improve Bun.stringWidth's algorithm

.gitmodules

@@ -82,4 +82,4 @@ url = https://github.com/litespeedtech/ls-hpack.git
 ignore = dirty
 depth = 1
 shallow = true
-fetchRecurseSubmodules = false
+fetchRecurseSubmodules = false

CMakeLists.txt

@@ -285,9 +285,11 @@ endif()
 set(ERROR_LIMIT 100 CACHE STRING "Maximum number of errors to show when compiling C++ code")
 
 set(ARCH x86_64)
+set(HOMEBREW_PREFIX "/usr/local")
 
 if(CMAKE_SYSTEM_PROCESSOR MATCHES "aarch64|arm64|arm")
     set(ARCH aarch64)
+    set(HOMEBREW_PREFIX "/opt/homebrew")
 endif()
 
 if(NOT CPU_TARGET)
@@ -463,6 +465,11 @@ if(NOT WEBKIT_DIR)
     endif()
 
     set(WEBKIT_INCLUDE_DIR "${BUN_WORKDIR}/bun-webkit/include")
+    if(APPLE)
+        set(ICU_INCLUDE_DIR "${HOMEBREW_PREFIX}/opt/icu4c/include")
+    else()
+        set(ICU_INCLUDE_DIR "${BUN_WORKDIR}/bun-webkit/include/wtf/unicode")
+    endif()
     set(WEBKIT_LIB_DIR "${BUN_WORKDIR}/bun-webkit/lib")
 elseif(WEBKIT_DIR STREQUAL "omit")
     message(STATUS "Not using WebKit. This is only valid if you are only trying to build Zig code")
@@ -879,6 +886,10 @@ else()
     add_compile_definitions("ASSERT_ENABLED=1")
 endif()
 
+if (ICU_INCLUDE_DIR)
+include_directories(${ICU_INCLUDE_DIR})
+endif()
+
 include_directories(
     ${CMAKE_CURRENT_SOURCE_DIR}/packages/
     ${CMAKE_CURRENT_SOURCE_DIR}/packages/bun-usockets

packages/bun-types/bun.d.ts

@@ -47,6 +47,14 @@ declare module "bun" {
    */
   function which(command: string, options?: { PATH?: string; cwd?: string }): string | null;
 
+  function stringWidth(
+    input: string,
+    options?: {
+      countAnsiEscapeCodes?: boolean;
+      ambiguousIsNarrow?: boolean;
+    },
+  ): number;
+
   export type ShellFunction = (input: Uint8Array) => Uint8Array;
 
   export type ShellExpression =

src/bun.js/ConsoleObject.zig

@@ -320,7 +320,7 @@ const TablePrinter = struct {
     fn updateColumnsForRow(this: *TablePrinter, columns: *std.ArrayList(Column), row_key: RowKey, row_value: JSValue) !void {
         // update size of "(index)" column
         const row_key_len: u32 = switch (row_key) {
-            .str => |value| @intCast(value.visibleWidthExcludeANSIColors()),
+            .str => |value| @intCast(value.visibleWidthExcludeANSIColors(false)),
             .num => |value| @truncate(bun.fmt.fastDigitCount(value)),
         };
         columns.items[0].width = @max(columns.items[0].width, row_key_len);
@@ -400,7 +400,7 @@ const TablePrinter = struct {
         try writer.writeAll("│");
         {
             const len: u32 = switch (row_key) {
-                .str => |value| @truncate(value.visibleWidthExcludeANSIColors()),
+                .str => |value| @truncate(value.visibleWidthExcludeANSIColors(false)),
                 .num => |value| @truncate(bun.fmt.fastDigitCount(value)),
             };
             const needed = columns.items[0].width -| len;
@@ -544,7 +544,7 @@ const TablePrinter = struct {
         {
             for (columns.items) |*col| {
                 // also update the col width with the length of the column name itself
-                col.width = @max(col.width, @as(u32, @intCast(col.name.visibleWidthExcludeANSIColors())));
+                col.width = @max(col.width, @as(u32, @intCast(col.name.visibleWidthExcludeANSIColors(false))));
             }
 
             try writer.writeAll("┌");
@@ -557,7 +557,7 @@ const TablePrinter = struct {
 
             for (columns.items, 0..) |col, i| {
                 if (i > 0) try writer.writeAll("│");
-                const len = col.name.visibleWidthExcludeANSIColors();
+                const len = col.name.visibleWidthExcludeANSIColors(false);
                 const needed = col.width -| len;
                 try writer.writeByteNTimes(' ', 1);
                 if (comptime enable_ansi_colors) {

src/bun.js/api/BunObject.zig

@@ -5163,19 +5163,24 @@ fn stringWidth(globalObject: *JSC.JSGlobalObject, callframe: *JSC.CallFrame) cal
     defer str.deref();
 
     var count_ansi_escapes = false;
+    var ambiguous_as_wide = false;
 
     if (options_object.isObject()) {
         if (options_object.getTruthy(globalObject, "countAnsiEscapeCodes")) |count_ansi_escapes_value| {
             if (count_ansi_escapes_value.isBoolean())
                 count_ansi_escapes = count_ansi_escapes_value.toBoolean();
         }
+        if (options_object.getTruthy(globalObject, "ambiguousIsNarrow")) |ambiguous_is_narrow| {
+            if (ambiguous_is_narrow.isBoolean())
+                ambiguous_as_wide = !ambiguous_is_narrow.toBoolean();
+        }
     }
 
     if (count_ansi_escapes) {
-        return JSC.jsNumber(str.visibleWidth());
+        return JSC.jsNumber(str.visibleWidth(ambiguous_as_wide));
     }
 
-    return JSC.jsNumber(str.visibleWidthExcludeANSIColors());
+    return JSC.jsNumber(str.visibleWidthExcludeANSIColors(ambiguous_as_wide));
 }
 
 /// EnvironmentVariables is runtime defined.

src/bun.js/bindings/BunObject.cpp

@@ -632,6 +632,7 @@ JSC_DEFINE_HOST_FUNCTION(functionHashCode,
     stdin                                          BunObject_getter_wrap_stdin                                         DontDelete|PropertyCallback
     stdout                                         BunObject_getter_wrap_stdout                                        DontDelete|PropertyCallback
     stringHashCode                                 functionHashCode                                                    DontDelete|Function 1
+    stringWidth                                    BunObject_callback_stringWidth                                      DontDelete|Function 2
     unsafe                                         BunObject_getter_wrap_unsafe                                        DontDelete|PropertyCallback
     version                                        constructBunVersion                                                 ReadOnly|DontDelete|PropertyCallback
     which                                          BunObject_callback_which                                            DontDelete|Function 1

src/bun.js/bindings/workaround-missing-symbols.cpp

@@ -1,7 +1,6 @@
 
 #if defined(WIN32)
 
-
 #include <cstdint>
 #include <algorithm>
 #include <sys/stat.h>
@@ -14,7 +13,7 @@
 #undef _environ
 #undef environ
 
-// Some libraries need these symbols. Windows makes it 
+// Some libraries need these symbols. Windows makes it
 extern "C" char** environ = nullptr;
 extern "C" char** _environ = nullptr;
 
@@ -76,7 +75,7 @@ extern "C" char* mkdtemp(char* template_name)
 {
     uv_fs_t req;
     int status_code = uv_fs_mkdtemp(uv_default_loop(), &req, template_name, nullptr);
-    
+
     if (status_code < 0)
         return nullptr;
     size_t outlen = std::min(strlen(req.path), strlen(template_name));
@@ -369,3 +368,10 @@ extern "C" int __ulock_wait2(uint32_t operation, void* addr, uint64_t value,
 }
 
 #endif
+
+#include <unicode/uchar.h>
+
+extern "C" bool icu_hasBinaryProperty(UChar32 cp, unsigned int prop)
+{
+    return u_hasBinaryProperty(cp, static_cast<UProperty>(prop));
+}

src/bun.js/webcore/encoding.zig

@@ -527,7 +527,7 @@ pub const TextDecoder = struct {
         while (remainder.len > 0) {
             switch (remainder[0]) {
                 0...127 => {
-                    const count: usize = if (strings.firstNonASCII16CheckMin(Slice, remainder, false)) |index| index + 1 else remainder.len;
+                    const count: usize = if (strings.firstNonASCII16IgnoreMin(Slice, remainder)) |index| index + 1 else remainder.len;
 
                     buffer.ensureUnusedCapacity(allocator, count) catch unreachable;