shithub: pdffs

Info  •  Files  •  Log

better api (less Pdf *pdf); eval more often; use null

--- a/array.c

+++ b/array.c

@@ -13,6 +13,7 @@

 	int c, noel;

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

+	o->pdf = pdf;

 	o->type = Oarray;

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

--- a/dict.c

+++ b/dict.c

@@ -18,6 +18,7 @@

 	k = v = nil;

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

 	o->type = Odict;

+	o->pdf = pdf;

 	for(nokey = 0;;){

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

 			goto err;

@@ -37,12 +38,12 @@

 			nokey = 1;

 			continue;

 		}

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

-			goto err;

 		if(k->type != Oname){

 			werrstr("expected name as a key");

 			goto err;

 		}

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

+			goto err;

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

 			goto err;

@@ -65,13 +66,58 @@

 }

 Object *

-pdfdictget(Object *o, char *name)

+dictget(Object *o, char *name)

 {

 	int i;

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

-		return nil;

+	o = 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++);

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

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

+}

+

+vlong

+dictint(Object *o, char *name)

+{

+	o = dictget(o, name);

+	return o->type == Onum ? o->num : 0;

+}

+

+char *

+dictstring(Object *o, char *name)

+{

+	o = dictget(o, name);

+	return o->type == Ostr ? o->str : "";

+}

+

+Object *

+dictdict(Object *o, char *name)

+{

+	o = dictget(o, name);

+	return o->type == Odict ? o : &null;

+}

+

+int

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

+{

+	int n;

+	Object *v;

+

+	o = dictget(o, name);

+	if(o->type != Oarray){

+		werrstr("not an array");

+		return -1;

+	}

+

+	for(n = 0; n < o->array.ne && n < nel; n++){

+		if((v = o->array.e[n])->type != Onum){

+			werrstr("not an integer array");

+			return -1;

+		}

+		el[n] = v->num;

+	}

+

+	return n;

 }

--- a/eval.c

+++ b/eval.c

@@ -3,32 +3,33 @@

 #include <bio.h>

 #include "pdf.h"

-int

-pdfeval(Pdf *pdf, Object *o)

+Object *

+pdfeval(Object *o)

 {

 	Object *d;

 	Xref *x;

 	int i;

-	if(o == nil || o->type != Oindir)

-		return 0;

+	if(o == nil)

+		return &null;

+	if(o->type != Oindir)

+		return o;

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

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

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

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

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

-		return -1;

+		return &null;

 	}

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

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

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

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

 		werrstr("xref seek failed");

-		return -1;

+		return &null;

 	}

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

-		return -1;

-	*o = *d;

-	memset(d, 0, sizeof(*d));

-	pdfobjfree(d);

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

+		return &null;

-	return 0;

+	pdfobjfree(o);

+

+	return d;

 }

--- a/main.c

+++ b/main.c

@@ -47,9 +47,9 @@

 		sysfatal("%r");

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

 		sysfatal("%r");

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

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

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

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

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

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

 	pdfclose(pdf);

--- a/object.c

+++ b/object.c

@@ -9,7 +9,7 @@

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

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

-static Object null = {

+Object null = {

 	.type = Onull,

 };

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

 Object *

 pdfobj(Pdf *pdf, void *b)

 {

-	Object *o, *o2, *m;

+	Object *o, *o2;

 	vlong off;

 	int c, tf;

 	Xref xref;

@@ -63,10 +63,6 @@

 				off = Boffset(b);

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

 					/* there IS a stream */

-					if((m = pdfdictget(o, "Length")) == nil || pdfeval(pdf, m) != 0 || m->type != Onum){

-						werrstr("stream has no valid /Length");

-						goto err;

-					}

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

 						goto err;

 					if(c != '\n'){

@@ -73,9 +69,9 @@

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

 						goto err;

 					}

-					o->type = Ostream;

-					o->stream.len = m->num;

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

+					o->type = Ostream;

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

 					return o;

 				}

 				Bseek(b, off, 0);

@@ -87,15 +83,21 @@

 	case '(':

 		Bungetc(b);

-		return pdfstring(b);

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

+			o->pdf = pdf;

+		return o;

 	case '/':

 		Bungetc(b);

-		return pdfname(b);

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

+			o->pdf = pdf;

+		return o;

 	case '[':

 		Bungetc(b);

-		return pdfarray(pdf, b);

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

+			o->pdf = pdf;

+		return o;

 	case 'n':

 		off = Boffset(b);

@@ -126,15 +128,14 @@

 		goto unexpected;

 bool:

 		Bungetc(b);

-		if((o = malloc(sizeof(*o))) == nil)

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

 			goto err;

 		o->type = Obool;

+		o->pdf = pdf;

 		o->bool = tf;

 		return o;

 	default:

-		if((o = malloc(sizeof(*o))) == nil)

-			goto err;

 		if(!isdigit(c)){

 unexpected:

 			Bungetc(b);

@@ -143,6 +144,9 @@

 		}

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

 		Bungetc(b);

+		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 */

--- a/pdf.c

+++ b/pdf.c

@@ -85,8 +85,8 @@

 		goto err;

 	}

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

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

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

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

 	pdfobjfree(o);

 	o = nil;

@@ -112,7 +112,7 @@

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

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

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

-	int i, n, off;

+	int i, n, off, w[3];

 	Stream *stream;

 	fmtinstall('T', Tfmt);

@@ -191,14 +191,18 @@

 		}

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

 		Bseek(b, xreftb, 0);

-		if((o = pdfobj(pdf, b)) == nil || (stream = streamopen(pdf, o)) == nil)

+		if((o = pdfobj(pdf, b)) == nil || (stream = streamopen(o)) == nil){

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

 			goto badxref;

+		}

+		if(dictints(o, "W", w, nelem(w)) != 3){

+			werrstr("W isn't 3 elements");

+			goto badxref;

+		}

 		streamclose(stream);

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

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

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

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

 	}

-	if(pdfeval(pdf, pdf->root) != 0 || pdfeval(pdf, pdf->info) != 0)

-		goto err;

 	fprint(2, "root %T\n", pdf->root);

 	fprint(2, "info %T\n", pdf->info);

--- a/pdf.h

+++ b/pdf.h

@@ -30,6 +30,7 @@

 struct Object {

 	int type;

+	Pdf *pdf;

 	union {

 		int bool;

 		double num;

@@ -88,6 +89,8 @@

 	void *bio;

 };

+extern Object null;

+

 Pdf *pdfopen(int fd);

 void pdfclose(Pdf *pdf);

@@ -95,17 +98,22 @@

 void pdfobjfree(Object *o);

 /*

- * If the object is indirect, resolve it. Operation is not recursive, ie

- * values of a dictionary won't be resolved automatically.

+ * Return a resolved object or &null if can't. Operation is

+ * not recursive, ie values of a dictionary won't be resolved

+ * automatically.

  */

-int pdfeval(Pdf *pdf, Object *o);

+Object *pdfeval(Object *o);

 int isws(int c);

 int isdelim(int c);

-Object *pdfdictget(Object *o, char *name);

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

+vlong dictint(Object *o, char *name);

+char *dictstring(Object *o, char *name);

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

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

-Stream *streamopen(Pdf *pdf, Object *o);

+Stream *streamopen(Object *o);

 void streamclose(Stream *s);

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

--- a/stream.c

+++ b/stream.c

@@ -14,7 +14,7 @@

 }

 Stream *

-streamopen(Pdf *pdf, Object *o)

+streamopen(Object *o)

 {

 	Stream *s;

 	Buffer b, x;

@@ -23,21 +23,19 @@

 	int i, nflts;

 	s = nil;

-	if(pdfeval(pdf, o) != 0 || o == nil || 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? */

 		return nil;

 	bufinit(&b, nil, 0);

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

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

 		return nil;

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

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

 		goto err;

 	bufdump(&b);

 	/* see if there are any filters */

-	if((of = pdfdictget(o, "Filter")) != nil){

-		if(pdfeval(pdf, of) != 0)

-			goto err;

-		if(of->type == Oname){ /* one filter */

+	if((of = dictget(o, "Filter")) != nil){

+		if(pdfeval(of)->type == Oname){ /* one filter */

 			flts = &of;

 			nflts = 1;

 		}else if(of->type == Oarray){ /* array of filters */

@@ -71,7 +69,7 @@

 	}

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

 	s->buf = b;

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

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

 	Biofn(s->bio, bufiof);

 	bufdump(&s->buf);