shithub: pdffs

--- a/buffer.c

+++ b/buffer.c

@@ -58,7 +58,7 @@

int

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

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

 	int n, end;

--- a/main.c

+++ b/main.c

@@ -17,8 +17,8 @@

 void

 threadmain(int argc, char **argv)

-	int fd;

 	Pdf *pdf;

+	Biobuf *b;

 	Object *v;

 	quotefmtinstall();

@@ -43,9 +43,9 @@

 	if(argc != 1)

 		usage();

-	if((fd = open(argv[0], OREAD)) < 0)

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

 		sysfatal("%r");

-	if((pdf = pdfopen(fd)) == nil)

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

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

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

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

--- a/misc.c

+++ b/misc.c

@@ -3,8 +3,36 @@

 #include <bio.h>

 #include "pdf.h"

-/* 7.2.2 whitespace */

+static char *otypes[] = {

+	[Obool] = "bool",

+	[Onum] = "num",

+	[Ostr] = "str",

+	[Oname] = "name",

+	[Oarray] = "array",

+	[Odict] = "dict",

+	[Ostream] = "stream",

+	[Onull] = "null",

+	[Oindir] = "indir",

+};

+Object null = {

+	.type = Onull,

+};

int

+Tfmt(Fmt *f)

+{

+	Object *o;

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

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

+		return fmtprint(f, "null");

+	if(o->type < 0 || o->type >= nelem(otypes))

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

+	return fmtprint(f, "%s", otypes[o->type]);

+}

+int

 isws(int c)

 	return /* \0 is missing on purpose */

@@ -12,7 +40,6 @@

 		c == ' ';

-/* 7.2.2 delimeters */

int

 isdelim(int c)

--- a/object.c

+++ b/object.c

@@ -9,38 +9,9 @@

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

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

-Object null = {

-	.type = Onull,

-};

-static char *otypes[] = {

-	[Obool] = "bool",

-	[Onum] = "num",

-	[Ostr] = "str",

-	[Oname] = "name",

-	[Oarray] = "array",

-	[Odict] = "dict",

-	[Ostream] = "stream",

-	[Onull] = "null",

-	[Oindir] = "indir",

-};

-int

-Tfmt(Fmt *f)

-{

-	Object *o;

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

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

-		return fmtprint(f, "null");

-	if(o->type < 0 || o->type >= nelem(otypes))

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

-	return fmtprint(f, "%s", otypes[o->type]);

-}

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

 Object *

-pdfobj(Pdf *pdf, void *b)

+pdfobj(Pdf *pdf, Biobuf *b)

 	Object *o, *o2;

 	vlong off;

@@ -61,6 +32,8 @@

 			if((o = pdfdict(pdf, b)) != 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])){

 					/* there IS a stream */

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

--- a/pdf.c

+++ b/pdf.c

@@ -4,6 +4,8 @@

 #include <ctype.h>

 #include "pdf.h"

+int Tfmt(Fmt *f);

/*

  * pre-1.5 xref section reader

  * PDF>=1.5 may have BOTH (or either) old xref format and xref streams

@@ -96,10 +98,9 @@

 Pdf *

-pdfopen(int fd)

+pdfopen(Biobuf *b)

 	Pdf *pdf;

-	Biobuf *b;

 	Object *o;

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

 	int xref0; /* 7.5.4 xref subsection first object number */

@@ -110,9 +111,8 @@

 	fmtinstall('T', Tfmt);

-	b = nil;

 	o = nil;

-	if((pdf = calloc(1, sizeof(*pdf))) == nil || (b = Bfdopen(fd, OREAD)) == nil)

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

 		goto err;

 	pdf->bio = b;

@@ -152,7 +152,7 @@

 		goto err;

 morexref:

-	off = Bseek(b, 0, 1);

+	off = Boffset(b);

 	n = sizeof(tmp)-1;

 	if((n = Bread(b, tmp, n)) < 16){

 badxref:

@@ -163,16 +163,14 @@

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

 		/* 7.5.4 xref */

 		x = tmp+4;

-		nxref = -1;

-		if((xref0 = strtol(x, &x, 10)) < 0 || (nxref = strtol(x, &x, 10)) < 1){

-			werrstr("xref0=%d nxref=%d", xref0, nxref);

-			goto badxref;

-		}

+		xref0 = strtol(x, &x, 10);

+		nxref = strtol(x, &x, 10);

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

 		for(; isws(*x) && x < tmp+n; x++);

 		n = x-tmp+off;

-		if(Bseek(b, n, 0) != n || xrefread(pdf, xref0, nxref) != 0)

+		if(Bseek(b, n, 0) != n)

+			goto badxref;

+		if(xref0 >= 0 && nxref > 0 && xrefread(pdf, xref0, nxref) != 0)

 			goto badxref;

 		goto morexref; /* there could be more updates, try it */

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

--- a/pdf.h

+++ b/pdf.h

@@ -19,6 +19,8 @@

 typedef struct Xref Xref;

 #pragma incomplete Filter

+#pragma varargck type "T" Object *

 struct Buffer {

 	uchar *b;

 	int ro;

@@ -71,7 +73,7 @@

};

 struct Pdf {

-	void *bio;

+	Biobuf *bio;

 	Xref *xref;

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

@@ -87,15 +89,23 @@

 struct Stream {

 	Buffer buf;

-	void *bio;

+	Biobuf *bio;

};

 extern Object null;

-Pdf *pdfopen(int fd);

+Pdf *pdfopen(Biobuf *b);

 void pdfclose(Pdf *pdf);

-Object *pdfobj(Pdf *pdf, void *b);

+/*

+ * Parse an object.

+ */

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

+/*

+ * Deallocate the object and all its children. Refcount is

+ * considered.

+ */

 void pdfobjfree(Object *o);

/*

@@ -105,9 +115,22 @@

*/

 Object *pdfeval(Object **o);

+/*

+ * Increment refcount of an object. Freeing an object (or its

+ * parent) decrements ref count.

+ */

 Object *pdfref(Object *o);

+/*

+ * Returns 1 if the character is whitespace. 0 otherwise.

+ * See 7.2.2.

+ */

 int isws(int c);

+/*

+ * Returns 1 if the character is a delimiter. 0 otherwise.

+ * See 7.2.2.

+ */

 int isdelim(int c);

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

@@ -127,10 +150,7 @@

 void buffree(Buffer *b);

 int bufeof(Buffer *b);

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

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

+int bufreadn(Buffer *b, Biobuf *bio, 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 "T" Object *

-int Tfmt(Fmt *f);

--- a/stream.c

+++ b/stream.c

@@ -8,7 +8,7 @@

 	Stream *s;

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

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

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

@@ -66,7 +66,7 @@

 		buffree(&b);

 		return nil;

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

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

 	s->buf = b;

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

 	Biofn(s->bio, bufiof);

--- a/string.c

+++ b/string.c

@@ -62,7 +62,7 @@

 		return nil;

 	for(paren = sz = 0;;){

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

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

 			break;

 		switch(c){

@@ -81,7 +81,7 @@

 			paren--;

 			if(paren < 1){

 				c = 0;

-				break;

+				goto end;

 			continue;

@@ -99,7 +99,7 @@

 				oct[i] = 0;

 				c = strtol(oct, nil, 8);

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

-				werrstr("unknown escape char %c", c);

+				werrstr("unknown escape char '%c'", c);

 				goto err;

 			}else if(c < 0){

 				continue;

@@ -108,13 +108,13 @@

 		default:

 			if(paren < 1){

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

+				werrstr("unexpected char %02x '%c'", (u8int)c, c);

 				goto err;

 			break;

-		if(c <= 0)

+		if(c < 0)

 			break;

 		if(sz+1 > maxsz){

 			maxsz *= 2;

@@ -124,7 +124,7 @@

 		s[sz++] = c;

+end:

 	if(paren != 0){

 		werrstr("bad paren");

 		goto err;