From c995d195adaf07811cf35d81423a5a4afa7d60cf Mon Sep 17 00:00:00 2001
From: Chi Tsai <chitsai@meta.com>
Date: Thu, 9 Jan 2025 05:15:51 -0800
Subject: [PATCH] Add getStringData and getPropNameIdData implementation for
 Hermes

Summary:
Adds the Hermes implementation for `getStringData` and `getPropNameIdData`,
which leverages the underlying UTF16 encoding

Grafted from 32084e2498f2b03ade118f1483b374dea4d7e711 (D65648690)
- Grafted path xplat/static_h to xplat/hermes

Reviewed By: tmikov

Differential Revision: D67412928

fbshipit-source-id: f9fa301e2fef49e7e77126aa65aa2303b8564497
---
 API/hermes/hermes.cpp     | 39 ++++++++++++++++++++
 unittests/API/APITest.cpp | 76 +++++++++++++++++++++++++++++++++++++++
 2 files changed, 115 insertions(+)

diff --git a/API/hermes/hermes.cpp b/API/hermes/hermes.cpp
index 300f7b9ae1a..22b3d4c3e6d 100644
--- a/API/hermes/hermes.cpp
+++ b/API/hermes/hermes.cpp
@@ -616,6 +616,16 @@ class HermesRuntimeImpl final : public HermesRuntime,
   std::u16string utf16(const jsi::String &str) override;
   std::u16string utf16(const jsi::PropNameID &sym) override;
 
+  void getStringData(
+      const jsi::String &str,
+      void *ctx,
+      void (*cb)(void *ctx, bool ascii, const void *data, size_t num)) override;
+
+  void getPropNameIdData(
+      const jsi::PropNameID &sym,
+      void *ctx,
+      void (*cb)(void *ctx, bool ascii, const void *data, size_t num)) override;
+
   jsi::Value createValueFromJsonUtf8(const uint8_t *json, size_t length)
       override;
 
@@ -1805,6 +1815,35 @@ std::u16string HermesRuntimeImpl::utf16(const jsi::PropNameID &sym) {
   return std::u16string(arrayRef.data(), arrayRef.size());
 }
 
+void HermesRuntimeImpl::getStringData(
+    const jsi::String &str,
+    void *ctx,
+    void (*cb)(void *ctx, bool ascii, const void *data, size_t num)) {
+  auto *stringPrim = phv(str).getString();
+  if (stringPrim->isASCII()) {
+    auto arrayRef = stringPrim->getStringRef<char>();
+    cb(ctx, true, arrayRef.data(), arrayRef.size());
+  } else {
+    auto arrayRef = stringPrim->getStringRef<char16_t>();
+    cb(ctx, false, arrayRef.data(), arrayRef.size());
+  }
+}
+
+void HermesRuntimeImpl::getPropNameIdData(
+    const jsi::PropNameID &sym,
+    void *ctx,
+    void (*cb)(void *ctx, bool ascii, const void *data, size_t num)) {
+  vm::SymbolID id = phv(sym).getSymbol();
+  auto *stringPrim = runtime_.getStringPrimFromSymbolID(id);
+  if (stringPrim->isASCII()) {
+    auto arrayRef = stringPrim->getStringRef<char>();
+    cb(ctx, true, arrayRef.data(), arrayRef.size());
+  } else {
+    auto arrayRef = stringPrim->getStringRef<char16_t>();
+    cb(ctx, false, arrayRef.data(), arrayRef.size());
+  }
+}
+
 jsi::Value HermesRuntimeImpl::createValueFromJsonUtf8(
     const uint8_t *json,
     size_t length) {
diff --git a/unittests/API/APITest.cpp b/unittests/API/APITest.cpp
index e4b91aef4c9..19cf3fb866a 100644
--- a/unittests/API/APITest.cpp
+++ b/unittests/API/APITest.cpp
@@ -1198,6 +1198,82 @@ TEST_P(HermesRuntimeTest, UTF16Test) {
   }
 }
 
+TEST_P(HermesRuntimeTest, GetStringDataTest) {
+  std::u16string buf;
+  auto cb = [&buf](bool ascii, const void *data, size_t num) {
+    // this callback copies the string content, but removes every 'o' character
+    if (ascii) {
+      const char *begin = (const char *)data;
+      const char *end = (const char *)data + num;
+      while (begin < end) {
+        char curr = begin[0];
+        if (curr != 'o') {
+          buf.push_back((char16_t)curr);
+        }
+        begin++;
+      }
+    } else {
+      const char16_t *begin = (const char16_t *)data;
+      const char16_t *end = (const char16_t *)data + num;
+      while (begin < end) {
+        char16_t curr = begin[0];
+        if (curr != 'o') {
+          buf.push_back(curr);
+        }
+        begin++;
+      }
+    }
+  };
+
+  String asciiString = String::createFromUtf8(*rt, "foobar");
+  asciiString.getStringData(*rt, cb);
+  EXPECT_EQ(buf, u"fbar");
+  buf.clear();
+
+  String utf16Str = String::createFromUtf8(*rt, "👍foobar你好");
+  utf16Str.getStringData(*rt, cb);
+  EXPECT_EQ(buf, u"👍fbar你好");
+  buf.clear();
+}
+
+TEST_P(HermesRuntimeTest, GetPropNameIdDataTest) {
+  std::u16string buf;
+  auto cb = [&buf](bool ascii, const void *data, size_t num) {
+    // this callback copies the string content, but removes every 'o' character
+    if (ascii) {
+      const char *begin = (const char *)data;
+      const char *end = (const char *)data + num;
+      while (begin < end) {
+        char curr = begin[0];
+        if (curr != 'o') {
+          buf.push_back((char16_t)curr);
+        }
+        begin++;
+      }
+    } else {
+      const char16_t *begin = (const char16_t *)data;
+      const char16_t *end = (const char16_t *)data + num;
+      while (begin < end) {
+        char16_t curr = begin[0];
+        if (curr != 'o') {
+          buf.push_back(curr);
+        }
+        begin++;
+      }
+    }
+  };
+
+  PropNameID ascii = PropNameID::forAscii(*rt, "foobar");
+  ascii.getPropNameIdData(*rt, cb);
+  EXPECT_EQ(buf, u"fbar");
+  buf.clear();
+
+  PropNameID utf16 = PropNameID::forUtf8(*rt, "👍foobar你好");
+  utf16.getPropNameIdData(*rt, cb);
+  EXPECT_EQ(buf, u"👍fbar你好");
+  buf.clear();
+}
+
 INSTANTIATE_TEST_CASE_P(
     Runtimes,
     HermesRuntimeTest,