Bug 1308021 - Replace "iw" with "he" in cld2 detectLanguage results. …

…r=florian I have a pull request for these changes upstream: CLD2Owners/cld2#50 MozReview-Commit-ID: LUqrA4Genv0
mozilla · Oct 7, 2016 · 78fe5d0 · 78fe5d0
1 parent 6668691
commit 78fe5d0
Show file tree

Hide file tree

Showing 7 changed files with 91 additions and 122 deletions.
diff --git a/browser/components/translation/cld2/cld-worker.js b/browser/components/translation/cld2/cld-worker.js
diff --git a/browser/components/translation/cld2/cld-worker.js.mem b/browser/components/translation/cld2/cld-worker.js.mem
diff --git a/browser/components/translation/cld2/internal/cld_generated_score_quad_octa_0122_2.cc b/browser/components/translation/cld2/internal/cld_generated_score_quad_octa_0122_2.cc
@@ -26,7 +26,7 @@ extern const short kAvgDeltaOctaScore[kAvgDeltaOctaScoreSize] = {
   1306,    0,    0,    0,  // 3 FINNISH fi
   1033,    0,    0,    0,  // 4 FRENCH fr
   1306,    0,    0,    0,  // 5 GERMAN de
-     0,    0,    0,  776,  // 6 HEBREW iw
+     0,    0,    0,  776,  // 6 HEBREW he
    960,    0,    0,    0,  // 7 ITALIAN it
      0,    0,    0, 3100,  // 8 Japanese ja
      0,    0,    0, 3669,  // 9 Korean ko
@@ -637,4 +637,3 @@ extern const short kAvgDeltaOctaScore[kAvgDeltaOctaScoreSize] = {
 };
 
 }       // End namespace CLD2
-
diff --git a/browser/components/translation/cld2/internal/compact_lang_det_hint_code.cc b/browser/components/translation/cld2/internal/compact_lang_det_hint_code.cc
@@ -174,7 +174,7 @@ static const LangTagLookup kCLDLangTagsHintTable1[kCLDTable1Size] = {
   {"haitian_creole", "ht", HAITIAN_CREOLE + W10, 0},
   {"hausa", "ha", HAUSA + W10, 0},
   {"hawaiian", "haw", HAWAIIAN + W10, 0},
-  {"hebrew", "iw", HEBREW + W10, 0},
+  {"hebrew", "he", HEBREW + W10, 0},
   {"hindi", "hi", HINDI + W10, MARATHI - W4},
   {"hn-in", "hi", HINDI + W10, MARATHI - W4},
   {"hungarian", "hu", HUNGARIAN + W10, 0},
@@ -447,9 +447,9 @@ static const LangTagLookup kCLDLangTagsHintTable2[kCLDTable2Size] = {
   {"ha", "ha", HAUSA + W10, 0},
   {"hat", "ht", HAITIAN_CREOLE + W10, 0},   // Haiti
   {"haw", "haw", HAWAIIAN + W10, 0},
-  {"hb", "iw", HEBREW + W10, 0},
-  {"he", "iw", HEBREW + W10, 0},
-  {"heb", "iw", HEBREW + W10, 0},
+  {"hb", "he", HEBREW + W10, 0},
+  {"he", "he", HEBREW + W10, 0},
+  {"heb", "he", HEBREW + W10, 0},
   {"hi", "hi", HINDI + W10, MARATHI - W4},
   {"hk", "zhT", CHINESE_T + W10, 0},          // Hong Kong
   {"hr", "hr", CROATIAN + W10, 0},
@@ -473,7 +473,7 @@ static const LangTagLookup kCLDLangTagsHintTable2[kCLDTable2Size] = {
   {"it", "it", ITALIAN + W10, 0},
   {"ita", "it", ITALIAN + W10, 0},
   {"iu", "iu,ik", INUKTITUT + W10, INUPIAK + W10},        // 1:2
-  {"iw", "iw", HEBREW + W10, 0},
+  {"iw", "he", HEBREW + W10, 0},
 
   {"ja", "ja", JAPANESE + W10, 0},
   {"jp", "ja", JAPANESE + W10, 0},          // Japan
@@ -1647,5 +1647,3 @@ string GetLangTagsFromHtml(const char* utf8_body, int32 utf8_body_len,
 }       // End namespace CLD2
 
 //==============================================================================
-
-
diff --git a/browser/components/translation/cld2/internal/generated_language.cc b/browser/components/translation/cld2/internal/generated_language.cc
@@ -32,7 +32,7 @@ extern const char* const kLanguageToName[kLanguageToNameSize] = {
   "FINNISH",               // 3 fi
   "FRENCH",                // 4 fr
   "GERMAN",                // 5 de
-  "HEBREW",                // 6 iw
+  "HEBREW",                // 6 he
   "ITALIAN",               // 7 it
   "Japanese",              // 8 ja
   "Korean",                // 9 ko
@@ -651,7 +651,7 @@ extern const char* const kLanguageToCode[kLanguageToCodeSize] = {
   "fi",    // 3 FINNISH
   "fr",    // 4 FRENCH
   "de",    // 5 GERMAN
-  "iw",    // 6 HEBREW
+  "he",    // 6 HEBREW
   "it",    // 7 ITALIAN
   "ja",    // 8 Japanese
   "ko",    // 9 Korean
@@ -1270,7 +1270,7 @@ extern const char* const kLanguageToCName[kLanguageToCNameSize] = {
   "FINNISH",               // 3 fi
   "FRENCH",                // 4 fr
   "GERMAN",                // 5 de
-  "HEBREW",                // 6 iw
+  "HEBREW",                // 6 he
   "ITALIAN",               // 7 it
   "JAPANESE",              // 8 ja
   "KOREAN",                // 9 ko
@@ -1890,7 +1890,7 @@ extern const FourScripts kLanguageToScripts[kLanguageToScriptsSize] = {
   {ULScript_Latin, None, None, None, },  // 3 fi
   {ULScript_Latin, None, None, None, },  // 4 fr
   {ULScript_Latin, None, None, None, },  // 5 de
-  {ULScript_Hebrew, None, None, None, },  // 6 iw
+  {ULScript_Hebrew, None, None, None, },  // 6 he
   {ULScript_Latin, None, None, None, },  // 7 it
   {ULScript_Hani, None, None, None, },  // 8 ja
   {ULScript_Hani, None, None, None, },  // 9 ko
@@ -2510,7 +2510,7 @@ extern const uint8 kLanguageToPLang[kLanguageToPLangSize] = {
     4,  // 3 fi
     5,  // 4 fr
     6,  // 5 de
-    1,  // 6 iw
+    1,  // 6 he
     7,  // 7 it
     2,  // 8 ja
     3,  // 9 ko

diff --git a/browser/components/translation/cld2/internal/generated_language.h b/browser/components/translation/cld2/internal/generated_language.h
@@ -35,7 +35,7 @@ typedef enum {
   FINNISH                      = 3,  // fi
   FRENCH                       = 4,  // fr
   GERMAN                       = 5,  // de
-  HEBREW                       = 6,  // iw
+  HEBREW                       = 6,  // he
   ITALIAN                      = 7,  // it
   JAPANESE                     = 8,  // ja
   KOREAN                       = 9,  // ko

diff --git a/browser/components/translation/test/unit/test_cld2.js b/browser/components/translation/test/unit/test_cld2.js