Calculate accurate per-glyph bounding boxes for fz_bound_text.

10 files changed, 155 insertions, 78 deletions

diff --git a/draw/draw_device.c b/draw/draw_device.c
index f07ad681..07a2bb82 100644
--- a/draw/draw_device.c
+++ b/draw/draw_device.c
@@ -589,7 +589,7 @@ fz_draw_clip_text(fz_device *devp, fz_text *text, fz_matrix ctm, int accumulate)
 	if (accumulate == 0)
 	{
 		/* make the mask the exact size needed */
-		bbox = fz_round_rect(fz_bound_text(text, ctm));
+		bbox = fz_round_rect(fz_bound_text(dev->ctx, text, ctm));
 		bbox = fz_intersect_bbox(bbox, state->scissor);
 	}
 	else
@@ -670,7 +670,7 @@ fz_draw_clip_stroke_text(fz_device *devp, fz_text *text, fz_stroke_state *stroke
 	fz_colorspace *model = state->dest->colorspace;
 
 	/* make the mask the exact size needed */
-	bbox = fz_round_rect(fz_bound_text(text, ctm));
+	bbox = fz_round_rect(fz_bound_text(dev->ctx, text, ctm));
 	bbox = fz_intersect_bbox(bbox, state->scissor);
 
 	mask = fz_new_pixmap_with_rect(dev->ctx, NULL, bbox);
diff --git a/fitz/dev_bbox.c b/fitz/dev_bbox.c
index 766adefa..9165407f 100644
--- a/fitz/dev_bbox.c
+++ b/fitz/dev_bbox.c
@@ -25,7 +25,7 @@ fz_bbox_fill_text(fz_device *dev, fz_text *text, fz_matrix ctm,
 	fz_colorspace *colorspace, float *color, float alpha)
 {
 	fz_bbox *result = dev->user;
-	fz_bbox bbox = fz_round_rect(fz_bound_text(text, ctm));
+	fz_bbox bbox = fz_round_rect(fz_bound_text(dev->ctx, text, ctm));
 	*result = fz_union_bbox(*result, bbox);
 }
 
@@ -34,7 +34,7 @@ fz_bbox_stroke_text(fz_device *dev, fz_text *text, fz_stroke_state *stroke, fz_m
 	fz_colorspace *colorspace, float *color, float alpha)
 {
 	fz_bbox *result = dev->user;
-	fz_bbox bbox = fz_round_rect(fz_bound_text(text, ctm));
+	fz_bbox bbox = fz_round_rect(fz_bound_text(dev->ctx, text, ctm));
 	*result = fz_union_bbox(*result, bbox);
 }
 
diff --git a/fitz/dev_list.c b/fitz/dev_list.c
index b77cb9e4..c1822227 100644
--- a/fitz/dev_list.c
+++ b/fitz/dev_list.c
@@ -320,7 +320,7 @@ fz_list_fill_text(fz_device *dev, fz_text *text, fz_matrix ctm,
 	node = fz_new_display_node(ctx, FZ_CMD_FILL_TEXT, ctm, colorspace, color, alpha);
 	fz_try(ctx)
 	{
-		node->rect = fz_bound_text(text, ctm);
+		node->rect = fz_bound_text(dev->ctx, text, ctm);
 		node->item.text = fz_clone_text(dev->ctx, text);
 	}
 	fz_catch(ctx)
@@ -341,7 +341,7 @@ fz_list_stroke_text(fz_device *dev, fz_text *text, fz_stroke_state *stroke, fz_m
 	node->item.text = NULL;
 	fz_try(ctx)
 	{
-		node->rect = fz_bound_text(text, ctm);
+		node->rect = fz_bound_text(dev->ctx, text, ctm);
 		node->item.text = fz_clone_text(dev->ctx, text);
 		node->stroke = fz_clone_stroke_state(dev->ctx, stroke);
 	}
@@ -361,7 +361,7 @@ fz_list_clip_text(fz_device *dev, fz_text *text, fz_matrix ctm, int accumulate)
 	node = fz_new_display_node(ctx, FZ_CMD_CLIP_TEXT, ctm, NULL, NULL, 0);
 	fz_try(ctx)
 	{
-		node->rect = fz_bound_text(text, ctm);
+		node->rect = fz_bound_text(dev->ctx, text, ctm);
 		node->item.text = fz_clone_text(dev->ctx, text);
 		node->flag = accumulate;
 		/* when accumulating, be conservative about culling */
@@ -384,7 +384,7 @@ fz_list_clip_stroke_text(fz_device *dev, fz_text *text, fz_stroke_state *stroke,
 	node = fz_new_display_node(ctx, FZ_CMD_CLIP_STROKE_TEXT, ctm, NULL, NULL, 0);
 	fz_try(ctx)
 	{
-		node->rect = fz_bound_text(text, ctm);
+		node->rect = fz_bound_text(dev->ctx, text, ctm);
 		node->item.text = fz_clone_text(dev->ctx, text);
 		node->stroke = fz_clone_stroke_state(dev->ctx, stroke);
 	}
@@ -404,7 +404,7 @@ fz_list_ignore_text(fz_device *dev, fz_text *text, fz_matrix ctm)
 	node = fz_new_display_node(ctx, FZ_CMD_IGNORE_TEXT, ctm, NULL, NULL, 0);
 	fz_try(ctx)
 	{
-		node->rect = fz_bound_text(text, ctm);
+		node->rect = fz_bound_text(dev->ctx, text, ctm);
 		node->item.text = fz_clone_text(dev->ctx, text);
 	}
 	fz_catch(ctx)
diff --git a/fitz/fitz.h b/fitz/fitz.h
index 73017e2f..ddaa6fe3 100644
--- a/fitz/fitz.h
+++ b/fitz/fitz.h
@@ -1042,11 +1042,10 @@ struct fz_font_s
 	void (*t3run)(void *xref, fz_obj *resources, fz_buffer *contents,
 		struct fz_device_s *dev, fz_matrix ctm);
 
-	fz_rect bbox;
+	fz_rect bbox;	/* font bbox is used only for t3 fonts */
 
-	/* substitute metrics */
 	int width_count;
-	int *width_table;
+	int *width_table; /* substitute metrics */
 };
 
 void fz_new_font_context(fz_context *ctx);
@@ -1061,7 +1060,9 @@ fz_font *fz_keep_font(fz_font *font);
 void fz_drop_font(fz_context *ctx, fz_font *font);
 
 void fz_debug_font(fz_font *font);
+
 void fz_set_font_bbox(fz_font *font, float xmin, float ymin, float xmax, float ymax);
+fz_rect fz_bound_glyph(fz_context *ctx, fz_font *font, int gid, fz_matrix trm);
 
 /*
  * Vector path buffer.
@@ -1158,9 +1159,9 @@ struct fz_text_s
 fz_text *fz_new_text(fz_context *ctx, fz_font *face, fz_matrix trm, int wmode);
 void fz_add_text(fz_context *ctx, fz_text *text, int gid, int ucs, float x, float y);
 void fz_free_text(fz_context *ctx, fz_text *text);
-void fz_debug_text(fz_text*, int indent);
-fz_rect fz_bound_text(fz_text *text, fz_matrix ctm);
+fz_rect fz_bound_text(fz_context *ctx, fz_text *text, fz_matrix ctm);
 fz_text *fz_clone_text(fz_context *ctx, fz_text *old);
+void fz_debug_text(fz_text*, int indent);
 
 /*
  * The shading code uses gouraud shaded triangle meshes.
diff --git a/fitz/res_font.c b/fitz/res_font.c
index 053b2a04..b224e7ac 100644
--- a/fitz/res_font.c
+++ b/fitz/res_font.c
@@ -38,8 +38,8 @@ fz_new_font(fz_context *ctx, char *name)
 
 	font->bbox.x0 = 0;
 	font->bbox.y0 = 0;
-	font->bbox.x1 = 1000;
-	font->bbox.y1 = 1000;
+	font->bbox.x1 = 1;
+	font->bbox.y1 = 1;
 
 	font->width_count = 0;
 	font->width_table = NULL;
@@ -216,10 +216,10 @@ fz_new_font_from_file(fz_context *ctx, char *path, int index)
 
 	font = fz_new_font(ctx, face->family_name);
 	font->ft_face = face;
-	font->bbox.x0 = face->bbox.xMin * 1000 / face->units_per_EM;
-	font->bbox.y0 = face->bbox.yMin * 1000 / face->units_per_EM;
-	font->bbox.x1 = face->bbox.xMax * 1000 / face->units_per_EM;
-	font->bbox.y1 = face->bbox.yMax * 1000 / face->units_per_EM;
+	font->bbox.x0 = (float) face->bbox.xMin / face->units_per_EM;
+	font->bbox.y0 = (float) face->bbox.yMin / face->units_per_EM;
+	font->bbox.x1 = (float) face->bbox.xMax / face->units_per_EM;
+	font->bbox.y1 = (float) face->bbox.yMax / face->units_per_EM;
 
 	return font;
 }
@@ -242,10 +242,10 @@ fz_new_font_from_memory(fz_context *ctx, unsigned char *data, int len, int index
 
 	font = fz_new_font(ctx, face->family_name);
 	font->ft_face = face;
-	font->bbox.x0 = face->bbox.xMin * 1000 / face->units_per_EM;
-	font->bbox.y0 = face->bbox.yMin * 1000 / face->units_per_EM;
-	font->bbox.x1 = face->bbox.xMax * 1000 / face->units_per_EM;
-	font->bbox.y1 = face->bbox.yMax * 1000 / face->units_per_EM;
+	font->bbox.x0 = (float) face->bbox.xMin / face->units_per_EM;
+	font->bbox.y0 = (float) face->bbox.yMin / face->units_per_EM;
+	font->bbox.x1 = (float) face->bbox.xMax / face->units_per_EM;
+	font->bbox.y1 = (float) face->bbox.yMax / face->units_per_EM;
 
 	return font;
 }
@@ -254,7 +254,7 @@ static fz_matrix
 fz_adjust_ft_glyph_width(fz_context *ctx, fz_font *font, int gid, fz_matrix trm)
 {
 	/* Fudge the font matrix to stretch the glyph if we've substituted the font. */
-	if (font->ft_substitute && gid < font->width_count)
+	if (font->ft_substitute && font->width_table && gid < font->width_count)
 	{
 		FT_Error fterr;
 		int subw;
@@ -504,6 +504,68 @@ fz_render_ft_stroked_glyph(fz_context *ctx, fz_font *font, int gid, fz_matrix tr
 	return pixmap;
 }
 
+static fz_rect
+fz_bound_ft_glyph(fz_context *ctx, fz_font *font, int gid, fz_matrix trm)
+{
+	FT_Face face = font->ft_face;
+	FT_Error fterr;
+	FT_BBox cbox;
+	FT_Matrix m;
+	FT_Vector v;
+	fz_rect bounds;
+
+	// TODO: stroke state
+	// TODO: refactor loading into fz_load_ft_glyph
+	// TODO: cache results
+
+	trm = fz_adjust_ft_glyph_width(ctx, font, gid, trm);
+
+	if (font->ft_italic)
+		trm = fz_concat(fz_shear(0.3f, 0), trm);
+
+	m.xx = trm.a * 64; /* should be 65536 */
+	m.yx = trm.b * 64;
+	m.xy = trm.c * 64;
+	m.yy = trm.d * 64;
+	v.x = trm.e * 64;
+	v.y = trm.f * 64;
+
+	fterr = FT_Set_Char_Size(face, 65536, 65536, 72, 72); /* should be 64, 64 */
+	if (fterr)
+		fz_warn(ctx, "freetype setting character size: %s", ft_error_string(fterr));
+	FT_Set_Transform(face, &m, &v);
+
+	fterr = FT_Load_Glyph(face, gid, FT_LOAD_NO_BITMAP | FT_LOAD_NO_HINTING);
+	if (fterr)
+	{
+		fz_warn(ctx, "freetype load glyph (gid %d): %s", gid, ft_error_string(fterr));
+		bounds.x0 = bounds.x1 = trm.e;
+		bounds.y0 = bounds.y1 = trm.f;
+		return bounds;
+	}
+
+	if (font->ft_bold)
+	{
+		float strength = fz_matrix_expansion(trm) * 0.04f;
+		FT_Outline_Embolden(&face->glyph->outline, strength * 64);
+		FT_Outline_Translate(&face->glyph->outline, -strength * 32, -strength * 32);
+	}
+
+	FT_Outline_Get_CBox(&face->glyph->outline, &cbox);
+	bounds.x0 = cbox.xMin / 64.0f;
+	bounds.y0 = cbox.yMin / 64.0f;
+	bounds.x1 = cbox.xMax / 64.0f;
+	bounds.y1 = cbox.yMax / 64.0f;
+
+	if (fz_is_empty_rect(bounds))
+	{
+		bounds.x0 = bounds.x1 = trm.e;
+		bounds.y0 = bounds.y1 = trm.f;
+	}
+
+	return bounds;
+}
+
 /*
  * Type 3 fonts...
  */
@@ -528,6 +590,13 @@ fz_new_type3_font(fz_context *ctx, char *name, fz_matrix matrix)
 	return font;
 }
 
+static fz_rect
+fz_bound_t3_glyph(fz_context *ctx, fz_font *font, int gid, fz_matrix trm)
+{
+	trm = fz_concat(font->t3matrix, trm);
+	return fz_transform_rect(trm, font->bbox);
+}
+
 fz_pixmap *
 fz_render_t3_glyph(fz_context *ctx, fz_font *font, int gid, fz_matrix trm, fz_colorspace *model)
 {
@@ -613,11 +682,21 @@ fz_debug_font(fz_font *font)
 		printf("\ttype3 matrix [%g %g %g %g]\n",
 			font->t3matrix.a, font->t3matrix.b,
 			font->t3matrix.c, font->t3matrix.d);
-	}
 
-	printf("\tbbox [%g %g %g %g]\n",
-		font->bbox.x0, font->bbox.y0,
-		font->bbox.x1, font->bbox.y1);
+		printf("\ttype3 bbox [%g %g %g %g]\n",
+			font->bbox.x0, font->bbox.y0,
+			font->bbox.x1, font->bbox.y1);
+	}
 
 	printf("}\n");
 }
+
+fz_rect
+fz_bound_glyph(fz_context *ctx, fz_font *font, int gid, fz_matrix trm)
+{
+	if (font->ft_face)
+		return fz_bound_ft_glyph(ctx, font, gid, trm);
+	if (font->t3procs)
+		return fz_bound_t3_glyph(ctx, font, gid, trm);
+	return fz_empty_rect;
+}
diff --git a/fitz/res_text.c b/fitz/res_text.c
index fae2606b..79a07e86 100644
--- a/fitz/res_text.c
+++ b/fitz/res_text.c
@@ -53,57 +53,44 @@ fz_clone_text(fz_context *ctx, fz_text *old)
 }
 
 fz_rect
-fz_bound_text(fz_text *text, fz_matrix ctm)
+fz_bound_text(fz_context *ctx, fz_text *text, fz_matrix ctm)
 {
-	fz_matrix trm;
+	fz_matrix tm, trm;
 	fz_rect bbox;
-	fz_rect fbox;
+	fz_rect gbox;
 	int i;
 
 	if (text->len == 0)
 		return fz_empty_rect;
 
-	/* find bbox of glyph origins in ctm space */
+	tm = text->trm;
 
-	bbox.x0 = bbox.x1 = text->items[0].x;
-	bbox.y0 = bbox.y1 = text->items[0].y;
+	tm.e = text->items[0].x;
+	tm.f = text->items[0].y;
+	trm = fz_concat(tm, ctm);
+	bbox = fz_bound_glyph(ctx, text->font, text->items[0].gid, trm);
 
 	for (i = 1; i < text->len; i++)
 	{
-		bbox.x0 = MIN(bbox.x0, text->items[i].x);
-		bbox.y0 = MIN(bbox.y0, text->items[i].y);
-		bbox.x1 = MAX(bbox.x1, text->items[i].x);
-		bbox.y1 = MAX(bbox.y1, text->items[i].y);
+		if (text->items[i].gid >= 0)
+		{
+			tm.e = text->items[i].x;
+			tm.f = text->items[i].y;
+			trm = fz_concat(tm, ctm);
+			gbox = fz_bound_glyph(ctx, text->font, text->items[i].gid, trm);
+
+			bbox.x0 = MIN(bbox.x0, gbox.x0);
+			bbox.y0 = MIN(bbox.y0, gbox.y0);
+			bbox.x1 = MAX(bbox.x1, gbox.x1);
+			bbox.y1 = MAX(bbox.y1, gbox.y1);
+		}
 	}
 
-	bbox = fz_transform_rect(ctm, bbox);
-
-	/* find bbox of font in trm * ctm space */
-
-	trm = fz_concat(text->trm, ctm);
-	trm.e = 0;
-	trm.f = 0;
-
-	fbox.x0 = text->font->bbox.x0 * 0.001f;
-	fbox.y0 = text->font->bbox.y0 * 0.001f;
-	fbox.x1 = text->font->bbox.x1 * 0.001f;
-	fbox.y1 = text->font->bbox.y1 * 0.001f;
-
-	fbox = fz_transform_rect(trm, fbox);
-
-	/* expand glyph origin bbox by font bbox */
-
-	bbox.x0 += fbox.x0;
-	bbox.y0 += fbox.y0;
-	bbox.x1 += fbox.x1;
-	bbox.y1 += fbox.y1;
-
-	/* add some fuzz at the edges, as font bbox is often not accurate.
-	 * Better to localise this fuzz in just this one place than have it
-	 * everywhere that looks at these results. If we ever change to using
-	 * freetype to exactly bound text, we can remove this. */
-	bbox.x0 -= 20; bbox.y0 -= 20;
-	bbox.x1 += 20; bbox.y1 += 20;
+	/* Compensate for the glyph cache limited positioning precision */
+	bbox.x0 -= 1;
+	bbox.y0 -= 1;
+	bbox.x1 += 1;
+	bbox.y1 += 1;
 
 	return bbox;
 }
diff --git a/pdf/pdf_font.c b/pdf/pdf_font.c
index 16223be7..76d4632a 100644
--- a/pdf/pdf_font.c
+++ b/pdf/pdf_font.c
@@ -1031,25 +1031,31 @@ pdf_load_font_descriptor(pdf_font_desc *fontdesc, pdf_xref *xref, fz_obj *dict,
 	}
 }
 
-static void
-pdf_make_width_table(fz_context *ctx, pdf_font_desc *fontdesc)
+static int
+pdf_count_font_glyphs(fz_context *ctx, pdf_font_desc *fontdesc)
 {
-	fz_font *font = fontdesc->font;
-	int i, k, cid, gid;
-
-	font->width_count = 0;
+	int i, k, n, cid, gid;
+	n = 0;
 	for (i = 0; i < fontdesc->hmtx_len; i++)
 	{
 		for (k = fontdesc->hmtx[i].lo; k <= fontdesc->hmtx[i].hi; k++)
 		{
 			cid = pdf_lookup_cmap(fontdesc->encoding, k);
 			gid = pdf_font_cid_to_gid(fontdesc, cid);
-			if (gid > font->width_count)
-				font->width_count = gid;
+			if (gid > n)
+				n = gid;
 		}
 	}
-	font->width_count ++;
+	return n + 1;
+}
+
+static void
+pdf_make_width_table(fz_context *ctx, pdf_font_desc *fontdesc)
+{
+	fz_font *font = fontdesc->font;
+	int i, k, cid, gid;
 
+	font->width_count = pdf_count_font_glyphs(ctx, fontdesc);
 	font->width_table = fz_malloc_array(ctx, font->width_count, sizeof(int));
 	fontdesc->size += font->width_count * sizeof(int);
 
diff --git a/pdf/pdf_interpret.c b/pdf/pdf_interpret.c
index 2172eb89..c8b65dd6 100644
--- a/pdf/pdf_interpret.c
+++ b/pdf/pdf_interpret.c
@@ -565,7 +565,7 @@ pdf_flush_text(pdf_csi *csi)
 
 	fz_try(ctx)
 	{
-		bbox = fz_bound_text(text, gstate->ctm);
+		bbox = fz_bound_text(ctx, text, gstate->ctm);
 
 		pdf_begin_group(csi, bbox);
 
diff --git a/pdf/pdf_type3.c b/pdf/pdf_type3.c
index 730a2754..98ebe934 100644
--- a/pdf/pdf_type3.c
+++ b/pdf/pdf_type3.c
@@ -40,6 +40,10 @@ pdf_load_type3_font(pdf_xref *xref, fz_obj *rdb, fz_obj *dict)
 
 		obj = fz_dict_gets(dict, "FontBBox");
 		bbox = pdf_to_rect(ctx, obj);
+		bbox.x0 *= 0.001;
+		bbox.y0 *= 0.001;
+		bbox.x1 *= 0.001;
+		bbox.y1 *= 0.001;
 
 		fontdesc->font = fz_new_type3_font(ctx, buf, matrix);
 		fontdesc->size += sizeof(fz_font) + 256 * (sizeof(fz_buffer*) + sizeof(float));
diff --git a/xps/xps_glyphs.c b/xps/xps_glyphs.c
index 030ca0e5..51cdaa77 100644
--- a/xps/xps_glyphs.c
+++ b/xps/xps_glyphs.c
@@ -544,7 +544,7 @@ xps_parse_glyphs(xps_document *doc, fz_matrix ctm,
 			fz_atof(origin_x_att), fz_atof(origin_y_att),
 			is_sideways, bidi_level, indices_att, unicode_att);
 
-	area = fz_bound_text(text, ctm);
+	area = fz_bound_text(doc->ctx, text, ctm);
 
 	xps_begin_opacity(doc, ctm, area, opacity_mask_uri, dict, opacity_att, opacity_mask_tag);