Merge pull request #73585 from kubamracek/embedded-string-unicode-tables

kubamracek · web-flow · commit 67e9df028e6f · 2024-05-27T21:59:08.000-07:00
[embedded] Provide Unicode data tables for embedded as a static library
diff --git a/stdlib/public/stubs/CMakeLists.txt b/stdlib/public/stubs/CMakeLists.txt
@@ -49,3 +49,63 @@ if("${SWIFT_PRIMARY_VARIANT_SDK}" IN_LIST SWIFT_DARWIN_PLATFORMS)
                APPEND_STRING PROPERTY COMPILE_FLAGS
                  "-fobjc-arc")
 endif()
+
+# Embedded Swift Unicode library
+if(SWIFT_SHOULD_BUILD_EMBEDDED_STDLIB)
+  add_custom_target(embedded-unicode ALL)
+
+  foreach(entry ${EMBEDDED_STDLIB_TARGET_TRIPLES})
+    string(REGEX REPLACE "[ \t]+" ";" list "${entry}")
+    list(GET list 0 arch)
+    list(GET list 1 mod)
+    list(GET list 2 triple)
+
+    if("${mod}" MATCHES "-windows-msvc$")
+      continue()
+    endif()
+
+    if (SWIFT_HOST_VARIANT STREQUAL "linux")
+      set(extra_c_compile_flags -ffreestanding)
+    elseif (SWIFT_HOST_VARIANT STREQUAL "macosx")
+      set(extra_c_compile_flags -D__MACH__ -D__APPLE__ -ffreestanding)
+    endif()
+    
+    set(SWIFT_SDK_embedded_ARCH_${mod}_MODULE "${mod}")
+    set(SWIFT_SDK_embedded_LIB_SUBDIR "embedded")
+    set(SWIFT_SDK_embedded_ARCH_${mod}_TRIPLE "${triple}")
+
+    add_swift_target_library_single(
+      embedded-unicode-${mod}
+      swiftUnicodeDataTables
+      STATIC
+      IS_FRAGILE
+
+      Unicode/UnicodeData.cpp
+      Unicode/UnicodeGrapheme.cpp
+      Unicode/UnicodeNormalization.cpp
+      Unicode/UnicodeScalarProps.cpp
+      Unicode/UnicodeWord.cpp
+
+      C_COMPILE_FLAGS ${extra_c_compile_flags}
+      MODULE_DIR "${CMAKE_BINARY_DIR}/lib/swift/embedded"
+      SDK "embedded"
+      ARCHITECTURE "${mod}"
+      DEPENDS embedded-stdlib-${mod}
+      INSTALL_IN_COMPONENT stdlib
+      )
+    swift_install_in_component(
+      TARGETS embedded-unicode-${mod}
+      DESTINATION "lib/swift/embedded/${mod}"
+      COMPONENT "stdlib"
+      )
+    swift_install_in_component(
+      FILES "${SWIFTLIB_DIR}/embedded/${mod}/libswiftUnicodeDataTables.a"
+      DESTINATION "lib/swift/embedded/${mod}/"
+      COMPONENT "stdlib"
+      PERMISSIONS OWNER_READ OWNER_WRITE OWNER_EXECUTE GROUP_READ GROUP_EXECUTE WORLD_READ WORLD_EXECUTE
+      )
+    set_property(TARGET embedded-unicode-${mod} PROPERTY OSX_ARCHITECTURES "${arch}")
+
+    add_dependencies(embedded-unicode embedded-unicode-${mod})
+  endforeach()
+endif()
diff --git a/stdlib/public/stubs/Unicode/UnicodeData.cpp b/stdlib/public/stubs/Unicode/UnicodeData.cpp
@@ -11,7 +11,7 @@
 //===----------------------------------------------------------------------===//
 
 #include "swift/shims/UnicodeData.h"
-#include <limits>
+#include <stdint.h>
 
 // Every 4 byte chunks of data that we need to hash (in this case only ever
 // scalars and levels who are all uint32), we need to calculate K. At the end
@@ -162,7 +162,7 @@ __swift_intptr_t _swift_stdlib_getScalarBitArrayIdx(__swift_uint32_t scalar,
   // If our chunk index is larger than the quick look indices, then it means
   // our scalar appears in chunks who are all 0 and trailing.
   if ((__swift_uint64_t) idx > quickLookSize - 1) {
-    return std::numeric_limits<__swift_intptr_t>::max();
+    return INTPTR_MAX;
   }
 
   // Our scalar actually exists in a quick look bit array that was implemented.
@@ -172,7 +172,7 @@ __swift_intptr_t _swift_stdlib_getScalarBitArrayIdx(__swift_uint32_t scalar,
   // (chunkSize) of the scalars being represented have no property and ours is
   // one of them.
   if ((quickLook & ((__swift_uint64_t) 1 << chunkBit)) == 0) {
-    return std::numeric_limits<__swift_intptr_t>::max();
+    return INTPTR_MAX;
   }
 
   // Ok, our scalar failed the quick look check. Go lookup our scalar in the
@@ -223,7 +223,7 @@ __swift_intptr_t _swift_stdlib_getScalarBitArrayIdx(__swift_uint32_t scalar,
   // If our scalar specifically is not turned on within our chunk's bit array,
   // then we know for sure that our scalar does not inhibit this property.
   if ((chunkWord & ((__swift_uint64_t) 1 << scalarSpecificBit)) == 0) {
-    return std::numeric_limits<__swift_intptr_t>::max();
+    return INTPTR_MAX;
   }
 
   // Otherwise, this scalar does have whatever property this scalar array is
diff --git a/stdlib/public/stubs/Unicode/UnicodeGrapheme.cpp b/stdlib/public/stubs/Unicode/UnicodeGrapheme.cpp
@@ -16,7 +16,7 @@
 #include "swift/Runtime/Debug.h"
 #endif
 #include "swift/shims/UnicodeData.h"
-#include <limits>
+#include <stdint.h>
 
 
 SWIFT_RUNTIME_STDLIB_INTERNAL
@@ -68,7 +68,7 @@ __swift_bool _swift_stdlib_isLinkingConsonant(__swift_uint32_t scalar) {
                                           _swift_stdlib_linkingConsonant,
                                           _swift_stdlib_linkingConsonant_ranks);
 
-  if (idx == std::numeric_limits<__swift_intptr_t>::max()) {
+  if (idx == INTPTR_MAX) {
     return false;
   }
 
diff --git a/stdlib/public/stubs/Unicode/UnicodeNormalization.cpp b/stdlib/public/stubs/Unicode/UnicodeNormalization.cpp
@@ -23,7 +23,7 @@
 #endif
 
 #include "swift/shims/UnicodeData.h"
-#include <limits>
+#include <stdint.h>
 
 SWIFT_RUNTIME_STDLIB_INTERNAL
 __swift_uint16_t _swift_stdlib_getNormData(__swift_uint32_t scalar) {
@@ -42,7 +42,7 @@ __swift_uint16_t _swift_stdlib_getNormData(__swift_uint32_t scalar) {
 
   // If we don't have an index into the data indices, then this scalar has no
   // normalization information.
-  if (dataIdx == std::numeric_limits<__swift_intptr_t>::max()) {
+  if (dataIdx == INTPTR_MAX) {
     return 0;
   }
 
@@ -91,7 +91,7 @@ __swift_uint32_t _swift_stdlib_getComposition(__swift_uint32_t x,
   auto realY = (array[0] << 11) >> 11;
 
   if (y != realY) {
-    return std::numeric_limits<__swift_uint32_t>::max();
+    return UINT32_MAX;
   }
 
   auto count = array[0] >> 21;
@@ -134,6 +134,6 @@ __swift_uint32_t _swift_stdlib_getComposition(__swift_uint32_t x,
   // If we made it out here, then our scalar was not found in the composition
   // array.
   // Return the max here to indicate that we couldn't find one.
-  return std::numeric_limits<__swift_uint32_t>::max();
+  return UINT32_MAX;
 #endif
 }
diff --git a/stdlib/public/stubs/Unicode/UnicodeScalarProps.cpp b/stdlib/public/stubs/Unicode/UnicodeScalarProps.cpp
@@ -26,7 +26,7 @@
 #endif
 
 #include "swift/shims/UnicodeData.h"
-#include <limits>
+#include <stdint.h>
 
 SWIFT_RUNTIME_STDLIB_INTERNAL
 __swift_uint64_t _swift_stdlib_getBinaryProperties(__swift_uint32_t scalar) {
@@ -124,7 +124,7 @@ __swift_uint8_t _swift_stdlib_getNumericType(__swift_uint32_t scalar) {
   // If we made it out here, then our scalar was not found in the composition
   // array.
   // Return the max here to indicate that we couldn't find one.
-  return std::numeric_limits<__swift_uint8_t>::max();
+  return UINT8_MAX;
 #endif
 }
 
@@ -153,7 +153,7 @@ const char *_swift_stdlib_getNameAlias(__swift_uint32_t scalar) {
                                                     _swift_stdlib_nameAlias,
                                                   _swift_stdlib_nameAlias_ranks);
 
-  if (dataIdx == std::numeric_limits<__swift_intptr_t>::max()) {
+  if (dataIdx == INTPTR_MAX) {
     return nullptr;
   }
 
@@ -171,7 +171,7 @@ __swift_int32_t _swift_stdlib_getMapping(__swift_uint32_t scalar,
                                                     _swift_stdlib_mappings,
                                                   _swift_stdlib_mappings_ranks);
 
-  if (dataIdx == std::numeric_limits<__swift_intptr_t>::max()) {
+  if (dataIdx == INTPTR_MAX) {
     return 0;
   }
 
@@ -219,7 +219,7 @@ const __swift_uint8_t *_swift_stdlib_getSpecialMapping(__swift_uint32_t scalar,
                                                  _swift_stdlib_special_mappings,
                                           _swift_stdlib_special_mappings_ranks);
 
-  if (dataIdx == std::numeric_limits<__swift_intptr_t>::max()) {
+  if (dataIdx == INTPTR_MAX) {
     return nullptr;
   }
 
@@ -261,7 +261,7 @@ __swift_intptr_t _swift_stdlib_getScalarName(__swift_uint32_t scalar,
 #else
   auto setOffset = _swift_stdlib_names_scalar_sets[scalar >> 7];
 
-  if (setOffset == std::numeric_limits<__swift_uint16_t>::max()) {
+  if (setOffset == UINT16_MAX) {
     return 0;
   }
 
@@ -385,7 +385,7 @@ __swift_uint16_t _swift_stdlib_getAge(__swift_uint32_t scalar) {
   // If we made it out here, then our scalar was not found in the composition
   // array.
   // Return the max here to indicate that we couldn't find one.
-  return std::numeric_limits<__swift_uint16_t>::max();
+  return UINT16_MAX;
 #endif
 }
 
@@ -427,7 +427,7 @@ __swift_uint8_t _swift_stdlib_getGeneralCategory(__swift_uint32_t scalar) {
   // If we made it out here, then our scalar was not found in the composition
   // array.
   // Return the max here to indicate that we couldn't find one.
-  return std::numeric_limits<__swift_uint8_t>::max();
+  return UINT8_MAX;
 #endif
 }
 
@@ -485,7 +485,7 @@ __swift_uint8_t _swift_stdlib_getScript(__swift_uint32_t scalar) {
   // all in the array. This should never happen because the array represents all
   // scalars from 0x0 to 0x10FFFF, but if somehow this branch gets reached,
   // return 255 to indicate a failure.
-  return std::numeric_limits<__swift_uint8_t>::max();
+  return UINT8_MAX;
 #endif
 }
 
@@ -501,7 +501,7 @@ const __swift_uint8_t *_swift_stdlib_getScriptExtensions(__swift_uint32_t scalar
   
   // If we don't have an index into the data indices, then this scalar has no
   // script extensions
-  if (dataIdx == std::numeric_limits<__swift_intptr_t>::max()) {
+  if (dataIdx == INTPTR_MAX) {
     return 0;
   }
   
diff --git a/stdlib/public/stubs/Unicode/UnicodeWord.cpp b/stdlib/public/stubs/Unicode/UnicodeWord.cpp
@@ -16,7 +16,7 @@
 #include "swift/Runtime/Debug.h"
 #endif
 #include "swift/shims/UnicodeData.h"
-#include <limits>
+#include <stdint.h>
 
 SWIFT_RUNTIME_STDLIB_INTERNAL
 __swift_uint8_t _swift_stdlib_getWordBreakProperty(__swift_uint32_t scalar) {
@@ -46,6 +46,6 @@ __swift_uint8_t _swift_stdlib_getWordBreakProperty(__swift_uint32_t scalar) {
   // If we made it out here, then our scalar was not found in the word
   // array (this occurs when a scalar doesn't map to any word break
   // property). Return the max value here to indicate .any.
-  return std::numeric_limits<__swift_uint8_t>::max();
+  return UINT8_MAX;
 #endif
 }
diff --git a/test/embedded/stdlib-strings-datatables.swift b/test/embedded/stdlib-strings-datatables.swift
@@ -1,25 +1,28 @@
-// Test String operations that require unicode data tables. This is not an executable test yet, because the data tables
-// are not available for linking yet.
-
-// RUN: %target-swift-frontend -emit-ir %s -enable-experimental-feature Embedded
+// RUN: %target-run-simple-swift(                            -enable-experimental-feature Embedded -runtime-compatibility-version none -wmo -Xlinker %swift_obj_root/lib/swift/embedded/%target-cpu-apple-macos/libswiftUnicodeDataTables.a) | %FileCheck %s
+// RUN: %target-run-simple-swift(-Osize -Xlinker -dead_strip -enable-experimental-feature Embedded -runtime-compatibility-version none -wmo -Xlinker %swift_obj_root/lib/swift/embedded/%target-cpu-apple-macos/libswiftUnicodeDataTables.a) | %FileCheck %s
 
 // REQUIRES: swift_in_compiler
+// REQUIRES: executable_test
 // REQUIRES: optimized_stdlib
-// REQUIRES: OS=macosx || OS=linux-gnu
+// REQUIRES: OS=macosx
 
 public func test1() {
   let string = "string"
   let other = "other"
   let appended = string + other
-  _ = appended
+  print(appended) // CHECK: stringother
 
   let _ = "aa" == "bb"
-  let dict: [String:Int] = [:]
-  _ = dict
+  var dict: [String:Int] = [:]
+  dict["aa"] = 42
+  print(dict["aa"]!) // CHECK: 42
 
-  let _ = "aaa".uppercased()
+  let u = "aaa".uppercased()
+  print(u) // CHECK: AAA
 
   let space: Character = " "
   let split = appended.split(separator: space)
-  _ = split
+  print(split[0]) // CHECK: stringother
 }
+
+test1()
diff --git a/test/embedded/stdlib-strings-unicode.swift b/test/embedded/stdlib-strings-unicode.swift
@@ -0,0 +1,39 @@
+// RUN: %target-run-simple-swift(                            -enable-experimental-feature Embedded -parse-as-library -runtime-compatibility-version none -wmo -Xlinker %swift_obj_root/lib/swift/embedded/%target-cpu-apple-macos/libswiftUnicodeDataTables.a) | %FileCheck %s
+// RUN: %target-run-simple-swift(-Osize -Xlinker -dead_strip -enable-experimental-feature Embedded -parse-as-library -runtime-compatibility-version none -wmo -Xlinker %swift_obj_root/lib/swift/embedded/%target-cpu-apple-macos/libswiftUnicodeDataTables.a) | %FileCheck %s
+
+// REQUIRES: swift_in_compiler
+// REQUIRES: executable_test
+// REQUIRES: optimized_stdlib
+// REQUIRES: OS=macosx
+
+@main
+struct Main {
+  static func main() {
+    let str = "Hello😊"
+    print(str) // CHECK: Hello😊
+    print(str.dropLast()) // CHECK: Hello
+    print(str.dropLast().count) // CHECK: 5
+
+    var dict: [String:String] = [:]
+    let c = "Cafe\u{301}"
+    let d = "Cafe\u{301}"
+    let e = "Café"
+    let f = "Caf\u{65}\u{301}"
+    let g = "Caf\u{e9}"
+    dict[c] = str
+    dict[d] = str
+    dict[e] = str
+    dict[f] = str
+    dict[g] = str
+    print(dict.count) // CHECK: 1
+    print(dict[f]!) // CHECK: Hello😊
+
+    var emoji = ""
+    // VAMPIRE, ZERO-WIDTH JOINER, FEMALE SIGN, VARIATION SELECTOR-16
+    emoji += "\u{1f9db}"
+    emoji += "\u{200d}"
+    emoji += "\u{2640}"
+    emoji += "\u{fe0f}"
+    print(emoji.count) // CHECK: 1
+  }
+}
diff --git a/test/embedded/unicode-dead-strip1.swift b/test/embedded/unicode-dead-strip1.swift
@@ -0,0 +1,40 @@
+// RUN: %target-swift-frontend -Osize -parse-as-library -enable-experimental-feature Embedded %s -c -o %t/a.o
+// RUN: %target-clang %t/a.o -o %t/a.out -dead_strip %swift_obj_root/lib/swift/embedded/%target-cpu-apple-macos/libswiftUnicodeDataTables.a
+// RUN: %llvm-nm --defined-only --format=just-symbols --demangle %t/a.out | grep swift_stdlib_ | sort | %FileCheck %s --check-prefix=INCLUDES
+// RUN: %llvm-nm --defined-only --format=just-symbols --demangle %t/a.out | grep swift_stdlib_ | sort | %FileCheck %s --check-prefix=EXCLUDES
+
+// REQUIRES: swift_in_compiler
+// REQUIRES: optimized_stdlib
+// REQUIRES: OS=macosx
+
+@main
+struct Main {
+  static func main() {
+    var dict: [String:String] = [:]
+    let c = "Cafe\u{301}"
+    let d = "Cafe\u{301}"
+    let e = "Café"
+    let f = "Caf\u{65}\u{301}"
+    let g = "Caf\u{e9}"
+    dict[c] = "x"
+    dict[d] = "x"
+    dict[e] = "x"
+    dict[f] = "x"
+    dict[g] = "x"
+    print(dict.count)
+    print(dict[f]!)
+  }
+}
+
+// The code uses String equality and hashing, should need the normalization, NFC, NFD tables, and not the others.
+// EXCLUDES-NOT: swift_stdlib_case
+// EXCLUDES-NOT: swift_stdlib_graphemeBreakProperties
+// EXCLUDES-NOT: swift_stdlib_linkingConsonant
+// EXCLUDES-NOT: swift_stdlib_mappings
+// EXCLUDES-NOT: swift_stdlib_names
+// INCLUDES:     swift_stdlib_nfc
+// INCLUDES:     swift_stdlib_nfd
+// INCLUDES:     swift_stdlib_normData
+// EXCLUDES-NOT: swift_stdlib_scripts
+// EXCLUDES-NOT: swift_stdlib_special_mappings
+// EXCLUDES-NOT: swift_stdlib_words
diff --git a/test/embedded/unicode-dead-strip2.swift b/test/embedded/unicode-dead-strip2.swift
diff --git a/test/embedded/unicode-dead-strip3.swift b/test/embedded/unicode-dead-strip3.swift
