garbage collect unreferenced objects

4 files changed, 111 insertions, 1 deletions

diff --git a/Jamfile b/Jamfile
index 7ce394ce..083b91a5 100644
--- a/Jamfile
+++ b/Jamfile
@@ -93,6 +93,7 @@ Library libmupdf :
 	mupdf/repair.c
 	mupdf/save.c
 	mupdf/xref.c
+	mupdf/doctor.c
 
 	mupdf/cmap.c
 	mupdf/font.c
diff --git a/include/mupdf.h b/include/mupdf.h
index 98f40c62..3f8b242c 100644
--- a/include/mupdf.h
+++ b/include/mupdf.h
@@ -122,6 +122,9 @@ fz_error *pdf_resolve(fz_obj **, pdf_xref*);
 fz_error *pdf_savestream(pdf_xref *xref, int oid, int gid, fz_buffer *buf);
 fz_error *pdf_deletestream(pdf_xref *xref, int oid, int gid);
 
+/* doctor.c */
+fz_error *pdf_garbagecollect(pdf_xref *xref);
+
 /* save.c */
 fz_error *pdf_saveincrementalpdf(pdf_xref *xref, char *path);
 fz_error *pdf_savepdf(pdf_xref *xref, char *path);
diff --git a/mupdf/doctor.c b/mupdf/doctor.c
new file mode 100644
index 00000000..fcc9032e
--- /dev/null
+++ b/mupdf/doctor.c
@@ -0,0 +1,104 @@
+#include <fitz.h>
+#include <mupdf.h>
+
+/*
+ * Garbage collection
+ */
+
+static fz_error *sweepref(pdf_xref *xref, fz_obj *ref);
+
+static fz_error *
+sweepobj(pdf_xref *xref, fz_obj *obj)
+{
+	fz_error *error;
+	int i;
+
+	if (fz_isdict(obj))
+	{
+		for (i = 0; i < fz_dictlen(obj); i++)
+		{
+			error = sweepobj(xref, fz_dictgetval(obj, i));
+			if (error)
+				return error;
+		}
+	}
+
+	if (fz_isarray(obj))
+	{
+		for (i = 0; i < fz_arraylen(obj); i++)
+		{
+			error = sweepobj(xref, fz_arrayget(obj, i));
+			if (error)
+				return error;
+		}
+	}
+
+	if (fz_isindirect(obj))
+		return sweepref(xref, obj);
+
+	return nil;
+}
+
+static fz_error *
+sweepref(pdf_xref *xref, fz_obj *ref)
+{
+	fz_error *error;
+	fz_obj *obj;
+	int oid;
+
+	oid = fz_toobjid(ref);
+
+	if (oid < 0 || oid >= xref->size)
+		return fz_throw("rangecheck: object number out of range");
+
+	if (xref->table[oid].mark)
+		return nil;
+
+	xref->table[oid].mark = 1;
+
+	error = pdf_loadobject(&obj, xref, ref, nil);
+	if (error)
+		return error;
+
+	error = sweepobj(xref, obj);
+	if (error)
+	{
+		fz_dropobj(obj);
+		return error;
+	}
+
+	fz_dropobj(obj);
+	return nil;
+}
+
+fz_error *
+pdf_garbagecollect(pdf_xref *xref)
+{
+	fz_error *error;
+	int i, g;
+
+	for (i = 0; i < xref->size; i++)
+		xref->table[i].mark = 0;
+
+	error = sweepobj(xref, xref->trailer);
+	if (error)
+		return error;
+
+	for (i = 0; i < xref->size; i++)
+	{
+		pdf_xrefentry *x = xref->table + i;
+		g = x->gen;
+		if (x->type == 'o')
+			g = 0;
+
+		if (!x->mark && x->type != 'f' && x->type != 'd')
+		{
+			error = pdf_deleteobject(xref, i, g);
+			if (error)
+				return error;
+		}
+	}
+
+	return nil;
+}
+
diff --git a/test/pdfclean.c b/test/pdfclean.c
index 7eb72cda..37d9de0d 100644
--- a/test/pdfclean.c
+++ b/test/pdfclean.c
@@ -122,10 +122,12 @@ int main(int argc, char **argv)
 		if (error) fz_abort(error);
 	}
 
+	if (dogc)
+		pdf_garbagecollect(xref);
+
 	if (doexpand)
 		expandstreams(xref);
 
-printf("saving %s...\n", outfile);
 	error = pdf_savepdf(xref, outfile);
 	if (error)
 		fz_abort(error);