shithub: pdffs

Info  •  Files  •  Log  •  Branches

fix tons of bugs, use proper streaming

--- a/array.c

+++ b/array.c

@@ -1,12 +1,11 @@

 #include <u.h>

 #include <libc.h>

-#include <bio.h>

 #include "pdf.h"

 /* 7.3.6 Array Objects */

 Object *

-pdfarray(Pdf *pdf, Biobuf *b)

+pdfarray(Pdf *pdf, Stream *s)

 {

 	Object *o, *m;

 	Object **a;

@@ -15,10 +14,10 @@

 	o = calloc(1, sizeof(*o));

 	o->pdf = pdf;

 	o->type = Oarray;

-	Bgetc(b); /* throw away '[' */

+	Sgetc(s); /* throw away '[' */

 	for(noel = 0;;){

-		if((c = Bgetc(b)) < 0 || c == ']')

+		if((c = Sgetc(s)) < 0 || c == ']')

 			break;

 		if(noel){

 			werrstr("no ']'");

@@ -25,8 +24,8 @@

 			goto err;

 		}

-		Bungetc(b);

-		if((m = pdfobj(pdf, b)) == nil){

+		Sungetc(s);

+		if((m = pdfobj(pdf, s)) == nil){

 			noel = 1;

 			continue;

 		}

@@ -65,7 +64,9 @@

 {

 	if(arraylen(o) <= i)

 		sysfatal("array: indexing out of range");

-	return o->type == Oarray ? o->array.e[i] : o;

+	o = o->type == Oarray ? o->array.e[i] : o;

+

+	return pdfeval(&o);

 }

 int

--- a/buffer.c

+++ b/buffer.c

@@ -64,7 +64,7 @@

 }

 int

-bufreadn(Buffer *b, Biobuf *bio, int sz)

+bufreadn(Buffer *b, Stream *s, int sz)

 {

 	int n, end;

@@ -71,7 +71,7 @@

 	if(bufgrow(b, sz) != 0)

 		return -1;

 	for(end = b->sz+sz; b->sz < end; b->sz += n){

-		if((n = Bread(bio, b->b+b->sz, sz)) < 1)

+		if((n = Sread(s, b->b+b->sz, sz)) < 1)

 			return -1;

 		sz -= n;

 	}

--- a/dict.c

+++ b/dict.c

@@ -1,12 +1,11 @@

 #include <u.h>

 #include <libc.h>

-#include <bio.h>

 #include "pdf.h"

 /* 7.3.7 Dictionary Objects */

 Object *

-pdfdict(Pdf *pdf, Biobuf *b)

+pdfdict(Pdf *pdf, Stream *s)

 {

 	Object *o, *k, *v;

 	KeyValue *kv;

@@ -13,7 +12,7 @@

 	int c, nokey;

 	/* skip '<<' */

-	Bseek(b, 2, 1);

+	Sseek(s, 2, 1);

 	k = v = nil;

 	o = calloc(1, sizeof(*o));

@@ -20,10 +19,10 @@

 	o->type = Odict;

 	o->pdf = pdf;

 	for(nokey = 0;;){

-		if((c = Bgetc(b)) < 0)

+		if((c = Sgetc(s)) < 0)

 			goto err;

 		if(c == '>'){

-			if(Bgetc(b) == '>')

+			if(Sgetc(s) == '>')

 				break;

 			werrstr("no '>>'");

 			goto err;

@@ -33,8 +32,8 @@

 			goto err;

 		}

-		Bungetc(b);

-		if((k = pdfobj(pdf, b)) == nil){

+		Sungetc(s);

+		if((k = pdfobj(pdf, s)) == nil){

 			nokey = 1;

 			continue;

 		}

@@ -42,7 +41,7 @@

 			werrstr("expected name as a key");

 			goto err;

 		}

-		if((v = pdfobj(pdf, b)) == nil)

+		if((v = pdfobj(pdf, s)) == nil)

 			goto err;

 		if((kv = realloc(o->dict.kv, (o->dict.nkv+1)*sizeof(KeyValue))) == nil)

@@ -73,9 +72,14 @@

 	pdfeval(&o);

 	if((o->type != Ostream && o->type != Odict) || name == nil)

 		return &null;

-	for(i = 0; i < o->dict.nkv && strcmp(name, o->dict.kv[i].key) != 0; i++);

+	for(i = 0; i < o->dict.nkv; i++){

+		if(strcmp(name, o->dict.kv[i].key) == 0){

+			o = pdfeval(i < o->dict.nkv ? &o->dict.kv[i].value : nil);

+			return o;

+		}

+	}

-	return pdfeval(i < o->dict.nkv ? &o->dict.kv[i].value : nil);

+	return &null;

 }

 vlong

--- a/eval.c

+++ b/eval.c

@@ -1,8 +1,65 @@

 #include <u.h>

 #include <libc.h>

-#include <bio.h>

 #include "pdf.h"

+static Object *

+evalobjstm(Pdf *pdf, Xref *x)

+{

+	Object *ostm, *o;

+	Stream *s;

+	Xref *xstm;

+	int i, off, nobj, first, index;

+

+	ostm = nil;

+	s = nil;

+	o = &null;

+	/* x is pointing at ObjStm, need to eval it to the actual object */

+	for(i = 0; i < pdf->nxref && pdf->xref[i].id != x->objstm; i++);

+	if(i >= pdf->nxref){

+		werrstr("no object id %d in xref", x->objstm);

+		goto err;

+	}

+	xstm = &pdf->xref[i];

+

+	if(Sseek(pdf->s, xstm->off, 0) != xstm->off){

+		werrstr("xref seek failed");

+		goto err;

+	}

+	if((ostm = pdfobj(pdf, pdf->s)) == nil)

+		goto err;

+	first = -1;

+	if((nobj = dictint(ostm, "N")) < 1 || (first = dictint(ostm, "First")) < 0){

+		werrstr("invalid ObjStm: nobj=%d first=%d", nobj, first);

+		goto err;

+	}

+

+	if((s = Sopen(ostm)) == nil)

+		goto err;

+	for(i = 0; i < nobj; i++){

+		Sgeti(s, &index);

+		Sgeti(s, &off);

+		if(x->id == index){

+			off += first;

+			if(Sseek(s, off, 0) != off){

+				werrstr("xref obj seek failed");

+				goto err;

+			}

+			if((o = pdfobj(pdf, s)) == nil)

+				goto err;

+			o = pdfeval(&o);

+			break;

+		}

+	}

+	Sclose(s);

+

+	return o;

+

+err:

+	pdfobjfree(ostm);

+	Sclose(s);

+	return &null;

+}

+

 Object *

 pdfeval(Object **oo)

 {

@@ -11,7 +68,7 @@

 	int i;

 	if(oo == nil)

-		return &null;

+		sysfatal("nil oo");

 	if(*oo == nil){

 		*oo = &null;

 		return &null;

@@ -20,19 +77,30 @@

 	if(o->type != Oindir)

 		return o;

-	for(i = 0; i < o->pdf->nxref && o->pdf->xref[i].id != o->indir.id; i++);

+	for(x = nil, i = 0; i < o->pdf->nxref; i++){

+		x = &o->pdf->xref[i];

+		if(x->id == o->indir.id)

+			break;

+	}

 	if(i >= o->pdf->nxref){

 		werrstr("no object id %d in xref", o->indir.id);

 		return &null;

 	}

-	x = &o->pdf->xref[i];

+	if(x->objstm > 0){

+		if((o = evalobjstm(o->pdf, x)) == &null)

+			werrstr("ObjStm: %r");

+		*oo = o;

+		return o;

+	}

-	if(Bseek(o->pdf->bio, x->off, 0) != x->off){

+	if(Sseek(o->pdf->s, x->off, 0) != x->off){

 		werrstr("xref seek failed");

 		return &null;

 	}

-	if((d = pdfobj(o->pdf, o->pdf->bio)) == nil)

+	if((d = pdfobj(o->pdf, o->pdf->s)) == nil){

+		werrstr("eval: %r [at %p]", (void*)x->off);

 		return &null;

+	}

 	*oo = d;

 	pdfobjfree(o);

--- a/f_flate.c

+++ b/f_flate.c

@@ -1,6 +1,5 @@

 #include <u.h>

 #include <libc.h>

-#include <bio.h>

 #include <flate.h>

 #include "pdf.h"

--- a/filter.c

+++ b/filter.c

@@ -1,6 +1,5 @@

 #include <u.h>

 #include <libc.h>

-#include <bio.h>

 #include "pdf.h"

 /* 7.4 Filters */

--- a/main.c

+++ b/main.c

@@ -20,6 +20,8 @@

 	Pdf *pdf;

 	Biobuf *b;

 	Object *v;

+	Stream *s;

+	int i, n;

 	quotefmtinstall();

 	inflateinit();

@@ -29,28 +31,33 @@

 		usage();

 	}ARGEND

-#ifdef TEST

-#define T(x) \

-	void x(void); \

-	x();

-

-	if(argc != 1){

-		T(test_pdfstring);

-		T(test_pdfname);

-		threadexitsall(nil);

-	}

-#endif

-

-	if(argc != 1)

+	if(argc < 1)

 		usage();

 	if((b = Bopen(argv[0], OREAD)) == nil)

 		sysfatal("%r");

 	if((pdf = pdfopen(b)) == nil)

 		sysfatal("%s: %r", argv[0]);

+	for(v = pdf->root, i = 1; i < argc; i++){

+		if(argv[i][0] == '['){

+			n = atoi(argv[i]+1);

+			v = arrayget(v, n);

+		}else if(argv[i][0] == '@' && argv[i][1] == 0 && v->type == Ostream){

+			if((s = Sopen(v)) == nil)

+				sysfatal("%r");

+			print("%.*s\n", s->buf.sz, s->buf.b);

+			Sclose(s);

+			break;

+		}else{

+			v = dictget(v, argv[i]);

+		}

+	}

+	print("%O\n", v);

+/*

 	if((v = dictget(pdf->info, "Creator")) != nil)

 		fprint(2, "creator: %s\n", v->str);

 	if((v = dictget(pdf->info, "Producer")) != nil)

 		fprint(2, "producer: %s\n", v->str);

+*/

 	pdfclose(pdf);

 	threadexitsall(nil);

--- a/misc.c

+++ b/misc.c

@@ -1,6 +1,5 @@

 #include <u.h>

 #include <libc.h>

-#include <bio.h>

 #include "pdf.h"

 static char *otypes[] = {

@@ -15,17 +14,58 @@

 	[Oindir] = "indir",

 };

-static char *xtypes[] = {

-	[Xusual] = "usual",

-	[Xuncompressed] = "uncompressed",

-	[Xcompressed] = "compressed",

-};

-

 Object null = {

 	.type = Onull,

 };

 int

+Ofmt(Fmt *f)

+{

+	Object *o;

+	int i;

+

+	o = va_arg(f->args, Object*);

+	if(o == nil || o == &null)

+		return fmtprint(f, "null");

+	switch(o->type){

+	case Obool:

+		return fmtprint(f, o->bool ? "true" : "false");

+

+	case Onum:

+		return fmtprint(f, "%g", o->num);

+

+	case Ostr:

+		if(isutf8(o->str, o->len))

+			return fmtprint(f, "%q", o->str);

+		return fmtprint(f, "<%.*H>", o->len, o->str);

+

+	case Oname:

+		return fmtprint(f, "/%s", o->name);

+

+	case Oarray:

+		fmtprint(f, "[");

+		for(i = 0; i < o->array.ne; i++)

+			fmtprint(f, "%s%O", i > 0 ? ", " : "", o->array.e[i]);

+		return fmtprint(f, "]");

+

+	case Ostream: /* FIXME dump the stream? */

+	case Odict:

+		fmtprint(f, "<<");

+		for(i = 0; i < o->dict.nkv; i++)

+			fmtprint(f, "%s%s = %O", i > 0 ? ", " : "", o->dict.kv[i].key, o->dict.kv[i].value);

+		return fmtprint(f, ">>%s", o->type == Ostream ? "+stream" : "");

+

+	case Onull:

+		return fmtprint(f, "null");

+

+	case Oindir:

+		return fmtprint(f, "@%d[gen=%d]", o->indir.id, o->indir.gen);

+

+	}

+	return fmtprint(f, "???");

+}

+

+int

 Tfmt(Fmt *f)

 {

 	Object *o;

@@ -45,15 +85,10 @@

 	x = va_arg(f->args, Xref);

-	switch(x.type){

-	case Xusual:

-		return fmtprint(f, "<%s id=%d gen=%d off=%d>", xtypes[x.type], x.id, x.gen, x.off);

-	case Xuncompressed:

-		return fmtprint(f, "<%s gen=%d off=%d>", xtypes[x.type], x.gen, x.off);

-	case Xcompressed:

-		return fmtprint(f, "<%s id=%d objnum=%d>", xtypes[x.type], x.id, x.objnum);

-	}

-	return -1;

+	if(x.objstm > 0)

+		return fmtprint(f, "<compressed id=%d objstm=%d index=%d>", x.id, x.objstm, x.index);

+

+	return fmtprint(f, "<uncompressed id=%d off=%d gen=%d>", x.id, x.off, x.gen);

 }

 int

@@ -74,13 +109,15 @@

 }

 int

-Bgetint(Biobuf *b, int *i)

+isutf8(char *s, int len)

 {

-	double d;

+	int i, n;

+	Rune r;

-	if(Bgetd(b, &d) != 1 || isNaN(d))

-		return -1;

-	*i = d;

+	for(i = 0; i < len; i += n, s += n){

+		if((n = chartorune(&r, s)) < 1 || r == Runeerror)

+			break;

+	}

-	return 1;

+	return i >= len;

 }

--- a/mkfile

+++ b/mkfile

@@ -1,6 +1,5 @@

 </$objtype/mkfile

-CFLAGS=$CFLAGS -DTEST

 TARG=pdffs

 OFILES=\

--- a/name.c

+++ b/name.c

@@ -1,18 +1,17 @@

 #include <u.h>

 #include <libc.h>

-#include <bio.h>

 #include "pdf.h"

 /* 7.3.5 Name Objects */

 Object *

-pdfname(Biobuf *b)

+pdfname(Stream *stream)

 {

 	Object *o;

 	char *s, *r, hex[3];

 	int c, sz, maxsz;

-	Bgetc(b); /* skip '/' */

+	Sgetc(stream); /* skip '/' */

 	maxsz = 32;

 	if((s = malloc(maxsz)) == nil)

@@ -19,14 +18,14 @@

 		goto err;

 	for(sz = 0;;){

-		if((c = Bgetc(b)) < 0){

-			if(c == Beof)

+		if((c = Sgetc(stream)) < 0){

+			if(c == -1)

 				break;

 			goto err;

 		}

 		if(isws(c) || isdelim(c)){

-			Bungetc(b);

+			Sungetc(stream);

 			break;

 		}

 		if(c < '!' || c > '~'){

@@ -34,10 +33,10 @@

 			goto err;

 		}

 		if(c == '#'){

-			if((c = Bgetc(b)) < 0)

+			if((c = Sgetc(stream)) < 0)

 				goto err;

 			hex[0] = c;

-			if((c = Bgetc(b)) < 0)

+			if((c = Sgetc(stream)) < 0)

 				goto err;

 			hex[1] = c;

 			if(dec16((uchar*)hex, 1, hex, 2) != 1){

@@ -67,65 +66,3 @@

 	free(s);

 	return nil;

 }

-

-#ifdef TEST

-static struct {

-	char *in;

-	char *out;

-}t[] = {

-	{"/SimpleName", "SimpleName"},

-	{"/.$()", ".$"},

-	{"/#30", "0"},

-	{"/#3", nil},

-	{"/#G0", nil},

-	{"/#", nil},

-	{"/Space Between", "Space"},

-	{"/Two/Names", "Two"},

-	{"/\xff", nil,},

-};

-

-static char *s;

-static int off, n;

-

-static int

-rd(Biobufhdr *, void *data, long sz)

-{

-	if(sz > n-off)

-		sz = n-off;

-	memmove(data, s+off, sz);

-	off += sz;

-	return sz;

-}

-

-void

-test_pdfname(void)

-{

-	Object *o;

-	Biobuf b;

-	int i;

-

-	fprint(2, "pdfname\n");

-	for(i = 0; i < nelem(t); i++){

-		s = t[i].in;

-		n = strlen(s);

-		off = 0;

-		Binit(&b, -1, OREAD);

-		Biofn(&b, rd);

-

-		fprint(2, "\t%d: ", i);

-		o = pdfname(&b);

-		if(o == nil && t[i].out != nil)

-			fprint(2, "ERROR: expected %q, got error: %r\n", t[i].out);

-		else if(o != nil && t[i].out == nil)

-			fprint(2, "ERROR: expected error, got %q\n", o->name);

-		else if(o == nil && t[i].out == nil)

-			fprint(2, "OK (%r)\n");

-		else if(strcmp(o->name, t[i].out) != 0)

-			fprint(2, "ERROR: expected %q, got %q\n", t[i].out, o->name);

-		else

-			fprint(2, "OK\n");

-		pdfobjfree(o);

-		Bterm(&b);

-	}

-}

-#endif

--- a/object.c

+++ b/object.c

@@ -1,106 +1,105 @@

 #include <u.h>

 #include <libc.h>

 #include <ctype.h>

-#include <bio.h>

 #include "pdf.h"

-Object *pdfstring(Biobuf *b);

-Object *pdfname(Biobuf *b);

-Object *pdfarray(Pdf *pdf, Biobuf *b);

-Object *pdfdict(Pdf *pdf, Biobuf *b);

+Object *pdfstring(Stream *s);

+Object *pdfname(Stream *s);

+Object *pdfarray(Pdf *pdf, Stream *s);

+Object *pdfdict(Pdf *pdf, Stream *s);

 /* General function to parse an object of any type. */

 Object *

-pdfobj(Pdf *pdf, Biobuf *b)

+pdfobj(Pdf *pdf, Stream *s)

 {

 	Object *o, *o2;

 	vlong off;

 	int c, tf;

 	Xref xref;

-	char s[16];

+	char b[16];

 	o = o2 = nil;

-	do; while(isws(c = Bgetc(b)));

+	do; while(isws(c = Sgetc(s)));

 	if(c < 0)

 		goto err;

 	switch(c){

 	case '<': /* dictionary or a string */

-		c = Bgetc(b);

+		c = Sgetc(s);

 		if(c == '<'){

-			Bseek(b, -2, 1);

-			if((o = pdfdict(pdf, b)) != nil){

+			Sseek(s, -2, 1);

+			if((o = pdfdict(pdf, s)) != nil){

 				/* check for attached stream */

-				off = Boffset(b);

-				do; while(isws(Bgetc(b)));

-				Bungetc(b);

-				if(Bread(b, s, 7) == 7 && memcmp(s, "stream", 6) == 0 && isws(c = s[6])){

+				off = Soffset(s);

+				do; while(isws(Sgetc(s)));

+				Sungetc(s);

+				if(Sread(s, b, 7) == 7 && memcmp(b, "stream", 6) == 0 && isws(c = b[6])){

 					/* there IS a stream */

-					if(c == '\r' && (c = Bgetc(b)) < 0)

+					if(c == '\r' && (c = Sgetc(s)) < 0)

 						goto err;

 					if(c != '\n'){

 						werrstr("stream has no newline after dict");

 						goto err;

 					}

-					o->stream.off = Boffset(b);

+					o->stream.off = Soffset(s);

 					o->type = Ostream;

 					o->stream.len = dictint(o, "Length");

 					return o;

 				}

-				Bseek(b, off, 0);

+				Sseek(s, off, 0);

 				return o;

 			}

 		}

-		Bungetc(b);

+		Sungetc(s);

 		/* fall through */

 	case '(':

-		Bungetc(b);

-		if((o = pdfstring(b)) != nil)

+		Sungetc(s);

+		if((o = pdfstring(s)) != nil)

 			o->pdf = pdf;

 		return o;

 	case '/':

-		Bungetc(b);

-		if((o = pdfname(b)) != nil)

+		Sungetc(s);

+		if((o = pdfname(s)) != nil)

 			o->pdf = pdf;

 		return o;

 	case '[':

-		Bungetc(b);

-		if((o = pdfarray(pdf, b)) != nil)

+		Sungetc(s);

+		if((o = pdfarray(pdf, s)) != nil)

 			o->pdf = pdf;

 		return o;

 	case 'n':

-		off = Boffset(b);

-		if(Bgetc(b) == 'u' && Bgetc(b) == 'l' && Bgetc(b) == 'l' && (isws(c = Bgetc(b)) || isdelim(c))){

-			Bungetc(b);

+		off = Soffset(s);

+		if(Sgetc(s) == 'u' && Sgetc(s) == 'l' && Sgetc(s) == 'l' && (isws(c = Sgetc(s)) || isdelim(c))){

+			Sungetc(s);

 			return &null;

 		}

-		Bseek(b, off, 0);

+		Sseek(s, off, 0);

 		c = 'f';

 		goto unexpected;

 	case 't':

-		off = Boffset(b);

+		off = Soffset(s);

 		tf = 1;

-		if(Bgetc(b) == 'r' && Bgetc(b) == 'u' && Bgetc(b) == 'e' && (isws(c = Bgetc(b)) || isdelim(c)))

+		if(Sgetc(s) == 'r' && Sgetc(s) == 'u' && Sgetc(s) == 'e' && (isws(c = Sgetc(s)) || isdelim(c)))

 			goto bool;

-		Bseek(b, off, 0);

+		Sseek(s, off, 0);

 		c = 't';

 		goto unexpected;

 	case 'f':

-		off = Boffset(b);

+		off = Soffset(s);

 		tf = 0;

-		if(Bgetc(b) == 'a' && Bgetc(b) == 'l' && Bgetc(b) == 's' && Bgetc(b) == 'e' && (isws(c = Bgetc(b)) || isdelim(c)))

+		if(Sgetc(s) == 'a' && Sgetc(s) == 'l' && Sgetc(s) == 's' && Sgetc(s) == 'e' && (isws(c = Sgetc(s)) || isdelim(c)))

 			goto bool;

-		Bseek(b, off, 0);

+		Sseek(s, off, 0);

 		c = 'f';

 		goto unexpected;

 bool:

-		Bungetc(b);

+		Sungetc(s);

 		if((o = calloc(1, sizeof(*o))) == nil)

 			goto err;

 		o->type = Obool;

@@ -111,20 +110,20 @@

 	default:

 		if(!isdigit(c)){

 unexpected:

-			Bungetc(b);

+			Sungetc(s);

 			werrstr("unexpected char '%c'", c);

 			goto err;

 		}

 		 /* it could be a number or an indirect object */

-		Bungetc(b);

+		Sungetc(s);

 		if((o = calloc(1, sizeof(*o))) == nil)

 			goto err;

 		o->pdf = pdf;

-		Bgetd(b, &o->num); /* get the first number */

-		off = Boffset(b); /* seek here if not an indirect object later */

+		Sgetd(s, &o->num); /* get the first number */

+		off = Soffset(s); /* seek here if not an indirect object later */

-		if((o2 = pdfobj(pdf, b)) != nil && o2->type == Onum){ /* second object is number too */

-			do; while(isws(c = Bgetc(b)));

+		if((o2 = pdfobj(pdf, s)) != nil && o2->type == Onum){ /* second object is number too */

+			do; while(isws(c = Sgetc(s)));

 			if(c < 0)

 				goto err;

 			if(c == 'R'){ /* indirect object */

@@ -134,13 +133,13 @@

 				pdfobjfree(o2);

 				return o;

 			}

-			if(c == 'o' && Bgetc(b) == 'b' && Bgetc(b) == 'j'){ /* object */

+			if(c == 'o' && Sgetc(s) == 'b' && Sgetc(s) == 'j'){ /* object */

 				xref.id = o->num;

 				xref.gen = o2->num;

 				/* FIXME put into a map */

 				pdfobjfree(o);

 				pdfobjfree(o2);

-				if((o = pdfobj(pdf, b)) != nil)

+				if((o = pdfobj(pdf, s)) != nil)

 					return o;

 				o2 = nil;

 			}

@@ -148,7 +147,7 @@

 		/* just a number, go back and return it */

 		o->type = Onum;

-		if(Bseek(b, off, 0) != off){

+		if(Sseek(s, off, 0) != off){

 			werrstr("seek failed");

 			goto err;

 		}

--- a/pdf.c

+++ b/pdf.c

@@ -1,6 +1,5 @@

 #include <u.h>

 #include <libc.h>

-#include <bio.h>

 #include <ctype.h>

 #include "pdf.h"

@@ -8,14 +7,16 @@

 trailerread(Pdf *pdf)

 {

 	Object *o;

+	int prev;

-	if((o = pdfobj(pdf, pdf->bio)) == nil)

+	if((o = pdfobj(pdf, pdf->s)) == nil)

 		goto err;

-

 	if(o->type != Odict){

 		werrstr("isn't a dictionary");

 		goto err;

 	}

+	if((prev = dictint(o, "Prev")) > 0 && (Sseek(pdf->s, prev, 0) < 0 || xrefreadold(pdf) != 0))

+		goto err;

 	pdf->root = pdfref(dictget(o, "Root"));

 	pdf->info = pdfref(dictget(o, "Info"));

@@ -28,24 +29,28 @@

 }

 Pdf *

-pdfopen(Biobuf *b)

+pdfopen(void *bio)

 {

 	Pdf *pdf;

 	Object *o;

+	Stream *stream;

 	char tmp[64], *s, *x;

 	int xreftb; /* 7.5.4 xref table offset from the beginning of the file */

 	int i, n, off;

+	fmtinstall('H', encodefmt);

+	fmtinstall('O', Ofmt);

 	fmtinstall('T', Tfmt);

 	fmtinstall(L'⊗', ⊗fmt);

 	o = nil;

-	if((pdf = calloc(1, sizeof(*pdf))) == nil)

+	pdf = nil;

+	if((stream = Sbio(bio)) == nil || (pdf = calloc(1, sizeof(*pdf))) == nil)

 		goto err;

-	pdf->bio = b;

+	pdf->s = stream;

 	/* check header */

-	if(Bread(b, tmp, 8) != 8 ||

+	if(Sread(stream, tmp, 8) != 8 ||

 	   strncmp(tmp, "%PDF-", 5) != 0 || !isdigit(tmp[5]) || tmp[6] != '.' || !isdigit(tmp[7])){

 		werrstr("not a pdf");

 		goto err;

@@ -55,8 +60,8 @@

 	/* read a block of data */

 	n = sizeof(tmp)-1;

-	Bseek(b, -n, 2);

-	if(Bread(b, tmp, n) != n){

+	Sseek(stream, -n, 2);

+	if(Sread(stream, tmp, n) != n){

 badtrailer:

 		werrstr("invalid trailer");

 		goto err;

@@ -75,25 +80,27 @@

 		goto badtrailer;

 	/* read xref */

-	if(Bseek(b, xreftb, 0) != xreftb){

+	if(Sseek(stream, xreftb, 0) != xreftb){

 		werrstr("xref position out of range");

 		goto err;

 	}

 	for(;;){

-		off = Boffset(b);

-		if(Bread(b, tmp, sizeof(tmp)) < 8){

+		while(isspace(Sgetc(stream)));

+		Sungetc(stream);

+		off = Soffset(stream);

+		if(Sread(stream, tmp, sizeof(tmp)) < 8){

 badxref:

 			werrstr("invalid xref: %r");

 			goto err;

 		}

 		if(memcmp(tmp, "xref", 4) == 0){

-			if(Bseek(b, -sizeof(tmp)+5, 1) < 0 || xrefreadold(pdf) != 0)

+			if(Sseek(stream, -sizeof(tmp), 1) < 0 || xrefreadold(pdf) != 0)

 				goto err;

 			/* there could be more updates, try it */

 		}else if(memcmp(tmp, "trailer", 7) == 0){ /* 7.5.5 file trailer */

 			/* move to the trailer dictionary */

 			n = off + 8;

-			if(Bseek(b, n, 0) != n || trailerread(pdf) != 0){

+			if(Sseek(stream, n, 0) != n || trailerread(pdf) != 0){

 				werrstr("invalid trailer: %r");

 				goto err;

 			}

@@ -100,7 +107,7 @@

 			/* trailer is supposed to be the last thing */

 			break;

 		}else if(isdigit(tmp[0])){ /* could be 7.5.8 xref stream (since PDF 1.5) */

-			if(Bseek(b, xreftb, 0) != xreftb)

+			if(Sseek(stream, xreftb, 0) != xreftb)

 				goto badxref;

 			if(xrefreadstream(pdf) != 0)

 				goto err;

@@ -110,13 +117,13 @@

 	/* root is required, info is optional */

 	if(pdf->root == &null){

-		werrstr("no root");

+		werrstr("no root: %r");

 		goto err;

 	}

 	return pdf;

 err:

-	werrstr("pdfopen: %r [at %p]", (void*)Boffset(b));

+	werrstr("pdfopen: %r [at %p]", (void*)Soffset(stream));

 	pdfclose(pdf);

 	pdfobjfree(o);

 	return nil;

@@ -127,8 +134,8 @@

 {

 	if(pdf == nil)

 		return;

-	if(pdf->bio != nil)

-		Bterm(pdf->bio);

+	if(pdf->s != nil)

+		Sclose(pdf->s);

 	free(pdf->xref);

 	free(pdf);

 }

--- a/pdf.h

+++ b/pdf.h

@@ -8,10 +8,6 @@

 	Ostream, /* 7.3.8 */

 	Onull,   /* 7.3.9 */

 	Oindir,  /* 7.3.10 */

-

-	Xusual = 0,

-	Xuncompressed,

-	Xcompressed,

 };

 typedef struct Buffer Buffer;

@@ -82,7 +78,7 @@

 };

 struct Pdf {

-	Biobuf *bio;

+	Stream *s;

 	Xref *xref;

 	int nxref; /* 7.5.4 xref subsection number of objects */

@@ -92,28 +88,34 @@

 struct Xref {

 	u32int id;

-	union{

-		u32int off;

-		u32int objnum;

+	union {

+		struct { /* uncompressed */

+			u32int off;

+			u16int gen;

+		};

+

+		struct { /* compressed, objstm > 0 */

+			u16int index; /* index within ObjStm */

+		};

 	};

-	u16int gen;

-	u16int type;

+	u16int objstm; /* > 0 means it's compressed and points to the ObjStm */

 };

 struct Stream {

 	Buffer buf;

-	Biobuf *bio;

+	void *bio;

+	int linelen;

 };

 extern Object null;

-Pdf *pdfopen(Biobuf *b);

+Pdf *pdfopen(void *bio);

 void pdfclose(Pdf *pdf);

 /*

  * Parse an object.

  */

-Object *pdfobj(Pdf *pdf, Biobuf *bio);

+Object *pdfobj(Pdf *pdf, Stream *s);

 /*

  * Deallocate the object and all its children. Refcount is

@@ -146,6 +148,8 @@

  */

 int isdelim(int c);

+int isutf8(char *s, int len);

+

 int arraylen(Object *o);

 Object *arrayget(Object *o, int i);

 int arrayint(Object *o, int i);

@@ -156,9 +160,19 @@

 Object *dictdict(Object *o, char *name);

 int dictints(Object *o, char *name, int *el, int nel);

-Stream *streamopen(Object *o);

-int streamsize(Stream *s);

-void streamclose(Stream *s);

+Stream *Sbio(void *bio);

+Stream *Sopen(Object *o);

+int Sread(Stream *s, void *b, int sz);

+int Sgetc(Stream *s);

+int Sungetc(Stream *s);

+int Ssize(Stream *s);

+int Soffset(Stream *s);

+int Sseek(Stream *s, int off, int whence);

+void Sclose(Stream *s);

+int Sgetd(Stream *s, double *d);

+int Sgeti(Stream *s, int *i);

+char *Srdstr(Stream *s, int delim, int zero);

+int Slinelen(Stream *s);

 Filter *filteropen(char *name, Object *o);

 int filterrun(Filter *f, Buffer *bi, Buffer *bo);

@@ -169,16 +183,17 @@

 int bufeof(Buffer *b);

 int bufleft(Buffer *b);

 uchar *bufdata(Buffer *b, int *sz);

-int bufreadn(Buffer *b, Biobuf *bio, int sz);

+int bufreadn(Buffer *b, Stream *s, int sz);

 int bufput(Buffer *b, uchar *d, int sz);

 int bufget(Buffer *b, uchar *d, int sz);

 void bufdump(Buffer *b);

+#pragma varargck type "O" Object*

 #pragma varargck type "T" Object*

 #pragma varargck type "⊗" Xref

+int Ofmt(Fmt *f);

 int Tfmt(Fmt *f);

 int ⊗fmt(Fmt *f);

-int Bgetint(Biobuf *b, int *i);

 int xrefreadold(Pdf *pdf);

 int xrefreadstream(Pdf *pdf);

--- a/stream.c

+++ b/stream.c

@@ -3,18 +3,20 @@

 #include <bio.h>

 #include "pdf.h"

-static int

-bufiof(Biobufhdr *b, void *data, long n)

+Stream *

+Sbio(void *bio)

 {

 	Stream *s;

-	s = (Stream*)((char*)b - sizeof(*s));

+	if((s = calloc(1, sizeof(*s))) == nil)

+		return nil;

+	s->bio = bio;

-	return bufget(&s->buf, data, n);

+	return s;

 }

 Stream *

-streamopen(Object *o)

+Sopen(Object *o)

 {

 	Stream *s;

 	Buffer b, x;

@@ -23,13 +25,15 @@

 	int i, nflts;

 	s = nil;

-	if(pdfeval(&o)->type != Ostream) /* FIXME open a string object as a stream as well? */

+	if(pdfeval(&o)->type != Ostream){ /* FIXME open a string object as a stream as well? */

+		werrstr("not a stream");

 		return nil;

+	}

 	bufinit(&b, nil, 0);

-	if(Bseek(o->pdf->bio, o->stream.off, 0) != o->stream.off)

+	if(Sseek(o->pdf->s, o->stream.off, 0) != o->stream.off)

 		return nil;

-	if(bufreadn(&b, o->pdf->bio, o->stream.len) < 0)

+	if(bufreadn(&b, o->pdf->s, o->stream.len) < 0)

 		goto err;

 	/* see if there are any filters */

@@ -64,14 +68,11 @@

 		}

 	}

-	if((s = calloc(1, sizeof(*s)+sizeof(Biobuf))) == nil){

+	if((s = calloc(1, sizeof(*s))) == nil){

 		buffree(&b);

 		return nil;

 	}

-	s->bio = (Biobuf*)(s+1);

 	s->buf = b;

-	Binit(s->bio, Bfildes(o->pdf->bio), OREAD);

-	Biofn(s->bio, bufiof);

 	return s;

 err:

@@ -82,18 +83,164 @@

 }

 int

-streamsize(Stream *s)

+Sread(Stream *s, void *b, int sz)

 {

+	return s->bio != nil ? Bread(s->bio, b, sz) : bufget(&s->buf, b, sz);

+}

+

+int

+Sgetc(Stream *s)

+{

+	int n;

+	uchar c;

+

+	if(s->bio != nil)

+		return Bgetc(s->bio);

+	if((n = bufget(&s->buf, &c, 1)) < 0)

+		return -2;

+

+	return n == 0 ? -1 : (int)c;

+}

+

+int

+Sungetc(Stream *s)

+{

+	return s->bio != nil ? Bungetc(s->bio) : Sseek(s, -1, 1);

+}

+

+int

+Soffset(Stream *s)

+{

+	return s->bio != nil ? Boffset(s->bio) : s->buf.off;

+}

+

+int

+Ssize(Stream *s)

+{

+	assert(s->bio == nil);

 	return bufleft(&s->buf);

 }

+struct sgetd

+{

+	Stream *s;

+	int eof;

+};

+

+static int

+Sgetdf(void *vp)

+{

+	int c;

+	struct sgetd *sg = vp;

+

+	c = Sgetc(sg->s);

+	if(c < 0)

+		sg->eof = 1;

+	return c;

+}

+

+int

+Sgetd(Stream *s, double *dp)

+{

+	double d;

+	struct sgetd b;

+

+	b.s = s;

+	b.eof = 0;

+	d = charstod(Sgetdf, &b);

+	if(b.eof)

+		return -1;

+	Sungetc(s);

+	*dp = d;

+

+	return 1;

+}

+

+int

+Sgeti(Stream *s, int *i)

+{

+	double d;

+	int res, c;

+

+	while((c = isws(Sgetc(s))));

+	if(c < 0)

+		return c;

+	Sungetc(s);

+	res = Sgetd(s, &d);

+	*i = d;

+

+	return res;

+}

+

+int

+Sseek(Stream *s, int off, int whence)

+{

+	if(s->bio != nil)

+		return Bseek(s->bio, off, whence);

+

+	if(whence == 1)

+		off += s->buf.off;

+	else if(whence == 2)

+		off += s->buf.sz;

+	if(off < 0){

+		werrstr("seek: %d < 0", off);

+		off = 0;

+	}else if(off > s->buf.sz){

+		werrstr("seek: %d > %d", off, s->buf.sz);

+		off = s->buf.sz;

+	}

+

+	s->buf.off = off;

+

+	return off;

+}

+

+char *

+Srdstr(Stream *s, int delim, int zero)

+{

+	int i, len;

+	char *line;

+

+	if(s->bio != nil){

+		line = Brdstr(s->bio, delim, zero);

+		s->linelen = Blinelen(s->bio);

+		return line;

+	}

+

+	for(i = s->buf.off; i < s->buf.sz;){

+		i++;

+		if(s->buf.b[i-1] == delim)

+			break;

+	}

+	if(i >= s->buf.sz)

+		return nil;

+	len = i - s->buf.off;

+	if((line = malloc(len+1)) == nil)

+		return nil;

+	memmove(line, s->buf.b+s->buf.off, len);

+	s->buf.off += len;

+	if(line[len-1] == delim && zero)

+		len--;

+	line[len] = 0;

+	s->linelen = len;

+

+	return line;

+}

+

+int

+Slinelen(Stream *s)

+{

+	return s->linelen;

+}

+

 void

-streamclose(Stream *s)

+Sclose(Stream *s)

 {

 	if(s == nil)

 		return;

 	buffree(&s->buf);

-	Bterm(s->bio);

+	if(s->bio != nil)

+		Bterm(s->bio);

 	free(s);

 }

--- a/string.c

+++ b/string.c

@@ -1,6 +1,5 @@

 #include <u.h>

 #include <libc.h>

-#include <bio.h>

 #include "pdf.h"

 /* 7.3.4 String Objects */

@@ -18,15 +17,15 @@

 };

 static Object *

-stringhex(Biobuf *b)

+stringhex(Stream *stream)

 {

 	char *s;

 	Object *o;

 	int len, n;

-	if((s = Brdstr(b, '>', 0)) == nil)

+	if((s = Srdstr(stream, '>', 0)) == nil)

 		return nil;

-	len = Blinelen(b) - 1;

+	len = Slinelen(stream) - 1;

 	if(s[len] != '>'){

 		werrstr("no '>'");

 		free(s);

@@ -50,7 +49,7 @@

 }

 Object *

-pdfstring(Biobuf *b)

+pdfstring(Stream *stream)

 {

 	Object *o;

 	char *s, *r;

@@ -62,14 +61,14 @@

 		return nil;

 	for(paren = sz = 0;;){

-		if((c = Bgetc(b)) < 0)

+		if((c = Sgetc(stream)) < 0)

 			break;

 		switch(c){

 		case '<':

 			if(sz == 0){

-				Bungetc(b);

-				return stringhex(b);

+				Sungetc(stream);

+				return stringhex(stream);

 			}

 			break;

@@ -86,16 +85,16 @@

 			continue;

 		case '\\':

-			if((c = Bgetc(b)) <= 0)

+			if((c = Sgetc(stream)) <= 0)

 				break;

 			if(c >= '0' && c <= '7'){ /* octal */

 				oct[0] = c;

-				for(i = 1; i < 3 && (c = Bgetc(b)) >= '0' && c <= '7'; i++)

+				for(i = 1; i < 3 && (c = Sgetc(stream)) >= '0' && c <= '7'; i++)

 					oct[i] = c;

 				if(c <= 0)

 					break;

 				if(c < '0' || c > '7')

-					Bungetc(b);

+					Sungetc(stream);

 				oct[i] = 0;

 				c = strtol(oct, nil, 8);

 			}else if(c >= nelem(esc) || (c = esc[c]) == 0){

@@ -147,77 +146,3 @@

 	werrstr("string: %r");

 	return nil;

 }

-

-#ifdef TEST

-static struct {

-	char *in;

-	char *out;

-}t[] = {

-	{"", nil},

-	{"(test, success)", "test, success"},

-	{"(simple string)", "simple string"},

-	{"(non-closed paren", nil},

-	{"wrong first char", nil},

-	{"(parens((()((())))()))", "parens"},

-	{"(\\0053)", "\x053"},

-	{"(\\053)", "+"},

-	{"(\\53)", "+"},

-	{"()", ""},

-	{")", nil},

-	{"(\\)\\()", ")("},

-	{"(\\\\)", "\\"},

-	{"a", nil},

-	{"(1\\\n2)", "12"},

-	{"<323130>", "210"},

-	{"<32313>", "210"},

-	{"<>", ""},

-	{"<", nil},

-	{"<zz>", nil},

-	{">", nil},

-};

-

-static char *s;

-static int off, n;

-

-static int

-rd(Biobufhdr *, void *data, long sz)

-{

-	if(sz > n-off)

-		sz = n-off;

-	memmove(data, s+off, sz);

-	off += sz;

-	return sz;

-}

-

-void

-test_pdfstring(void)

-{

-	Object *o;

-	Biobuf b;

-	int i;

-

-	fprint(2, "pdfstring\n");

-	for(i = 0; i < nelem(t); i++){

-		s = t[i].in;

-		n = strlen(s);

-		off = 0;

-		Binit(&b, -1, OREAD);

-		Biofn(&b, rd);

-

-		fprint(2, "\t%d: ", i);

-		o = pdfstring(&b);

-		if(o == nil && t[i].out != nil)

-			fprint(2, "ERROR: expected %q, got error: %r\n", t[i].out);

-		else if(o != nil && t[i].out == nil)

-			fprint(2, "ERROR: expected error, got %q\n", o->str);

-		else if(o == nil && t[i].out == nil)

-			fprint(2, "OK (%r)\n");

-		else if(strcmp(o->str, t[i].out) != 0)

-			fprint(2, "ERROR: expected %q, got %q\n", t[i].out, o->str);

-		else

-			fprint(2, "OK\n");

-		pdfobjfree(o);

-		Bterm(&b);

-	}

-}

-#endif

--- a/xref.c

+++ b/xref.c

@@ -1,6 +1,5 @@

 #include <u.h>

 #include <libc.h>

-#include <bio.h>

 #include <ctype.h>

 #include "pdf.h"

@@ -18,18 +17,21 @@

 	char *s, *e;

 	Xref *x;

-	if(Bgetint(pdf->bio, &xref0) != 1 || xref0 < 0){

+	Sseek(pdf->s, 4, 1);	

+	if(Sgeti(pdf->s, &xref0) != 1 || xref0 < 0){

 		werrstr("invalid xref0");

 		return -1;

 	}

-	if(Bgetint(pdf->bio, &nxref) != 1 || nxref < 0){

+	if(Sgeti(pdf->s, &nxref) != 1 || nxref < 0){

 		werrstr("invalid nxref");

 		return -1;

 	}

+	if(nxref < 1)

+		return 0;

 	/* skip whitespace and move to the first subsection */

-	while(isspace(Bgetc(pdf->bio)));

-	Bungetc(pdf->bio);

+	do; while(isspace(Sgetc(pdf->s)));

+	Sungetc(pdf->s);

 	s = nil;

 	if((x = realloc(pdf->xref, (pdf->nxref + nxref)*sizeof(Xref))) == nil)

@@ -41,12 +43,13 @@

 	if((s = malloc(sz)) == nil)

 		goto err;

 	for(i = 0; i < sz; i += n){

-		if((n = Bread(pdf->bio, s+i, sz-i)) < 1)

+		if((n = Sread(pdf->s, s+i, sz-i)) < 1)

 			goto err;

 	}

 	/* store non-free objects only */

 	newnxref = pdf->nxref;

+	xref.objstm = 0;

 	for(e = s, i = 0; i < nxref; i++, e += 20){

 		if(!isspace(e[10]) || !isspace(e[18]) || !isspace(e[19])){

 			werrstr("invalid xref line (%d/%d)", i, nxref);

@@ -54,8 +57,7 @@

 		}

 		xref.id = xref0 + i;

 		xref.off = strtoul(e, nil, 10);

-		/* FIXME xref.gen */

-		xref.type = Xusual;

+		xref.gen = strtoul(e+11, nil, 10);

 		/* search in already existing xrefs, update if found */

 		for(j = 0; j < pdf->nxref; j++){

@@ -107,19 +109,24 @@

 int

 xrefreadstream(Pdf *pdf)

 {

-	Object *o;

+	Object *o, *p, *index;

 	Stream *s;

 	Xref *x;

 	uchar buf[32];

-	int w[8], nw, i, c, n, nxref, newnxref, prev, extra;

+	int w[8], nw, c, n, nxref, newnxref, prev, extra;

+	int i, ni, nsubsec, subsec;

 	s = nil;

-	if((o = pdfobj(pdf, pdf->bio)) == nil){

+	if((o = pdfobj(pdf, pdf->s)) == nil){

 		werrstr("xref stream obj: %r");

 		goto err;

 	}

-	if((prev = dictint(o, "Prev")) > 0){

-		if(Bseek(pdf->bio, prev, 0) != prev){

+

+	index = dictget(o, "Index"); /* 7.5.8.2 subsection indexing */

+	nsubsec = arraylen(index) / 2;

+

+	if((prev = dictint(o, "Prev")) > 0){ /* 7.5.8.2 previous xref stream */

+		if(Sseek(pdf->s, prev, 0) != prev){

 			werrstr("xref stream prev seek failed");

 			goto err;

 		}

@@ -128,7 +135,7 @@

 			return -1;

 		}

 	}

-	if((s = streamopen(o)) == nil){

+	if((s = Sopen(o)) == nil){

 		werrstr("failed to stream xref: %r");

 		goto err;

 	}

@@ -143,11 +150,11 @@

 		werrstr("W is beyond imaginable: %d bytes", n);

 		goto err;

 	}

-	if((nxref = streamsize(s)/n) < 1){

+	if((nxref = Ssize(s)/n) < 1){

 		werrstr("no xref elements in the stream");

 		goto err;

 	}

-	extra = streamsize(s) % (nxref*n);

+	extra = Ssize(s) % (nxref*n);

 	if(extra != 0)

 		fprint(2, "extra %d bytes in xref stream", extra);

@@ -156,34 +163,41 @@

 		goto err;

 	pdf->xref = x;

 	x += pdf->nxref;

-	while(Bread(s->bio, buf, n) == n){ /* stop on short read or error */

+	i = 0;

+	for(ni = subsec = 0; Sread(s, buf, n) == n; ni--, i++){ /* stop on short read or error */

+		if(ni == 0 && nsubsec > 0){

+			i = arrayint(index, subsec*2+0); /* index of the first object */

+			ni = arrayint(index, subsec*2+1); /* number of objects in the subsection */

+			subsec++;

+		}

+

 		c = getint(buf, w[0], 1); /* default type is 1 */

-		if(c == 1){ /* not compressed */

+		if(c == 1){ /* uncompressed */

+			x->objstm = 0;

+			x->id = i;

 			x->off = getint(buf+w[0], w[1], 0);

 			x->gen = getint(buf+w[0]+w[1], w[2], 0);

-			x->type = Xuncompressed;

 			pdf->nxref++;

-			fprint(2, "xref %⊗\n", *x);

 			x++;

 		}else if(c == 2){ /* compressed */

-			x->objnum = getint(buf+w[0], w[1], 0);

-			x->id = getint(buf+w[0]+w[1], w[2], 0);

-			x->type = Xcompressed;

+			x->id = i;

+			x->objstm = getint(buf+w[0], w[1], 0);

+			x->index = getint(buf+w[0]+w[1], w[2], 0);

 			pdf->nxref++;

-			fprint(2, "xref %⊗\n", *x);

 			x++;

 		}

 	}

-	streamclose(s);

-	pdf->root = pdfref(dictget(o, "Root"));

-	pdf->info = pdfref(dictget(o, "Info"));

+	Sclose(s);

+	if((p = dictget(o, "Root")) != &null)

+		pdf->root = pdfref(p);

+	if((p = dictget(o, "Info")) != &null)

+		pdf->info = pdfref(p);

 	pdfobjfree(o);

 	return 0;

 err:

-	streamclose(s);

+	Sclose(s);

 	pdfobjfree(o);

 	return -1;

 }

-