Select and copy structured text by lines.

3 files changed, 275 insertions, 60 deletions

diff --git a/source/fitz/stext-device.c b/source/fitz/stext-device.c
index cae3c583..cd657acc 100644
--- a/source/fitz/stext-device.c
+++ b/source/fitz/stext-device.c
@@ -643,14 +643,13 @@ fz_stext_close_device(fz_context *ctx, fz_device *dev)
 
 	for (block = page->first_block; block; block = block->next)
 	{
-		if (block->type == FZ_STEXT_BLOCK_TEXT)
+		if (block->type != FZ_STEXT_BLOCK_TEXT)
+			continue;
+		for (line = block->u.t.first_line; line; line = line->next)
 		{
-			for (line = block->u.t.first_line; line; line = line->next)
-			{
-				for (ch = line->first_char; ch; ch = ch->next)
-					fz_union_rect(&line->bbox, &ch->bbox);
-				fz_union_rect(&block->bbox, &line->bbox);
-			}
+			for (ch = line->first_char; ch; ch = ch->next)
+				fz_union_rect(&line->bbox, &ch->bbox);
+			fz_union_rect(&block->bbox, &line->bbox);
 		}
 	}
 
diff --git a/source/fitz/stext-output.c b/source/fitz/stext-output.c
index 6a23cc6d..492885cb 100644
--- a/source/fitz/stext-output.c
+++ b/source/fitz/stext-output.c
@@ -244,8 +244,6 @@ static void fz_print_stext_block_as_xhtml(fz_context *ctx, fz_output *out, fz_st
 
 	for (line = block->u.t.first_line; line; line = line->next)
 	{
-		if (line != block->u.t.first_line)
-			fz_write_string(ctx, out, "\n");
 		for (ch = line->first_char; ch; ch = ch->next)
 		{
 			int ch_sup = detect_super_script(line, ch);
diff --git a/source/fitz/stext-search.c b/source/fitz/stext-search.c
index c8fbdc0b..b2247b33 100644
--- a/source/fitz/stext-search.c
+++ b/source/fitz/stext-search.c
@@ -1,6 +1,8 @@
 #include "mupdf/fitz.h"
 
 #include <string.h>
+#include <limits.h>
+#include <stdio.h>
 
 static inline int fz_tolower(int c)
 {
@@ -159,21 +161,68 @@ fz_search_stext_page(fz_context *ctx, fz_stext_page *text, const char *needle, f
 	return hit_count;
 }
 
-int
-fz_highlight_selection(fz_context *ctx, fz_stext_page *page, fz_rect rect, fz_rect *hit_bbox, int hit_max)
+static float dist2(float a, float b)
+{
+	return a * a + b * b;
+}
+
+static int line_length(fz_stext_line *line)
 {
-	fz_rect linebox;
-	fz_stext_block *block;
-	fz_stext_line *line;
 	fz_stext_char *ch;
-	int hit_count;
+	int n = 0;
+	for (ch = line->first_char; ch; ch = ch->next)
+		++n;
+	return n;
+}
 
-	float x0 = rect.x0;
-	float x1 = rect.x1;
-	float y0 = rect.y0;
-	float y1 = rect.y1;
+static int find_closest_in_line(fz_stext_line *line, int idx, fz_point p)
+{
+	fz_stext_char *ch;
+	float closest_dist = 1e30;
+	int closest_idx = idx;
 
-	hit_count = 0;
+	if (line->dir.x > line->dir.y)
+	{
+		if (p.y < line->bbox.y0)
+			return idx;
+		if (p.y > line->bbox.y1)
+			return idx + line_length(line);
+	}
+	else
+	{
+		if (p.x < line->bbox.x0)
+			return idx + line_length(line);
+		if (p.x > line->bbox.x1)
+			return idx;
+	}
+
+	for (ch = line->first_char; ch; ch = ch->next)
+	{
+		float mid_x = (ch->bbox.x0 + ch->bbox.x1) / 2;
+		float mid_y = (ch->bbox.y0 + ch->bbox.y1) / 2;
+		float this_dist = dist2(p.x - mid_x, p.y - mid_y);
+		if (this_dist < closest_dist)
+		{
+			closest_dist = this_dist;
+			if (line->dir.x > line->dir.y)
+				closest_idx = (p.x < mid_x) ? idx : idx+1;
+			else
+				closest_idx = (p.y < mid_y) ? idx : idx+1;
+		}
+		++idx;
+	}
+	return closest_idx;
+}
+
+static int find_closest_in_page(fz_stext_page *page, fz_point p)
+{
+	fz_stext_block *block;
+	fz_stext_line *line;
+	fz_stext_line *closest_line = NULL;
+	int closest_idx = 0;
+	float closest_dist = 1e30;
+	float this_dist;
+	int idx = 0;
 
 	for (block = page->first_block; block; block = block->next)
 	{
@@ -181,77 +230,246 @@ fz_highlight_selection(fz_context *ctx, fz_stext_page *page, fz_rect rect, fz_re
 			continue;
 		for (line = block->u.t.first_line; line; line = line->next)
 		{
-			linebox = fz_empty_rect;
-			for (ch = line->first_char; ch; ch = ch->next)
+			fz_rect box = line->bbox;
+			if (p.x >= box.x0 && p.x <= box.x1)
 			{
-				if (ch->bbox.x1 >= x0 && ch->bbox.x0 <= x1 && ch->bbox.y1 >= y0 && ch->bbox.y0 <= y1)
-				{
-					if (ch->bbox.y0 != linebox.y0 || fz_abs(ch->bbox.x0 - linebox.x1) > 5)
-					{
-						if (!fz_is_empty_rect(&linebox) && hit_count < hit_max)
-							hit_bbox[hit_count++] = linebox;
-						linebox = ch->bbox;
-					}
-					else
-					{
-						fz_union_rect(&linebox, &ch->bbox);
-					}
-				}
+				if (p.y < box.y0)
+					this_dist = dist2(box.y0 - p.y, 0);
+				else if (p.y > box.y1)
+					this_dist = dist2(p.y - box.y1, 0);
+				else
+					this_dist = 0;
 			}
-			if (!fz_is_empty_rect(&linebox) && hit_count < hit_max)
-				hit_bbox[hit_count++] = linebox;
+			else if (p.y >= box.y0 && p.y <= box.y1)
+			{
+				if (p.x < box.x0)
+					this_dist = dist2(box.x0 - p.x, 0);
+				else if (p.x > box.x1)
+					this_dist = dist2(p.x - box.x1, 0);
+				else
+					this_dist = 0;
+			}
+			else
+			{
+				float dul = dist2(p.x - box.x0, p.y - box.y0);
+				float dur = dist2(p.x - box.x1, p.y - box.y0);
+				float dll = dist2(p.x - box.x0, p.y - box.y1);
+				float dlr = dist2(p.x - box.x1, p.y - box.y1);
+				this_dist = fz_min(fz_min(dul, dur), fz_min(dll, dlr));
+			}
+			if (this_dist < closest_dist)
+			{
+				closest_dist = this_dist;
+				closest_line = line;
+				closest_idx = idx;
+			}
+			idx += line_length(line);
 		}
 	}
 
-	return hit_count;
+	if (closest_line)
+		return find_closest_in_line(closest_line, closest_idx, p);
+	return 0;
 }
 
-char *
-fz_copy_selection(fz_context *ctx, fz_stext_page *page, fz_rect rect)
+struct callbacks
+{
+	void (*on_char)(fz_context *ctx, void *arg, fz_stext_line *ln, fz_stext_char *ch);
+	void (*on_line)(fz_context *ctx, void *arg, fz_stext_line *ln);
+	void *arg;
+};
+
+static void
+fz_enumerate_selection(fz_context *ctx, fz_stext_page *page, fz_point a, fz_point b, struct callbacks *cb)
 {
-	fz_buffer *buffer;
-	int c, seen = 0;
-	unsigned char *s;
 	fz_stext_block *block;
 	fz_stext_line *line;
 	fz_stext_char *ch;
+	int idx, start, end;
+	int hit_count;
+	int inside;
 
-	float x0 = rect.x0;
-	float x1 = rect.x1;
-	float y0 = rect.y0;
-	float y1 = rect.y1;
+	start = find_closest_in_page(page, a);
+	end = find_closest_in_page(page, b);
 
-	buffer = fz_new_buffer(ctx, 1024);
+	if (start > end)
+		idx = start, start = end, end = idx;
+
+	if (start == end)
+		return;
 
+	hit_count = 0;
+	inside = 0;
+	idx = 0;
 	for (block = page->first_block; block; block = block->next)
 	{
 		if (block->type != FZ_STEXT_BLOCK_TEXT)
 			continue;
 		for (line = block->u.t.first_line; line; line = line->next)
 		{
-			if (seen)
+			for (ch = line->first_char; ch; ch = ch->next)
 			{
-				fz_append_byte(ctx, buffer, '\n');
+				if (!inside)
+					if (idx == start)
+						inside = 1;
+				if (inside)
+					cb->on_char(ctx, cb->arg, line, ch);
+				if (++idx == end)
+					return;
 			}
+			if (inside)
+				cb->on_line(ctx, cb->arg, line);
+		}
+	}
+}
 
-			seen = 0;
+struct highlight
+{
+	int len, cap;
+	fz_rect *box;
+};
 
-			for (ch = line->first_char; ch; ch = ch->next)
+static void on_highlight_char(fz_context *ctx, void *arg, fz_stext_line *line, fz_stext_char *ch)
+{
+	struct highlight *hits = arg;
+	float vfuzz = ch->size * 0.1f;
+	float hfuzz = ch->size * 0.5f;
+	fz_rect bbox;
+
+	if (line->dir.x > line->dir.y)
+	{
+		bbox.x0 = ch->bbox.x0;
+		bbox.x1 = ch->bbox.x1;
+		bbox.y0 = line->bbox.y0;
+		bbox.y1 = line->bbox.y1;
+	}
+	else
+	{
+		bbox.x0 = line->bbox.x0;
+		bbox.x1 = line->bbox.x1;
+		bbox.y0 = ch->bbox.y0;
+		bbox.y1 = ch->bbox.y1;
+	}
+
+	if (hits->len > 0)
+	{
+		fz_rect *end = &hits->box[hits->len-1];
+		if (fz_abs(bbox.y0 - end->y0) < vfuzz && fz_abs(bbox.y1 - end->y1) < vfuzz)
+		{
+			if (bbox.x1 < end->x0)
 			{
-				c = ch->c;
-				if (c < 32)
-					c = FZ_REPLACEMENT_CHARACTER;
-				if (ch->bbox.x1 >= x0 && ch->bbox.x0 <= x1 && ch->bbox.y1 >= y0 && ch->bbox.y0 <= y1)
+				if (end->x0 - bbox.x1 < hfuzz)
 				{
-					fz_append_rune(ctx, buffer, c);
-					seen = 1;
+					end->x0 = bbox.x0;
+					return;
 				}
 			}
-
-			seen = (seen && line == block->u.t.last_line);
+			else if (bbox.x0 > end->x1)
+			{
+				if (bbox.x0 - end->x1 < hfuzz)
+				{
+					end->x1 = bbox.x1;
+					return;
+				}
+			}
+			else
+			{
+				end->x0 = fz_min(bbox.x0, end->x0);
+				end->x1 = fz_max(bbox.x1, end->x1);
+				return;
+			}
+		}
+		if (fz_abs(bbox.x0 - end->x0) < vfuzz && fz_abs(bbox.x1 - end->x1) < vfuzz)
+		{
+			if (bbox.y1 < end->y0)
+			{
+				if (end->y0 - bbox.y1 < hfuzz)
+				{
+					end->y0 = bbox.y0;
+					return;
+				}
+			}
+			else if (bbox.y0 > end->y1)
+			{
+				if (bbox.y0 - end->y1 < hfuzz)
+				{
+					end->y1 = bbox.y1;
+					return;
+				}
+			}
+			else
+			{
+				end->y0 = fz_min(bbox.y0, end->y0);
+				end->y1 = fz_max(bbox.y1, end->y1);
+				return;
+			}
 		}
 	}
 
+	if (hits->len < hits->cap)
+		hits->box[hits->len++] = bbox;
+}
+
+static void on_highlight_line(fz_context *ctx, void *arg, fz_stext_line *line)
+{
+}
+
+int
+fz_highlight_selection(fz_context *ctx, fz_stext_page *page, fz_point a, fz_point b, fz_rect *hit_bbox, int hit_max)
+{
+	struct callbacks cb;
+	struct highlight hits;
+
+	hits.len = 0;
+	hits.cap = hit_max;
+	hits.box = hit_bbox;
+
+	cb.on_char = on_highlight_char;
+	cb.on_line = on_highlight_line;
+	cb.arg = &hits;
+
+	fz_enumerate_selection(ctx, page, a, b, &cb);
+
+	return hits.len;
+}
+
+static void on_copy_char(fz_context *ctx, void *arg, fz_stext_line *line, fz_stext_char *ch)
+{
+	fz_buffer *buffer = arg;
+	int c = ch->c;
+	if (c < 32)
+		c = FZ_REPLACEMENT_CHARACTER;
+	fz_append_rune(ctx, buffer, c);
+}
+
+static void on_copy_line_crlf(fz_context *ctx, void *arg, fz_stext_line *line)
+{
+	fz_buffer *buffer = arg;
+	fz_append_byte(ctx, buffer, '\r');
+	fz_append_byte(ctx, buffer, '\n');
+}
+
+static void on_copy_line_lf(fz_context *ctx, void *arg, fz_stext_line *line)
+{
+	fz_buffer *buffer = arg;
+	fz_append_byte(ctx, buffer, '\n');
+}
+
+char *
+fz_copy_selection(fz_context *ctx, fz_stext_page *page, fz_point a, fz_point b, int crlf)
+{
+	struct callbacks cb;
+	fz_buffer *buffer;
+	unsigned char *s;
+
+	buffer = fz_new_buffer(ctx, 1024);
+
+	cb.on_char = on_copy_char;
+	cb.on_line = crlf ? on_copy_line_crlf : on_copy_line_lf;
+	cb.arg = buffer;
+
+	fz_enumerate_selection(ctx, page, a, b, &cb);
+
 	fz_terminate_buffer(ctx, buffer);
 	fz_buffer_extract(ctx, buffer, &s); /* take over the data */
 	fz_drop_buffer(ctx, buffer);