[PATCH] D92180: Speedup some unicode rendering

[serge via Phabricator via llvm-commits]

serge-sans-paille updated this revision to Diff 308372.
serge-sans-paille added a comment.

Make the ascii test for isprintable locale-independent


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D92180/new/

https://reviews.llvm.org/D92180

Files:
  llvm/lib/Support/Unicode.cpp
  llvm/unittests/Support/UnicodeTest.cpp


Index: llvm/unittests/Support/UnicodeTest.cpp
===================================================================
--- llvm/unittests/Support/UnicodeTest.cpp
+++ llvm/unittests/Support/UnicodeTest.cpp
@@ -7,6 +7,7 @@
 //===----------------------------------------------------------------------===//
 
 #include "llvm/Support/Unicode.h"
+#include "llvm/Support/ConvertUTF.h"
 #include "gtest/gtest.h"
 
 namespace llvm {
@@ -23,6 +24,7 @@
   EXPECT_EQ(6, columnWidthUTF8("abcdef"));
 
   EXPECT_EQ(-1, columnWidthUTF8("\x01"));
+  EXPECT_EQ(-1, columnWidthUTF8("\t"));
   EXPECT_EQ(-1, columnWidthUTF8("aaaaaaaaaa\x01"));
   EXPECT_EQ(-1, columnWidthUTF8("\342\200\213")); // 200B ZERO WIDTH SPACE
 
@@ -84,6 +86,18 @@
   EXPECT_TRUE(isPrintable(0x20000));  // CJK UNIFIED IDEOGRAPH-20000
 
   EXPECT_FALSE(isPrintable(0x10FFFF)); // noncharacter
+
+  // test the validity of a fast path in columnWidthUTF8
+  for (unsigned char c = 0; c < 128; ++c) {
+    const UTF8 buf8[1] = {c};
+    const UTF8 *Target8 = &buf8[0];
+    UTF32 buf32[1];
+    UTF32 *Target32 = &buf32[0];
+    auto status = ConvertUTF8toUTF32(&Target8, Target8 + 1, &Target32,
+                                     Target32 + 1, strictConversion);
+    EXPECT_TRUE(status == conversionOK);
+    EXPECT_TRUE((bool)isprint(c) == (bool)isPrintable(buf32[0]));
+  }
 }
 
 } // namespace
Index: llvm/lib/Support/Unicode.cpp
===================================================================
--- llvm/lib/Support/Unicode.cpp
+++ llvm/lib/Support/Unicode.cpp
@@ -348,11 +348,22 @@
   return 1;
 }
 
+static bool isprintableascii(char c) { return c > 31 && c < 127; }
+
 int columnWidthUTF8(StringRef Text) {
   unsigned ColumnWidth = 0;
   unsigned Length;
   for (size_t i = 0, e = Text.size(); i < e; i += Length) {
     Length = getNumBytesForUTF8(Text[i]);
+
+    // fast path for ASCII characters
+    if (Length == 1) {
+      if (!isprintableascii(Text[i]))
+        return ErrorNonPrintableCharacter;
+      ColumnWidth += 1;
+      continue;
+    }
+
     if (Length <= 0 || i + Length > Text.size())
       return ErrorInvalidUTF8;
     UTF32 buf[1];


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D92180.308372.patch
Type: text/x-patch
Size: 2123 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20201130/d429d402/attachment.bin>