Run PPB_CharSet_Dev logic directly in the plugin process.

ppapi/shared_impl/char_set_impl.cc

+// Copyright (c) 2011 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "ppapi/shared_impl/char_set_impl.h"
+
+#include "base/i18n/icu_string_conversions.h"
+#include "ppapi/c/ppb_core.h"
+#include "unicode/ucnv.h"
+#include "unicode/ucnv_cb.h"
+#include "unicode/ucnv_err.h"
+#include "unicode/ustring.h"
+
+namespace pp {
+namespace shared_impl {
+
+namespace {
+
+// Converts the given PP error handling behavior to the version in base,
+// placing the result in |*result| and returning true on success. Returns false
+// if the enum is invalid.
+bool PPToBaseConversionError(PP_CharSet_ConversionError on_error,
+                             base::OnStringConversionError::Type* result) {
+  switch (on_error) {
+    case PP_CHARSET_CONVERSIONERROR_FAIL:
+      *result = base::OnStringConversionError::FAIL;
+      return true;
+    case PP_CHARSET_CONVERSIONERROR_SKIP:
+      *result = base::OnStringConversionError::SKIP;
+      return true;
+    case PP_CHARSET_CONVERSIONERROR_SUBSTITUTE:
+      *result = base::OnStringConversionError::SUBSTITUTE;
+      return true;
+    default:
+      return false;
+  }
+}
+
+}  // namespace
+
+// static
+// The "substitution" behavior of this function does not match the
+// implementation in base, so we partially duplicate the code from
+// icu_string_conversions.cc with the correct error handling setup required
+// by the PPAPI interface.
+char* CharSetImpl::UTF16ToCharSet(const PPB_Core* core,
+                                  const uint16_t* utf16,
+                                  uint32_t utf16_len,
+                                  const char* output_char_set,
+                                  PP_CharSet_ConversionError on_error,
+                                  uint32_t* output_length) {
+  if (!core || !utf16 || !output_char_set || !output_length)
+    return NULL;
+
+  *output_length = 0;
+
+  UErrorCode status = U_ZERO_ERROR;
+  UConverter* converter = ucnv_open(output_char_set, &status);
+  if (!U_SUCCESS(status))
+    return NULL;
+
+  int encoded_max_length = UCNV_GET_MAX_BYTES_FOR_STRING(utf16_len,
+      ucnv_getMaxCharSize(converter));
+
+  // Setup our error handler.
+  switch (on_error) {
+    case PP_CHARSET_CONVERSIONERROR_FAIL:
+      ucnv_setFromUCallBack(converter, UCNV_FROM_U_CALLBACK_STOP, 0,
+                            NULL, NULL, &status);
+      break;
+    case PP_CHARSET_CONVERSIONERROR_SKIP:
+      ucnv_setFromUCallBack(converter, UCNV_FROM_U_CALLBACK_SKIP, 0,
+                            NULL, NULL, &status);
+      break;
+    case PP_CHARSET_CONVERSIONERROR_SUBSTITUTE: {
+      // ICU sets the substitution char for some character sets (like latin1)
+      // to be the ASCII "substitution character" (26). We want to use '?'
+      // instead for backwards-compat with Windows behavior.
+      char subst_chars[32];
+      int8_t subst_chars_len = 32;
+      ucnv_getSubstChars(converter, subst_chars, &subst_chars_len, &status);
+      if (subst_chars_len == 1 && subst_chars[0] == 26) {
+        // Override to the question mark character if possible. When using
+        // setSubstString, the input is a Unicode character. The function will
+        // try to convert it to the destination character set and fail if that
+        // can not be converted to the destination character set.
+        //
+        // We just ignore any failure. If the dest char set has no
+        // representation for '?', then we'll just stick to the ICU default
+        // substitution character.
+        UErrorCode subst_status = U_ZERO_ERROR;
+        UChar question_mark = '?';
+        ucnv_setSubstString(converter, &question_mark, 1, &subst_status);
+      }
+
+      ucnv_setFromUCallBack(converter, UCNV_FROM_U_CALLBACK_SUBSTITUTE, 0,
+                            NULL, NULL, &status);
+      break;
+    }
+    default:
+      return NULL;
+  }
+
+  // ucnv_fromUChars returns size not including terminating null.
+  char* encoded = static_cast<char*>(core->MemAlloc(encoded_max_length + 1));
+  int actual_size = ucnv_fromUChars(converter, encoded,
+      encoded_max_length, reinterpret_cast<const UChar*>(utf16), utf16_len,
+      &status);
+  ucnv_close(converter);
+  if (!U_SUCCESS(status)) {
+    core->MemFree(encoded);
+    return NULL;
+  }
+  encoded[actual_size] = 0;
+  *output_length = actual_size;
+  return encoded;
+}
+
+// static
+uint16_t* CharSetImpl::CharSetToUTF16(const PPB_Core* core,
+                                      const char* input,
+                                      uint32_t input_len,
+                                      const char* input_char_set,
+                                      PP_CharSet_ConversionError on_error,
+                                      uint32_t* output_length) {
+  if (!core || !input || !input_char_set || !output_length)
+    return NULL;
+
+  *output_length = 0;
+
+  base::OnStringConversionError::Type base_on_error;
+  if (!PPToBaseConversionError(on_error, &base_on_error))
+    return NULL;  // Invalid enum value.
+
+  // We can convert this call to the implementation in base to avoid code
+  // duplication, although this does introduce an extra copy of the data.
+  string16 output;
+  if (!base::CodepageToUTF16(std::string(input, input_len), input_char_set,
+                             base_on_error, &output))
+    return NULL;
+
+  uint16_t* ret_buf = static_cast<uint16_t*>(
+      core->MemAlloc((output.size() + 1) * sizeof(uint16_t)));
+  if (!ret_buf)
+    return NULL;
+
+  *output_length = static_cast<uint32_t>(output.size());
+  memcpy(ret_buf, output.c_str(), (output.size() + 1) * sizeof(uint16_t));
+  return ret_buf;
+}
+
+}  // namespace shared_impl
+}  // namespace pp