pdftool.c

来自「SumatraPDF是一款小型开源的pdf阅读工具。虽然玲珑小巧（只有800多K」· C语言 代码 · 共 1,217 行 · 第 1/2 页

	fz_md5init(&digest);

	drawloadpage(pagenum, loadtimes);

	if (benchmark)
		gettime(&start);

	ctm = fz_identity();
	ctm = fz_concat(ctm, fz_translate(0, -drawpage->mediabox.y1));
	ctm = fz_concat(ctm, fz_scale(drawzoom, -drawzoom));
	ctm = fz_concat(ctm, fz_rotate(drawrotate + drawpage->rotate));

	bbox = fz_roundrect(fz_transformaabb(ctm, drawpage->mediabox));
	w = bbox.x1 - bbox.x0;
	h = bbox.y1 - bbox.y0;
	bh = h / drawbands;

	if (drawpattern)
	{
		sprintf(name, drawpattern, drawcount++);
		fd = open(name, O_BINARY|O_WRONLY|O_CREAT|O_TRUNC, 0666);
		if (fd < 0)
			die(fz_throw("ioerror: could not open file '%s'", name));

		sprintf(pnmhdr, "P6\n%d %d\n255\n", w, h);
		write(fd, pnmhdr, strlen(pnmhdr));
	}

	error = fz_newpixmap(&pix, bbox.x0, bbox.y0, w, bh, 4);
	if (error)
		die(error);

	memset(pix->samples, 0xff, pix->h * pix->w * pix->n);

	for (b = 0; b < drawbands; b++)
	{
		if (drawbands > 1)
			fprintf(stderr, "drawing band %d / %d\n", b + 1, drawbands);

		error = fz_rendertreeover(drawgc, pix, drawpage->tree, ctm);
		if (error)
			die(error);

		if (drawpattern)
		{
			for (y = 0; y < pix->h; y++)
			{
				unsigned char *src = pix->samples + y * pix->w * 4;
				unsigned char *dst = src;

				for (x = 0; x < pix->w; x++)
				{
					dst[x * 3 + 0] = src[x * 4 + 1];
					dst[x * 3 + 1] = src[x * 4 + 2];
					dst[x * 3 + 2] = src[x * 4 + 3];
				}

				write(fd, dst, pix->w * 3);

				memset(src, 0xff, pix->w * 4);
			}
		}

		fz_md5update(&digest, pix->samples, pix->h * pix->w * 4);

		pix->y += bh;
		if (pix->y + pix->h > bbox.y1)
			pix->h = bbox.y1 - pix->y;
	}

	fz_droppixmap(pix);

	{
	    unsigned char buf[16];
	    fz_md5final(&digest, buf);
	    for (i = 0; i < 16; i++)
		fprintf(stderr, "%02x", buf[i]);
	}

	if (drawpattern)
		close(fd);

	drawfreepage();

	if (benchmark)
	{
	    gettime(&end);
	    elapsed = end - start;

	    if (elapsed < drawtimes->min)
	    {
		drawtimes->min = elapsed;
		drawtimes->minpage = pagenum;
	    }
	    if (elapsed > drawtimes->max)
	    {
		drawtimes->max = elapsed;
		drawtimes->maxpage = pagenum;
	    }
	    drawtimes->avg += elapsed;
	    drawtimes->pages++;

	    fprintf(stderr, " time %.3fs",
		    elapsed / 1000000.0);
	}

	fprintf(stderr, "\n");
}

void
drawtxt(int pagenum)
{
	fz_error *error;
	pdf_textline *line;
	fz_matrix ctm;

	drawloadpage(pagenum, NULL);

	ctm = fz_concat(
			fz_translate(0, -drawpage->mediabox.y1),
			fz_scale(drawzoom, -drawzoom));

	error = pdf_loadtextfromtree(&line, drawpage->tree, ctm);
	if (error)
		die(error);

	pdf_debugtextline(line);
	pdf_droptextline(line);

	drawfreepage();
}

void
drawxml(int pagenum)
{
	drawloadpage(pagenum, NULL);
	fz_debugtree(drawpage->tree);
	drawfreepage();
}

void
drawpages(char *pagelist)
{
	int page, spage, epage;
	char *spec, *dash;
	struct benchmark loadtimes, drawtimes;

	if (!src)
		drawusage();

	if (benchmark)
	{
		memset(&loadtimes, 0x00, sizeof (loadtimes));
		loadtimes.min = LONG_MAX;
		memset(&drawtimes, 0x00, sizeof (drawtimes));
		drawtimes.min = LONG_MAX;
	}

	spec = strsep(&pagelist, ",");
	while (spec)
	{
		dash = strchr(spec, '-');

		if (dash == spec)
			spage = epage = 1;
		else
			spage = epage = atoi(spec);

		if (dash)
		{
			if (strlen(dash) > 1)
				epage = atoi(dash + 1);
			else
				epage = pdf_getpagecount(srcpages);
		}

		if (spage > epage)
			page = spage, spage = epage, epage = page;

		if (spage < 1)
			spage = 1;
		if (epage > pdf_getpagecount(srcpages))
			epage = pdf_getpagecount(srcpages);

		printf("Drawing pages %d-%d...\n", spage, epage);
		for (page = spage; page <= epage; page++)
		{
			switch (drawmode)
			{
			case DRAWPNM: drawpnm(page, &loadtimes, &drawtimes); break;
			case DRAWTXT: drawtxt(page); break;
			case DRAWXML: drawxml(page); break;
			}
		}

		spec = strsep(&pagelist, ",");
	}

	if (benchmark)
	{
		if (loadtimes.pages > 0)
		{
			loadtimes.avg /= loadtimes.pages;
			drawtimes.avg /= drawtimes.pages;

			printf("benchmark[load]: min: %6.3fs (page % 4d), avg: %6.3fs, max: %6.3fs (page % 4d)\n",
				loadtimes.min / 1000000.0, loadtimes.minpage,
				loadtimes.avg / 1000000.0,
				loadtimes.max / 1000000.0, loadtimes.maxpage);
			printf("benchmark[draw]: min: %6.3fs (page % 4d), avg: %6.3fs, max: %6.3fs (page % 4d)\n",
				drawtimes.min / 1000000.0, drawtimes.minpage,
				drawtimes.avg / 1000000.0,
				drawtimes.max / 1000000.0, drawtimes.maxpage);
		}
	}
}

void
drawmain(int argc, char **argv)
{
	fz_error *error;
	char *password = "";
	int c;
	enum { NO_FILE_OPENED, NO_PAGES_DRAWN, DREW_PAGES } state;

	while ((c = getopt(argc, argv, "b:d:o:r:txm")) != -1)
	{
		switch (c)
		{
		case 'b': drawbands = atoi(optarg); break;
		case 'd': password = optarg; break;
		case 'o': drawpattern = optarg; break;
		case 'r': drawzoom = atof(optarg) / 72.0; break;
		case 't': drawmode = DRAWTXT; break;
		case 'x': drawmode = DRAWXML; break;
		case 'm': benchmark = 1; break;
		default:
			  drawusage();
			  break;
		}
	}

	if (optind == argc)
		drawusage();

	error = fz_newrenderer(&drawgc, pdf_devicergb, 0, 1024 * 512);
	if (error)
		die(error);

	state = NO_FILE_OPENED;
	while (optind < argc)
	{
		if (strstr(argv[optind], ".pdf"))
		{
			if (state == NO_PAGES_DRAWN)
				drawpages("1-");

			opensrc(argv[optind], password, 1);
			state = NO_PAGES_DRAWN;
		}
		else
		{
			drawpages(argv[optind]);
			state = DREW_PAGES;
		}
		optind++;
	}

	if (state == NO_PAGES_DRAWN)
		drawpages("1-");

	closesrc();

	fz_droprenderer(drawgc);
}

/* --------------------------------------------------------------------- */

/*
 * Edit tool.
 * Copy or impose pages from other pdf files into output pdf.
 */

/* for each source pdf, build a list of objects to transplant.
 * for each source pdf, do the transplants at the end of object collecting.
 * build a new page tree structure for output.
 * change page nodes into xobjects for over and n-up modes.
 * create new page nodes.
 * create new page tree.
 */

enum { COPY, OVER, NUP2, NUP4, NUP8 };

pdf_xref *editxref = nil;
fz_obj *editpagelist = nil;
fz_obj *editmodelist = nil;
fz_obj *editobjects = nil;
int editmode = COPY;

void
editusage(void)
{
	fprintf(stderr, "usage: mupdftool edit [-o file.pdf] [mode file.pdf pages ... ]\n");
	fprintf(stderr, "  mode is one of: copy over 2up 4up 8up\n");
	fprintf(stderr, "  pages is a comma separated list of ranges\n");
	fprintf(stderr, "  example:\n");
	fprintf(stderr, "    mupdftool edit -o output.pdf copy one.pdf 1-3,5,9 two.pdf 1-\n");
	exit(1);
}

void
editcopy(int pagenum)
{
	fz_error *error;
	fz_obj *obj;
	fz_obj *ref;
	fz_obj *num;

	printf("copy %s page %d\n", srcname, pagenum);

	ref = srcpages->pref[pagenum - 1];
	obj = pdf_getpageobject(srcpages, pagenum - 1);

	fz_dictdels(obj, "Parent");
	/*
	fz_dictdels(obj, "B");
	fz_dictdels(obj, "PieceInfo");
	fz_dictdels(obj, "Metadata");
	fz_dictdels(obj, "Annots");
	fz_dictdels(obj, "Tabs");
	*/

	pdf_updateobject(src, fz_tonum(ref), fz_togen(ref), obj);

	error = fz_arraypush(editobjects, ref);
	if (error)
		die(error);

	error = fz_newint(&num, editmode);
	if (error)
		die(error);

	error = fz_arraypush(editmodelist, num);
	if (error)
		die(error);

	fz_dropobj(num);
}

void
editflushobjects(void)
{
	fz_error *error;
	fz_obj *results;
	int i;

	error = pdf_transplant(editxref, src, &results, editobjects);
	if (error)
		die(error);

	for (i = 0; i < fz_arraylen(results); i++)
	{
		error = fz_arraypush(editpagelist, fz_arrayget(results, i));
		if (error)
			die(error);
	}

	fz_dropobj(results);
}

void
editflushpagetree(void)
{

	/* TODO: merge pages where editmode != COPY by turning them into XObjects
	   and creating a new page object with resource dictionary and content
	   stream placing the xobjects on the page. */
}

void
editflushcatalog(void)
{
	fz_error *error;
	int rootnum, rootgen;
	int listnum, listgen;
	fz_obj *listref;
	fz_obj *obj;
	int i;

	/* Create page tree and add back-links */

	error = pdf_allocobject(editxref, &listnum, &listgen);
	if (error)
		die(error);

	error = fz_packobj(&obj, "<</Type/Pages/Count %i/Kids %o>>",
			fz_arraylen(editpagelist),
			editpagelist);
	if (error)
		die(error);

	pdf_updateobject(editxref, listnum, listgen, obj);

	fz_dropobj(obj);

	error = fz_newindirect(&listref, listnum, listgen);
	if (error)
		die(error);

	for (i = 0; i < fz_arraylen(editpagelist); i++)
	{
		int num = fz_tonum(fz_arrayget(editpagelist, i));
		int gen = fz_togen(fz_arrayget(editpagelist, i));

		error = pdf_loadobject(&obj, editxref, num, gen);
		if (error)
			die(error);

		error = fz_dictputs(obj, "Parent", listref);
		if (error)
			die(error);

		pdf_updateobject(editxref, num, gen, obj);

		fz_dropobj(obj);
	}

	/* Create catalog */

	error = pdf_allocobject(editxref, &rootnum, &rootgen);
	if (error)
		die(error);

	error = fz_packobj(&obj, "<</Type/Catalog/Pages %r>>", listnum, listgen);
	if (error)
		die(error);

	pdf_updateobject(editxref, rootnum, rootgen, obj);

	fz_dropobj(obj);

	/* Create trailer */

	error = fz_packobj(&editxref->trailer, "<</Root %r>>", rootnum, rootgen);
	if (error)
		die(error);
}

void
editpages(char *pagelist)
{
	int page, spage, epage;
	char *spec, *dash;

	if (!src)
		editusage();

	spec = strsep(&pagelist, ",");
	while (spec)
	{
		dash = strchr(spec, '-');

		if (dash == spec)
			spage = epage = 1;
		else
			spage = epage = atoi(spec);

		if (dash)
		{
			if (strlen(dash) > 1)
				epage = atoi(dash + 1);
			else
				epage = pdf_getpagecount(srcpages);
		}

		if (spage > epage)
			page = spage, spage = epage, epage = page;

		for (page = spage; page <= epage; page++)
		{
			if (page < 1 || page > pdf_getpagecount(srcpages))
				continue;
			editcopy(page);
		}

		spec = strsep(&pagelist, ",");
	}
}

void
editmain(int argc, char **argv)
{
	char *outfile = "out.pdf";
	fz_error *error;
	int c;

	while ((c = getopt(argc, argv, "o:")) != -1)
	{
		switch (c)
		{
		case 'o':
			outfile = optarg;
			break;
		default:
			editusage();
			break;
		}
	}

	if (optind == argc)
		editusage();

	error = pdf_newxref(&editxref);
	if (error)
		die(error);

	error = pdf_initxref(editxref);
	if (error)
		die(error);

	error = fz_newarray(&editpagelist, 100);
	if (error)
		die(error);

	error = fz_newarray(&editmodelist, 100);
	if (error)
		die(error);

	while (optind < argc)
	{
		if (strstr(argv[optind], ".pdf"))
		{
			if (editobjects)
				editflushobjects();

			opensrc(argv[optind], "", 1);

			error = fz_newarray(&editobjects, 100);
			if (error)
				die(error);
		}
		else if (!strcmp(argv[optind], "copy"))
			editmode = COPY;
		else if (!strcmp(argv[optind], "over"))
			editmode = OVER;
		else if (!strcmp(argv[optind], "2up"))
			editmode = NUP2;
		else if (!strcmp(argv[optind], "4up"))
			editmode = NUP4;
		else if (!strcmp(argv[optind], "8up"))
			editmode = NUP8;
		else
			editpages(argv[optind]);
		optind++;
	}

	if (editobjects)
		editflushobjects();

	closesrc();

	editflushpagetree();
	editflushcatalog();

	error = pdf_savexref(editxref, outfile, nil);
	if (error)
		die(error);

	pdf_closexref(editxref);
}

/* --------------------------------------------------------------------- */

/*
 * Main!
 */

void
mainusage(void)
{
	fprintf(stderr, "usage: mupdftool <command> [options...]\n");
	fprintf(stderr, "  command is one of: show, draw, clean, edit\n");
	exit(1);
}

int
main(int argc, char **argv)
{
	if (argc >= 2)
	{
		optind = 2;
		if (!strcmp(argv[1], "show"))
			showmain(argc, argv);
		else if (!strcmp(argv[1], "draw"))
			drawmain(argc, argv);
		else if (!strcmp(argv[1], "clean"))
			cleanmain(argc, argv);
		else if (!strcmp(argv[1], "edit"))
			editmain(argc, argv);
		else
			mainusage();
	}
	else
		mainusage();
	return 0;
}