Update mutool clean sanitize to clean annotations too.

4 files changed, 56 insertions, 1 deletions

diff --git a/include/mupdf/pdf/page.h b/include/mupdf/pdf/page.h
index 1b8fa755..f1caf936 100644
--- a/include/mupdf/pdf/page.h
+++ b/include/mupdf/pdf/page.h
@@ -122,6 +122,33 @@ void pdf_clean_page_contents(fz_context *ctx, pdf_document *doc, pdf_page *page,
 	pdf_page_contents_process_fn *proc, void *proc_arg, int ascii);
 
 /*
+	pdf_clean_annot_contents: Clean a loaded annotations rendering operations,
+	with an optional post processing step.
+
+	Each appearance stream in the annotation is processed.
+
+	Firstly, this filters the PDF operators used to avoid (some cases
+	of) repetition, and leaves the page in a balanced state with an
+	unchanged top level matrix etc. At the same time, the resources
+	used by the page contents are collected.
+
+	Next, the resources themselves are cleaned (as appropriate) in the
+	same way.
+
+	Next, an optional post processing stage is called.
+
+	Finally, the updated stream of operations is reinserted into the
+	appearance stream.
+
+	annot: An annotation loaded by pdf_load_annot.
+
+	cookie: A pointer to an optional fz_cookie structure that can be used
+	to track progress, collect errors etc.
+*/
+void pdf_clean_annot_contents(fz_context *ctx, pdf_document *doc, pdf_annot *annot, fz_cookie *cookie,
+	pdf_page_contents_process_fn *proc, void *proc_arg, int ascii);
+
+/*
 	Presentation interface.
 */
 fz_transition *pdf_page_presentation(fz_context *ctx, pdf_page *page, float *duration);
diff --git a/source/pdf/pdf-clean.c b/source/pdf/pdf-clean.c
index 2c8ade73..ecadc7f2 100644
--- a/source/pdf/pdf-clean.c
+++ b/source/pdf/pdf-clean.c
@@ -287,3 +287,24 @@ void pdf_clean_page_contents(fz_context *ctx, pdf_document *doc, pdf_page *page,
 		fz_rethrow_message(ctx, "Failed while cleaning page");
 	}
 }
+
+void pdf_clean_annot_contents(fz_context *ctx, pdf_document *doc, pdf_annot *annot, fz_cookie *cookie, pdf_page_contents_process_fn *proc_fn, void *proc_arg, int ascii)
+{
+	pdf_obj *ap;
+	int i, n;
+
+	ap = pdf_dict_get(ctx, annot->obj, PDF_NAME_AP);
+	if (ap == NULL)
+		return;
+
+	n = pdf_dict_len(ctx, ap);
+	for (i = 0; i < n; i++)
+	{
+		pdf_obj *v = pdf_dict_get_val(ctx, ap, i);
+
+		if (v == NULL)
+			continue;
+
+		pdf_clean_stream_object(ctx, doc, v, NULL, cookie, 1, 1);
+	}
+}
diff --git a/source/pdf/pdf-interpret.c b/source/pdf/pdf-interpret.c
index 3a386614..cff3f197 100644
--- a/source/pdf/pdf-interpret.c
+++ b/source/pdf/pdf-interpret.c
@@ -1258,7 +1258,7 @@ pdf_process_annot(fz_context *ctx, pdf_processor *proc, pdf_document *doc, pdf_p
 		proc->op_cm(ctx, proc,
 				annot->matrix.a, annot->matrix.b, annot->matrix.c,
 				annot->matrix.d, annot->matrix.e, annot->matrix.f);
-		proc->op_Do_form(ctx, proc, "Annot", annot->ap, page->resources);
+		proc->op_Do_form(ctx, proc, NULL, annot->ap, page->resources);
 		proc->op_Q(ctx, proc);
 	}
 }
diff --git a/source/pdf/pdf-write.c b/source/pdf/pdf-write.c
index 388a5820..8e686218 100644
--- a/source/pdf/pdf-write.c
+++ b/source/pdf/pdf-write.c
@@ -2656,8 +2656,15 @@ static void sanitize(fz_context *ctx, pdf_document *doc, int ascii)
 
 	for (i = 0; i < n; i++)
 	{
+		pdf_annot *annot;
 		pdf_page *page = pdf_load_page(ctx, doc, i);
 		pdf_clean_page_contents(ctx, doc, page, NULL, NULL, NULL, ascii);
+
+		for (annot = pdf_first_annot(ctx, page); annot != NULL; annot = pdf_next_annot(ctx, annot))
+		{
+			pdf_clean_annot_contents(ctx, doc, annot, NULL, NULL, NULL, ascii);
+		}
+
 		fz_drop_page(ctx, &page->super);
 	}
 }