/***************************************************************************

	"vnconvert" converts Vietnamese text files between different
	formats. It is intended to run on Unix platforms. The user
	shoul recompile on his/her specific platform. A very simple 
	"Makefile" is provided.

	This program is similar to the VPS "convert" for DOS, and in
	fact, is based on the coding table provided by Vietnamese
	Professional Society (VPS).

	This program is also based on the TRICHLOR "vn7to8" package and 
	takes advantage of its header files, converting functions between 
	VISCII and VIQR formats.

	This program considers only the standard VISCII and VIQR formats.

	The package consists of the following files: 

		- vnconvert.c  (this file)
		- conv.c
		- vn78.c

		- vnkeys.h
		- conv.h
		- ctype.h
		- vn78.h
		- charset.h       
		- charsetv.h   
		- charseta.h     
		- portable.h      
		- vncompos.h

		- Makefile

	Run "make: to install "vnconvert".
	
	There may be a lot of things which can be improved and any comments
	or suggestions are welcome. Please forward them to 

		Binh Do <binhdo@cs.ubc.ca>

*****************************************************************************/

#include <sys/file.h>
#include <sys/stat.h>
#include <stdio.h>
#include "vnkeys.h" 
#include "conv.h"


main()

{
	u_char	buf[BUFSIZ];
	char	bufline[255];
	char	input_file[255], output_file[255];
	int fd_in, fd_out;
	int n, k, icontinue;
	int from, to;

        u_char   esc = VK_DEFAULT_ESC;
        u_char   bs = VK_DEFAULT_BS;
        Vk_State        state = VK_ST_VIETNAMESE;
        u_char   obuf[4*BUFSIZ], newbuf[4*BUFSIZ];
        int  i, count;
        Vk_Fsm  id;
        u_char FAR  *ic, *oc;

	struct 	stat 	stbuf;
	short  mode;


        /* Read in input_file, output_file and their formats */

        id = vk_init(state, esc, bs);
        vk_set(id, 1, VK_SET_BATCH);



	while (1)  {

		/* Read in input_file, output_file and their formats */
		printf("\nVNCONVERT converts Vietnamese text files between the following formats :\n\n");
		printf("\t0 - VISCII \t\t\t 7 - VN Thin-Art (PC)\n");
		printf("\t1 - VNCII (VPSWin)\t\t 8 - Viet-Toan\n");
		printf("\t2 - VPSedit (DOS)\t\t 9 - Plain Vietnamese\n");
		printf("\t3 - VNI\t\t\t\t10 - Standard VIQR\n");
		printf("\t4 - VNLab/VNU\n");
		printf("\t5 - VNU (Windows)\n");
		printf("\t6 - VN Thin-Art (Mac)\n\n");

		printf("Input-file name (or Enter to quit): ");
		gets(bufline);
		if (strlen(bufline) == 0)
			break;
		sscanf(bufline, "%s", input_file);

		printf("currently in the format number: ");
		*bufline = '\0'; 
		while (strlen(bufline)==0) {
			gets(bufline);
			sscanf(bufline, "%d", &from);
		}

		printf("\nOutput-file name: ");
		*bufline = '\0';
                while (strlen(bufline)==0) {
                        gets(bufline);
                        sscanf(bufline, "%s", output_file); 
                }

		printf("to the format number: ");
		*bufline = '\0';
                while (strlen(bufline)==0) {
                        gets(bufline);
                        sscanf(bufline, "%d", &to);
                }


		/* Check for possible errors */
		icontinue = 0;

        	if (stat(input_file, &stbuf) == -1) {
			fprintf(stderr, "cannot access \"%s\"\n", input_file);
			icontinue = 1;
		}
		else if ((stbuf.st_mode & S_IFMT) == S_IFDIR) {
                     	 	fprintf(stderr, "\"%s\" is a directory\n", 
						input_file);
				icontinue = 1;
		     }

		mode = stbuf.st_mode;

		if (from < 0  || from > FORMATS) {
			fprintf(stderr, "the format number (%d) of input file is out of range [0 - %d]\n", FORMATS, from);
			icontinue = 1;
		}

		if (to < 0  || to > FORMATS) {
                        fprintf(stderr, "the format number (%d) of output file is out of range [0 - %d]\n", FORMATS, to);
                        icontinue = 1;
                }

		if (stat(output_file, &stbuf) != -1)  {
			if ((stbuf.st_mode & S_IFMT) == S_IFDIR) {
				fprintf(stderr, "\"%s\" is a directory\n",
						output_file);
			
				icontinue = 1;
			}
			else {
				fprintf(stderr, "\"%s\" already exists. Overwrite it (y/n)? ", output_file);
				gets(buf);
				if (toupper(buf[0]) != 'Y')  
					icontinue = 1;
			}
		}

		if (from == to)  {   /* no need to convert */
			fprintf(stderr, "Use \"cp\" to copy files of same formats\n");
			icontinue = 1;
		}

		
		/* And procesing the files as follows:

			for formats 0 - 9, just find the corresponding
			chars in the table and done. (9 should be only
			destination, not source).

			If one of the formats is 10 (VIQR), then we must go 
			thru format 0 (VISCII). 
			Should be improved here  */ 

		if (icontinue) 
			continue;
		else {	
			if ((fd_in = open(input_file, 0)) == -1) {
				fprintf(stderr, "cannot open \"%s\"\n",
					input_file);
				continue;
			}

			if ((fd_out = creat(output_file, mode))==-1) { 
				fprintf(stderr, " cannot create \"%s\"\n", 
					output_file);
				continue;
			}
		}

		if (from == PLAIN_format)    /* nonsense */
			to = from;
		

		if (from != VIQR_format) {
			if (to != VIQR_format) {
				while ((n = read(fd_in, buf, BUFSIZ)) > 0)  {
                			for (k=0; k<n; ++k)  {
                       		 		*(buf+k) = 
						    convert_char(*(buf+k), 
								 from, to);
                			}
                			write(fd_out, buf, n);
        			}
			}
			else   { /* convert to VISCII and then to VIQR */
				while ((n = read(fd_in, buf, BUFSIZ-1)) > 0) {
					buf[n] = '\0';
					if (from != VISCII_format) 
						for (k=0; k<n; ++k)  
                                                    *(buf+k) = 
                                                         convert_char(*(buf+k),
                                                          from, VISCII_format);
                                        	
					k = vk_8toqr(buf, obuf, esc, state);
			                write(fd_out, obuf, k);
				}
			}
		}
		else  {    /* if from VIQR, */    
		   while ((n = read(fd_in, buf, BUFSIZ)) > 0)  {
                	ic = buf;
                	oc = obuf;
                	for (i = 0; i < n; i++) {
                        	vk_step(id, *ic++, oc, (int FAR *)&count);
                        	if (count) 
                               	 	oc += count;
                	}
			if (to == VISCII_format)
                		write(fd_out, obuf, oc-obuf);
			else {
				k = oc - obuf;
				for (i=0; i<k; i++) 
				   *(obuf+i) = convert_char(*(obuf+i),
						            VISCII_format, to);
				write(fd_out, obuf, k);
			}
			
		    }
		} 

        	close (fd_in);
        	close (fd_out);

	}


}