|
Table of Content | Chapter Thirteen (Part 7) |
|
Table of Content | Chapter Thirteen (Part 7) |
Starting with DOS v2.0, Microsoft introduced a set of file handling procedures which (finally) made disk file access bearable under MS-DOS. Not only bearable, but actually easy to use! The following sections describe the use of these commands to access files on a disk drive.
File commands which deal with filenames (Create, Open, Delete, Rename, and others) are passed the address of a zero-terminated pathname. Those that actually open a file (Create and Open) return a file handle as the result (assuming, of course, that there wasn't an error). This file handle is used with other calls (read, write, seek, close, etc.) to gain access to the file you've opened. In this respect, a file handle is not unlike a file variable in Pascal. Consider the following Microsoft/Turbo Pascal code:
program DemoFiles; var F:TEXT;
begin
assign(f,'FileName.TXT');
rewrite(f);
writeln(f,'Hello there');
close(f);
end.
The file variable "f" is used in this Pascal example in much the same way that a file handle is used in an assembly language program - to gain access to the file that was created in the program.
All the following DOS filing commands return an error status in the carry flag. If the carry flag is clear when DOS returns to your program, then the operation was completed successfully. If the carry flag is set upon return, then some sort of error has occurred and the AX register contains the error number. The actual error return values will be discussed along with each function in the following sections.
Function (ah): 3Dh
Entry parameters:
al- file access value
0- File opened for reading
1- File opened for writing
2- File opened for reading and writing
ds:dx- Point at a zero terminated string containing the filename.
Exit parameters:
If the carry is set, ax contains one of the following error codes:
2- File not found
4- Too many open files
5- Access denied
12- Invalid access
If the carry is clear,
ax contains the file handle value assigned by DOS.
A file must be opened before you can access it. The open command opens a file that already exists. This makes it quite similar to Pascal's Reset procedure. Attempting to open a file that doesn't exist produces an error. Example:
lea dx, Filename ;Assume DS points at segment
mov ah, 3dh ; of filename
mov al, 0 ;Open for reading.
int 21h
jc OpenError
mov FileHandle, ax
If an error occurs while opening a file, the file will not be opened. You should always check for an error after executing a DOS open command, since continuing to operate on the file which hasn't been properly opened will produce disastrous consequences. Exactly how you handle an open error is up to you, but at the very least you should print an error message and give the user the opportunity to specify a different filename.
If the open command completes without generating an error,
DOS returns a file handle for that file in the ax register. Typically, you
should save this value away somewhere so you can use it when accessing the file later on.
Function (ah): 3Ch
Entry parameters:
ds:dx- Address of zero terminated pathname
cx- File attribute
Exit parameters:
If the carry is set, ax contains one of the following error codes:
3- Path not found
4- Too many open files
5- Access denied
If the carry is clear,
ax is returned containing the file handle
Create opens a new file for output. As with the OPEN
command, ds:dx points at a zero terminated string containing the filename.
Since this call creates a new file, DOS assumes that you're opening the file for writing
only. Another parameter, passed in cx, is the initial file attribute
settings. The L.O. six bits of cx contain the following values:
Bit Meaning if equal to one 0 File is a Read-Only file 1 File is a hidden file 2 File is a system file 3 File is a volume label name 4 File is a subdirectory 5 File has been archived
In general, you shouldn't set any of these bits. Most
normal files should be created with a file attribute of zero. Therefore, the cx register
should be loaded with zero before calling the create function.
Upon exit, the carry flag is set if an error occurs. The "Path not found" error requires some additional explanation. This error is generated, not if the file isn't found (which would be most of the time since this command is typically used to create a new file), but if a subdirectory in the pathname cannot be found.
If the carry flag is clear when DOS returns to your
program, then the file has been properly opened for output and the ax register
contains the file handle for this file.
Function (ah): 3Eh
Entry parameters:
bx- File Handle
Exit parameters:
If the carry flag is set,
ax contains 6, the only possible error,
which is an invalid handle error.
This call is used to close a file opened with the Open or
Create commands above. It is passed the file handle in the bx register and,
assuming the file handle is valid, closes the specified file.
You should close all files your program uses as soon as you're through with them to avoid disk file corruption in the event the user powers the system down or resets the machine while your files are left open.
Note that quitting to DOS (or aborting to DOS by pressing control-C or control-break) automatically closes all open files. However, you should never rely on this feature since doing so is an extremely poor programming practice.
Function (ah): 3Fh
Entry parameters:
bx- File handle
cx- Number of bytes to read
ds:dx- Array large enough to hold bytes read
Exit parameters:
If the carry flag is set,
ax contains one of the following error codes
5- Access denied
6- Invalid handle
If the carry flag is clear,
ax contains the number of bytes actually read from the file.
The read function is used to read some number of bytes from
a file. The actual number of bytes is specified by the cx register upon entry
into DOS. The file handle, which specifies the file from which the bytes are to be read,
is passed in the bx register. The ds:dx register contains the
address of a buffer into which the bytes read from the file are to be stored.
On return, if there wasn't an error, the ax register
contains the number of bytes actually read. Unless the end of file (EOF) was reached, this
number will match the value passed to DOS in the cx register. If the end of
file has been reached, the value return in ax will be somewhere between zero
and the value passed to DOS in the cx register. This is the only test for the
EOF condition.
Example: This example opens a file and reads it to the EOF
mov ah, 3dh ;Open the file
mov al, 0 ;Open for reading
lea dx, Filename ;Presume DS points at filename
int 21h ; segment.
jc BadOpen
mov FHndl, ax ;Save file handle
LP: mov ah,3fh ;Read data from the file
lea dx, Buffer ;Address of data buffer
mov cx, 1 ;Read one byte
mov bx, FHndl ;Get file handle value
int 21h
jc ReadError
cmp ax, cx ;EOF reached?
jne EOF
mov al, Buffer ;Get character read
putc ;Print it
jmp LP ;Read next byte
EOF: mov bx, FHndl
mov ah, 3eh ;Close file
int 21h
jc CloseError
This code segment will read the entire file whose
(zero-terminated) filename is found at address "Filename" in the current data
segment and write each character in the file to the standard output device using the UCR
StdLib putc routine. Be forewarned that one-character-at-a-time I/O such as
this is extremely slow. We'll discuss better ways to quickly read a file a little later in
this chapter.
Function (ah): 40h
Entry parameters:
bx- File handle
cx- Number of bytes to write
ds:dx- Address of buffer containing data to write
Exit parameters:
If the carry is set,
ax contains one of the following error codes
5- Accessed denied
6- Invalid handle
If the carry is clear on return,
ax contains the number of bytes actually written to the file.
This call is almost the converse of the read command
presented earlier. It writes the specified number of bytes at ds:dx to the
file rather than reading them. On return, if the number of bytes written to the file is
not equal to the number originally specified in the cx register, the disk is
full and this should be treated as an error.
If cx contains zero when this function is
called, DOS will truncate the file to the current file position (i.e., all data following
the current position in the file will be deleted).
3.3.8.6 Seek (Move File Pointer)
Function (ah): 42h Entry parameters:
al- Method of moving
0- Offset specified is from the beginning of the file.
1- Offset specified is distance from the current file pointer.
2- The pointer is moved to the end of the file minus the specified offset.
bx- File handle.
cx:dx- Distance to move, in bytes.
Exit parameters:
If the carry is set,
ax contains one of the following error codes
1- Invalid function
6- Invalid handle
If the carry is clear,
dx:ax contains the new file position
This command is used to move the file pointer around in a
random access file. There are three methods of moving the file pointer, an absolute
distance within the file (if al=0), some positive distance from the current
file position (if al=1), or some distance from the end of the file (if al=2).
If AL doesn't contain 0, 1, or 2, DOS will return an invalid function error. If this call
is successfully completed, the next byte read or written will occur at the specified
location.
Note that DOS treats cx:dx as an unsigned
integer. Therefore, a single seek command cannot be used to move backwards in the file.
Instead, method #0 must be used to position the file pointer at some absolute position in
the file. If you don't know where you currently are and you want to move back 256 bytes,
you can use the following code:
mov ah, 42h ;Seek command
mov al, 1 ;Move from current location
xor cx, cx ;Zero out CX and DX so we
xor dx, dx ; stay right here
mov bx, FileHandle
int 21h
jc SeekError
sub ax, 256 ;DX:AX now contains the
sbb dx, 0 ; current file position, so
mov cx, dx ; compute a location 256
mov dx, ax ; bytes back.
mov ah, 42h
mov al, 0 ;Absolute file position
int 21h ;BX still contains handle.
13.3.8.7 Set Disk Transfer Address (DTA)
Function (ah): 1Ah Entry parameters:
ds:dx- Pointer to DTA
Exit parameters:
None
This command is called "Set Disk Transfer Address" because it was (is) used with the original DOS v1.0 file functions. We wouldn't normally consider this function except for the fact that it is also used by functions 4Eh and 4Fh (described next) to set up a pointer to a 43-byte buffer area. If this function isn't executed before executing functions 4Eh or 4Fh, DOS will use the default buffer space at PSP:80h.
Function (ah): 4Eh
Entry parameters:
cx- Attributes
ds:dx- Pointer to filename
Exit parameters:
If carry is set,
ax contains one of the following error codes
2- File not found
18- No more files
The Find First File and Find Next File (described next) functions are used to search for files specified using ambiguous file references. An ambiguous file reference is any filename containing the "*" and "?" wildcard characters. The Find First File function is used to locate the first such filename within a specified directory, the Find Next File function is used to find successive entries in the directory.
Generally, when an ambiguous file reference is provided, the Find First File command is issued to locate the first occurrence of the file, and then a loop is used, calling Find Next File, to locate all other occurrences of the file within that loop until there are no more files (error #18). Whenever Find First File is called, it sets up the following information at the DTA:
Offset Description 0 Reserved for use by Find Next File 21 Attribute of file found 22 Time stamp of file 24 Date stamp of file 26 File size in bytes 30 Filename and extension (zero terminated)
(The offsets are decimal)
Assuming Find First File doesn't return some sort of error, the name of the first file matching the ambiguous file description will appear at offset 30 in the DTA.
Note: if the specified pathname doesn't contain any wildcard characters, then Find First File will return the exact filename specified, if it exists. Any subsequent call to Find Next File will return an error.
The cx register contains the search attributes
for the file. Normally, cx should contain zero. If non-zero, Find First File
(and Find Next File) will include file names which have the specified attributes as well
as all normal file names.
Function (ah): 4Fh
Entry parameters:
none
Exit parameters:
If the carry is set,
then there aren't any more files and
ax will be returned containing 18.
The Find Next File function is used to search for additional file names matching an ambiguous file reference after a call to Find First File. The DTA must point at a data record set up by the Find First File function.
Example: The following code lists the names of all the files in the current directory that end with ".EXE". Presumably, the variable "DTA" is in the current data segment:
mov ah, 1Ah ;Set DTA
lea dx, DTA
int 21h
xor cx, cx ;No attributes.
lea dx, FileName
mov ah, 4Eh ;Find First File
int 21h
jc NoMoreFiles ;If error, we're done
DirLoop: lea si, DTA+30 ;Address of filename
cld
PrtName: lodsb
test al, al ;Zero byte?
jz NextEntry
putc ;Print this character
jmp PrtName
NextEntry: mov ah, 4Fh ;Find Next File
int 21h
jnc DirLoop ;Print this name
13.3.8.10 Delete File
Function (ah): 41h
Entry parameters:
ds:dx- Address of pathname to delete
Exit parameters:
If carry set,
ax contains one of the following error codes
2- File not found
5- Access denied
This function will delete the specified file from the directory. The filename must be an unambiguous filename (i.e., it cannot contain any wildcard characters).
Function (ah): 56h Entry parameters:
ds:dx- Pointer to pathname of existing file
es:di- Pointer to new pathname
Exit parameters:
If carry set,
ax contains one of the following error codes
2- File not found
5- Access denied
17- Not the same device
This command serves two purposes: it allows you to rename one file to another and it allows you to move a file from one directory to another (as long as the two subdirectories are on the same disk).
Example: Rename "MYPGM.EXE" to "YOURPGM.EXE"
; Assume ES and DS both point at the current data segment
; containing the filenames.
lea dx, OldName
lea di, NewName
mov ah, 56h
int 21h
jc BadRename
.
.
.
OldName byte "MYPGM.EXE",0
NewName byte "YOURPGM.EXE",0
Example #2: Move a filename from one directory to another:
; Assume ES and DS both point at the current data segment
; containing the filenames.
lea dx, OldName
lea di, NewName
mov ah, 56h
int 21h
jc BadRename
.
.
.
OldName byte "\DIR1\MYPGM.EXE",0
NewName byte "\DIR2\MYPGM.EXE",0
13.3.8.12 Change/Get File Attributes
Function (ah): 43h
Entry parameters:
al- Subfunction code
0- Return file attributes in cx
1- Set file attributes to those in cx
cx- Attribute to be set if AL=01
ds:dx- address of pathname
Exit parameters:
If carry set,
ax contains one of the following error codes:
1- Invalid function
3- Pathname not found
5- Access denied
If the carry is clear and the subfunction was zero,
cx will contain the file's attributes.
This call is useful for setting/resetting and reading a file's attribute bits. It can be used to set a file to read-only, set/clear the archive bit, or otherwise mess around with the file attributes.
13.3.8.13 Get/Set File Date and Time
Function (ah): 57h
Entry parameters:
al- Subfunction code
0- Get date and time
1- Set date and time
bx- File handle
cx- Time to be set (if AL=01)
dx- Date to be set (if AL=01)
Exit parameters:
If carry set, ax contains one of the following error codes
1- Invalid subfunction
6- Invalid handle
If the carry is clear,
cx/dx is set to the time/date if al=00
This call sets the "last-write" date/time for the specified file. The file must be open (using open or create) before using this function. The date will not be recorded until the file is closed.
The following tables briefly list many of the other DOS calls. For more information on the use of these DOS functions consult the Microsoft MS-DOS Programmer's Reference or the MS-DOS Technical Reference.
| Function #(AH) | Input Parameters |
Output Parameters |
Description |
|---|---|---|---|
| 39h | ds:dx- pointer to zero terminated pathname. |
- | Create Directory: Creates a new directory with the specified name. |
| 3Ah | ds:dx- pointer to zero terminated pathname. |
- | Remove Directory: Deletes the directory with the specified pathname. Error if directory is not empty or the specified directory is the current directory. |
| 3Bh | ds:dx- pointer to zero terminated pathname. |
- | Change Directory: Changes the default directory to the specified pathname. |
| 45h | bx- file handle |
ax- new handle |
Duplicate File Handle: creates a copy of a file handle so a program can access a file using two separate file variables. This allows the program to close the file with one handle and continue accessing it with the other. |
| 46h | bx- file handlecx- duplicate handle |
- |
Force Duplicate File Handle: Like function 45h above, except you specify which handle (in cx) you want to refer to the existing file (specified by bx). |
| 47h | ds:si- pointer to bufferdl- drive |
- | Get Current Directory: Stores a string containing the current pathname (terminated with a zero) starting at location ds:si. These registers must point at a buffer containing at least 64 bytes. The dl register specifies the drive number (0=default, 1=A, 2=B, 3=C, etc.). |
| 5Ah | cx- attributesds:dx- pointer to temporary path. |
ax- handle |
Create Temporary File: Creates a file with a unique name in the directory specified by the zero terminated string at which ds:dx points. There must be at least 13 zero bytes beyond the end of the pathname because this function will store the generated filename at the end of the pathname. The attributes are the same as for the Create call. |
| 5Bh | cx- attributesds:dx- pointer to zero terminated pathname. |
ax- handle |
Create New File: Like the create call, but this call insists that the file not exist. It returns an error if the file exists (rather than deleting the old file). |
| 67h | bx- handles |
- | Set Maximum Handle Count: This function sets the maximum number of handles a program can use at any one given time. |
| 68h | bx- handle |
- | Commit File: Flushes all data to a file without closing it, ensuring that the file's data is current and consistent. |
| Function #(AH) |
Input Parameters | Output Parameters | Description |
|---|---|---|---|
| 25h | al- interrupt # ds:dx- pointer
to interrupt service routine. |
- |
Set Interrupt Vector: Stores the specified address in ds:dx into the interrupt vector table at the entry specified by the al register. |
| 30h | - |
al- major version ah- minor
version bh- Version flag bl:cx- 24 bit serial number |
Get Version Number: Returns the current version number of DOS (or value set by SETVER). |
| 33h | al- 0 |
dl- break flag (0=off, 1=on) |
Get Break Flag: Returns the status of the DOS break flag. If on, MS-DOS checks for ctrl-C when processing any DOS command; if off, MS-DOS only checks on functions 1-0Ch. |
| 33h | al- 1 dl- break flag. |
- |
Set Break Flag: Sets the MS-DOS break flag according to the value in dl (see function above for details). |
| 33h | al- 6 |
bl- major version bh- minor
version dl- revision dh- version flags |
Get MS-DOS Version: Returns the "real" version number, not the one set by the SETVER command. Bits three and four of the version flags are one if DOS is in ROM or DOS is in high memory, respectively. |
| 34h | - |
es:bx- pointer to InDOS flag. |
Get InDOS Flag Address: Returns the address of the InDOS flag. This flag helps prevent reentrancy in TSR applications |
| 35h | al- interrupt # |
es:bx- pointer to interrupt service routine. |
Get Interrupt Vector: Returns a pointer to the interrupt service routine for the specified interrupt number. See function 25h above for more details. |
| 44h | al- subcode Other parameters! |
- |
Device Control: This is a whole family of additional DOS commands to control various devices. See the DOS programmer's reference manual for more details. |
| 4Dh | - |
al- return value ah- termination
method |
Get Child Program Return Value: Returns the last result code from a child program in al. The ah register contains the termination method, which is one of the following values: 0-normal, 1-ctrl-C, 2-critical device error, 3-terminate and stay resident. |
| 50h | bx- PSP address |
- |
Set PSP Address: Set DOS' current PSP address to the value specified in the bx register. |
| 51h | - | bx- PSP address |
Get PSP Address: Returns a pointer to the current PSP in the bx register. |
| 59h | - | ax- extended error bh- error
class bl- error action ch- error location |
Get Extended Error: Returns additional information when an error occurs on a DOS call. See the DOS programmer's guide for more details on these errors and how to handle them. |
| 5Dh | al- 0Ah ds:si- pointer to extended error
structure. |
- |
Set Extended Error: copies the data from the extended error structure to DOS' internal record. |
In addition to the above commands, there are several additional DOS calls that deal with networks and international character sets. See the MS-DOS reference for more details.
|
Table of Content | Chapter Thirteen (Part 7) |
Chapter Thirteen: MS-DOS, PC-BIOS and
File I/O (Part 6)
28 SEP 1996