[llvm] [Support] formatv: non-negative-plus for integral numbers (PR #185008)

[Konrad Kleine via llvm-commits]

================
@@ -402,18 +402,63 @@ TEST(FormatAndFormatvTest, EquivalentHexFormatting) {
   uint8_t HexDigits = 10;
   uint64_t N = 255;
 
+  // with format()
+  // -------------
   // Here's the old format() way of printing a hex number with
   // dynamic width and precision, both being the same.
-  EXPECT_EQ(
-      "0x00000000ff",
-      printToString(100, format("0x%*.*" PRIx64, HexDigits, HexDigits, N)));
+  EXPECT_EQ("0x00000000ff",
+            printToString(format("0x%*.*" PRIx64, HexDigits, HexDigits, N)));
 
+  // with formatv()
+  // --------------
   // Now, do the same with formatv()
   EXPECT_EQ("0x00000000ff",
             formatv("0x{0:x-}", fmt_align(N, AlignStyle::Right, HexDigits, '0'))
                 .str());
 }
 
+TEST(FormatAndFormatvTest, NonNegativePlusInteger) {
+  // with format()
+  // -------------
+  EXPECT_EQ("-255", printToString(format("%+d", -255)));
+  EXPECT_EQ("+255", printToString(format("%+d", 255)));
+  EXPECT_EQ("+0", printToString(format("%+d", 0)));
+  EXPECT_EQ("-7", printToString(format("%+d", -7)));
+
+  // with formatv()
+  // --------------
+  // Check that +d works for signed and unsigned integral values.
+  // Ensure the default is not changes (a + is not added without requesting it).
+  EXPECT_EQ("+1", formatv("{0:+d}", static_cast<unsigned int>(1)).str());
+  EXPECT_EQ("+2", formatv("{0:+d}", static_cast<int>(2)).str());
+  EXPECT_EQ("+3", formatv("{0:+d}", static_cast<unsigned long>(3)).str());
+  EXPECT_EQ("+4", formatv("{0:+d}", static_cast<long>(4)).str());
+  EXPECT_EQ("+5", formatv("{0:+d}", static_cast<unsigned long long>(5)).str());
+  EXPECT_EQ("+6", formatv("{0:+d}", static_cast<long long>(6)).str());
+  EXPECT_EQ("-7", formatv("{0:+d}", static_cast<int>(-7)).str());
+  EXPECT_EQ("-8", formatv("{0:+d}", static_cast<long>(-8)).str());
+  EXPECT_EQ("-9", formatv("{0:+d}", static_cast<long long>(-9)).str());
+
+  EXPECT_EQ("11", formatv("{0:d}", static_cast<unsigned int>(11)).str());
+  EXPECT_EQ("22", formatv("{0:d}", static_cast<int>(22)).str());
+  EXPECT_EQ("33", formatv("{0:d}", static_cast<unsigned long>(33)).str());
+  EXPECT_EQ("44", formatv("{0:d}", static_cast<long>(44)).str());
+  EXPECT_EQ("55", formatv("{0:d}", static_cast<unsigned long long>(55)).str());
+  EXPECT_EQ("66", formatv("{0:d}", static_cast<long long>(66)).str());
+  EXPECT_EQ("-77", formatv("{0:d}", static_cast<int>(-77)).str());
+  EXPECT_EQ("-88", formatv("{0:d}", static_cast<long>(-88)).str());
+  EXPECT_EQ("-99", formatv("{0:d}", static_cast<long long>(-99)).str());
+
+  // Ensure that 0 is also prefixed with + (old behaviour from format(), see
+  // above).
+  EXPECT_EQ("+0", formatv("{0:+d}", 0).str());
+  EXPECT_EQ("0", formatv("{0:d}", 0).str());
+
+  // Ensure that an empty or otherwise empty format string is also working as
+  // expected
+  EXPECT_EQ("+333", formatv("{0:+}", 333).str());
+  EXPECT_EQ("444", formatv("{0:}", 444).str());
+}
----------------
kwk wrote:

> We're only supporting this for integer outputs. Add a test or two (+ve/-ve number) with a hex format. Presumably it does nothing / does something silly but that's fine we just want to record what it was doing at this time.

What do you mean by this?

> (+ve/-ve number)

https://github.com/llvm/llvm-project/pull/185008