Ensure altered objects are moved to the incremental xref section

5 files changed, 136 insertions, 22 deletions

diff --git a/source/pdf/pdf-annot.c b/source/pdf/pdf-annot.c
index 5e50f0a2..9c93ac93 100644
--- a/source/pdf/pdf-annot.c
+++ b/source/pdf/pdf-annot.c
@@ -705,7 +705,6 @@ pdf_create_annot(pdf_document *doc, pdf_page *page, fz_annot_type type)
 
 		annot = fz_malloc_struct(ctx, pdf_annot);
 		annot->page = page;
-		annot->obj = pdf_keep_obj(annot_obj);
 		annot->rect = rect;
 		annot->pagerect = rect;
 		annot->ap = NULL;
@@ -721,6 +720,7 @@ pdf_create_annot(pdf_document *doc, pdf_page *page, fz_annot_type type)
 		pdf_update_object(doc, ind_obj_num, annot_obj);
 		ind_obj = pdf_new_indirect(doc, ind_obj_num, 0);
 		pdf_array_push(annot_arr, ind_obj);
+		annot->obj = pdf_keep_obj(ind_obj);
 
 		/*
 			Linking must be done after any call that might throw because
@@ -794,11 +794,13 @@ pdf_delete_annot(pdf_document *doc, pdf_page *page, pdf_annot *annot)
 					pdf_array_push(annot_arr, obj);
 			}
 
-			/*
-			Overwrite "Annots" in the page dictionary, which has the
-			side-effect of releasing the last reference to old_annot_arr
-			*/
-			pdf_dict_puts(page->me, "Annots", annot_arr);
+			if (pdf_is_indirect(old_annot_arr))
+				pdf_update_object(doc, pdf_to_num(old_annot_arr), annot_arr);
+			else
+				pdf_dict_puts(page->me, "Annots", annot_arr);
+
+			if (pdf_is_indirect(annot->obj))
+				pdf_delete_object(doc, pdf_to_num(annot->obj));
 		}
 		fz_always(ctx)
 		{
@@ -985,7 +987,7 @@ pdf_set_annot_appearance(pdf_document *doc, pdf_annot *annot, fz_rect *rect, fz_
 
 		/* See if there is a current normal appearance */
 		ap_obj = pdf_dict_getp(obj, "AP/N");
-		if (!pdf_is_stream(doc, pdf_to_num(obj), pdf_to_gen(obj)))
+		if (!pdf_is_stream(doc, pdf_to_num(ap_obj), pdf_to_gen(ap_obj)))
 			ap_obj = NULL;
 
 		if (ap_obj == NULL)
@@ -995,6 +997,7 @@ pdf_set_annot_appearance(pdf_document *doc, pdf_annot *annot, fz_rect *rect, fz_
 		}
 		else
 		{
+			pdf_xref_ensure_incremental_object(doc, pdf_to_num(ap_obj));
 			pdf_dict_puts_drop(ap_obj, "Rect", pdf_new_rect(doc, &trect));
 			pdf_dict_puts_drop(ap_obj, "Matrix", pdf_new_matrix(doc, &mat));
 		}
diff --git a/source/pdf/pdf-object.c b/source/pdf/pdf-object.c
index 037f1e16..88af2639 100644
--- a/source/pdf/pdf-object.c
+++ b/source/pdf/pdf-object.c
@@ -34,6 +34,7 @@ struct pdf_obj_s
 	unsigned char kind;
 	unsigned char flags;
 	pdf_document *doc;
+	int parent_num;
 	union
 	{
 		int b;
@@ -71,6 +72,7 @@ pdf_new_null(pdf_document *doc)
 	obj->refs = 1;
 	obj->kind = PDF_NULL;
 	obj->flags = 0;
+	obj->parent_num = 0;
 	return obj;
 }
 
@@ -84,6 +86,7 @@ pdf_new_bool(pdf_document *doc, int b)
 	obj->refs = 1;
 	obj->kind = PDF_BOOL;
 	obj->flags = 0;
+	obj->parent_num = 0;
 	obj->u.b = b;
 	return obj;
 }
@@ -98,6 +101,7 @@ pdf_new_int(pdf_document *doc, int i)
 	obj->refs = 1;
 	obj->kind = PDF_INT;
 	obj->flags = 0;
+	obj->parent_num = 0;
 	obj->u.i = i;
 	return obj;
 }
@@ -112,6 +116,7 @@ pdf_new_real(pdf_document *doc, float f)
 	obj->refs = 1;
 	obj->kind = PDF_REAL;
 	obj->flags = 0;
+	obj->parent_num = 0;
 	obj->u.f = f;
 	return obj;
 }
@@ -126,6 +131,7 @@ pdf_new_string(pdf_document *doc, const char *str, int len)
 	obj->refs = 1;
 	obj->kind = PDF_STRING;
 	obj->flags = 0;
+	obj->parent_num = 0;
 	obj->u.s.len = len;
 	memcpy(obj->u.s.buf, str, len);
 	obj->u.s.buf[len] = '\0';
@@ -142,6 +148,7 @@ pdf_new_name(pdf_document *doc, const char *str)
 	obj->refs = 1;
 	obj->kind = PDF_NAME;
 	obj->flags = 0;
+	obj->parent_num = 0;
 	strcpy(obj->u.n, str);
 	return obj;
 }
@@ -156,6 +163,7 @@ pdf_new_indirect(pdf_document *doc, int num, int gen)
 	obj->refs = 1;
 	obj->kind = PDF_INDIRECT;
 	obj->flags = 0;
+	obj->parent_num = 0;
 	obj->u.r.num = num;
 	obj->u.r.gen = gen;
 	return obj;
@@ -442,6 +450,7 @@ pdf_new_array(pdf_document *doc, int initialcap)
 	obj->refs = 1;
 	obj->kind = PDF_ARRAY;
 	obj->flags = 0;
+	obj->parent_num = 0;
 
 	obj->u.a.len = 0;
 	obj->u.a.cap = initialcap > 1 ? initialcap : 6;
@@ -519,6 +528,24 @@ pdf_array_get(pdf_obj *obj, int i)
 	return obj->u.a.items[i];
 }
 
+static void object_altered(pdf_obj *obj, pdf_obj *val)
+{
+	/*
+	parent_num = 0 while an object is being parsed from the file.
+	No further action is necessary.
+	*/
+	if (obj->parent_num == 0 || obj->doc->freeze_updates)
+		return;
+
+	/*
+	Otherwise we need to ensure that the containing hierarchy of objects
+	has been moved to the incremental xref section and the newly linked
+	object needs to record the parent_num
+	*/
+	pdf_xref_ensure_incremental_object(obj->doc, obj->parent_num);
+	pdf_set_objects_parent_num(val, obj->parent_num);
+}
+
 void
 pdf_array_put(pdf_obj *obj, int i, pdf_obj *item)
 {
@@ -537,6 +564,8 @@ pdf_array_put(pdf_obj *obj, int i, pdf_obj *item)
 		pdf_drop_obj(obj->u.a.items[i]);
 		obj->u.a.items[i] = pdf_keep_obj(item);
 	}
+
+	object_altered(obj, item);
 }
 
 void
@@ -555,6 +584,8 @@ pdf_array_push(pdf_obj *obj, pdf_obj *item)
 		obj->u.a.items[obj->u.a.len] = pdf_keep_obj(item);
 		obj->u.a.len++;
 	}
+
+	object_altered(obj, item);
 }
 
 void
@@ -593,6 +624,8 @@ pdf_array_insert(pdf_obj *obj, pdf_obj *item)
 		obj->u.a.items[0] = pdf_keep_obj(item);
 		obj->u.a.len++;
 	}
+
+	object_altered(obj, item);
 }
 
 int
@@ -723,6 +756,7 @@ pdf_new_dict(pdf_document *doc, int initialcap)
 	obj->refs = 1;
 	obj->kind = PDF_DICT;
 	obj->flags = 0;
+	obj->parent_num = 0;
 
 	obj->u.d.len = 0;
 	obj->u.d.cap = initialcap > 1 ? initialcap : 10;
@@ -987,6 +1021,8 @@ pdf_dict_put(pdf_obj *obj, pdf_obj *key, pdf_obj *val)
 		obj->u.d.items[i].v = pdf_keep_obj(val);
 		obj->u.d.len ++;
 	}
+
+	object_altered(obj, val);
 }
 
 void
@@ -1136,6 +1172,8 @@ pdf_dict_dels(pdf_obj *obj, const char *key)
 			obj->u.d.len --;
 		}
 	}
+
+	object_altered(obj, NULL);
 }
 
 void
@@ -1274,6 +1312,31 @@ pdf_drop_obj(pdf_obj *obj)
 		fz_free(obj->doc->ctx, obj);
 }
 
+void
+pdf_set_objects_parent_num(pdf_obj *obj, int num)
+{
+	int n, i;
+
+	if (!obj)
+		return;
+
+	obj->parent_num = num;
+
+	switch(obj->kind)
+	{
+	case PDF_ARRAY:
+		n = pdf_array_len(obj);
+		for (i = 0; i < n; i++)
+			pdf_set_objects_parent_num(pdf_array_get(obj, i), num);
+		break;
+	case PDF_DICT:
+		n = pdf_dict_len(obj);
+		for (i = 0; i < n; i++)
+			pdf_set_objects_parent_num(pdf_dict_get_val(obj, i), num);
+		break;
+	}
+}
+
 pdf_obj *pdf_new_obj_from_str(pdf_document *doc, const char *src)
 {
 	pdf_obj *result;
diff --git a/source/pdf/pdf-repair.c b/source/pdf/pdf-repair.c
index bf4bcf89..3db32de0 100644
--- a/source/pdf/pdf-repair.c
+++ b/source/pdf/pdf-repair.c
@@ -256,6 +256,8 @@ pdf_repair_xref(pdf_document *doc, pdf_lexbuf *buf)
 	fz_var(obj);
 
 	doc->dirty = 1;
+	/* Can't support incremental update after repair */
+	doc->freeze_updates = 1;
 
 	fz_seek(doc->file, 0, 0);
 
diff --git a/source/pdf/pdf-write.c b/source/pdf/pdf-write.c
index 79271499..9d768d0f 100644
--- a/source/pdf/pdf-write.c
+++ b/source/pdf/pdf-write.c
@@ -2225,6 +2225,7 @@ void pdf_write_document(pdf_document *doc, char *filename, fz_write_options *fz_
 	if (!doc)
 		return;
 
+	doc->freeze_updates = 1;
 	ctx = doc->ctx;
 
 	pdf_finish_edit(doc);
@@ -2356,6 +2357,7 @@ void pdf_write_document(pdf_document *doc, char *filename, fz_write_options *fz_
 		pdf_drop_obj(opts.hints_length);
 		page_objects_list_destroy(ctx, opts.page_object_lists);
 		fclose(opts.out);
+		doc->freeze_updates = 0;
 	}
 	fz_catch(ctx)
 	{
diff --git a/source/pdf/pdf-xref.c b/source/pdf/pdf-xref.c
index 91083468..a67f6bc7 100644
--- a/source/pdf/pdf-xref.c
+++ b/source/pdf/pdf-xref.c
@@ -147,26 +147,39 @@ pdf_xref_entry *pdf_get_xref_entry(pdf_document *doc, int i)
 	return &doc->xref_sections[0].table[i];
 }
 
-/* Used when altering a document */
-static pdf_xref_entry *pdf_get_new_xref_entry(pdf_document *doc, int i)
+/*
+		Ensure we have an incremental xref section where we can store
+		updated versions of indirect objects
+*/
+static void ensure_incremental_xref(pdf_document *doc)
 {
 	fz_context *ctx = doc->ctx;
-	pdf_xref *xref;
+	pdf_xref *xref, *pxref;
 
-	/* Make a new final xref section if we haven't already */
 	if (!doc->xref_altered)
 	{
+		pdf_xref_entry *new_table;
 		doc->xref_sections = fz_resize_array(ctx, doc->xref_sections, doc->num_xref_sections + 1, sizeof(pdf_xref));
-		memmove(&doc->xref_sections[1], &doc->xref_sections[0], doc->num_xref_sections * sizeof(pdf_xref));
-		doc->num_xref_sections++;
 		xref = &doc->xref_sections[0];
-		xref->len = 0;
-		xref->table = NULL;
-		xref->trailer = pdf_keep_obj(doc->xref_sections[1].trailer);
-		/* All new levels must be at least as big as the level before */
-		pdf_resize_xref(ctx, xref, fz_maxi(i+1, doc->xref_sections[1].len));
+		pxref = &doc->xref_sections[1];
+		new_table = fz_calloc(ctx, xref->len, sizeof(pdf_xref_entry));
+		memmove(pxref, xref, doc->num_xref_sections * sizeof(pdf_xref));
+		doc->num_xref_sections++;
+		/* xref->len is already correct */
+		xref->table = new_table;
+		xref->trailer = pdf_keep_obj(pxref->trailer);
 		doc->xref_altered = 1;
 	}
+}
+
+/* Used when altering a document */
+static pdf_xref_entry *pdf_get_incremental_xref_entry(pdf_document *doc, int i)
+{
+	fz_context *ctx = doc->ctx;
+	pdf_xref *xref;
+
+	/* Make a new final xref section if we haven't already */
+	ensure_incremental_xref(doc);
 
 	xref = &doc->xref_sections[0];
 	if (i >= xref->len)
@@ -175,6 +188,36 @@ static pdf_xref_entry *pdf_get_new_xref_entry(pdf_document *doc, int i)
 	return &xref->table[i];
 }
 
+/* Ensure that an object has been cloned into the incremental xref section */
+void pdf_xref_ensure_incremental_object(pdf_document *doc, int num)
+{
+	fz_context *ctx = doc->ctx;
+	pdf_xref_entry *new_entry, *old_entry;
+	int i;
+
+	/* Make sure we have created an xref section for incremental updates */
+	ensure_incremental_xref(doc);
+
+	/* Search for the section that contains this object */
+	for (i = 0; i < doc->num_xref_sections; i++)
+	{
+		pdf_xref *xref = &doc->xref_sections[i];
+		if (num >= 0 && num < xref->len && xref->table[num].type)
+			break;
+	}
+
+	/* If we don't find it, or it's already in the incremental section, return */
+	if (i == 0 || i == doc->num_xref_sections)
+		return;
+
+	/* Move the object to the incremental section */
+	old_entry = &doc->xref_sections[i].table[num];
+	new_entry = pdf_get_incremental_xref_entry(doc, num);
+	*new_entry = *old_entry;
+	old_entry->obj = NULL;
+	old_entry->stm_buf = NULL;
+}
+
 void pdf_replace_xref(pdf_document *doc, pdf_xref_entry *entries, int n)
 {
 	fz_context *ctx = doc->ctx;
@@ -1213,7 +1256,6 @@ pdf_cache_object(pdf_document *doc, int num, int gen)
 	if (x->type == 'f')
 	{
 		x->obj = pdf_new_null(doc);
-		return;
 	}
 	else if (x->type == 'n')
 	{
@@ -1259,6 +1301,8 @@ pdf_cache_object(pdf_document *doc, int num, int gen)
 	{
 		fz_throw(ctx, FZ_ERROR_GENERIC, "cannot find object in xref (%d %d R)", num, gen);
 	}
+
+	pdf_set_objects_parent_num(x->obj, num);
 }
 
 pdf_obj *
@@ -1337,7 +1381,7 @@ pdf_create_object(pdf_document *doc)
 	/* TODO: reuse free object slots by properly linking free object chains in the ofs field */
 	pdf_xref_entry *entry;
 	int num = pdf_xref_len(doc);
-	entry = pdf_get_new_xref_entry(doc, num);
+	entry = pdf_get_incremental_xref_entry(doc, num);
 	entry->type = 'f';
 	entry->ofs = -1;
 	entry->gen = 0;
@@ -1358,7 +1402,7 @@ pdf_delete_object(pdf_document *doc, int num)
 		return;
 	}
 
-	x = pdf_get_new_xref_entry(doc, num);
+	x = pdf_get_incremental_xref_entry(doc, num);
 
 	fz_drop_buffer(doc->ctx, x->stm_buf);
 	pdf_drop_obj(x->obj);
@@ -1382,7 +1426,7 @@ pdf_update_object(pdf_document *doc, int num, pdf_obj *newobj)
 		return;
 	}
 
-	x = pdf_get_new_xref_entry(doc, num);
+	x = pdf_get_incremental_xref_entry(doc, num);
 
 	pdf_drop_obj(x->obj);