ASCIIHexEncode inline images during sanitization if do_ascii is set.

3 files changed, 48 insertions, 31 deletions

diff --git a/source/pdf/pdf-clean.c b/source/pdf/pdf-clean.c
index 90de0444..3d34c1a1 100644
--- a/source/pdf/pdf-clean.c
+++ b/source/pdf/pdf-clean.c
@@ -1,7 +1,7 @@
 #include "mupdf/pdf.h"
 
 static void
-pdf_clean_stream_object(fz_context *ctx, pdf_document *doc, pdf_obj *obj, pdf_obj *orig_res, fz_cookie *cookie, int own_res)
+pdf_clean_stream_object(fz_context *ctx, pdf_document *doc, pdf_obj *obj, pdf_obj *orig_res, fz_cookie *cookie, int own_res, int ascii)
 {
 	pdf_processor *proc_buffer = NULL;
 	pdf_processor *proc_filter = NULL;
@@ -30,7 +30,7 @@ pdf_clean_stream_object(fz_context *ctx, pdf_document *doc, pdf_obj *obj, pdf_ob
 
 		res = pdf_new_dict(ctx, doc, 1);
 
-		proc_buffer = pdf_new_buffer_processor(ctx, buffer);
+		proc_buffer = pdf_new_buffer_processor(ctx, buffer, ascii);
 		proc_filter = pdf_new_filter_processor(ctx, proc_buffer, doc, orig_res, res);
 
 		pdf_process_contents(ctx, proc_filter, doc, orig_res, obj, cookie);
@@ -58,7 +58,7 @@ pdf_clean_stream_object(fz_context *ctx, pdf_document *doc, pdf_obj *obj, pdf_ob
 }
 
 static void
-pdf_clean_type3(fz_context *ctx, pdf_document *doc, pdf_obj *obj, pdf_obj *orig_res, fz_cookie *cookie)
+pdf_clean_type3(fz_context *ctx, pdf_document *doc, pdf_obj *obj, pdf_obj *orig_res, fz_cookie *cookie, int ascii)
 {
 	pdf_processor *proc_buffer = NULL;
 	pdf_processor *proc_filter = NULL;
@@ -90,7 +90,7 @@ pdf_clean_type3(fz_context *ctx, pdf_document *doc, pdf_obj *obj, pdf_obj *orig_
 			fz_buffer *buffer = fz_new_buffer(ctx, 1024);
 			fz_try(ctx)
 			{
-				proc_buffer = pdf_new_buffer_processor(ctx, buffer);
+				proc_buffer = pdf_new_buffer_processor(ctx, buffer, ascii);
 				proc_filter = pdf_new_filter_processor(ctx, proc_buffer, doc, orig_res, res);
 
 				pdf_process_contents(ctx, proc_filter, doc, orig_res, val, cookie);
@@ -126,7 +126,7 @@ pdf_clean_type3(fz_context *ctx, pdf_document *doc, pdf_obj *obj, pdf_obj *orig_
 	}
 }
 
-void pdf_clean_page_contents(fz_context *ctx, pdf_document *doc, pdf_page *page, fz_cookie *cookie, pdf_page_contents_process_fn *proc_fn, void *proc_arg)
+void pdf_clean_page_contents(fz_context *ctx, pdf_document *doc, pdf_page *page, fz_cookie *cookie, pdf_page_contents_process_fn *proc_fn, void *proc_arg, int ascii)
 {
 	pdf_processor *proc_buffer = NULL;
 	pdf_processor *proc_filter = NULL;
@@ -151,7 +151,7 @@ void pdf_clean_page_contents(fz_context *ctx, pdf_document *doc, pdf_page *page,
 	{
 		res = pdf_new_dict(ctx, doc, 1);
 
-		proc_buffer = pdf_new_buffer_processor(ctx, buffer);
+		proc_buffer = pdf_new_buffer_processor(ctx, buffer, ascii);
 		proc_filter = pdf_new_filter_processor(ctx, proc_buffer, doc, page->resources, res);
 
 		pdf_process_contents(ctx, proc_filter, doc, page->resources, page->contents, cookie);
@@ -193,7 +193,7 @@ void pdf_clean_page_contents(fz_context *ctx, pdf_document *doc, pdf_page *page,
 					continue;
 
 				/* Transparency group XObject */
-				pdf_clean_stream_object(ctx, doc, o, page->resources, cookie, 1);
+				pdf_clean_stream_object(ctx, doc, o, page->resources, cookie, 1, ascii);
 			}
 		}
 
@@ -213,7 +213,7 @@ void pdf_clean_page_contents(fz_context *ctx, pdf_document *doc, pdf_page *page,
 				if (!pat)
 					continue;
 				if (pdf_to_int(ctx, pdf_dict_get(ctx, pat, PDF_NAME_PatternType)) == 1)
-					pdf_clean_stream_object(ctx, doc, pat, page->resources, cookie, 0);
+					pdf_clean_stream_object(ctx, doc, pat, page->resources, cookie, 0, ascii);
 			}
 		}
 
@@ -233,7 +233,7 @@ void pdf_clean_page_contents(fz_context *ctx, pdf_document *doc, pdf_page *page,
 				if (!pdf_name_eq(ctx, PDF_NAME_Form, pdf_dict_get(ctx, xobj, PDF_NAME_Subtype)))
 					continue;
 
-				pdf_clean_stream_object(ctx, doc, xobj, page->resources, cookie, 1);
+				pdf_clean_stream_object(ctx, doc, xobj, page->resources, cookie, 1, ascii);
 			}
 		}
 
@@ -250,7 +250,7 @@ void pdf_clean_page_contents(fz_context *ctx, pdf_document *doc, pdf_page *page,
 
 				if (pdf_name_eq(ctx, PDF_NAME_Type3, pdf_dict_get(ctx, o, PDF_NAME_Subtype)))
 				{
-					pdf_clean_type3(ctx, doc, o, page->resources, cookie);
+					pdf_clean_type3(ctx, doc, o, page->resources, cookie, ascii);
 				}
 			}
 		}
diff --git a/source/pdf/pdf-op-buffer.c b/source/pdf/pdf-op-buffer.c
index cdea4c78..d4f56286 100644
--- a/source/pdf/pdf-op-buffer.c
+++ b/source/pdf/pdf-op-buffer.c
@@ -6,6 +6,7 @@ struct pdf_output_processor_s
 {
 	pdf_processor super;
 	fz_output *out;
+	int ahxencode;
 	int extgstate;
 };
 
@@ -543,6 +544,7 @@ static void
 pdf_out_BI(fz_context *ctx, pdf_processor *proc, fz_image *img)
 {
 	fz_output *out = ((pdf_output_processor*)proc)->out;
+	int ahx = ((pdf_output_processor*)proc)->ahxencode;
 	fz_compressed_buffer *cbuf;
 	fz_buffer *buf;
 	int i;
@@ -588,15 +590,14 @@ pdf_out_BI(fz_context *ctx, pdf_processor *proc, fz_image *img)
 		break;
 
 	case FZ_IMAGE_JPEG:
-		fz_printf(ctx, out, "/F/DCT\n");
+		fz_printf(ctx, out, ahx ? "/F[/AHx/DCT]\n" : "/F/DCT\n");
 		if (cbuf->params.u.jpeg.color_transform != -1)
 			fz_printf(ctx, out, "/DP<</ColorTransform %d>>\n",
 				cbuf->params.u.jpeg.color_transform);
 		break;
 
 	case FZ_IMAGE_FAX:
-		fz_printf(ctx, out, "/F/CCF\n");
-		fz_printf(ctx, out, "/DP<<\n");
+		fz_printf(ctx, out, ahx ? "/F[/AHx/CCF]\n/DP[null<<\n" : "/F/CCF\n/DP<<\n");
 		fz_printf(ctx, out, "/K %d\n", cbuf->params.u.fax.k);
 		if (cbuf->params.u.fax.columns != 1728)
 			fz_printf(ctx, out, "/Columns %d\n", cbuf->params.u.fax.columns);
@@ -613,21 +614,23 @@ pdf_out_BI(fz_context *ctx, pdf_processor *proc, fz_image *img)
 		if (cbuf->params.u.fax.damaged_rows_before_error > 0)
 			fz_printf(ctx, out, "/DamagedRowsBeforeError %d\n",
 				cbuf->params.u.fax.damaged_rows_before_error);
-		fz_printf(ctx, out, ">>\n");
+		fz_printf(ctx, out, ahx ? ">>]\n" : ">>\n");
 		break;
 
 	case FZ_IMAGE_RAW:
+		if (ahx)
+			fz_printf(ctx, out, "/F/AHx\n");
 		break;
 
 	case FZ_IMAGE_RLD:
-		fz_printf(ctx, out, "/F/RL\n");
+		fz_printf(ctx, out, ahx ? "/F[/AHx/RL]\n" : "/F/RL\n");
 		break;
 
 	case FZ_IMAGE_FLATE:
-		fz_printf(ctx, out, "/F/Fl\n");
+		fz_printf(ctx, out, ahx ? "/F[/AHx/Fl]\n" : "/F/Fl\n");
 		if (cbuf->params.u.flate.predictor > 1)
 		{
-			fz_printf(ctx, out, "/DP<<\n");
+			fz_printf(ctx, out, ahx ? "/DP[null<<\n" : "/DP<<\n");
 			fz_printf(ctx, out, "/Predictor %d\n", cbuf->params.u.flate.predictor);
 			if (cbuf->params.u.flate.columns != 1)
 				fz_printf(ctx, out, "/Columns %d\n", cbuf->params.u.flate.columns);
@@ -635,15 +638,15 @@ pdf_out_BI(fz_context *ctx, pdf_processor *proc, fz_image *img)
 				fz_printf(ctx, out, "/Colors %d\n", cbuf->params.u.flate.colors);
 			if (cbuf->params.u.flate.bpc != 8)
 				fz_printf(ctx, out, "/BitsPerComponent %d\n", cbuf->params.u.flate.bpc);
-			fz_printf(ctx, out, ">>\n");
+			fz_printf(ctx, out, ahx ? ">>]\n" : ">>\n");
 		}
 		break;
 
 	case FZ_IMAGE_LZW:
-		fz_printf(ctx, out, "/F/LZW\n");
+		fz_printf(ctx, out, ahx ? "/F[/AHx/LZW]\n" : "/F/LZW\n");
 		if (cbuf->params.u.lzw.predictor > 1)
 		{
-			fz_printf(ctx, out, "/DP<<\n");
+			fz_printf(ctx, out, ahx ? "/DP[<<null\n" : "/DP<<\n");
 			fz_printf(ctx, out, "/Predictor %d\n", cbuf->params.u.lzw.predictor);
 			if (cbuf->params.u.lzw.columns != 1)
 				fz_printf(ctx, out, "/Columns %d\n", cbuf->params.u.lzw.columns);
@@ -653,13 +656,28 @@ pdf_out_BI(fz_context *ctx, pdf_processor *proc, fz_image *img)
 				fz_printf(ctx, out, "/BitsPerComponent %d\n", cbuf->params.u.lzw.bpc);
 			if (cbuf->params.u.lzw.early_change != 1)
 				fz_printf(ctx, out, "/EarlyChange %d\n", cbuf->params.u.lzw.early_change);
-			fz_printf(ctx, out, ">>\n");
+			fz_printf(ctx, out, ahx ? ">>]\n" : ">>\n");
 		}
 		break;
 	}
 
 	fz_printf(ctx, out, "ID\n");
-	fz_write(ctx, out, buf->data, buf->len);
+	if (ahx)
+	{
+		for (i = 0; i < buf->len; ++i)
+		{
+			int c = buf->data[i];
+			fz_putc(ctx, out, "0123456789abcdef"[(c >> 4) & 0xf]);
+			fz_putc(ctx, out, "0123456789abcdef"[c & 0xf]);
+			if ((i & 31) == 31)
+				fz_putc(ctx, out, '\n');
+		}
+		fz_putc(ctx, out, '>');
+	}
+	else
+	{
+		fz_write(ctx, out, buf->data, buf->len);
+	}
 	fz_printf(ctx, out, "\nEI\n");
 }
 
@@ -749,7 +767,7 @@ pdf_drop_imp_output_processor(fz_context *ctx, pdf_processor *proc)
 }
 
 static pdf_processor *
-pdf_new_output_processor(fz_context *ctx, fz_output *out)
+pdf_new_output_processor(fz_context *ctx, fz_output *out, int ahxencode)
 {
 	pdf_output_processor *proc = pdf_new_processor(ctx, sizeof *proc);
 	{
@@ -867,18 +885,19 @@ pdf_new_output_processor(fz_context *ctx, fz_output *out)
 	}
 
 	proc->out = out;
+	proc->ahxencode = ahxencode;
 
 	return (pdf_processor*)proc;
 }
 
 pdf_processor *
-pdf_new_buffer_processor(fz_context *ctx, fz_buffer *buffer)
+pdf_new_buffer_processor(fz_context *ctx, fz_buffer *buffer, int ahxencode)
 {
 	pdf_processor *proc;
 	fz_output *out = fz_new_output_with_buffer(ctx, buffer);
 	fz_try(ctx)
 	{
-		proc = pdf_new_output_processor(ctx, out);
+		proc = pdf_new_output_processor(ctx, out, ahxencode);
 	}
 	fz_catch(ctx)
 	{
diff --git a/source/pdf/pdf-write.c b/source/pdf/pdf-write.c
index 20c9bfca..6ad308f1 100644
--- a/source/pdf/pdf-write.c
+++ b/source/pdf/pdf-write.c
@@ -2548,7 +2548,7 @@ static void complete_signatures(fz_context *ctx, pdf_document *doc, pdf_write_op
 	}
 }
 
-static void sanitise(fz_context *ctx, pdf_document *doc)
+static void sanitize(fz_context *ctx, pdf_document *doc, int ascii)
 {
 	int n = pdf_count_pages(ctx, doc);
 	int i;
@@ -2556,9 +2556,7 @@ static void sanitise(fz_context *ctx, pdf_document *doc)
 	for (i = 0; i < n; i++)
 	{
 		pdf_page *page = pdf_load_page(ctx, doc, i);
-
-		pdf_clean_page_contents(ctx, doc, page, NULL, NULL, NULL);
-
+		pdf_clean_page_contents(ctx, doc, page, NULL, NULL, NULL, ascii);
 		fz_drop_page(ctx, &page->super);
 	}
 }
@@ -2580,9 +2578,9 @@ void pdf_write_document(fz_context *ctx, pdf_document *doc, char *filename, fz_w
 
 	doc->freeze_updates = 1;
 
-	/* Sanitise the operator streams */
+	/* Sanitize the operator streams */
 	if (fz_opts->do_clean)
-		sanitise(ctx, doc);
+		sanitize(ctx, doc, fz_opts->do_ascii);
 
 	pdf_finish_edit(ctx, doc);
 	presize_unsaved_signature_byteranges(ctx, doc);