URL
https://opencores.org/ocsvn/openrisc/openrisc/trunk
Subversion Repositories openrisc
[/] [openrisc/] [trunk/] [gnu-stable/] [gcc-4.5.1/] [gcc/] [ada/] [widechar.adb] - Rev 826
Compare with Previous | Blame | View Log
------------------------------------------------------------------------------ -- -- -- GNAT COMPILER COMPONENTS -- -- -- -- W I D E C H A R -- -- -- -- B o d y -- -- -- -- Copyright (C) 1992-2009 Free Software Foundation, Inc. -- -- -- -- GNAT is free software; you can redistribute it and/or modify it under -- -- terms of the GNU General Public License as published by the Free Soft- -- -- ware Foundation; either version 3, or (at your option) any later ver- -- -- sion. GNAT is distributed in the hope that it will be useful, but WITH- -- -- OUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY -- -- or FITNESS FOR A PARTICULAR PURPOSE. -- -- -- -- As a special exception under Section 7 of GPL version 3, you are granted -- -- additional permissions described in the GCC Runtime Library Exception, -- -- version 3.1, as published by the Free Software Foundation. -- -- -- -- You should have received a copy of the GNU General Public License and -- -- a copy of the GCC Runtime Library Exception along with this program; -- -- see the files COPYING3 and COPYING.RUNTIME respectively. If not, see -- -- <http://www.gnu.org/licenses/>. -- -- -- -- GNAT was originally developed by the GNAT team at New York University. -- -- Extensive contributions were provided by Ada Core Technologies Inc. -- -- -- ------------------------------------------------------------------------------ -- Note: this package uses the generic subprograms in System.Wch_Cnv, which -- completely encapsulate the set of wide character encoding methods, so no -- modifications are required when adding new encoding methods. with Opt; use Opt; with System.WCh_Cnv; use System.WCh_Cnv; with System.WCh_Con; use System.WCh_Con; package body Widechar is --------------------------- -- Is_Start_Of_Wide_Char -- --------------------------- function Is_Start_Of_Wide_Char (S : Source_Buffer_Ptr; P : Source_Ptr) return Boolean is begin case Wide_Character_Encoding_Method is -- For Hex mode, just test for an ESC character. The ESC character -- cannot appear in any other context in a legal Ada program. when WCEM_Hex => return S (P) = ASCII.ESC; -- For brackets, just test ["x where x is a hex character. This is -- sufficient test, since this sequence cannot otherwise appear in a -- legal Ada program. when WCEM_Brackets => return P <= S'Last - 2 and then S (P) = '[' and then S (P + 1) = '"' and then (S (P + 2) in '0' .. '9' or else S (P + 2) in 'a' .. 'f' or else S (P + 2) in 'A' .. 'F'); -- All other encoding methods use the upper bit set in the first -- character to uniquely represent a wide character. when WCEM_Upper | WCEM_Shift_JIS | WCEM_EUC | WCEM_UTF8 => return S (P) >= Character'Val (16#80#); end case; end Is_Start_Of_Wide_Char; ----------------- -- Length_Wide -- ----------------- function Length_Wide return Nat is begin return WC_Longest_Sequence; end Length_Wide; --------------- -- Scan_Wide -- --------------- procedure Scan_Wide (S : Source_Buffer_Ptr; P : in out Source_Ptr; C : out Char_Code; Err : out Boolean) is P_Init : constant Source_Ptr := P; Chr : Character; function In_Char return Character; -- Function to obtain characters of wide character escape sequence ------------- -- In_Char -- ------------- function In_Char return Character is begin P := P + 1; return S (P - 1); end In_Char; function WC_In is new Char_Sequence_To_UTF_32 (In_Char); -- Start of processing for Scan_Wide begin Chr := In_Char; -- Scan out the wide character. If the first character is a bracket, -- we allow brackets encoding regardless of the standard encoding -- method being used, but otherwise we use this standard method. if Chr = '[' then C := Char_Code (WC_In (Chr, WCEM_Brackets)); else C := Char_Code (WC_In (Chr, Wide_Character_Encoding_Method)); end if; Err := False; Wide_Char_Byte_Count := Wide_Char_Byte_Count + Nat (P - P_Init - 1); exception when Constraint_Error => C := Char_Code (0); P := P - 1; Err := True; end Scan_Wide; -------------- -- Set_Wide -- -------------- procedure Set_Wide (C : Char_Code; S : in out String; P : in out Natural) is procedure Out_Char (C : Character); -- Procedure to store one character of wide character sequence -------------- -- Out_Char -- -------------- procedure Out_Char (C : Character) is begin P := P + 1; S (P) := C; end Out_Char; procedure WC_Out is new UTF_32_To_Char_Sequence (Out_Char); -- Start of processing for Set_Wide begin WC_Out (UTF_32_Code (C), Wide_Character_Encoding_Method); end Set_Wide; --------------- -- Skip_Wide -- --------------- procedure Skip_Wide (S : String; P : in out Natural) is P_Init : constant Natural := P; function Skip_Char return Character; -- Function to skip one character of wide character escape sequence --------------- -- Skip_Char -- --------------- function Skip_Char return Character is begin P := P + 1; return S (P - 1); end Skip_Char; function WC_Skip is new Char_Sequence_To_UTF_32 (Skip_Char); Discard : UTF_32_Code; pragma Warnings (Off, Discard); -- Start of processing for Skip_Wide begin Discard := WC_Skip (Skip_Char, Wide_Character_Encoding_Method); Wide_Char_Byte_Count := Wide_Char_Byte_Count + Nat (P - P_Init - 1); end Skip_Wide; --------------- -- Skip_Wide -- --------------- procedure Skip_Wide (S : Source_Buffer_Ptr; P : in out Source_Ptr) is P_Init : constant Source_Ptr := P; function Skip_Char return Character; -- Function to skip one character of wide character escape sequence --------------- -- Skip_Char -- --------------- function Skip_Char return Character is begin P := P + 1; return S (P - 1); end Skip_Char; function WC_Skip is new Char_Sequence_To_UTF_32 (Skip_Char); Discard : UTF_32_Code; pragma Warnings (Off, Discard); -- Start of processing for Skip_Wide begin Discard := WC_Skip (Skip_Char, Wide_Character_Encoding_Method); Wide_Char_Byte_Count := Wide_Char_Byte_Count + Nat (P - P_Init - 1); end Skip_Wide; end Widechar;