New thin layer of API extension chrome.i18n.detectLanguage

chrome/browser/extensions/api/i18n/i18n_api.cc

Index: chrome/browser/extensions/api/i18n/i18n_api.cc
diff --git a/chrome/browser/extensions/api/i18n/i18n_api.cc b/chrome/browser/extensions/api/i18n/i18n_api.cc
index ca40229ba1a03de856fde71207b23ff3386653e6..f0e553d05f2ea60a0f6fe0394bbede121cefdb1d 100644
--- a/chrome/browser/extensions/api/i18n/i18n_api.cc
+++ b/chrome/browser/extensions/api/i18n/i18n_api.cc
@@ -5,7 +5,6 @@
 #include "chrome/browser/extensions/api/i18n/i18n_api.h"
 
 #include <algorithm>
-#include <string>
 #include <vector>
 
 #include "base/lazy_instance.h"
@@ -59,4 +58,108 @@ bool I18nGetAcceptLanguagesFunction::RunSync() {
   return true;
 }
 
+bool I18nDetectLanguageFunction::RunAsync() {
+  scoped_ptr<api::i18n::DetectLanguage::Params> params(
+      api::i18n::DetectLanguage::Params::Create(*args_));
+  EXTENSION_FUNCTION_VALIDATE(params.get());
+
+  if (!params->text.get()) {
+    // text was not provided so return
+    return false;
+  }
+
+  GetLanguage(*params->text);
+  return true;
+}
+
+void I18nDetectLanguageFunction::GetLanguage(const std::string& text) {
+  // Sets up the input String
+  const std::string utf8_text(text);
+  const int num_utf8_bytes = static_cast<int>(utf8_text.size());
+  const char* raw_utf8_bytes = utf8_text.c_str();
+
+  // TODO(mcindy): improve this by providing better CLD hints
+  const char tld_hint[] = "";  // asummed no cld hints provided
+  int encoding_hint = CLD2::UNKNOWN_ENCODING;
+  CLD2::Language language_hint = CLD2::UNKNOWN_LANGUAGE;
+  CLD2::CLDHints cldhints = {NULL, tld_hint, encoding_hint, language_hint};
+
+  const bool is_plain_text = true;
+  CLD2::Language language3[3];
+  int percent3[3];
+  int text_bytes;  // amount of non-tag/letters-only text (assumed 0)
+  bool is_reliable = false;
+
+  int cld_language = 0;
+  int flags = 0;
+  int valid_prefix_bytes = false;
+  double normalized_score3[3];
+
+  cld_language = CLD2::ExtDetectLanguageSummaryCheckUTF8(
+                    raw_utf8_bytes,
+                    num_utf8_bytes,
+                    is_plain_text,
+                    &cldhints,
+                    flags,
+                    language3,
+                    percent3,
+                    normalized_score3,
+                    NULL,  // assumed no ResultChunkVector is used
+                    &text_bytes,
+                    &is_reliable,
+                    &valid_prefix_bytes);
+
+  DetectedLanguage detected_languages[3];
+  InitDetectedLanguages(language3, percent3, detected_languages);
+
+  SendLanguagesResult(detected_languages, is_reliable);
+}
+
+void I18nDetectLanguageFunction::SendLanguagesResult(
+                DetectedLanguage* detected_languages,
+                const bool is_reliable) {
+  base::DictionaryValue* result = new base::DictionaryValue();
+  result->Set("is_reliable", new base::FundamentalValue(is_reliable));
+
+  base::ListValue* languages = new base::ListValue();
+  // Set up three languages returned by CLD
+  for (int i = 0; i <3; i++) {
+    base::DictionaryValue* language = new base::DictionaryValue();
+    language->Set("language",
+                  new base::StringValue(detected_languages[i].language));
+    language->Set("percentage",
+                  new base::FundamentalValue(detected_languages[i].percentage));
+    languages->Set(i, language);
+  }

[not at google - send to devlin, 2015/07/08 19:34:53]

The i18n API is compiled to a typed struct, see
out/{Debug,Release}/gen/chrome/common/extensions/api/i18n.{h,cc} - better to use
that than setting Values directly.

[amalika, 2015/07/09 23:00:02]

Done.

+
+  result->Set("languages", languages);
+
+  SetResult(result);
+  SendResponse(true);
+}
+
+void I18nDetectLanguageFunction::InitDetectedLanguages(
+                             CLD2::Language* languages,
+                             int* percent3,
+                             DetectedLanguage detected_languages[3]) {

[not at google - send to devlin, 2015/07/08 19:34:53]

Is there a way to implement this stuff without passing around an array of
precisely size 3? It's a bit of an odd interface IMO. At least define 3 as a
constant somewhere.

[amalika, 2015/07/08 21:58:03]

Done.

+  for (int i = 0; i< 3; i++) {
+    std::string language_code = "";
+
+    // Convert LanguageCode 'zh' to 'zh-CN' and 'zh-Hant' to 'zh-TW' for
+    // Translate server usage. see DetermineTextLanguage in
+    // components/translate/core/language_detection/language_detection_util.cc
+    if (languages[i] == CLD2::CHINESE) {
+      language_code = "zh-CN";
+    } else if (languages[i] == CLD2::CHINESE_T) {
+      language_code = "zh-TW";
+    } else {
+      language_code =
+          CLD2::LanguageCode(static_cast<CLD2::Language>(languages[i]));
+    }
+
+    detected_languages[i].language = language_code;
+    detected_languages[i].percentage = percent3[i];
+  }
+}
+
 }  // namespace extensions