NS_IMETHOD NormalizeUnicodeNFKC( const nsAString& aSrc, nsAString& aDest);
+
+   // low-level access to the composition data needed for HarfBuzz callbacks
+   static bool Compose(PRUint32 a, PRUint32 b, PRUint32 *ab);
+   static bool DecomposeNonRecursively(PRUint32 comp, PRUint32 *c1, PRUint32 *c2);
 };
 
 #endif //nsUnicodeNormalizer_h__
diff --git a/intl/unicharutil/src/nsUnicodeNormalizer.cpp b/intl/unicharutil/src/nsUnicodeNormalizer.cpp
index 97da73e9a5e4..c63339f63d26 100644
--- a/intl/unicharutil/src/nsUnicodeNormalizer.cpp
+++ b/intl/unicharutil/src/nsUnicodeNormalizer.cpp
@@ -713,3 +713,30 @@ nsUnicodeNormalizer::NormalizeUnicodeNFKC( const nsAString& aSrc, nsAString& aDe
   return mdn_normalize(true, true, aSrc, aDest);
 }
 
+bool
+nsUnicodeNormalizer::Compose(PRUint32 a, PRUint32 b, PRUint32 *ab)
+{
+  return mdn__unicode_compose(a, b, ab) == NS_OK;
+}
+
+bool
+nsUnicodeNormalizer::DecomposeNonRecursively(PRUint32 c, PRUint32 *c1, PRUint32 *c2)
+{
+  // We can't use mdn__unicode_decompose here, because that does a recursive
+  // decomposition that may yield more than two characters, but the harfbuzz
+  // callback wants just a single-step decomp that is guaranteed to produce
+  // no more than two characters. So we do a low-level lookup in the table
+  // of decomp sequences.
+  const PRUint32 *seq;
+  PRUint32 seqidx = decompose_char(c, &seq);
+  if (seqidx == 0 || ((seqidx & DECOMP_COMPAT) != 0)) {
+    return false;
+  }
+  *c1 = *seq & ~END_BIT;
+  if (*seq & END_BIT) {
+    *c2 = 0;
+  } else {
+    *c2 = *++seq & ~END_BIT;
+  }
+  return true;
+}