Add 'text' option to pdf_apply_redactions().

This allows for the default (current) operation of removing
text that intersects with redaction regions, as well as the
new option of leaving text alone.

This is obviously insecure, but given the increasing use of
redaction operations as a mechanism for removing arbitrary
content, this should be useful.

include/mupdf/pdf/page.h

@@ -265,11 +265,23 @@ enum {
 	PDF_REDACT_LINE_ART_REMOVE_IF_TOUCHED
 };
 
+enum {
+	/* Remove any text that overlaps with the redaction region,
+	 * however slightly. This is the default option, and is the
+	 * correct option for secure behaviour. */
+	PDF_REDACT_TEXT_REMOVE,
+	/* Do not remove any text at all as part of this redaction
+	 * operation. Using this option is INSECURE! Use at your own
+	 * risk. */
+	PDF_REDACT_TEXT_NONE
+};
+
 typedef struct
 {
 	int black_boxes;
 	int image_method;
 	int line_art;
+	int text;
 } pdf_redact_options;
 
 int pdf_redact_page(fz_context *ctx, pdf_document *doc, pdf_page *page, pdf_redact_options *opts);

platform/java/jni/pdfannotation.c

@@ -1583,11 +1583,11 @@ FUN(PDFAnnotation_getHiddenForEditing)(JNIEnv *env, jobject self)
 }
 
 JNIEXPORT jboolean JNICALL
-FUN(PDFAnnotation_applyRedaction)(JNIEnv *env, jobject self, jboolean blackBoxes, jint imageMethod, jint lineArt)
+FUN(PDFAnnotation_applyRedaction)(JNIEnv *env, jobject self, jboolean blackBoxes, jint imageMethod, jint lineArt, jint text)
 {
 	fz_context *ctx = get_context(env);
 	pdf_annot *annot = from_PDFAnnotation_safe(env, self);
-	pdf_redact_options opts = { blackBoxes, imageMethod, lineArt };
+	pdf_redact_options opts = { blackBoxes, imageMethod, lineArt, text };
 	jboolean redacted = JNI_FALSE;
 
 	if (!ctx || !annot) return JNI_FALSE;

platform/java/jni/pdfpage.c

@@ -72,12 +72,12 @@ FUN(PDFPage_update)(JNIEnv *env, jobject self)
 }
 
 JNIEXPORT jboolean JNICALL
-FUN(PDFPage_applyRedactions)(JNIEnv *env, jobject self, jboolean blackBoxes, jint imageMethod, jint lineArt)
+FUN(PDFPage_applyRedactions)(JNIEnv *env, jobject self, jboolean blackBoxes, jint imageMethod, jint lineArt, jint text)
 {
 	fz_context *ctx = get_context(env);
 	pdf_page *page = from_PDFPage(env, self);
 	jboolean redacted = JNI_FALSE;
-	pdf_redact_options opts = { blackBoxes, imageMethod, lineArt };
+	pdf_redact_options opts = { blackBoxes, imageMethod, lineArt, text };
 
 	if (!ctx || !page) return JNI_FALSE;
 

platform/java/mupdf_native.c

@@ -652,6 +652,9 @@ static int check_enums()
 	valid &= com_artifex_mupdf_fitz_PDFPage_REDACT_LINE_ART_REMOVE_IF_TOUCHED == PDF_REDACT_LINE_ART_REMOVE_IF_TOUCHED;
 	valid &= com_artifex_mupdf_fitz_PDFPage_REDACT_LINE_ART_REMOVE_IF_COVERED == PDF_REDACT_LINE_ART_REMOVE_IF_COVERED;
 
+	valid &= com_artifex_mupdf_fitz_PDFPage_REDACT_TEXT_REMOVE == PDF_REDACT_TEXT_REMOVE;
+	valid &= com_artifex_mupdf_fitz_PDFPage_REDACT_TEXT_NONE == PDF_REDACT_TEXT_NONE;
+
 	return valid ? 1 : 0;
 }
 

platform/java/src/com/artifex/mupdf/fitz/PDFAnnotation.java

@@ -346,15 +346,20 @@ public void setAppearance(Image image) {
 
 	public boolean applyRedaction(boolean blackBoxes)
 	{
-		return applyRedaction(blackBoxes, PDFPage.REDACT_IMAGE_PIXELS, PDFPage.REDACT_LINE_ART_NONE);
+		return applyRedaction(blackBoxes, PDFPage.REDACT_IMAGE_PIXELS, PDFPage.REDACT_LINE_ART_NONE, PDFPage.REDACT_TEXT_REMOVE);
 	}
 
 	public boolean applyRedaction(boolean blackBoxes, int imageMethod)
 	{
-		return applyRedaction(blackBoxes, imageMethod, PDFPage.REDACT_LINE_ART_NONE);
+		return applyRedaction(blackBoxes, imageMethod, PDFPage.REDACT_LINE_ART_NONE, PDFPage.REDACT_TEXT_REMOVE);
 	}
 
-	public native boolean applyRedaction(boolean blackBoxes, int imageMethod, int lineArt);
+	public boolean applyRedaction(boolean blackBoxes, int imageMethod, int lineArt)
+	{
+		return applyRedaction(blackBoxes, imageMethod, lineArt, PDFPage.REDACT_TEXT_REMOVE);
+	}
+
+	public native boolean applyRedaction(boolean blackBoxes, int imageMethod, int lineArt, int text);
 
 	/** @deprecated use getBorderWidth instead. */
 	public float getBorder() { return getBorderWidth(); }

platform/java/src/com/artifex/mupdf/fitz/PDFPage.java

@@ -44,14 +44,21 @@ public class PDFPage extends Page
 	public static final int REDACT_LINE_ART_REMOVE_IF_COVERED = 1;
 	public static final int REDACT_LINE_ART_REMOVE_IF_TOUCHED = 2;
 
-	public native boolean applyRedactions(boolean blackBoxes, int imageMethod, int lineArt);
+	public static final int REDACT_TEXT_REMOVE = 0;
+	public static final int REDACT_TEXT_NONE = 1;
+
+	public native boolean applyRedactions(boolean blackBoxes, int imageMethod, int lineArt, int text);
 
 	public boolean applyRedactions() {
-		return applyRedactions(true, REDACT_IMAGE_PIXELS, REDACT_LINE_ART_NONE);
+		return applyRedactions(true, REDACT_IMAGE_PIXELS, REDACT_LINE_ART_NONE, REDACT_TEXT_REMOVE);
 	}
 
 	public boolean applyRedactions(boolean blackBoxes, int imageMethod) {
-		return applyRedactions(blackBoxes, imageMethod, REDACT_LINE_ART_NONE);
+		return applyRedactions(blackBoxes, imageMethod, REDACT_LINE_ART_NONE, REDACT_TEXT_REMOVE);
+	}
+
+	public boolean applyRedactions(boolean blackBoxes, int imageMethod, int lineArt) {
+		return applyRedactions(blackBoxes, imageMethod, lineArt, REDACT_TEXT_REMOVE);
 	}
 
 	public native boolean update();

source/pdf/pdf-clean.c

@@ -449,6 +449,7 @@ struct redact_filter_state {
 	pdf_page *page;
 	pdf_annot *target; // NULL if all
 	int line_art;
+	int text;
 };
 
 static void
@@ -989,6 +990,7 @@ void init_redact_filter(fz_context *ctx, pdf_redact_options *redact_opts, struct
 	int black_boxes = redact_opts ? redact_opts->black_boxes : 0;
 	int image_method = redact_opts ? redact_opts->image_method : PDF_REDACT_IMAGE_PIXELS;
 	int line_art = redact_opts ? redact_opts->line_art : PDF_REDACT_LINE_ART_NONE;
+	int text = redact_opts ? redact_opts->text : PDF_REDACT_TEXT_REMOVE;
 
 	memset(&red->filter_opts, 0, sizeof red->filter_opts);
 	memset(&red->sanitize_opts, 0, sizeof red->sanitize_opts);
@@ -1001,9 +1003,11 @@ void init_redact_filter(fz_context *ctx, pdf_redact_options *redact_opts, struct
 	if (black_boxes)
 		red->filter_opts.complete = pdf_redact_end_page;
 	red->line_art = line_art;
+	red->text = text;
 
 	red->sanitize_opts.opaque = red;
-	red->sanitize_opts.text_filter = pdf_redact_text_filter;
+	if (text == PDF_REDACT_TEXT_REMOVE)
+		red->sanitize_opts.text_filter = pdf_redact_text_filter;
 	if (image_method == PDF_REDACT_IMAGE_PIXELS)
 		red->sanitize_opts.image_filter = pdf_redact_image_filter_pixels;
 	if (image_method == PDF_REDACT_IMAGE_REMOVE)

source/tools/murun.c

@@ -7899,6 +7899,7 @@ static void ffi_PDFPage_applyRedactions(js_State *J)
 	if (js_isdefined(J, 1)) opts.black_boxes = js_toboolean(J, 1);
 	if (js_isdefined(J, 2)) opts.image_method = js_tointeger(J, 2);
 	if (js_isdefined(J, 3)) opts.line_art = js_tointeger(J, 3);
+	if (js_isdefined(J, 4)) opts.text = js_tointeger(J, 4);
 	fz_try(ctx)
 		pdf_redact_page(ctx, page->doc, page, &opts);
 	fz_catch(ctx)
@@ -10444,7 +10445,7 @@ int murun_main(int argc, char **argv)
 		jsB_propfun(J, "PDFPage.createAnnotation", ffi_PDFPage_createAnnotation, 1);
 		jsB_propfun(J, "PDFPage.deleteAnnotation", ffi_PDFPage_deleteAnnotation, 1);
 		jsB_propfun(J, "PDFPage.update", ffi_PDFPage_update, 0);
-		jsB_propfun(J, "PDFPage.applyRedactions", ffi_PDFPage_applyRedactions, 3);
+		jsB_propfun(J, "PDFPage.applyRedactions", ffi_PDFPage_applyRedactions, 4);
 		jsB_propfun(J, "PDFPage.process", ffi_PDFPage_process, 1);
 		jsB_propfun(J, "PDFPage.toPixmap", ffi_PDFPage_toPixmap, 6);
 		jsB_propfun(J, "PDFPage.getTransform", ffi_PDFPage_getTransform, 0);