commit 72c389d3e539e6d1b8586dab2e48de3c05cbb457
from: Sven M. Hallberg <pesco@khjk.org>
date: Fri Aug 12 17:48:03 2022 UTC

Merge branch 'selfref' into ostream

commit - f962b66b483de1dea4a80f6ed68c9f7d77123065
commit + 72c389d3e539e6d1b8586dab2e48de3c05cbb457
blob - 0dd4fce48f341be7f85ad3987366a1ff1e491739
blob + 55f85b3810ad6fabfb96f75b9922fd15217f2e35
--- pdf.c
+++ pdf.c
@@ -2813,6 +2813,8 @@ parse_objstm_obj(struct Env *aux, size_t nr, size_t st
 }
 
 /* resolve a parsed object to its final semantic value */
+	static void *INVALID = &INVALID;	/* a reserved dummy pointer */
+		// XXX move this back into the function
 const HParsedToken *
 resolve(struct Env *aux, const HParsedToken *v)
 {
@@ -2823,8 +2825,10 @@ resolve(struct Env *aux, const HParsedToken *v)
 	 * The null object maps to NULL. This makes it equivalent to a
 	 * nonexistent object and covers the handling of (indirect) null values
 	 * in dictionaries (treating them as if the entry did not exist).
+	 *
+	 * Likewise, we map invalid (e.g. circular) references to NULL.
 	 */
-	if (v == NULL || v->token_type == TT_Null)
+	if (v == NULL || v == INVALID || v->token_type == TT_Null)
 		return NULL;
 
 	/* other direct objects pass through */
@@ -2841,8 +2845,7 @@ resolve(struct Env *aux, const HParsedToken *v)
 	if (ent->obj != NULL)
 		return resolve(aux, ent->obj);
 
-	/* parse the object and memoize */
-	ent->obj = v;				/* break loops */
+	/* parse the object */
 	switch (ent->type)
 	{
 	case XR_FREE:
@@ -2850,12 +2853,12 @@ resolve(struct Env *aux, const HParsedToken *v)
 	case XR_INUSE:
 		if (ent->n.gen != r->gen)
 			return NULL;		/* obj nr reused */
-		ent->obj = parse_obj(aux, r->nr, r->gen, ent->n.offs);
+		v = parse_obj(aux, r->nr, r->gen, ent->n.offs);
 		break;
 	case XR_OBJSTM:
 		if (r->gen != 0)
 			return NULL;		/* invalid entry! */
-		ent->obj = parse_objstm_obj(aux, r->nr, ent->o.stm, ent->o.idx);
+		v = parse_objstm_obj(aux, r->nr, ent->o.stm, ent->o.idx);
 		break;
 	default:
 		/*
@@ -2865,7 +2868,13 @@ resolve(struct Env *aux, const HParsedToken *v)
 		return NULL;
 	}
 
-	return resolve(aux, ent->obj);
+	/* resolve recursively and memoize */
+	ent->obj = INVALID;			/* break loops */
+	if ((v = resolve(aux, v)) != NULL)
+		ent->obj = v;
+
+	assert(v != INVALID);			/* our dummy does not escape */
+	return v;
 }
 
 
@@ -4384,6 +4393,7 @@ p_stream_data__m(HAllocator *mm__, const Dict *dict, s
 	const HParsedToken *v;
 
 	v = dictentry(dict, "Type");
+	v = resolve(aux, v);
 	if (v == NULL || v->token_type != TT_BYTES)	// XXX -> custom type
 		return NULL;				/* no /Type field */
 
blob - /dev/null
blob + be67696abfef0b34980551ecc6ad61dce956dc6d (mode 644)
--- /dev/null
+++ test/valid/selfref.pdf
@@ -0,0 +1,79 @@
+%PDF-1.4
+1 0 obj
+    << /Type /Catalog
+       /Outlines 2 0 R
+       /Pages 3 0 R
+    >>
+endobj
+
+2 0 obj
+    << /Type /Outlines
+       /Count 0
+    >>
+endobj
+
+3 0 obj
+    << /Type /Pages
+       /Kids [4 0 R]
+       /Count 1
+    >>
+endobj
+
+4 0 obj
+    << /Type /Page
+       /Parent 3 0 R
+       /MediaBox [0 0 612 792]
+       /Contents 5 0 R
+       /Resources << /ProcSet 6 0 R
+                     /Font << /F1 7 0 R >>
+                  >>
+    >>
+endobj
+
+5 0 obj
+    << /Length 76 /Type 8 0 R >>
+stream
+    BT
+        /F1 24 Tf
+        100 100 Td
+        (Hello World) Tj
+    ET
+endstream
+endobj
+
+6 0 obj
+    [/PDF /Text]
+endobj
+
+7 0 obj
+    << /Type /Font
+       /Subtype /Type1
+       /Name /F1
+       /BaseFont /Helvetica
+       /Encoding /MacRomanEncoding
+    >>
+endobj
+
+8 0 obj
+    8 0 R
+endobj
+
+xref
+0 9
+0000000000 65535 f 
+0000000009 00000 n 
+0000000097 00000 n 
+0000000159 00000 n 
+0000000239 00000 n 
+0000000456 00000 n 
+0000000598 00000 n 
+0000000631 00000 n 
+0000000776 00000 n 
+
+trailer
+    << /Size 9
+       /Root 1 0 R
+    >>
+startxref
+802
+%%EOF