xlhtml.c~

excel to html

			{
				int d1, d2;
				if (sscanf(argv[i] + 4, "%d-%d", &d1, &d2) != 2)
				{
					fprintf(stderr, "column range %s not valid, expected -xc:FIRST-LAST\n", argv[i] + 4);
					display_usage();
				}
				xc1 = (S16)d1;
				xc2 = (S16)d2;
				Xtract = 1;
				if (xc1 > xc2)
				{
					fprintf(stderr, "last column must be >= the first\n");
					exit(1);
				}
			}
			else if (strncmp(argv[i], "-xp:", 4) == 0)
			{
				Xtract = 1;
				xp = (S16)atoi(&(argv[i][4]));
				if (xp < 0)
				{
					fprintf(stderr, "Negative numbers are illegal.\n");
					exit(1);
				}
			}
			else if (strncmp(argv[i], "-xr:", 4) == 0)
			{
				char *ptr, *buf;
				Xtract = 1;
				buf = strdup(argv[i]);
				ptr = strrchr(buf, '-');
				xr2 = (S16)atoi(ptr+1);
				*ptr = 0;
				ptr = strchr(buf, ':');
				xr1 = (S16)atoi(ptr+1);
				free(buf);
				if (xr1 > xr2)
				{
					fprintf(stderr, "row's 2nd digit must be >= the first\n");
					exit(1);
				}
			}
			else
				display_usage();
		}
		if (strcmp(filename, "-v") == 0)
		{
			print_version();
			exit(0);
		}
		if (strcmp(filename, "--version") == 0)
		{
			print_version();
			exit(0);
		}

		if (strcmp(filename, "--help") == 0)
			display_usage();
		if (strcmp(filename, "-?") == 0)
			display_usage();
	}
	if (Ascii)
	{	/* Disable it if DumpPage or Xtract isn't used... */
		if (!(DumpPage||Xtract))
			Ascii = 0;
	}
	if (Xtract)
		trim_edges = 0;		/* No trimming when extracting... */
	if (OutputXML)
		aggressive = 0;

	/* Init arrays... */
	ws_array = (work_sheet **)malloc(max_worksheets * sizeof(work_sheet*));
	for (i=0; i<(int)max_worksheets; i++)
		ws_array[i] = 0;

	str_array = (uni_string **)malloc(max_strings*sizeof(uni_string *));
	for (i=0; i<(int)max_strings; i++)
		str_array[i] = 0;

	font_array = (font_attr **)malloc(max_fonts * sizeof(font_attr *));
	f_cnt = (fnt_cnt *)malloc(max_fonts * sizeof(fnt_cnt));
	for (i=0; i<(int)max_fonts; i++)
	{	/* I assume these won't fail since we are just starting up... */
		font_array[i] = 0;
		f_cnt[i].name = 0;
	}
	xf_array = (xf_attr **)malloc(max_xformats * sizeof(xf_attr *));
	for (i=0; i<(int)max_xformats; i++)
		xf_array[i] = 0;

	uni_string_clear(&author);
	uni_string_clear(&default_font);
	umask(GLOBAL_UMASK);

#if defined( __WIN32__ ) || defined( __BORLANDC__ )
	{
		char *ptr = strchr(filename, ':');
		if (ptr)
		{
			int len;
			char new_drive[MAXPATH];
			fnsplit(filename, new_drive, 0, 0, 0);
			if (new_drive[0] >= 'a')
				setdisk(new_drive[0] - 'a');
			else
				setdisk(new_drive[0] - 'A');
			ptr++;	/* Get past the colon */
			len = strlen(ptr);
			memmove(filename, ptr, len);
			filename[len] = 0;
		}
	}
#endif
	/* If successful, this calls scan_file to extract the work book... */
	cfs = cole_mount(filename, &colerrno);
	if (cfs == NULL)
	{
		cole_perror (NULL, colerrno);
		exit(1);
	}
	while (cole_locate_filename (cfs, SectionName[f_ptr], NULL, scan_file, &colerrno))
	{
		if (f_ptr)
		{	/* Two strikes...we're out! */
			cole_perror (PRGNAME, colerrno);
			if (colerrno == COLE_EFILENOTFOUND)
				fprintf(stderr, "Section: Workbook\n");
			break;
		}
		else
			f_ptr++;
	}

	if (cole_umount (cfs, &colerrno))
	{
		cole_perror (PRGNAME, colerrno);
		exit(1);
	}

	/* For some reason, this loop core dumps on some */
	/* files from Central Europe...so it's commented out for now */
/*	for (i=0; i<max_strings; i++)
	{
		if (str_array[i])
		{
			if (str_array[i]->str)
				free(str_array[i]->str);
			free(str_array[i]);
		}
	}	*/

	for (i=0; i<(int)max_fonts; i++)
	{
		if (font_array[i])
		{
			if (font_array[i]->name.str)
				free(font_array[i]->name.str);
			free(font_array[i]);
			if (f_cnt[i].name)
			{
				if (f_cnt[i].name->str)
					free(f_cnt[i].name->str);
				free(f_cnt[i].name);
			}
		}
	}
	free(font_array);
	free(f_cnt);
	for (i=0; i<(int)max_worksheets; i++)
	{
		if (ws_array[i])
		{
			if (ws_array[i]->ws_title.str)
				free(ws_array[i]->ws_title.str);
			if (ws_array[i]->c_array)
			{
				for (j=0; j<ws_array[i]->max_rows; j++)
				{
					for (k=0; k<ws_array[i]->max_cols; k++)
					{
						if (ws_array[i]->c_array[(j*ws_array[i]->max_cols)+k])
						{
							if (ws_array[i]->c_array[(j*ws_array[i]->max_cols)+k]->ustr.str)
								free(ws_array[i]->c_array[(j*ws_array[i]->max_cols)+k]->ustr.str);
							if (ws_array[i]->c_array[(j*ws_array[i]->max_cols)+k]->ustr.fmt_run)
								free(ws_array[i]->c_array[(j*ws_array[i]->max_cols)+k]->ustr.fmt_run);
							if (ws_array[i]->c_array[(j*ws_array[i]->max_cols)+k]->h_link.str)
								free(ws_array[i]->c_array[(j*ws_array[i]->max_cols)+k]->h_link.str);
							free(ws_array[i]->c_array[(j*ws_array[i]->max_cols)+k]);
						}
					}
				}
				free(ws_array[i]->c_array);
			}
			free(ws_array[i]);
		}
	}

	for (i=0; i<(int)max_xformats; i++)
	{
		if (xf_array[i])
			free(xf_array[i]);
	}
	free(xf_array);

	if (numCustomColors)
	{
		for (i=0; i<numCustomColors; i++)
			free(customColors[i]);
		free(customColors);
	}

	if (default_font.str)
		free(default_font.str);
	if (author.str)
		free(author.str);
	if (title)
		free(title);
	if (lastUpdated)
		free(lastUpdated);

	return 0;
}

static void print_version(void)
{
	printf("xlhtml %s\nCopyright (c) 1999-2002, Charles Wyble\n" 
	"Released under GPL.\n", VERSION);
	exit(1);
}

void scan_file(COLEDIRENT *cde, void *_info)
{
	U32 count = 0;
	U16 length=0, target=0, opcode=0, version=0;
	U8 buf[16];
	COLEFILE *cf;
	COLERRNO err;

	cf = cole_fopen_direntry(cde, &err);
	if (cf == 0)
	{	/* error abort processing */
		cole_perror (PRGNAME, err);
		return;
	}

	/* Read & process the file... */
	while (cole_fread(cf, buf, 1, &err))
	{
		if (count > 3)
			main_line_processor(opcode, version, count-4, target, buf[0]);
		else if (count == 0)
		{	/* Init everything */
			length = 0;
			opcode = (U16)buf[0];
			target = 80;	/* ficticious number */
		}
		else if (count == 1)
			version = (U16)buf[0];
		else if (count == 2)
			length = (U16)buf[0];
		else if (count == 3)
		{
			length |= (U16)(buf[0]<<8);
			target = length;
		}

		if (count == (U32)(target+3))
			count = 0;
		else
			count++;
		if (MaxColExceeded || MaxRowExceeded || MaxWorksheetsExceeded)
			break;	/* We're outta memory and therefore...done */

	}
	cole_fclose(cf, &err);

	if (Ascii)
	{
		if (DumpPage)
		{	/* Output the XLS Parameters */
			int i;
			printf("There are %d pages total.\n", sheet_count+1);
			for (i=0; i<=sheet_count; i++)
			{
				printf("Page:%d Name:%s MaxRow:%ld MaxCol:%d\n", i,
					ws_array[i]->ws_title.str ? (char *)ws_array[i]->ws_title.str : "(Unknown Page)",
					ws_array[i]->biggest_row, ws_array[i]->biggest_col);
			}
		}
		else if (Xtract)
			OutputPartialTableAscii();
	}
	else
	{
		if (DumpPage)
		{	/* Output the XLS Parameters */
			int i;
			output_header();
			printf("<p>There are %d pages total.</p>\n", sheet_count+1);
			for (i=0; i<=sheet_count; i++)
			{
				printf("<p>Page:%d Name:%s MaxRow:%ld MaxCol:%d</p>\n", i,
					ws_array[i]->ws_title.str ? (char *)ws_array[i]->ws_title.str : "(Unknown Page)",
					ws_array[i]->biggest_row, ws_array[i]->biggest_col);
			}
			output_footer();
		}
		else
		{
			if( OutputXML )
				OutputTableXML();
			else
				OutputTableHTML();
		}
	}
}

static void OutputPartialTableAscii(void)
{
	int i, j, k;

	SetupExtraction();

	/* Here's where we dump the Html Page out */
	for (i=first_sheet; i<=last_sheet; i++)	/* For each worksheet */
	{
		if (ws_array[i] == 0)
			continue;
		if ((ws_array[i]->biggest_row == -1)||(ws_array[i]->biggest_col == -1))
			continue;
		if (ws_array[i]->c_array == 0)
			continue;

		/* Now dump the table */
		for (j=ws_array[i]->first_row; j<=ws_array[i]->biggest_row; j++)
		{
			for (k=ws_array[i]->first_col; k<=ws_array[i]->biggest_col; k++)
			{
				int safe, numeric=0;
				cell *c = ws_array[i]->c_array[(j*ws_array[i]->max_cols)+k]; /* This stuff happens for each cell... */

				if (c)
				{
					numeric = IsCellNumeric(c);
					if (!numeric && Csv)
						printf("\"");
					safe = IsCellSafe(c);

					if (c->ustr.str)
					{
						if (safe)
							output_formatted_data(&(c->ustr), xf_array[c->xfmt]->fmt_idx, numeric, IsCellFormula(c));
						else
							OutputString(&(c->ustr));
					}
					else if (!Csv)
						printf(" ");	/* Empty cell... */
				}
				else
				{   	/* Empty cell... */
					if (!Csv)
						printf(" ");
					else
						printf("\"");
				}
				if (ws_array[i]->c_array[(j*ws_array[i]->max_cols)+k])	/* Honor Column spanning ? */
				{
					if (ws_array[i]->c_array[(j*ws_array[i]->max_cols)+k]->colspan != 0)
						k += ws_array[i]->c_array[(j*ws_array[i]->max_cols)+k]->colspan-1;
				}
				if (!numeric && Csv)
					printf("\"");

				if (Csv && (k < ws_array[i]->biggest_col))
				{	/* big cheat here: quoting everything! */
					putchar(',');	/* Csv Cell Separator */
				}
				else
				{
					if (( !Csv )&&( k != ws_array[i]->biggest_col ))
						putchar('\t');	/* Ascii Cell Separator */
				}
			}
			if (Csv)
				printf("\r\n");
			else
				putchar(0x0A);		/* Row Separator */
		}
		if (!Csv)
			printf("\n\n");		 	/* End of Table 2 LF-CR */
	}
}

static void do_cr(void)
{
	if (!aggressive)
		putchar('\n');
}

static void OutputTableHTML(void)
{
	int i, j, k;

	output_header();
	if (center_tables)
	{
		printf("<CENTER>");
		do_cr();
	}

	SetupExtraction();

	/* Here's where we dump the Html Page out */
	for (i=first_sheet; i<=last_sheet; i++)	/* For each worksheet */
	{
		update_default_font(i);
		if (ws_array[i] == 0)
			continue;
		if ((ws_array[i]->biggest_row == -1)||(ws_array[i]->biggest_col == -1))
			continue;
		if (ws_array[i]->c_array == 0)
			continue;
		trim_sheet_edges(i);

		/* Print its name */
		if (next_ws_title > 0)
		{
			if (ws_array[i]->ws_title.str)
			{
				printf("<H1><CENTER>");
				OutputString(&ws_array[i]->ws_title);
				printf("</CENTER></H1><br>");
				do_cr();
			}
			else
			{
				printf("<H1><CENTER>(Unknown Page)</CENTER></H1><br>");
				do_cr();
			}
		}

		/* Now dump the table */
		printf("<FONT FACE=\"");
		OutputString(&default_font);
		if (default_fontsize != 3)
			printf("\" SIZE=\"%d", default_fontsize);
		printf("\">");
		do_cr();
		printf("<TABLE BORDER=\"1\" CELLSPACING=\"2\">");
		do_cr();
		for (j=ws_array[i]->first_row; j<=ws_array[i]->biggest_row; j++)
		{
			update_default_alignment(i, j);
			printf("<TR");
			if (null_string((U8 *)default_alignment))
				printf(">");
			else
			{
				if (strcmp(default_alignment, "left") != 0)
					printf(" ALIGN=\"%s\"", default_alignment);
				if (!aggressive)
					printf(" VALIGN=\"bottom\">\n");
				else
					printf(">");
			}
			for (k=ws_array[i]->first_col; k<=ws_array[i]->biggest_col; k++)
			{
				output_cell(ws_array[i]->c_array[(j*ws_array[i]->max_cols)+k],0); /* This stuff happens for each cell... */
				if (ws_array[i]->c_array[(j*ws_array[i]->max_cols)+k])
				{
					if (ws_array[i]->c_array[(j*ws_array[i]->max_cols)+k]->colspan != 0)