ICU4C lang names working now... (unicode-org#145)

* ICU4C LangNames - almost working * More updates for CPP lang_names * More updates for CPP lang_names - add ICU74 data * CPP Langnames - use getDisplayNames * CPP Langnames - working * Fix byte count to extract * Clean up C++ code and revise testplay error handling * Fixing failures in testplay
sven-oly · Dec 18, 2023 · c5dde43 · c5dde43
1 parent 7fbd3a4
commit c5dde43
Show file tree

Hide file tree

Showing 7 changed files with 194,319 additions and 47 deletions.
diff --git a/executors/cpp/Makefile b/executors/cpp/Makefile
@@ -15,7 +15,7 @@
 TARGET=executor
 
 # All object files (C or C++)
-OBJECTS=main.o coll.o util.o capi.o
+OBJECTS=main.o coll.o langnames.o util.o capi.o
 # OBJECTS=main.o coll.o numfmt.o util.o langnames.o capi.o
 
 #### rules

diff --git a/executors/cpp/langnames.cpp b/executors/cpp/langnames.cpp
@@ -29,38 +29,45 @@ using std::string;
 
 
 const string test_langnames (json_object *json_in) {
-  cout << "# LANGNAME: " << json_object_to_json_string(json_in);
+  UErrorCode status = U_ZERO_ERROR;
 
   json_object *label_obj = json_object_object_get(json_in, "label");
   string label_string = json_object_get_string(label_obj);
-  cout << "# LABEL: " << json_object_get_string(label_obj);
 
-  json_object *return_json = json_object_new_object();
-  json_object_object_add(return_json, "label", label_obj);
-  json_object_object_add(return_json,
-                         "result",
-                         json_object_new_string(result.c_str()));
 
+  // The locale in which the name is given.
   json_object *locale_label_obj = json_object_object_get(json_in, "locale_label");
   string locale_string = json_object_get_string(locale_label_obj);
-  cout << "# locale label: " << json_object_get_string(locale_label_obj);
 
+  // The language's name to be displayed.
   json_object *language_label_obj = json_object_object_get(json_in, "language_label");
   string language_label_string = json_object_get_string(language_label_obj);
-  cout << "# language label: " << json_object_get_string(language_label_obj);
 
-  Locale display_locale(locale_string);
-  UnicodeString display_lang(lange_label_string);
+  Locale displayLocale(locale_string.c_str());
+
+  Locale testLocale(language_label_string.c_str());
 
-  UnicodeString &name = getDisplayLanguage(const display_locale, UnicodeString &display_lang);
+  UnicodeString testLang;
+
+  testLocale.getDisplayName(displayLocale, testLang);
+  // The following gives only the language name, without region information
+  /// testLocale.getDisplayLanguage(displayLocale, testLang);
 
   json_object *return_json = json_object_new_object();
   json_object_object_add(return_json, "label", label_obj);
-  json_object_object_add(return_json,
-                         "result",
-                         json_object_new_string(name.c_str()));
- string return_str = json_object_to_json_string(return_json);
 
+  char test_result_string[1000] = "";
+
+  int32_t chars_out = testLang.extract(test_result_string, 1000, nullptr, status);
+  if (U_FAILURE(status)) {
+    json_object_object_add(return_json,
+                           "error", json_object_new_string("langnames extract error"));
+  } else {
+    json_object_object_add(return_json,
+                           "result",
+                           json_object_new_string(test_result_string));
+  }
 
-  return result_string;
+  string return_string = json_object_to_json_string(return_json);
+  return return_string;
 }
diff --git a/executors/cpp/main.cpp b/executors/cpp/main.cpp
@@ -40,16 +40,16 @@ using std::string;
 
 // Test functions
 extern const string test_collator(json_object *json_in);
-// extern const string test_numfmt(json_object *json_in);
-// extern const string test_langnames(json_object *json_in);
+extern const string test_langnames(json_object *json_in);
 
 std::string supported_tests[6] = {
   "collation_short"
   // "decimal_fmt",
   // "number_fmt",
-  // "display_names",
-  // "language_display_name",
-  // "likely_subtags"
+  //"display_names",
+  "lang_names",
+  "language_display_name",
+  "likely_subtags"
 };
 
 
@@ -95,8 +95,7 @@ int main(int argc, const char** argv)
       // Parse the JSON data.
       // Get the test type and call the test function.
 
-      std::string outputLine = "# BAD TEST ";
-      // !!! cout << "# input line: " << line << endl;
+      std::string outputLine;
 
       struct json_object *json_input;
       json_input = json_tokener_parse(line.c_str());
@@ -109,11 +108,17 @@ int main(int argc, const char** argv)
       // else if (test_type == "number_fmt") {
       //   outputLine = test_numfmt(json_input);
       //   // }
-      // else if (test_type == "language_display_name") {
-      //   outputLine = test_langnames(json_input);
-      // }
-else {
-        std::string outputLine = "# BAD TEST " + test_type;
+      else if (test_type == "lang_names") {
+        outputLine = test_langnames(json_input);
+      }
+      else if (test_type == "likely_subtags") {
+        // !!!       outputLine = test_likely_subtags(json_input);
+      }
+      else {
+        outputLine =  "# BAD TEST " + test_type;
+        //       "{\"error\": \"unknown test type\"," +
+        //       "\"test_type\":" +  test_type + "," +
+        //       "\"unsupported_test:\"" + test_type + "}";
       }
 
       // TODO: Instead of a string, get JSON output and stringify it.

diff --git a/run_config.json b/run_config.json
@@ -9,9 +9,10 @@
       "icu_version": "icu71",
       "exec": "cpp",
       "test_type": [
-        "collation_short"
+        "collation_short",
+        "lang_names"
       ],
-      "per_execution": 100
+      "per_execution": 10000
     }
   },
   {
@@ -24,9 +25,10 @@
       "icu_version": "icu72",
       "exec": "cpp",
       "test_type": [
-        "collation_short"
+        "collation_short",
+        "lang_names"
       ],
-      "per_execution": 100
+      "per_execution": 10000
     }
   },
   {
@@ -39,24 +41,26 @@
       "icu_version": "icu73",
       "exec": "cpp",
       "test_type": [
-        "collation_short"
+        "collation_short",
+        "lang_names"
       ],
-      "per_execution": 100
+      "per_execution": 10000
     }
   },
   {
     "prereq": {
-      "name": "Get ICU4x",
+      "name": "Get ICU4C 74",
       "version": "74.1",
       "command": "bash ../executors/cpp/set_icu4c_binary.sh ../gh-cache/icu4c-74_1-Ubuntu22.04-x64.tgz"
     },
     "run": {
       "icu_version": "icu74",
       "exec": "cpp",
       "test_type": [
-        "collation_short"
+        "collation_short",
+        "lang_names"
       ],
-      "per_execution": 100
+      "per_execution": 10000
     }
   },
   {

diff --git a/testdriver/testplan.py b/testdriver/testplan.py
@@ -520,18 +520,15 @@ def send_one_line(self, input_line):
                 logging.error(' !!!!!! exec_list = %s\n  input_line = %s' % (self.exec_list, input_line))
                 logging.error(' !!!!!! %s' % self.run_error_message)
 
-                # TODO!!!! Return an error for the offending line instead of failing for the whole batch
-
-                return None
-                input = json.loads(input_line.replace('#EXIT', ''))
-                error_result = {'label': input['label'],
-                                'input_data': input,
+                # Handle problems with decoding errors and other unknowns.
+                error_result = {'label': 'UNKNOWN',
+                                'input_data': input_line,
                                 'error': self.run_error_message
-                }
+                                }
                 return json.dumps(error_result)
         except BaseException as err:
             logging.error('!!! send_one_line fails: input => %s<. Err = %s', input_line, err)
-            input = json.loads(input_line.replace('#EXIT', ''))
+            input = json.loads(input_line.replace('#EXIT', '').strip())
             error_result = {'label': input['label'],
                             'input_data': input,
                             'error': err