static void
fz_divide_text_chars(fz_text_span **last, int n, fz_bbox bbox)
{
fz_text_span *span = *last;
int i, x;
x = span->len - n;
if (x >= 0)
for (i = 0; i < n; i++)
span->text[x + i].bbox = fz_split_bbox(bbox, i, n);
}
static void
fz_add_text_char(fz_context *ctx, fz_text_span **last, fz_font *font, float size, int wmode, int c, fz_bbox bbox)
{
fz_text_span *span = *last;
if (!span->font)
{
span->font = fz_keep_font(ctx, font);
span->size = size;
}
if ((span->font != font || span->size != size || span->wmode != wmode) && c != 32)
{
span = fz_new_text_span(ctx);
span->font = fz_keep_font(ctx, font);
span->size = size;
span->wmode = wmode;
(*last)->next = span;
*last = span;
}
switch (c)
{
case -1: /* ignore when one unicode character maps to multiple glyphs */
break;
case 0xFB00: /* ff */
fz_add_text_char_imp(ctx, span, 'f', fz_split_bbox(bbox, 0, 2));
fz_add_text_char_imp(ctx, span, 'f', fz_split_bbox(bbox, 1, 2));
break;
case 0xFB01: /* fi */
fz_add_text_char_imp(ctx, span, 'f', fz_split_bbox(bbox, 0, 2));
fz_add_text_char_imp(ctx, span, 'i', fz_split_bbox(bbox, 1, 2));
break;
case 0xFB02: /* fl */
fz_add_text_char_imp(ctx, span, 'f', fz_split_bbox(bbox, 0, 2));
fz_add_text_char_imp(ctx, span, 'l', fz_split_bbox(bbox, 1, 2));
break;
case 0xFB03: /* ffi */
fz_add_text_char_imp(ctx, span, 'f', fz_split_bbox(bbox, 0, 3));
fz_add_text_char_imp(ctx, span, 'f', fz_split_bbox(bbox, 1, 3));
fz_add_text_char_imp(ctx, span, 'i', fz_split_bbox(bbox, 2, 3));
break;
case 0xFB04: /* ffl */
fz_add_text_char_imp(ctx, span, 'f', fz_split_bbox(bbox, 0, 3));
fz_add_text_char_imp(ctx, span, 'f', fz_split_bbox(bbox, 1, 3));
fz_add_text_char_imp(ctx, span, 'l', fz_split_bbox(bbox, 2, 3));
break;
case 0xFB05: /* long st */
case 0xFB06: /* st */
fz_add_text_char_imp(ctx, span, 's', fz_split_bbox(bbox, 0, 2));
fz_add_text_char_imp(ctx, span, 't', fz_split_bbox(bbox, 1, 2));
break;
default:
fz_add_text_char_imp(ctx, span, c, bbox);
break;
}
}
static void
fz_text_extract(fz_context *ctx, fz_text_device *dev, fz_text *text, const fz_matrix *ctm, fz_text_style *style)
{
fz_point *pen = &dev->point;
fz_font *font = text->font;
FT_Face face = font->ft_face;
fz_matrix tm = text->trm;
fz_matrix trm;
float size;
float adv;
fz_rect rect;
fz_point dir, ndir;
fz_point delta, ndelta;
float dist, dot;
float ascender = 1;
float descender = 0;
int multi;
int i, j, err;
if (text->len == 0)
return;
if (font->ft_face)
{
fz_lock(ctx, FZ_LOCK_FREETYPE);
err = FT_Set_Char_Size(font->ft_face, 64, 64, 72, 72);
if (err)
fz_warn(ctx, "freetype set character size: %s", ft_error_string(err));
ascender = (float)face->ascender / face->units_per_EM;
descender = (float)face->descender / face->units_per_EM;
fz_unlock(ctx, FZ_LOCK_FREETYPE);
}
else if (font->t3procs && !fz_is_empty_rect(&font->bbox))
{
ascender = font->bbox.y1;
descender = font->bbox.y0;
}
rect = fz_empty_rect;
/* SumatraPDF: TODO: make this depend on the per-glyph displacement vector */
if (text->wmode == 0)
{
dir.x = 1;
dir.y = 0;
}
else
{
dir.x = 0;
dir.y = 1;
}
tm.e = 0;
tm.f = 0;
fz_concat(&trm, &tm, ctm);
fz_transform_vector(&dir, &trm);
dist = sqrtf(dir.x * dir.x + dir.y * dir.y);
ndir.x = dir.x / dist;
ndir.y = dir.y / dist;
size = fz_matrix_expansion(&trm);
for (i = 0; i < text->len; i++)
{
/* Calculate new pen location and delta */
tm.e = text->items[i].x;
tm.f = text->items[i].y;
fz_concat(&trm, &tm, ctm);
delta.x = pen->x - trm.e;
delta.y = pen->y - trm.f;
if (pen->x == -1 && pen->y == -1)
delta.x = delta.y = 0;
dist = sqrtf(delta.x * delta.x + delta.y * delta.y);
/* Add space and newlines based on pen movement */
if (dist > 0)
{
/* SumatraPDF: don't add spaces for large overlapping glyphs */
fz_text_span *last = &dev->cur_span;
if (last->len == 0 && dev->cur_line.len > 0)
last = &dev->cur_line.spans[dev->cur_line.len - 1];
ndelta.x = delta.x / dist;
ndelta.y = delta.y / dist;
dot = ndelta.x * ndir.x + ndelta.y * ndir.y;
/* SumatraPDF: don't merge multiple lines into one */
if (dist > size * LINE_DIST && hypotf(delta.y * ndir.x, delta.x * ndir.y) > size * 0.5f)
{
fz_flush_text_line(ctx, dev, style);
dev->lastchar = ' ';
}
else
/* SumatraPDF: use 0.95f instead of 0.9995f for slightly better results */
if (fabsf(dot) > 0.95f && dist > size * SPACE_DIST && dist < size * SPACE_MAX_DIST)
{
if (dev->lastchar != ' ' &&
/* SumatraPDF: don't add spaces before spaces or for large overlapping glyphs */
text->items[i].ucs != ' ' && !fz_maps_into_rect(trm, last->text[last->len - 1].bbox))
{
fz_rect spacerect;
spacerect.x0 = -0.2f;
spacerect.y0 = descender;
spacerect.x1 = 0;
spacerect.y1 = ascender;
fz_transform_rect(&spacerect, &trm);
fz_add_text_char(ctx, dev, style, ' ', spacerect);
dev->lastchar = ' ';
}
}
else if (dist > size * LINE_DIST)
{
fz_flush_text_line(ctx, dev, style);
dev->lastchar = ' ';
}
}
/* Calculate bounding box and new pen position based on font metrics */
if (font->ft_face)
{
FT_Fixed ftadv = 0;
int mask = FT_LOAD_NO_BITMAP | FT_LOAD_NO_HINTING | FT_LOAD_IGNORE_TRANSFORM;
/* TODO: freetype returns broken vertical metrics */
/* if (text->wmode) mask |= FT_LOAD_VERTICAL_LAYOUT; */
fz_lock(ctx, FZ_LOCK_FREETYPE);
err = FT_Set_Char_Size(font->ft_face, 64, 64, 72, 72);
if (err)
fz_warn(ctx, "freetype set character size: %s", ft_error_string(err));
FT_Get_Advance(font->ft_face, text->items[i].gid, mask, &ftadv);
adv = ftadv / 65536.0f;
fz_unlock(ctx, FZ_LOCK_FREETYPE);
rect.x0 = 0;
rect.y0 = descender;
rect.x1 = adv;
rect.y1 = ascender;
}
/* SumatraPDF: TODO: this check might no longer be needed */
else if (text->items[i].gid < 256)
{
adv = font->t3widths[text->items[i].gid];
rect.x0 = 0;
rect.y0 = descender;
rect.x1 = adv;
rect.y1 = ascender;
}
fz_transform_rect(&rect, &trm);
/* cf. http://code.google.com/p/sumatrapdf/issues/detail?id=1839 */
if (font->ft_face)
{
fz_rect bbox;
fz_bound_glyph(ctx, font, text->items[i].gid, &trm, &bbox);
rect.y0 = fz_min(rect.y0, bbox.y0);
rect.y1 = fz_max(rect.y1, bbox.y1);
}
pen->x = trm.e + dir.x * adv;
pen->y = trm.f + dir.y * adv;
/* Check for one glyph to many char mapping */
for (j = i + 1; j < text->len; j++)
if (text->items[j].gid >= 0)
break;
multi = j - i;
if (multi == 1)
{
fz_add_text_char(ctx, dev, style, text->items[i].ucs, rect);
}
else
{
for (j = 0; j < multi; j++)
{
fz_rect part = fz_split_bbox(rect, j, multi);
fz_add_text_char(ctx, dev, style, text->items[i + j].ucs, part);
}
i += j - 1;
}
dev->lastchar = text->items[i].ucs;
}
}
static void
fz_add_text_char(fz_context *ctx, fz_text_device *dev, fz_text_style *style, int c, fz_rect bbox)
{
switch (c)
{
case -1: /* ignore when one unicode character maps to multiple glyphs */
break;
case 0xFB00: /* ff */
fz_add_text_char_imp(ctx, dev, style, 'f', fz_split_bbox(bbox, 0, 2));
fz_add_text_char_imp(ctx, dev, style, 'f', fz_split_bbox(bbox, 1, 2));
break;
case 0xFB01: /* fi */
fz_add_text_char_imp(ctx, dev, style, 'f', fz_split_bbox(bbox, 0, 2));
fz_add_text_char_imp(ctx, dev, style, 'i', fz_split_bbox(bbox, 1, 2));
break;
case 0xFB02: /* fl */
fz_add_text_char_imp(ctx, dev, style, 'f', fz_split_bbox(bbox, 0, 2));
fz_add_text_char_imp(ctx, dev, style, 'l', fz_split_bbox(bbox, 1, 2));
break;
case 0xFB03: /* ffi */
fz_add_text_char_imp(ctx, dev, style, 'f', fz_split_bbox(bbox, 0, 3));
fz_add_text_char_imp(ctx, dev, style, 'f', fz_split_bbox(bbox, 1, 3));
fz_add_text_char_imp(ctx, dev, style, 'i', fz_split_bbox(bbox, 2, 3));
break;
case 0xFB04: /* ffl */
fz_add_text_char_imp(ctx, dev, style, 'f', fz_split_bbox(bbox, 0, 3));
fz_add_text_char_imp(ctx, dev, style, 'f', fz_split_bbox(bbox, 1, 3));
fz_add_text_char_imp(ctx, dev, style, 'l', fz_split_bbox(bbox, 2, 3));
break;
case 0xFB05: /* long st */
case 0xFB06: /* st */
fz_add_text_char_imp(ctx, dev, style, 's', fz_split_bbox(bbox, 0, 2));
fz_add_text_char_imp(ctx, dev, style, 't', fz_split_bbox(bbox, 1, 2));
break;
default:
fz_add_text_char_imp(ctx, dev, style, c, bbox);
break;
}
}