Simplify stext structure and device.

* Use pool allocator and linked lists for all levels.
* Remove separate fz_stext_sheet struct.
* Remove unused 'script' style.
* Remove 'span' level items.
* Detect visual/logical RTL layouts.
* Detect indented paragraphs.

1 files changed, 34 insertions, 75 deletions

diff --git a/source/fitz/util.c b/source/fitz/util.c
index 6f900174..d6a7f317 100644
--- a/source/fitz/util.c
+++ b/source/fitz/util.c
@@ -267,7 +267,7 @@ fz_new_pixmap_from_page_number(fz_context *ctx, fz_document *doc, int number, co
 }
 
 fz_stext_page *
-fz_new_stext_page_from_display_list(fz_context *ctx, fz_display_list *list, fz_stext_sheet *sheet, const fz_stext_options *options)
+fz_new_stext_page_from_display_list(fz_context *ctx, fz_display_list *list, const fz_stext_options *options)
 {
 	fz_stext_page *text;
 	fz_device *dev;
@@ -279,7 +279,7 @@ fz_new_stext_page_from_display_list(fz_context *ctx, fz_display_list *list, fz_s
 	text = fz_new_stext_page(ctx, fz_bound_display_list(ctx, list, &mediabox));
 	fz_try(ctx)
 	{
-		dev = fz_new_stext_device(ctx, sheet, text, options);
+		dev = fz_new_stext_device(ctx, text, options);
 		fz_run_display_list(ctx, list, dev, &fz_identity, NULL, NULL);
 		fz_close_device(ctx, dev);
 	}
@@ -297,7 +297,7 @@ fz_new_stext_page_from_display_list(fz_context *ctx, fz_display_list *list, fz_s
 }
 
 fz_stext_page *
-fz_new_stext_page_from_page(fz_context *ctx, fz_page *page, fz_stext_sheet *sheet, const fz_stext_options *options)
+fz_new_stext_page_from_page(fz_context *ctx, fz_page *page, const fz_stext_options *options)
 {
 	fz_stext_page *text;
 	fz_device *dev;
@@ -309,7 +309,7 @@ fz_new_stext_page_from_page(fz_context *ctx, fz_page *page, fz_stext_sheet *shee
 	text = fz_new_stext_page(ctx, fz_bound_page(ctx, page, &mediabox));
 	fz_try(ctx)
 	{
-		dev = fz_new_stext_device(ctx, sheet, text, options);
+		dev = fz_new_stext_device(ctx, text, options);
 		fz_run_page(ctx, page, dev, &fz_identity, NULL);
 		fz_close_device(ctx, dev);
 	}
@@ -327,14 +327,14 @@ fz_new_stext_page_from_page(fz_context *ctx, fz_page *page, fz_stext_sheet *shee
 }
 
 fz_stext_page *
-fz_new_stext_page_from_page_number(fz_context *ctx, fz_document *doc, int number, fz_stext_sheet *sheet, const fz_stext_options *options)
+fz_new_stext_page_from_page_number(fz_context *ctx, fz_document *doc, int number, const fz_stext_options *options)
 {
 	fz_page *page;
 	fz_stext_page *text;
 
 	page = fz_load_page(ctx, doc, number);
 	fz_try(ctx)
-		text = fz_new_stext_page_from_page(ctx, page, sheet, options);
+		text = fz_new_stext_page_from_page(ctx, page, options);
 	fz_always(ctx)
 		fz_drop_page(ctx, page);
 	fz_catch(ctx)
@@ -345,24 +345,14 @@ fz_new_stext_page_from_page_number(fz_context *ctx, fz_document *doc, int number
 int
 fz_search_display_list(fz_context *ctx, fz_display_list *list, const char *needle, fz_rect *hit_bbox, int hit_max)
 {
-	fz_stext_sheet *sheet = NULL;
-	fz_stext_page *text = NULL;
+	fz_stext_page *text;
 	int count;
 
-	fz_var(sheet);
-	fz_var(text);
-
+	text = fz_new_stext_page_from_display_list(ctx, list, NULL);
 	fz_try(ctx)
-	{
-		sheet = fz_new_stext_sheet(ctx);
-		text = fz_new_stext_page_from_display_list(ctx, list, sheet, NULL);
 		count = fz_search_stext_page(ctx, text, needle, hit_bbox, hit_max);
-	}
 	fz_always(ctx)
-	{
 		fz_drop_stext_page(ctx, text);
-		fz_drop_stext_sheet(ctx, sheet);
-	}
 	fz_catch(ctx)
 		fz_rethrow(ctx);
 	return count;
@@ -371,24 +361,14 @@ fz_search_display_list(fz_context *ctx, fz_display_list *list, const char *needl
 int
 fz_search_page(fz_context *ctx, fz_page *page, const char *needle, fz_rect *hit_bbox, int hit_max)
 {
-	fz_stext_sheet *sheet = NULL;
-	fz_stext_page *text = NULL;
+	fz_stext_page *text;
 	int count;
 
-	fz_var(sheet);
-	fz_var(text);
-
+	text = fz_new_stext_page_from_page(ctx, page, NULL);
 	fz_try(ctx)
-	{
-		sheet = fz_new_stext_sheet(ctx);
-		text = fz_new_stext_page_from_page(ctx, page, sheet, NULL);
 		count = fz_search_stext_page(ctx, text, needle, hit_bbox, hit_max);
-	}
 	fz_always(ctx)
-	{
 		fz_drop_stext_page(ctx, text);
-		fz_drop_stext_sheet(ctx, sheet);
-	}
 	fz_catch(ctx)
 		fz_rethrow(ctx);
 	return count;
@@ -411,14 +391,15 @@ fz_search_page_number(fz_context *ctx, fz_document *doc, int number, const char
 }
 
 fz_buffer *
-fz_new_buffer_from_stext_page(fz_context *ctx, fz_stext_page *text, const fz_rect *sel, int crlf)
+fz_new_buffer_from_stext_page(fz_context *ctx, fz_stext_page *page, const fz_rect *sel, int crlf)
 {
 	fz_buffer *buf;
 	fz_rect hitbox;
 	float x0, y0, x1, y1;
-	int block_num;
+	fz_stext_block *block;
+	fz_stext_line *line;
+	fz_stext_char *ch;
 	int need_newline;
-	int i;
 
 	need_newline = 0;
 
@@ -438,45 +419,33 @@ fz_new_buffer_from_stext_page(fz_context *ctx, fz_stext_page *text, const fz_rec
 	buf = fz_new_buffer(ctx, 256);
 	fz_try(ctx)
 	{
-		for (block_num = 0; block_num < text->len; block_num++)
+		for (block = page->first_block; block; block = block->next)
 		{
-			fz_stext_line *line;
-			fz_stext_block *block;
-			fz_stext_span *span;
-
-			if (text->blocks[block_num].type != FZ_PAGE_BLOCK_TEXT)
+			if (block->type != FZ_STEXT_BLOCK_TEXT)
 				continue;
 
-			block = text->blocks[block_num].u.text;
-			for (line = block->lines; line < block->lines + block->len; line++)
+			for (line = block->u.t.first_line; line; line = line->next)
 			{
 				int saw_text = 0;
-				for (span = line->first_span; span; span = span->next)
+				for (ch = line->first_char; ch; ch = ch->next)
 				{
-					if (span->spacing > 1)
-						fz_append_byte(ctx, buf, ' ');
-					for (i = 0; i < span->len; i++)
+					int c = ch->c;
+					fz_stext_char_bbox(ctx, &hitbox, line, ch);
+					if (c < 32)
+						c = 0xFFFD;
+					if (hitbox.x1 >= x0 && hitbox.x0 <= x1 && hitbox.y1 >= y0 && hitbox.y0 <= y1)
 					{
-						int c;
-						fz_stext_char_bbox(ctx, &hitbox, span, i);
-						c = span->text[i].c;
-						if (c < 32)
-							c = 0xFFFD;
-						if (hitbox.x1 >= x0 && hitbox.x0 <= x1 && hitbox.y1 >= y0 && hitbox.y0 <= y1)
+						saw_text = 1;
+						if (need_newline)
 						{
-							saw_text = 1;
-							if (need_newline)
-							{
-								if (crlf)
-									fz_append_byte(ctx, buf, '\r');
-								fz_append_byte(ctx, buf, '\n');
-								need_newline = 0;
-							}
-							fz_append_rune(ctx, buf, c);
+							if (crlf)
+								fz_append_byte(ctx, buf, '\r');
+							fz_append_byte(ctx, buf, '\n');
+							need_newline = 0;
 						}
+						fz_append_rune(ctx, buf, c);
 					}
 				}
-
 				if (saw_text)
 					need_newline = 1;
 			}
@@ -494,42 +463,32 @@ fz_new_buffer_from_stext_page(fz_context *ctx, fz_stext_page *text, const fz_rec
 fz_buffer *
 fz_new_buffer_from_display_list(fz_context *ctx, fz_display_list *list, const fz_rect *sel, int crlf, const fz_stext_options *options)
 {
-	fz_stext_sheet *sheet;
 	fz_stext_page *text;
 	fz_buffer *buf;
 
-	sheet = fz_new_stext_sheet(ctx);
+	text = fz_new_stext_page_from_display_list(ctx, list, options);
 	fz_try(ctx)
-	{
-		text = fz_new_stext_page_from_display_list(ctx, list, sheet, options);
 		buf = fz_new_buffer_from_stext_page(ctx, text, sel, crlf);
-	}
 	fz_always(ctx)
-		fz_drop_stext_sheet(ctx, sheet);
+		fz_drop_stext_page(ctx, text);
 	fz_catch(ctx)
 		fz_rethrow(ctx);
-	fz_drop_stext_page(ctx, text);
 	return buf;
 }
 
 fz_buffer *
 fz_new_buffer_from_page(fz_context *ctx, fz_page *page, const fz_rect *sel, int crlf, const fz_stext_options *options)
 {
-	fz_stext_sheet *sheet;
 	fz_stext_page *text;
 	fz_buffer *buf;
 
-	sheet = fz_new_stext_sheet(ctx);
+	text = fz_new_stext_page_from_page(ctx, page, options);
 	fz_try(ctx)
-	{
-		text = fz_new_stext_page_from_page(ctx, page, sheet, options);
 		buf = fz_new_buffer_from_stext_page(ctx, text, sel, crlf);
-	}
 	fz_always(ctx)
-		fz_drop_stext_sheet(ctx, sheet);
+		fz_drop_stext_page(ctx, text);
 	fz_catch(ctx)
 		fz_rethrow(ctx);
-	fz_drop_stext_page(ctx, text);
 	return buf;
 }