Simtel MSDOS 1992 June

home *** CD-ROM | disk | FTP | other *** search

/ Simtel MSDOS 1992 June / SIMTEL_0692.cdr / msdos / txtutl / concat10.arc / CONCAT10.ASM < prev next >

Wrap

Assembly Source File | 1989-05-02 | 11.8 KB | 506 lines

TITLE Concat ;Input: ; Read in multiple files (as named on the DOS cmd line). ; Wildcards and full pathing are supported. ; ;Output: ; To StdOut (e.g., via redirection at the DOS command line). ; (No ^Z is appended to the output.) ; ;End of File: ; Assumes the first ^Z in each file is EOF (and the rest is garbage), ; so will discard everything after that first ^Z. ;Notes: ; Makes no attempt to add terminating CR/LF to each file ; or any other such cleanup. ;Released to the public domain ;David Kirschbaum ;Toad Hall ;kirsch@braggvax.ARPA CR equ 0DH LF equ 0AH CTRLZ equ 1AH ;Ctrl-Z STDOUT equ 1 ;DOS StdOut STDERR equ 2 ;DOS StdErr (screen) DTA_TYPE struc reserved db 21 DUP(?) ;Used by DOS for find next attrib db ? ;file attribute time dw ? ;file time stamp date dw ? ;file date stamp filesize dw ?,? ;file size in bytes (long integer) filename db 13 DUP(?) ;AsciiZ filename string DTA_TYPE ends CSEG segment public 'CODE' assume CS:CSEG,DS:CSEG org 80H dta label DTA_TYPE nchar db ? ;for Args params db ? ; org fcb_dta + 26 ;dta filesize ;filesize dw ?,? ; ; org dta + 30 ;DTA.filename ;dtaname db ? ;for wildcard expansion org 100H ;yep, a .COM file Concat proc near jmp Start ;skip over data usagemsg db 'CONCAT - Concatenate text files to StdOut',CR,LF db 'Usage: CONCAT [d:\]file1.txt file2.txt >output.txt',CR,LF db 'Input: Wildcards and full pathing permitted.',CR,LF db 'Output: DOS Standard Output (redirectable).',CR,LF db 'David Kirschbaum, Toad Hall, v1.0' crlf db CR,LF USAGEMSGLEN equ $ - usagemsg CRLFLEN equ $ - crlf notfoundmsg db ' not found',CR,LF NOTFOUNDLEN equ $ - notfoundmsg ;msg length openerrmsg db ' File open error',CR,LF OPENERRLEN equ $ - openerrmsg readerrmsg db ' File read error',CR,LF READERRLEN equ $ - readerrmsg writerrmsg db ' File write error',CR,LF WRITERRLEN equ $ - writerrmsg zerofilemsg db ' Empty file, skipped',CR,LF ZEROFILELEN equ $ - zerofilemsg handle dw 0 ;input file handle Concat endp ;Returns position of char in string ;DI -> string start or end ;AL = char to seek ;Returns psn in CX ;It'll work in reverse also if you enter with DI -> string end ;and STD instead of CLD Find_Char proc near mov cx,0FFFFH ;max scan repne scasb not cx ;CX=char psn ret Find_Char endp ;Display an AsciiZ string to StdErr ;Enters with SI -> AsciiZ string ;Destroys AX,DI ;Returns CX = string length, DX, SI -> AsciiZ string Pr_AsciiZ proc near xor al,al ;find terminating 0 mov di,si ;from string start cld ;go forward call Find_Char ;CX = psn of 0, SI unchanged dec cx ;- 1 for actual length mov dx,si ;string start mov bx,STDERR ;output to StdErr mov ah,40H ;write to file/device int 21H ret Pr_AsciiZ endp ;Separate a drive:\path from an arg filename. ;SI -> arg (an AsciiZ string) ;Pointer pathend -> beyond last char in path (filename[0] if none) Split_Path proc near mov di,si ;from string start xor al,al ;find AsciiZ arg end cld ;go forward call Find_Char ;CX=psn of 0, DI -> beyond 0 dec di ;DI -> AsciiZ 0 dec di ;DI -> last char in filename mov al,'\' ;scan for subdirectories std ;backwards towards start repne scasb ;CX has filename length cld ;insure forward again or cx,cx ;any path? jnz Got_Path ;yep ;No '\',how about ':'? xor cx,cx ;assume no path, no move mov di,si ;string start cmp byte ptr 1[di],':' ;divider should be after drive char jnz No_Path ;nope, DI -> filename's first char ;In either case, DI should be pointing to the char ;BEFORE the path or drive separator. Got_Path: add di,2 ;bump beyond '\' or ':' mov cx,di ;path\filename end sub cx,si ;- path end = bytes to move No_Path: mov di,offset path ;move into path jcxz No_Path_Move rep movsb ;copy args path into buffer No_Path_Move: mov word ptr pathptr,di ;pathptr -> char beyond args path ;where we'll move the filename Split_Path endp ;Move our expanded filename after our arg path. Add_Path proc near mov si,offset dta.filename ;expanded filename mov di,si ;Find_Char wants it in DI xor al,al ;get AsciiZ name length cld ;forward call Find_Char ;CX = char psn, SI unchanged mov di,word ptr pathptr ;-> arg path + 1 (or start) rep movsb ;move in the name ;(include the AsciiZ 0) mov si,offset path ;-> path start ret Add_Path endp ;Displays expanded filename ;Copies the wildcard file to StdOut ;Enter with SI -> expanded filename ;On return: ; handle = file handle (or 0 if never opened) ; CF set = error ; DX=CR/LF or error msg (if error) ; CX = msg length Copy_File proc near mov handle,0 ;insure file handle is cleared call Add_Path ;add arg path to expanded filename ;(we could be doing a Find Next) ;returns SI, DX -> path:\name call Pr_AsciiZ ;display AsciiZ name ;DX -> AsciiZ path:\name mov ax,3D00H ;open for read only int 21H jb Open_Err ;failed mov handle,ax ;save the handle mov si,offset dta.filesize ;DTA filesize long integer lodsw ;get filesize.lo or ax,[si] ;test filesize.hi jz Zero_File ;empty file, forget it ReadLup: mov dx,offset readbuff ;file read buffer mov cx,READBUFFSIZE ;bytes to read mov bx,handle ;file handle mov ah,3FH ;read from file/device int 21H jb Read_Err ;read failed mov cx,ax ;bytes read jcxz File_Eof1 ;nothing read, it's EOF mov bx,ax ;remember bytes read mov di,dx ;DI -> read buffer mov al,CTRLZ ;scan for terminating Ctrl Z repne scasb jz File_EOF ;it's logical EOF mov cx,bx ;restore nr bytes read mov bx,STDOUT ;output to StdOut mov ah,40H ;write to file/device int 21H jnb ReadLup ;wrote ok, keep going Write_Err: mov dx,offset writerrmsg ;' File write error' mov cx,WRITERRLEN ret ;CF set Zero_File: mov dx,offset zerofilemsg ;' Zero file size .. skipped' mov cx,ZEROFILELEN ret ;CF clear Read_Err: mov dx,offset readerrmsg ;' File read error' mov cx,READERRLEN ret ;CF set File_Eof: inc cx ;adjust for that Ctrl Z sub bx,cx ; bytes read - Ctrl Z psn mov cx,bx ;= bytes to write mov bx,STDOUT ;output to StdOut mov ah,40H ;write to file/device int 21H jb Write_Err ;final write failed File_Eof1: mov dx,offset crlf ;new line after filename mov cx,CRLFLEN ret ;CF should be clear Open_Err: mov dx,offset openerrmsg ;'File open error' mov cx,OPENERRLEN ret ;CF set Copy_File endp ;Do a wildcard copy. ;SI -> AsciiZ filename OutPut proc near mov dx,si ;DX = AsciiZ filename ptr xor cx,cx ;no special attributes mov ah,4EH ;find first int 21H or ax,ax ;find anything? jz Found_First ;yep call Pr_AsciiZ ;display filename (still in SI) mov dx,offset notfoundmsg ;' not found' mov cx,NOTFOUNDLEN ;msg length stc ;CF set to exit jmp short Not_Found ;display error msg, return ;Remember, a Find First or Find Next will give us an expanded filename ;(in the DTA) but will NOT give us a path! ;We have to process the arg (SI still points to it) and split off ;the path from the filename. Found_First: call Split_Path ;separate arg path from arg ;arg filename (for Find First only). ;fall thru to... Find_Next: ;our wildcard loop call Copy_File ;open, copy the file ;On return: ; handle = file handle (or 0 if never opened) ; CF set = error ; DX=CR/LF or error msg (if error) ; CX = msg length Not_Found: pushf ;save copy flags mov bx,STDERR ;Std Err handle mov ah,40H ;write to file/device int 21H mov bx,handle ;restore input file handle or bx,bx ;ever opened? jz No_Handle ;nope mov ah,3EH ;close input file int 21H No_Handle: popf ;restore copy flags jb OutPutX ;some sort of error ;Now for the Find Next mov ah,4FH ;continue file search int 21H cmp al,18 ;no more files to be found? jnz Find_Next ;nope, found one, continue OutPutX: ret ;all done with this file name OutPut endp Start proc near call _Args ;process cmdline into argv array or ax,ax ;any args? jnz Concat_Lup ;yep, concat them mov dx,offset usagemsg ;'Usage:' mov cx,USAGEMSGLEN mov bx,STDERR mov ah,40H ;write to file/device int 21H jmp short Concat_Term ;terminate Concat_Lup: mov si,word ptr argv[2] ;snarf an arg or si,si ;null arg? jz Concat_Done ;yep call OutPut ;copy file to StdOut call _Shift ;shift the args down jmp Concat_Lup Concat_Done: Concat_Term: mov ax,4C00H int 21H Start endp ;---- args.asm ---------------------------------------------------------- ; ; Args parses the command line into a unix-style parameter array. ; Argc and Argv are just as in C; argc = # of params on cmd line, ; argv is an array of pointers to strings (null terminated, of course). ; Because MS-DOS doesn't pass us the name used to invoke the program, ; argv[0] always points to a null string. ; _Shift updates argc. ; ; Do not call NEW before calling _Args. ; ;-------------------------------------------------------------------------- ; Maximum number of parameters MAXPARMS equ 32 Args2 proc near jmp Start Args2 endp _Args proc near ; initialize cld ;insure fwd mov di,offset argv ;first arg mov cx,MAXPARMS ;max parms allowed xor ax,ax rep stosw ;point all args at null mov bx, 2 ; argc = 0 (will sub 2 from bx at end) mov si,offset nchar ;cmdline char count lodsb ;snarf, bump SI ptr mov cl,al ;into CX for counter xor ch,ch jcxz A_Done ; no arg chars -> we're done. mov di,offset argbuff ;big arg buffer ; Move arguments out of default DTA. push cx push di mov dx,2000H + 'a' ;handy constant A_UpperLup: lodsb ;snarf PSP cmdline char cmp al,CR ;CR? jz Stuff0 ;yep cmp al,dh ;space? jnz A_NotSpace ;nope, more checking Stuff0: xor al,al ;replace with a 0 jmp short A_Stuff A_NotSpace: cmp al,dl ;'a' ;lower case? jb A_Stuff ;nope sub al,dh ;20H ;uppercase it A_Stuff: stosb ;stuff arg char into argbuff loop A_UpperLup xor al,al stosb ;insure terminating 0 pop si pop cx ; Big loop- find arguments... A_ParmL: lodsb ; al = [si++] or al,al ;former space separator? loopnz A_ParmL ;keep gobbling until we find one jcxz A_Done ; found no unblank chars -> we're done. mov word ptr argv[bx],si ; save pointer to this string add bx,2 ; argc++; cmp bx, MAXPARMS*2 jae A_Done ;done or cx,cx ;stop any CX wraparound to 0FFFFH! loopnz A_ParmL ;decr CX, reloop A_Done: ; All done finding parms; now share argc with caller. mov ax,bx sub ax, 2 ;adjust for argv[0] ;(traditionally program name) shr ax, 1 mov word ptr argc,ax ;save argc ret _Args endp ;---- _Shift: -------------------------------------------- ; Shifts %2 to %1, %3 to %2, etc. Leaves %0 alone. ; Works by shuffling argv[*]. _Shift proc near cld mov si, offset argv[4] mov di, offset argv[2] mov cx,word ptr argc ;move this many args rep movsw dec word ptr argc ;decr arg counter ret _Shift endp ;---- parameter count, array ---------------- ; Pointers to up to MAXPARMS parameter strings are held here. even ;keep things neat argc equ $ ;really just a byte ;same as dw ? argv equ argc + 2 ;dynamic argv array of pointers ;same as dw MAXPARMS DUP(?) argbuff equ argv + (MAXPARMS ShL 1) ;dynamic args buffer ;same as db 128 DUP(?) readbuff equ argbuff + 128 ;dynamic read buffer ;same as db 4096 DUP(?) READBUFFSIZE equ 4096 ;4Kb for now pathptr equ readbuff + 4096 ;pointer to path start ;same as dw ? path equ pathptr + 2 ;path:\filename buffer ;same as db 128 dup(?) CSEG ENDS end Concat