CONSTRAINT_ERROR on reading a file containing “[”

问题:

I am reading a simple text file. Everything works as it should, except when an open bracket ("[") character is encountered. Then I get a CONSTRAINT_ERROR.

My function is:

----------------------------------------------
-- GET_FILE_CONTENTS
function Get_File_Contents (File_Name : in String)
    return String_Array is
    -- Loads the entire file into a dynamically sized
    -- array of Unbounded_Wide_String.

    -- The line count is used to dynamically size the array.
    Line_Count : Natural
               := 0;

    File : Ada.Wide_Text_IO.File_Type;
begin
    -- Get the line count before opening the file.
    Line_Count := Get_File_Line_Count (File_Name);

    Ada.Wide_Text_IO.Open (File,
                           In_File,
                           File_Name);
    declare
        Lines : String_Array (1 .. Line_Count);
    begin

        -- Step through the file and save each line.
        for Current_Line in reverse 1 .. Line_Count loop
            Lines(Current_Line) := To_Unbounded_Wide_String (Ada.Wide_Text_IO.Get_Line (File));
        end loop;

    -- Remember to close the file.
    Ada.Wide_Text_IO.Close (File);
    return Lines;

end;
end Get_File_Contents;

The CONSTRAINT_ERROR is raised in "s-wchcnv.adb:207". The relevant part of the file is this

when WCEM_Brackets =>
    if C /= '[' then
       return Character'Pos (C);
    end if;

    if In_Char /= '"' then
       raise Constraint_Error; <======= CONSTRAINT_ERROR
    end if;

This is the info I found:

--  Note on the use of brackets encoding (WCEM_Brackets). The brackets
--  encoding method is ambiguous in the context of this function, since
--  there is no way to tell if ["1234"] is eight unencoded characters or
--  one encoded character. In the context of Ada sources, any sequence
--  starting [" must be the start of an encoding (since that sequence is
--  not valid in Ada source otherwise). The routines in this package use
--  the same approach. If the input string contains the sequence [" then
--  this is assumed to be the start of a brackets encoding sequence, and
--  if it does not match the syntax, an error is raised.

---

How do I turn this off, so that "[" aren't being interpreted as WCEM?

Edit:

Compiler version is GNAT GPL 2014 (20140331) Operating system is Windows 7 x64

Small program that reproduces the CONSTRAINT_ERROR:

with Ada.Wide_Text_IO; use Ada.Wide_Text_IO;

procedure Main is
    File : File_Type;
    File_Name : String := "Data.txt";
begin
    Open (File,
          In_File,
          File_Name);

    while not End_Of_File (File) loop
    declare
        Line : Wide_String := Get_Line (File);
    begin
        Put_Line (Line);
    end;
    end loop;

    Close (File);
end Main;

A "Data.txt" file needs to be near the executable.

If "Data.txt" contains text as below, no problem.

Hello
abc

Add "[" and a CONSTRAINT_ERROR is raised:

Hello[
abc

回答1:

Use the "Form" parameter to the Open call to specify the character encoding as something other than "WCEM=b" for bracket.

with Ada.Wide_Text_IO; use Ada.Wide_Text_IO;

procedure Main is
    File : File_Type;
    File_Name : String := "Data.txt";
begin
    Open (File,
          In_File,
          File_Name, 
          Form => "WCEM=8");

    while not End_Of_File (File) loop
    declare
        Line : Wide_String := Get_Line (File);
    begin
        Put_Line (Line);
    end;
    end loop;

    Close (File);
end Main;

For the relevant documentation, see here

Keith Thompson deserves the credit for finding the relevant documentation.