pdf_open.c

SumatraPDF是一款小型开源的pdf阅读工具。虽然玲珑小巧（只有800多KB）

#include "fitz.h"
#include "mupdf.h"

static inline int iswhite(int ch)
{
	return
		ch == '\000' || ch == '\011' || ch == '\012' ||
		ch == '\014' || ch == '\015' || ch == '\040';
}

/*
 * magic version tag and startxref
 */

static fz_error *
loadversion(pdf_xref *xref)
{
	fz_error *error;
	char buf[20];

	error = fz_seek(xref->file, 0, 0);
	if (error)
		return fz_rethrow(error, "cannot seek to beginning of file");

	error = fz_readline(xref->file, buf, sizeof buf);
	if (error)
		return fz_rethrow(error, "cannot read version marker");
	if (memcmp(buf, "%PDF-", 5) != 0)
		return fz_throw("cannot recognize version marker");

	xref->version = (int) (atof(buf + 5) * 10.0 + 0.5);

	pdf_logxref("version %d.%d\n", xref->version / 10, xref->version % 10);

	return fz_okay;
}

static fz_error *
readstartxref(pdf_xref *xref)
{
	fz_error *error;
	unsigned char buf[1024];
	int t, n;
	int i;

	error = fz_seek(xref->file, 0, 2);
	if (error)
		return fz_rethrow(error, "cannot seek to end of file");

	t = MAX(0, fz_tell(xref->file) - ((int)sizeof buf));
	error = fz_seek(xref->file, t, 0);
	if (error)
		return fz_rethrow(error, "cannot seek to offset %d", t);

	error = fz_read(&n, xref->file, buf, sizeof buf);
	if (error)
		return fz_rethrow(error, "cannot read from file");

	for (i = n - 9; i >= 0; i--)
	{
		if (memcmp(buf + i, "startxref", 9) == 0)
		{
			i += 9;
			while (iswhite(buf[i]) && i < n)
				i ++;
			xref->startxref = atoi((char*)(buf + i));
			return fz_okay;
		}
	}

	return fz_throw("cannot find startxref");
}

/*
 * trailer dictionary
 */

static fz_error *
readoldtrailer(pdf_xref *xref, char *buf, int cap)
{
	fz_error *error;
	int ofs, len;
	char *s;
	int n;
	int t;
	pdf_token_e tok;
	int c;

	pdf_logxref("load old xref format trailer\n");

	error = fz_readline(xref->file, buf, cap);
	if (error)
		return fz_rethrow(error, "cannot read xref marker");
	if (strncmp(buf, "xref", 4) != 0)
		return fz_throw("cannot find xref marker");

	while (1)
	{
		c = fz_peekbyte(xref->file);
		if (!(c >= '0' && c <= '9'))
			break;

		error = fz_readline(xref->file, buf, cap);
		if (error)
			return fz_rethrow(error, "cannot read xref count");

		s = buf;
		ofs = atoi(strsep(&s, " "));
		len = atoi(strsep(&s, " "));

		/* broken pdfs where the section is not on a separate line */
		if (s && *s != '\0')
		{
			error = fz_seek(xref->file, -(2 + strlen(s)), 1);
			if (error)
				return fz_rethrow(error, "cannot seek in file");
		}

		t = fz_tell(xref->file);
		if (t < 0)
			return fz_throw("cannot tell in file");

		error = fz_seek(xref->file, t + 20 * len, 0);
		if (error)
			return fz_rethrow(error, "cannot seek in file");
	}

	error = fz_readerror(xref->file);
	if (error)
		return fz_rethrow(error, "cannot read from file");

	error = pdf_lex(&tok, xref->file, buf, cap, &n);
	if (error)
		return fz_rethrow(error, "cannot parse trailer");
	if (tok != PDF_TTRAILER)
		return fz_throw("expected trailer marker");

	error = pdf_lex(&tok, xref->file, buf, cap, &n);
	if (error)
		return fz_rethrow(error, "cannot parse trailer");
	if (tok != PDF_TODICT)
		return fz_throw("expected trailer dictionary");

	error = pdf_parsedict(&xref->trailer, xref->file, buf, cap);
	if (error)
		return fz_rethrow(error, "cannot parse trailer");
	return fz_okay;
}

static fz_error *
readnewtrailer(pdf_xref *xref, char *buf, int cap)
{
	fz_error *error;

	pdf_logxref("load new xref format trailer\n");

	error = pdf_parseindobj(&xref->trailer, xref->file, buf, cap, nil, nil, nil);
	if (error)
		return fz_rethrow(error, "cannot parse trailer (compressed)");
	return fz_okay;
}

static fz_error *
readtrailer(pdf_xref *xref, char *buf, int cap)
{
	fz_error *error;
	int c;

	error = fz_seek(xref->file, xref->startxref, 0);
	if (error)
		return fz_rethrow(error, "cannot seek to startxref");

	c = fz_peekbyte(xref->file);
	error = fz_readerror(xref->file);
	if (error)
		return fz_rethrow(error, "cannot read trailer");

	if (c == 'x')
	{
		error = readoldtrailer(xref, buf, cap);
		if (error)
			return fz_rethrow(error, "cannot read trailer");
	}
	else if (c >= '0' && c <= '9')
	{
		error = readnewtrailer(xref, buf, cap);
		if (error)
			return fz_rethrow(error, "cannot read trailer");
	}
	else
	{
		return fz_throw("cannot recognize xref format");
	}

	return fz_okay;
}

/*
 * xref tables
 */

static fz_error *
readoldxref(fz_obj **trailerp, pdf_xref *xref, char *buf, int cap)
{
	fz_error *error;
	int ofs, len;
	char *s;
	int n;
	pdf_token_e tok;
	int i;
	int c;

	pdf_logxref("load old xref format\n");

	error = fz_readline(xref->file, buf, cap);
	if (error)
		return fz_rethrow(error, "cannot read xref marker");
	if (strncmp(buf, "xref", 4) != 0)
		return fz_throw("cannot find xref marker");

	while (1)
	{
		c = fz_peekbyte(xref->file);
		if (!(c >= '0' && c <= '9'))
			break;

		error = fz_readline(xref->file, buf, cap);
		if (error)
			return fz_rethrow(error, "cannot read xref count");

		s = buf;
		ofs = atoi(strsep(&s, " "));
		len = atoi(strsep(&s, " "));

		/* broken pdfs where the section is not on a separate line */
		if (s && *s != '\0')
		{
			fz_warn("broken xref section. proceeding anyway.");
			error = fz_seek(xref->file, -(2 + strlen(s)), 1);
			if (error)
				return fz_rethrow(error, "cannot seek to xref");
		}

		/* broken pdfs where size in trailer undershoots
		   entries in xref sections */
		if ((ofs + len) > xref->cap)
		{
			fz_warn("broken xref section, proceeding anyway.");
			xref->cap = ofs + len;
			xref->table = fz_realloc(xref->table, xref->cap * sizeof(pdf_xrefentry));
			if (!xref->table)
				return fz_throw("outofmem: xref table");
		}

		if ((ofs + len) > xref->len)
		{
			for (i = xref->len; i < (ofs + len); i++)
			{
				xref->table[i].ofs = 0;
				xref->table[i].gen = 0;
				xref->table[i].type = 0;
				xref->table[i].mark = 0;
				xref->table[i].stmbuf = nil;
				xref->table[i].stmofs = 0;
				xref->table[i].obj = nil;
			}
			xref->len = ofs + len;
		}

		for (i = 0; i < len; i++)
		{
			error = fz_read(&n, xref->file, (unsigned char *) buf, 20);
			if (error)
				return fz_rethrow(error, "cannot read xref table");
			if (!xref->table[ofs + i].type)
			{
				s = buf;
				xref->table[ofs + i].ofs = atoi(s);
				xref->table[ofs + i].gen = atoi(s + 11);
				xref->table[ofs + i].type = s[17];
			}
		}
	}

	error = pdf_lex(&tok, xref->file, buf, cap, &n);
	if (error)
		return fz_rethrow(error, "cannot parse trailer");
	if (tok != PDF_TTRAILER)
		return fz_throw("expected trailer marker");

	error = pdf_lex(&tok, xref->file, buf, cap, &n);
	if (error)
		return fz_rethrow(error, "cannot parse trailer");
	if (tok != PDF_TODICT)
		return fz_throw("expected trailer dictionary");

	error = pdf_parsedict(trailerp, xref->file, buf, cap);
	if (error)
		return fz_rethrow(error, "cannot parse trailer");
	return fz_okay;
}

static fz_error *
readnewxrefsection(pdf_xref *xref, fz_stream *stm, int i0, int i1, int w0, int w1, int w2)
{
	fz_error *error;
	int i, n;

	if (i0 < 0 || i0 + i1 > xref->len)
		return fz_throw("xref stream has too many entries");

	for (i = i0; i < i0 + i1; i++)
	{
		int a = 0;
		int b = 0;
		int c = 0;

		if (fz_peekbyte(stm) == EOF)
		{
			error = fz_readerror(stm);
			if (error)
				return fz_rethrow(error, "truncated xref stream");
			return fz_throw("truncated xref stream");
		}

		for (n = 0; n < w0; n++)
			a = (a << 8) + fz_readbyte(stm);
		for (n = 0; n < w1; n++)
			b = (b << 8) + fz_readbyte(stm);
		for (n = 0; n < w2; n++)
			c = (c << 8) + fz_readbyte(stm);

		error = fz_readerror(stm);
		if (error)
			return fz_rethrow(error, "truncated xref stream");

		if (!xref->table[i].type)
		{
			int t = w0 ? a : 1;
			xref->table[i].type = t == 0 ? 'f' : t == 1 ? 'n' : t == 2 ? 'o' : 0;
			xref->table[i].ofs = w2 ? b : 0;
			xref->table[i].gen = w1 ? c : 0;
		}
	}

	return fz_okay;
}

static fz_error *
readnewxref(fz_obj **trailerp, pdf_xref *xref, char *buf, int cap)
{
	fz_error *error;
	fz_stream *stm;
	fz_obj *trailer;
	fz_obj *index;
	fz_obj *obj;
	int oid, gen, stmofs;
	int size, w0, w1, w2;
	int t;

	pdf_logxref("load new xref format\n");

	error = pdf_parseindobj(&trailer, xref->file, buf, cap, &oid, &gen, &stmofs);
	if (error)
		return fz_rethrow(error, "cannot parse compressed xref stream object");

	if (oid < 0 || oid >= xref->len)
	{
		if (oid == xref->len && oid < xref->cap)
		{
		    /* allow broken pdf files that have off-by-one errors in the xref */
		    fz_warn("object id (%d %d R) out of range (0..%d)", oid, gen, xref->len - 1);
		    xref->len ++;
		}