Distinguish separate search hit results.

Add a new optional argument to mark search hits to all of the search functions.

With the marker each individual hit on the page can be identified. A quad that
starts a new array has a matching 1 in the 'hits' array. Each continuation quad
is marked with a 0 instead.

This is the minimally invasive API change, it just adds an optional extra int
array for recording the hit markers.

The Java and mutool run APIs have been changed to return an array of array of
quads for the search results.

docs/api/changes.html

@@ -39,6 +39,35 @@ <h2>Why does the API change?</h2>
 Note, that we only list changes in existing APIs here, not all additions
 to the API.
 
+<h2>Changes from 1.19</h2>
+
+<p>
+The search API has been extended with one optional parameter: an integer array.
+If the 'hit_mark' array is provided it must be the same length as the quad array.
+Each entry corresponds to the quad of the same index.
+It will be filled in with 1 for each quad that starts a new search hit, and 0 for
+each quad that is a continuation of a search hit
+(for example when a match spans a line-break).
+
+<ul>
+<li>int fz_search_stext_page(ctx, text, needle, <i>int *hit_mark</i>, fz_quad *hit_bbox, int hit_max);
+<li>int fz_search_display_list(ctx, list, needle, <i>int *hit_mark</i>, fz_quad *hit_bbox, int hit_max);
+<li>int fz_search_page(ctx, page, needle, <i>int *hit_mark</i>, fz_quad *hit_bbox, int hit_max);
+<li>int fz_search_page_number(ctx, doc, page, needle, <i>int *hit_mark</i>, fz_quad *hit_bbox, int hit_max);
+<li>int fz_search_chapter_page_number(ctx, doc, c, p, needle, <i>int *hit_mark</i>, fz_quad *hit_bbox, int hit_max);
+</ul>
+
+<p>
+The Java search functions have also changed.
+They now return an array of array of quads instead of an array of quads.
+
+<ul>
+<li>Quad[][] StructuredText.search(String needle);
+<li>Quad[][] DisplayList.search(String needle);
+<li>Quad[][] Page.search(String needle);
+<li>Quad[][] Document.search(int chapter, int page, String needle);
+</ul>
+
 <h2>Changes from 1.18 to 1.19</h2>
 
 <p>We were inconsistent with the behaviour of fz_create_ and pdf_create_

include/mupdf/fitz/structured-text.h

@@ -244,7 +244,7 @@ void fz_print_stext_page_as_text(fz_context *ctx, fz_output *out, fz_stext_page
 	NOTE: This is an experimental interface and subject to change
 	without notice.
 */
-int fz_search_stext_page(fz_context *ctx, fz_stext_page *text, const char *needle, fz_quad *quads, int max_quads);
+int fz_search_stext_page(fz_context *ctx, fz_stext_page *text, const char *needle, int *hit_mark, fz_quad *hit_bbox, int hit_max);
 
 /**
 	Return a list of quads to highlight lines inside the selection

include/mupdf/fitz/util.h

@@ -95,10 +95,10 @@ fz_buffer *fz_new_buffer_from_display_list(fz_context *ctx, fz_display_list *lis
 	Record the hits in the hit_bbox array and return the number of
 	hits. Will stop looking once it has filled hit_max rectangles.
 */
-int fz_search_page(fz_context *ctx, fz_page *page, const char *needle, fz_quad *hit_bbox, int hit_max);
-int fz_search_page_number(fz_context *ctx, fz_document *doc, int number, const char *needle, fz_quad *hit_bbox, int hit_max);
-int fz_search_chapter_page_number(fz_context *ctx, fz_document *doc, int chapter, int page, const char *needle, fz_quad *hit_bbox, int hit_max);
-int fz_search_display_list(fz_context *ctx, fz_display_list *list, const char *needle, fz_quad *hit_bbox, int hit_max);
+int fz_search_page(fz_context *ctx, fz_page *page, const char *needle, int *hit_mark, fz_quad *hit_bbox, int hit_max);
+int fz_search_page_number(fz_context *ctx, fz_document *doc, int number, const char *needle, int *hit_mark, fz_quad *hit_bbox, int hit_max);
+int fz_search_chapter_page_number(fz_context *ctx, fz_document *doc, int chapter, int page, const char *needle, int *hit_mark, fz_quad *hit_bbox, int hit_max);
+int fz_search_display_list(fz_context *ctx, fz_display_list *list, const char *needle, int *hit_mark, fz_quad *hit_bbox, int hit_max);
 
 /**
 	Parse an SVG document into a display-list.

platform/gl/gl-annotate.c

@@ -1299,7 +1299,7 @@ static int mark_search_step(int cancel)
 		return -1;
 	}
 
-	count = fz_search_page_number(ctx, (fz_document*)pdf, rds_state.i-1, search_needle, quads, nelem(quads));
+	count = fz_search_page_number(ctx, (fz_document*)pdf, rds_state.i-1, search_needle, NULL, quads, nelem(quads));
 	if (count > 0)
 	{
 		pdf_page *page = pdf_load_page(ctx, pdf, rds_state.i-1);

platform/gl/gl-main.c

@@ -2534,7 +2534,7 @@ void do_main(void)
 			search_hit_count = fz_search_chapter_page_number(ctx, doc,
 				search_page.chapter, search_page.page,
 				search_needle,
-				search_hit_quads, nelem(search_hit_quads));
+				NULL, search_hit_quads, nelem(search_hit_quads));
 			trace_action("hits = doc.loadPage(%d).search(%q);\n", fz_page_number_from_location(ctx, doc, search_page), search_needle);
 			trace_action("print('Search page %d:', repr(%q), hits.length, repr(hits));\n", fz_page_number_from_location(ctx, doc, search_page), search_needle);
 			if (search_hit_count)

platform/java/example/Viewer.java

@@ -61,7 +61,7 @@ public class Viewer extends Frame implements WindowListener, ActionListener, Ite
 	protected int searchDirection = 1;
 	protected Location searchPage = null;
 	protected Location searchHitPage = null;
-	protected Quad[] searchHits = new Quad[0];
+	protected Quad[][] searchHits = new Quad[0][];
 	protected EventQueue eq = null;
 	protected SearchTask searchTask = null;
 
@@ -224,8 +224,10 @@ public void paint(Graphics g) {
 			if (currentPage.equals(searchHitPage) && searchHits != null) {
 				g2d.setColor(new Color(1, 0, 0, 0.4f));
 				for (int i = 0; i < searchHits.length; ++i) {
-					Rect hit = new Rect(searchHits[i]).transform(pageCTM);
-					g2d.fillRect((int)hit.x0, (int)hit.y0, (int)(hit.x1-hit.x0), (int)(hit.y1-hit.y0));
+					for (int k = 0; k < searchHits[i].length; ++k) {
+						Rect hit = new Rect(searchHits[i][k]).transform(pageCTM);
+						g2d.fillRect((int)hit.x0, (int)hit.y0, (int)(hit.x1-hit.x0), (int)(hit.y1-hit.y0));
+					}
 				}
 			}
 
@@ -763,7 +765,7 @@ public void run() {
 			while (!done && !isCancelled() && System.currentTimeMillis() < executionUntil) {
 				searchHits = doc.search(searchPage.chapter, searchPage.page, needle);
 				if (searchHits != null && searchHits.length > 0) {
-					searchHitPage = new Location(searchPage, searchHits[0].ul_x, searchHits[0].ul_y);
+					searchHitPage = new Location(searchPage, searchHits[0][0].ul_x, searchHits[0][0].ul_y);
 					jumpToLocation(searchHitPage);
 					done = true;
 				}

platform/java/jni/displaylist.c

@@ -139,7 +139,8 @@ FUN(DisplayList_search)(JNIEnv *env, jobject self, jstring jneedle)
 {
 	fz_context *ctx = get_context(env);
 	fz_display_list *list = from_DisplayList(env, self);
-	fz_quad hits[256];
+	fz_quad hits[500];
+	int marks[500];
 	const char *needle = NULL;
 	int n = 0;
 
@@ -150,11 +151,11 @@ FUN(DisplayList_search)(JNIEnv *env, jobject self, jstring jneedle)
 	if (!needle) return NULL;
 
 	fz_try(ctx)
-		n = fz_search_display_list(ctx, list, needle, hits, nelem(hits));
+		n = fz_search_display_list(ctx, list, needle, marks, hits, nelem(hits));
 	fz_always(ctx)
 		(*env)->ReleaseStringUTFChars(env, jneedle, needle);
 	fz_catch(ctx)
 		jni_rethrow(env, ctx);
 
-	return to_QuadArray_safe(ctx, env, hits, n);
+	return to_SearchHits_safe(ctx, env, marks, hits, n);
 }

platform/java/jni/document.c

@@ -1006,7 +1006,8 @@ FUN(Document_search)(JNIEnv *env, jobject self, jint chapter, jint page, jstring
 {
 	fz_context *ctx = get_context(env);
 	fz_document *doc = from_Document(env, self);
-	fz_quad hits[256];
+	fz_quad hits[500];
+	int marks[500];
 	const char *needle = NULL;
 	int n = 0;
 
@@ -1017,11 +1018,11 @@ FUN(Document_search)(JNIEnv *env, jobject self, jint chapter, jint page, jstring
 	if (!needle) return 0;
 
 	fz_try(ctx)
-		n = fz_search_chapter_page_number(ctx, doc, chapter, page, needle, hits, nelem(hits));
+		n = fz_search_chapter_page_number(ctx, doc, chapter, page, needle, marks, hits, nelem(hits));
 	fz_always(ctx)
 		(*env)->ReleaseStringUTFChars(env, jneedle, needle);
 	fz_catch(ctx)
 		jni_rethrow(env, ctx);
 
-	return to_QuadArray_safe(ctx, env, hits, n);
+	return to_SearchHits_safe(ctx, env, marks, hits, n);
 }

platform/java/jni/page.c

@@ -267,7 +267,8 @@ FUN(Page_search)(JNIEnv *env, jobject self, jstring jneedle)
 {
 	fz_context *ctx = get_context(env);
 	fz_page *page = from_Page(env, self);
-	fz_quad hits[256];
+	fz_quad hits[500];
+	int marks[500];
 	const char *needle = NULL;
 	int n = 0;
 
@@ -278,13 +279,13 @@ FUN(Page_search)(JNIEnv *env, jobject self, jstring jneedle)
 	if (!needle) return 0;
 
 	fz_try(ctx)
-		n = fz_search_page(ctx, page, needle, hits, nelem(hits));
+		n = fz_search_page(ctx, page, needle, marks, hits, nelem(hits));
 	fz_always(ctx)
 		(*env)->ReleaseStringUTFChars(env, jneedle, needle);
 	fz_catch(ctx)
 		jni_rethrow(env, ctx);
 
-	return to_QuadArray_safe(ctx, env, hits, n);
+	return to_SearchHits_safe(ctx, env, marks, hits, n);
 }
 
 JNIEXPORT jobject JNICALL

platform/java/jni/structuredtext.c

@@ -37,7 +37,8 @@ FUN(StructuredText_search)(JNIEnv *env, jobject self, jstring jneedle)
 {
 	fz_context *ctx = get_context(env);
 	fz_stext_page *text = from_StructuredText(env, self);
-	fz_quad hits[256];
+	fz_quad hits[500];
+	int marks[500];
 	const char *needle = NULL;
 	int n = 0;
 
@@ -48,13 +49,13 @@ FUN(StructuredText_search)(JNIEnv *env, jobject self, jstring jneedle)
 	if (!needle) return NULL;
 
 	fz_try(ctx)
-		n = fz_search_stext_page(ctx, text, needle, hits, nelem(hits));
+		n = fz_search_stext_page(ctx, text, needle, marks, hits, nelem(hits));
 	fz_always(ctx)
 		(*env)->ReleaseStringUTFChars(env, jneedle, needle);
 	fz_catch(ctx)
 		jni_rethrow(env, ctx);
 
-	return to_QuadArray_safe(ctx, env, hits, n);
+	return to_SearchHits_safe(ctx, env, marks, hits, n);
 }
 
 JNIEXPORT jobject JNICALL

platform/java/jni/wrap.c

@@ -403,6 +403,65 @@ static inline jobjectArray to_QuadArray_safe(fz_context *ctx, JNIEnv *env, const
 	return arr;
 }
 
+static int count_next_hits(const int *marks, int a, int end)
+{
+	int b = a + 1;
+	while (b < end && !marks[b])
+		++b;
+	return b - a;
+}
+
+/* Array of Array of Quad */
+static inline jobjectArray to_SearchHits_safe(fz_context *ctx, JNIEnv *env, const int *marks, const fz_quad *quads, jint n)
+{
+	jobjectArray toparr, arr;
+	int i, k, a, m;
+
+	if (!ctx || !marks || !quads)
+		return NULL;
+
+	/* Count total number of search hits */
+	i = a = 0;
+	while (a < n)
+	{
+		a += count_next_hits(marks, a, n);
+		++i;
+	}
+
+	toparr = (*env)->NewObjectArray(env, i, cls_ArrayOfQuad, NULL);
+	if (!toparr || (*env)->ExceptionCheck(env))
+		return NULL;
+
+	i = a = 0;
+	while (a < n)
+	{
+		m = count_next_hits(marks, a, n);
+
+		arr = (*env)->NewObjectArray(env, m, cls_Quad, NULL);
+		if (!arr || (*env)->ExceptionCheck(env))
+			return NULL;
+		(*env)->SetObjectArrayElement(env, toparr, i++, arr);
+		if ((*env)->ExceptionCheck(env))
+			return NULL;
+
+		for (k = 0; k < m; ++k) {
+			jobject jquad = to_Quad_safe(ctx, env, quads[a+k]);
+			if (!jquad || (*env)->ExceptionCheck(env))
+				return NULL;
+			(*env)->SetObjectArrayElement(env, arr, k, jquad);
+			if ((*env)->ExceptionCheck(env))
+				return NULL;
+			(*env)->DeleteLocalRef(env, jquad);
+		}
+
+		(*env)->DeleteLocalRef(env, arr);
+
+		a += m;
+	}
+
+	return toparr;
+}
+
 static inline jobject to_Rect_safe(fz_context *ctx, JNIEnv *env, fz_rect rect)
 {
 	if (!ctx) return NULL;

platform/java/mupdf_native.c

@@ -87,6 +87,7 @@ static JavaVM *jvm = NULL;
 
 /* All the cached classes/mids/fids we need. */
 
+static jclass cls_ArrayOfQuad;
 static jclass cls_Buffer;
 static jclass cls_ColorSpace;
 static jclass cls_Context;
@@ -938,6 +939,8 @@ static int find_fids(JNIEnv *env)
 	fid_Quad_lr_y = get_field(&err, env, "lr_y", "F");
 	mid_Quad_init = get_method(&err, env, "<init>", "(FFFFFFFF)V");
 
+	cls_ArrayOfQuad = get_class(&err, env, "[L"PKG"Quad;");
+
 	cls_Rect = get_class(&err, env, PKG"Rect");
 	fid_Rect_x0 = get_field(&err, env, "x0", "F");
 	fid_Rect_x1 = get_field(&err, env, "x1", "F");
@@ -1090,6 +1093,7 @@ static void jni_detach_thread(jboolean detach)
 
 static void lose_fids(JNIEnv *env)
 {
+	(*env)->DeleteGlobalRef(env, cls_ArrayOfQuad);
 	(*env)->DeleteGlobalRef(env, cls_Buffer);
 	(*env)->DeleteGlobalRef(env, cls_ColorSpace);
 	(*env)->DeleteGlobalRef(env, cls_Context);

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

@@ -53,7 +53,7 @@ public StructuredText toStructuredText() {
 		return toStructuredText(null);
 	}
 
-	public native Quad[] search(String needle);
+	public native Quad[][] search(String needle);
 
 	public native void run(Device dev, Matrix ctm, Rect scissor, Cookie cookie);
 

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

@@ -191,7 +191,7 @@ public int pageNumberFromLocation(Location loc) {
 		return -1;
 	}
 
-	public native Quad[] search(int chapter, int page, String needle);
+	public native Quad[][] search(int chapter, int page, String needle);
 
 	public native Location resolveLink(String uri);
 	public Location resolveLink(Outline link) {

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

@@ -68,7 +68,7 @@ public StructuredText toStructuredText() {
 		return toStructuredText(null);
 	}
 
-	public native Quad[] search(String needle);
+	public native Quad[][] search(String needle);
 
 	public native byte[] textAsHtml();
 

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

@@ -46,7 +46,7 @@ private StructuredText(long p) {
 		pointer = p;
 	}
 
-	public native Quad[] search(String needle);
+	public native Quad[][] search(String needle);
 	public native Quad[] highlight(Point a, Point b);
 	public native Quad snapSelection(Point a, Point b, int mode);
 	public native String copy(Point a, Point b);

platform/x11/pdfapp.c

@@ -1186,7 +1186,7 @@ static void pdfapp_search_in_direction(pdfapp_t *app, enum panning *panto, int d
 			pdfapp_showpage(app, 1, 0, 0, 0, 1);
 		}
 
-		app->hit_count = fz_search_stext_page(app->ctx, app->page_text, app->search, app->hit_bbox, nelem(app->hit_bbox));
+		app->hit_count = fz_search_stext_page(app->ctx, app->page_text, app->search, NULL, app->hit_bbox, nelem(app->hit_bbox));
 		if (app->hit_count > 0)
 		{
 			*panto = dir == 1 ? PAN_TO_TOP : PAN_TO_BOTTOM;