+/** @file\r
+ Language Library implementation that provides functions for language conversion\r
+ between ISO 639-2 and RFC 4646 language codes.\r
+\r
+ Copyright (c) 2009, Intel Corporation<BR>\r
+ All rights reserved. This program and the accompanying materials\r
+ are licensed and made available under the terms and conditions of the BSD License\r
+ which accompanies this distribution. The full text of the license may be found at\r
+ http://opensource.org/licenses/bsd-license.php\r
+\r
+ THE PROGRAM IS DISTRIBUTED UNDER THE BSD LICENSE ON AN "AS IS" BASIS,\r
+ WITHOUT WARRANTIES OR REPRESENTATIONS OF ANY KIND, EITHER EXPRESS OR IMPLIED.\r
+\r
+**/\r
+\r
+#include <Library/BaseLib.h>\r
+#include <Library/DebugLib.h>\r
+#include <Library/MemoryAllocationLib.h>\r
+#include <Library/LanguageLib.h>\r
+\r
+//\r
+// Lookup table of ISO639-2 3 character language codes to ISO 639-1 2 character language codes\r
+// Each entry is 5 CHAR8 values long. The first 3 CHAR8 values are the ISO 639-2 code.\r
+// The last 2 CHAR8 values are the ISO 639-1 code.\r
+//\r
+// ISO 639-2 B codes and deprecated ISO 639-1 codes are not supported.\r
+//\r
+// Commonly used language codes such as English and French are put in the front of the table for quick match.\r
+//\r
+GLOBAL_REMOVE_IF_UNREFERENCED CONST CHAR8 mIso639ToRfc4646ConversionTable[] =\r
+"\\r
+engen\\r
+frafr\\r
+aaraa\\r
+abkab\\r
+aveae\\r
+afraf\\r
+akaak\\r
+amham\\r
+argan\\r
+araar\\r
+asmas\\r
+avaav\\r
+aymay\\r
+azeaz\\r
+bakba\\r
+belbe\\r
+bulbg\\r
+bihbh\\r
+bisbi\\r
+bambm\\r
+benbn\\r
+bodbo\\r
+brebr\\r
+bosbs\\r
+catca\\r
+chece\\r
+chach\\r
+cosco\\r
+crecr\\r
+cescs\\r
+chucu\\r
+chvcv\\r
+cymcy\\r
+danda\\r
+deude\\r
+divdv\\r
+dzodz\\r
+eweee\\r
+ellel\\r
+epoeo\\r
+spaes\\r
+estet\\r
+euseu\\r
+fasfa\\r
+fulff\\r
+finfi\\r
+fijfj\\r
+faofo\\r
+fryfy\\r
+glega\\r
+glagd\\r
+glggl\\r
+grngn\\r
+gujgu\\r
+glvgv\\r
+hauha\\r
+hebhe\\r
+hinhi\\r
+hmoho\\r
+hrvhr\\r
+hatht\\r
+hunhu\\r
+hyehy\\r
+herhz\\r
+inaia\\r
+indid\\r
+ileie\\r
+iboig\\r
+iiiii\\r
+ipkik\\r
+idoio\\r
+islis\\r
+itait\\r
+ikuiu\\r
+jpnja\\r
+javjv\\r
+katka\\r
+konkg\\r
+kikki\\r
+kuakj\\r
+kazkk\\r
+kalkl\\r
+khmkm\\r
+kankn\\r
+korko\\r
+kaukr\\r
+kasks\\r
+kurku\\r
+komkv\\r
+corkw\\r
+kirky\\r
+latla\\r
+ltzlb\\r
+luglg\\r
+limli\\r
+linln\\r
+laolo\\r
+litlt\\r
+lublu\\r
+lavlv\\r
+mlgmg\\r
+mahmh\\r
+mrimi\\r
+mkdmk\\r
+malml\\r
+monmn\\r
+marmr\\r
+msams\\r
+mltmt\\r
+myamy\\r
+nauna\\r
+nobnb\\r
+ndend\\r
+nepne\\r
+ndong\\r
+nldnl\\r
+nnonn\\r
+norno\\r
+nblnr\\r
+navnv\\r
+nyany\\r
+ocioc\\r
+ojioj\\r
+ormom\\r
+orior\\r
+ossos\\r
+panpa\\r
+plipi\\r
+polpl\\r
+pusps\\r
+porpt\\r
+quequ\\r
+rohrm\\r
+runrn\\r
+ronro\\r
+rusru\\r
+kinrw\\r
+sansa\\r
+srdsc\\r
+sndsd\\r
+smese\\r
+sagsg\\r
+sinsi\\r
+slksk\\r
+slvsl\\r
+smosm\\r
+snasn\\r
+somso\\r
+sqisq\\r
+srpsr\\r
+sswss\\r
+sotst\\r
+sunsu\\r
+swesv\\r
+swasw\\r
+tamta\\r
+telte\\r
+tgktg\\r
+thath\\r
+tirti\\r
+tuktk\\r
+tgltl\\r
+tsntn\\r
+tonto\\r
+turtr\\r
+tsots\\r
+tattt\\r
+twitw\\r
+tahty\\r
+uigug\\r
+ukruk\\r
+urdur\\r
+uzbuz\\r
+venve\\r
+vievi\\r
+volvo\\r
+wlnwa\\r
+wolwo\\r
+xhoxh\\r
+yidyi\\r
+yoryo\\r
+zhaza\\r
+zhozh\\r
+zulzu\\r
+";\r
+\r
+/**\r
+ Converts upper case ASCII characters in an ASCII string to lower case ASCII \r
+ characters in an ASCII string.\r
+\r
+ If a an ASCII character in Source is in the range 'A'..'Z', then it is converted \r
+ to an ASCII character in the range 'a'..'z' in Destination. Otherwise, no \r
+ conversion is performed. Length ASCII characters from Source are convertered and\r
+ stored in Destination.\r
+\r
+ @param Destination An ASCII string to store the results of the conversion.\r
+ @param Source The source ASCII string of the conversion.\r
+ @param Length The number of ASCII characters to convert.\r
+\r
+**/\r
+VOID\r
+EFIAPI\r
+InternalLanguageLibToLower (\r
+ OUT CHAR8 *Destination,\r
+ IN CONST CHAR8 *Source,\r
+ IN UINTN Length\r
+ )\r
+{\r
+ for (; Length > 0; Length--, Destination++, Source++) {\r
+ *Destination = (*Source >= 'A' && *Source <= 'Z') ? *Source + ('a' - 'A') : *Source;\r
+ }\r
+}\r
+\r
+/**\r
+ Convert an ISO 639-2 language code to a RFC 4646 language code.\r
+ If the ISO 639-2 language code has a corresponding ISO 639-1 code, then the ISO 639-1\r
+ code is returned. Else the original ISO 639-2 code is returned. The returned RFC 4646\r
+ language code is composed of only a primary language subtag.\r
+\r
+ If Iso639Language is NULL, then ASSERT.\r
+ If Rfc4646Language is NULL, then ASSERT.\r
+\r
+ @param[out] Rfc4646Language Pointers to a buffer large enough for an ASCII string\r
+ which reprsents a RFC 4646 language code containging only\r
+ either a ISO 639-1 or ISO 639-2 primary language subtag.\r
+ This string is Null-terminated.\r
+ @param[in] Iso639Language Pointer to a 3-letter ASCII string which represents\r
+ an ISO 639-2 language code. This string is not required\r
+ to be Null-terminated.\r
+\r
+ @retval TRUE The ISO 639-2 language code was converted to a ISO 639-1 code.\r
+ @retval FALSE The language code does not have corresponding ISO 639-1 code.\r
+\r
+**/\r
+BOOLEAN\r
+EFIAPI\r
+ConvertIso639ToRfc4646 (\r
+ OUT CHAR8 *Rfc4646Language,\r
+ IN CONST CHAR8 *Iso639Language\r
+ )\r
+{\r
+ CONST CHAR8 *Match;\r
+ \r
+ ASSERT (Iso639Language != NULL);\r
+ ASSERT (Rfc4646Language != NULL);\r
+\r
+ //\r
+ // Convert first 3 characters of Iso639Language to lower case ASCII characters in Rfc4646Language\r
+ //\r
+ InternalLanguageLibToLower (Rfc4646Language, Iso639Language, 3);\r
+ Rfc4646Language[3] = '\0';\r
+\r
+ Match = mIso639ToRfc4646ConversionTable;\r
+ do {\r
+ Match = AsciiStrStr (Match, Rfc4646Language);\r
+ if (Match == NULL) {\r
+ return FALSE;\r
+ }\r
+ if (((Match - mIso639ToRfc4646ConversionTable) % 5) == 0) {\r
+ break;\r
+ }\r
+ ++Match;\r
+ } while (TRUE);\r
+ Rfc4646Language[0] = Match[3];\r
+ Rfc4646Language[1] = Match[4];\r
+ Rfc4646Language[2] = '\0';\r
+ return TRUE;\r
+}\r
+\r
+/**\r
+ Convert a RFC 4646 language code to an ISO 639-2 language code. The primary language\r
+ subtag of the RFC 4646 code must be either an ISO 639-1 or 639-2 code. If the primary\r
+ language subtag is an ISO 639-1 code, then it is converted to its corresponding ISO 639-2\r
+ code (T code if applies). Else the ISO 639-2 code is returned.\r
+\r
+ If Rfc4646Language is NULL, then ASSERT.\r
+ If Iso639Language is NULL, then ASSERT.\r
+\r
+ @param[out] Iso639Language Pointers to a buffer large enough for a 3-letter ASCII string\r
+ which reprsents an ISO 639-2 language code. The string is Null-terminated.\r
+ @param[in] Rfc4646Language Pointer to a RFC 4646 language code string. This string is terminated\r
+ by a NULL or a ';' character.\r
+\r
+ @retval TRUE Language code converted successfully.\r
+ @retval FALSE The RFC 4646 language code is invalid or unsupported.\r
+\r
+**/\r
+BOOLEAN\r
+EFIAPI\r
+ConvertRfc4646ToIso639 (\r
+ OUT CHAR8 *Iso639Language,\r
+ IN CONST CHAR8 *Rfc4646Language\r
+ )\r
+{\r
+ CONST CHAR8 *Match;\r
+ \r
+ ASSERT (Rfc4646Language != NULL);\r
+ ASSERT (Iso639Language != NULL);\r
+\r
+ //\r
+ // RFC 4646 language code check before determining \r
+ // if the primary language subtag is ISO 639-1 or 639-2 code\r
+ //\r
+ if (Rfc4646Language[0] == '\0' || Rfc4646Language[1] == '\0') {\r
+ return FALSE;\r
+ }\r
+ \r
+ //\r
+ // Check if the primary language subtag is ISO 639-1 code\r
+ //\r
+ if (Rfc4646Language[2] == ';' || Rfc4646Language[2] == '-' || Rfc4646Language[2] == '\0') {\r
+ //\r
+ // Convert first 2 characters of Rfc4646Language to lower case ASCII characters in Iso639Language\r
+ //\r
+ InternalLanguageLibToLower (Iso639Language, Rfc4646Language, 2);\r
+ //\r
+ // Convert ISO 639-1 code to ISO 639-2 code\r
+ //\r
+ Iso639Language[2] = '\0';\r
+ Match = mIso639ToRfc4646ConversionTable;\r
+ do {\r
+ Match = AsciiStrStr (Match, Iso639Language);\r
+ if (Match == NULL) {\r
+ return FALSE;\r
+ }\r
+ if (((Match - mIso639ToRfc4646ConversionTable) % 5) == 3) {\r
+ break;\r
+ }\r
+ ++Match;\r
+ } while (TRUE);\r
+ Rfc4646Language = Match - 3;\r
+ } else if (!(Rfc4646Language[3] == ';' || Rfc4646Language[3] == '-' || Rfc4646Language[3] == '\0')) {\r
+ return FALSE;\r
+ }\r
+ Iso639Language[0] = Rfc4646Language[0];\r
+ Iso639Language[1] = Rfc4646Language[1];\r
+ Iso639Language[2] = Rfc4646Language[2];\r
+ Iso639Language[3] = '\0';\r
+ return TRUE; \r
+}\r
+\r
+/**\r
+ Convert ISO 639-2 language codes to RFC 4646 codes and return the converted codes.\r
+ Caller is responsible for freeing the allocated buffer.\r
+\r
+ If Iso639Languages is NULL, then ASSERT.\r
+\r
+ @param[in] Iso639Languages Pointers to a Null-terminated ISO 639-2 language codes string containing\r
+ one or more ISO 639-2 3-letter language codes.\r
+ \r
+ @retval NULL Invalid ISO 639-2 language code found.\r
+ @retval NULL Out of memory.\r
+ @retval !NULL Pointer to the allocate buffer containing the Null-terminated converted language codes string.\r
+ This string is composed of one or more RFC4646 language codes each of which has only\r
+ ISO 639-1 2-letter primary language subtag.\r
+\r
+**/\r
+CHAR8 *\r
+EFIAPI\r
+ConvertLanguagesIso639ToRfc4646 (\r
+ IN CONST CHAR8 *Iso639Languages\r
+ )\r
+{\r
+ UINTN Length;\r
+ UINTN Iso639Index;\r
+ UINTN Rfc4646Index;\r
+ CHAR8 *Rfc4646Languages;\r
+ \r
+ ASSERT (Iso639Languages != NULL);\r
+ \r
+ //\r
+ // The length of ISO 639-2 lanugage codes string must be multiple of 3\r
+ //\r
+ Length = AsciiStrLen (Iso639Languages);\r
+ if (Length % 3) {\r
+ return NULL;\r
+ }\r
+ \r
+ //\r
+ // Allocate buffer for RFC 4646 language codes string\r
+ //\r
+ Rfc4646Languages = AllocatePool (Length + (Length / 3));\r
+ if (Rfc4646Languages == NULL) {\r
+ return NULL;\r
+ }\r
+\r
+ for (Iso639Index = 0, Rfc4646Index = 0; Iso639Languages[Iso639Index] != '\0'; Iso639Index += 3) {\r
+ if (ConvertIso639ToRfc4646 (&Rfc4646Languages[Rfc4646Index], &Iso639Languages[Iso639Index])) {\r
+ Rfc4646Index += 2;\r
+ } else {\r
+ Rfc4646Index += 3;\r
+ }\r
+ Rfc4646Languages[Rfc4646Index++] = ';';\r
+ }\r
+ Rfc4646Languages[Rfc4646Index - 1] = '\0';\r
+ return Rfc4646Languages;\r
+}\r
+\r
+/**\r
+ Convert RFC 4646 language codes to ISO 639-2 codes and return the converted codes.\r
+ The primary language subtag of the RFC 4646 code must be either an ISO 639-1 or 639-2 code.\r
+ Caller is responsible for freeing the allocated buffer.\r
+\r
+ If Rfc4646Languages is NULL, then ASSERT.\r
+\r
+ @param[in] Rfc4646Languages Pointers to a Null-terminated RFC 4646 language codes string containing\r
+ one or more RFC 4646 language codes.\r
+ \r
+ @retval NULL Invalid or unsupported RFC 4646 language code found.\r
+ @retval NULL Out of memory.\r
+ @retval !NULL Pointer to the allocate buffer containing the Null-terminated converted language codes string.\r
+ This string is composed of one or more ISO 639-2 language codes.\r
+\r
+**/\r
+CHAR8 *\r
+EFIAPI\r
+ConvertLanguagesRfc4646ToIso639 (\r
+ IN CONST CHAR8 *Rfc4646Languages\r
+ )\r
+{\r
+ UINTN NumLanguages;\r
+ UINTN Iso639Index;\r
+ UINTN Rfc4646Index;\r
+ CHAR8 *Iso639Languages;\r
+\r
+ ASSERT (Rfc4646Languages != NULL);\r
+\r
+ //\r
+ // Determine the number of languages in the RFC 4646 language codes string\r
+ //\r
+ for (Rfc4646Index = 0, NumLanguages = 1; Rfc4646Languages[Rfc4646Index] != '\0'; Rfc4646Index++) {\r
+ if (Rfc4646Languages[Rfc4646Index] == ';') {\r
+ NumLanguages++;\r
+ }\r
+ }\r
+ \r
+ //\r
+ // Allocate buffer for ISO 639-2 language codes string\r
+ //\r
+ Iso639Languages = AllocateZeroPool (NumLanguages * 3 + 1);\r
+ if (Iso639Languages == NULL) {\r
+ return NULL;\r
+ }\r
+\r
+ //\r
+ // Do the conversion for each RFC 4646 language code\r
+ //\r
+ for (Rfc4646Index = 0, Iso639Index = 0; Rfc4646Languages[Rfc4646Index] != '\0';) {\r
+ if (ConvertRfc4646ToIso639 (&Iso639Languages[Iso639Index], &Rfc4646Languages[Rfc4646Index])) {\r
+ Iso639Index += 3;\r
+ } else {\r
+ FreePool (Iso639Languages);\r
+ return NULL;\r
+ }\r
+ //\r
+ // Locate next language code\r
+ //\r
+ while (Rfc4646Languages[Rfc4646Index] != ';' && Rfc4646Languages[Rfc4646Index] != '\0') {\r
+ Rfc4646Index++;\r
+ }\r
+ if (Rfc4646Languages[Rfc4646Index] == ';') {\r
+ Rfc4646Index++;\r
+ }\r
+ }\r
+ Iso639Languages[Iso639Index] = '\0';\r
+ return Iso639Languages;\r
+}\r
+\r
projects / mirror_edk2.git / commitdiff