Determine language code and type of format for address.

third_party/libaddressinput/chromium/cpp/src/address_ui.cc

Index: third_party/libaddressinput/chromium/cpp/src/address_ui.cc
diff --git a/third_party/libaddressinput/chromium/cpp/src/address_ui.cc b/third_party/libaddressinput/chromium/cpp/src/address_ui.cc
index c4eeb99c94e879914bd32ebd345747ca8b1c52b1..b3a4fb5892049394fa5b328f3f991e4f9f5056dc 100644
--- a/third_party/libaddressinput/chromium/cpp/src/address_ui.cc
+++ b/third_party/libaddressinput/chromium/cpp/src/address_ui.cc
@@ -17,12 +17,15 @@
 #include <libaddressinput/address_field.h>
 #include <libaddressinput/address_ui_component.h>
 
+#include <algorithm>
+#include <cstddef>
 #include <set>
 #include <string>
 #include <vector>
 
 #include "grit.h"
 #include "grit/libaddressinput_strings.h"
+#include "normalize_language_code.h"
 #include "region_data_constants.h"
 #include "rule.h"
 
@@ -58,6 +61,54 @@ int GetMessageIdForField(AddressField field,
   }
 }
 
+// Returns the BCP 47 language code that should be used to format the address
+// that the user entered.
+//
+// If the rule does not contain information about languages, then returns the
+// UI language.
+//
+// If the UI language is either the default language for the country, one of the
+// languages for rules, or one of the languages for address input, then returns
+// this UI language. If there're no matches, then picks one of the languages in
+// the rule and returns it.
+//
+// If |using_latin_format| is true and the UI language does not match any
+// languages in the rule, then appends "-latn" to the normalized version of the
+// language for the rule.
+std::string GetComponentsLanguageCode(
+    const std::vector<std::string>& rule_languages,
+    const std::vector<std::string>& input_languages,
+    const std::string& norm_default_lang,

[Evan Stade, 2014/03/21 23:44:54]

ditto

[please use gerrit instead, 2014/03/22 01:34:30]

How would I distinguish normalized UI language from the actual UI language?

I search for normalized UI language, but return the actual UI language. For
example, if the region supports "zh", but the UI language is "zh-hans", then I
return "zh-hans" here.

[Evan Stade, 2014/03/24 19:04:15]

maybe it's confusing because there's too many parameters. Why do you need to
pass "zh-hans" as well as "zh"? That just opens the door for problematic
situations where, e.g., they don't match.

[please use gerrit instead, 2014/03/24 22:56:29]

Made the function less confusing by passing in only |rule| and
|ui_language_code|.

+    const std::string& norm_ui_lang,
+    const std::string& ui_lang,
+    bool using_latin_format) {
+  if (std::find(rule_languages.begin(), rule_languages.end(), norm_ui_lang) !=
+          rule_languages.end() ||
+     std::find(rule_languages.begin(), rule_languages.end(), ui_lang) !=
+          rule_languages.end() ||
+      std::find(input_languages.begin(), input_languages.end(), norm_ui_lang) !=
+          input_languages.end() ||
+      std::find(input_languages.begin(), input_languages.end(), ui_lang) !=
+          input_languages.end() ||
+      norm_default_lang == norm_ui_lang) {
+    return ui_lang;
+  }
+
+  std::string rule_lang;
+  if (!norm_default_lang.empty()) {
+    rule_lang = norm_default_lang;
+  } else if (!rule_languages.empty()) {
+    rule_lang = rule_languages[0];
+  } else if (!input_languages.empty()) {
+    rule_lang = input_languages[0];
+  } else {
+    return ui_lang;
+  }
+
+  return using_latin_format ? NormalizeLanguageCode(rule_lang) + "-latn"
+                            : rule_lang;
+}
+
 }  // namespace
 
 const std::vector<std::string>& GetRegionCodes() {
@@ -65,7 +116,9 @@ const std::vector<std::string>& GetRegionCodes() {
 }
 
 std::vector<AddressUiComponent> BuildComponents(
-    const std::string& region_code) {
+    const std::string& region_code,
+    const std::string& ui_language_code,
+    std::string* components_language_code) {
   std::vector<AddressUiComponent> result;
 
   Rule rule;
@@ -75,13 +128,28 @@ std::vector<AddressUiComponent> BuildComponents(
     return result;
   }
 
+  std::string norm_ui_lang = NormalizeLanguageCode(ui_language_code);
+  std::string norm_region_lang = NormalizeLanguageCode(rule.GetLanguage());
+
+  const std::vector<std::vector<FormatElement> >& latin_format =
+      rule.GetLatinFormat();
+  bool use_latin_format =
+      norm_ui_lang != norm_region_lang && !latin_format.empty();
+  const std::vector<std::vector<FormatElement> >& format = use_latin_format
+      ? latin_format : rule.GetFormat();
+  if (components_language_code != NULL) {
+    *components_language_code = GetComponentsLanguageCode(
+        rule.GetLanguages(), rule.GetInputLanguages(), norm_region_lang,
+        norm_ui_lang, ui_language_code, use_latin_format);
+  }
+
   // For avoiding showing an input field twice, when the field is displayed
   // twice on an envelope.
   std::set<AddressField> fields;
 
   for (std::vector<std::vector<FormatElement> >::const_iterator
-           line_it = rule.GetFormat().begin();
-       line_it != rule.GetFormat().end();
+           line_it = format.begin();
+       line_it != format.end();
        ++line_it) {
     int num_fields_this_row = 0;
     for (std::vector<FormatElement>::const_iterator element_it =