Bigelow & Holmes have prepared new versions of the Go Fonts. The version
numbers have changed from 2.004 to 2.008. ChangeLog:
* New! 2 new fonts, bringing the total to 12: Go Smallcaps and Go
Smallcaps Italic.
* New! U+FFFD REPLACEMENT CHARACTER glyph added.
* U+00C5 LATIN CAPITAL LETTER A WITH RING ABOVE ring tweaked.
* U+00F0 LATIN SMALL LETTER ETH glyph lightened.
* Superfluous U+0005 <control> glyph removed.
* Split integral, chart draw, block, shade and related glyphs now span
the full height and have fewer seams when tiled.
* Raised most capital letter diacritics and some lowercase diacritics.
* Mono character widths fixed to be uniform.
* Underline position adjusted.
* SFNT glyph order and cmap table optimized.
* SFNT glyph names (post table) regularized, consistent with the AGLFN.
* ttfautohint upgraded to the latest version, from 1.5 to 1.6.
Change-Id: I0985f07d58b984010199dd723ccdf5892ee59c78
Reviewed-on: https://go-review.googlesource.com/38870
Reviewed-by: Rob Pike <r@golang.org>
Recent commits (probably 792d36e1 "support TrueType compound glyphs" and
1995ed1a "support TrueType glyph transformations") let us decode all the
glyphs in the Adobe TrueType fonts, but I forgot to update the test
coverage in those commits.
Change-Id: I7b135005cd67082f84aee16d296debdedcc9488e
Reviewed-on: https://go-review.googlesource.com/38281
Reviewed-by: David Crawshaw <crawshaw@golang.org>
This will lessen the diff in a subsequent commit, which introduces a
call stack.
Change-Id: I478857675f47b53e6bd3ee270087a00b9d1d4d34
Reviewed-on: https://go-review.googlesource.com/38279
Reviewed-by: Nigel Tao <nigeltao@golang.org>
The format is relatively complicated but rarely seen. It doesn't seem
worth the effort. Ignoring it is what Microsoft Windows and FreeType do.
Change-Id: I4415bd591c832650066de1d4e035b20a7230a4da
Reviewed-on: https://go-review.googlesource.com/38273
Reviewed-by: David Crawshaw <crawshaw@golang.org>
Also add tests for Apple proprietary fonts.
Change-Id: I5ce8efa2397bb01c5255d956a77c955ba1383105
Reviewed-on: https://go-review.googlesource.com/38272
Reviewed-by: David Crawshaw <crawshaw@golang.org>
This lets us load a glyph at e.g. 12 pixels per em.
Change-Id: I048b3db89af8670782953a8361afe0e6373df9b0
Reviewed-on: https://go-review.googlesource.com/37175
Reviewed-by: David Crawshaw <crawshaw@golang.org>
The latest version of that TTF file fixes the inconsistency between the
TTF and OTF versions.
Change-Id: I0b86f6ddc3f0326e538bdd18fe097be9ad8533ae
Reviewed-on: https://go-review.googlesource.com/37173
Reviewed-by: David Crawshaw <crawshaw@golang.org>
This is based on the post table in the sfnt file.
Change-Id: I11f7a9bd9024cfc8f92adc5abb4d5356521f0df7
Reviewed-on: https://go-review.googlesource.com/36972
Reviewed-by: David Crawshaw <crawshaw@golang.org>
Two recent commits ("proprietary fonts" and "cmap format 12") each
passed all of its own tests, but the combination wasn't tested until
both were submitted.
Change-Id: Ic4c2ae8deb1e4623ca5543672dc46d55bfce91a4
Reviewed-on: https://go-review.googlesource.com/36372
Reviewed-by: David Crawshaw <crawshaw@golang.org>
These are analogous to the image.Point and image.Rectangle methods in
the standard library.
Change-Id: If8997421a9aeb31d29cfe5eefb79cc481a39df82
Reviewed-on: https://go-review.googlesource.com/34753
Reviewed-by: David Crawshaw <crawshaw@golang.org>
The bounding box of a string does not necessarily contain the origin.
Prior to this commit, BoundString(etc, "x") would call grow exactly
once, with the first argument being the (empty) zero rectangle.
Change-Id: Id7d4f6c250ac0749f6dae19d11f4e97f9c6f45bc
Reviewed-on: https://go-review.googlesource.com/34674
Reviewed-by: Dave Day <djd@golang.org>
Also make all of the Buffer fields non-exported. LoadGlyph now returns
the segments instead of setting Buffer.Segments for the caller to use.
Change-Id: I3f87070da5e8f014f88dbca70b62a4cd30e3fd66
Reviewed-on: https://go-review.googlesource.com/34532
Reviewed-by: Dave Day <djd@golang.org>
Parsing the per-glyph Charstrings will be a follow-up commit.
Change-Id: I52f849a77dd7fa14b6e07420820bdfb4347c2438
Reviewed-on: https://go-review.googlesource.com/33593
Reviewed-by: Dave Day <djd@golang.org>
A face's glyph height (the height of the glyph image) isn't necessarily
the inter-line spacing: f.Ascent + f.Descent isn't always f.Height.
Similarly, a face's glyph image's left side isn't always dot.X.
Change-Id: I86594c52b8d31bf652ac34a695e9270ac635a5bc
Reviewed-on: https://go-review.googlesource.com/24481
Reviewed-by: David Crawshaw <crawshaw@golang.org>
It can be useful to measure some text without having to set up a Drawer.
Change-Id: I18b7b1fecc32ca69b7644d598ed21462e7c41edd
Reviewed-on: https://go-review.googlesource.com/21785
Reviewed-by: Andrew Gerrand <adg@golang.org>
Reviewed-by: David Crawshaw <crawshaw@golang.org>
Also remove the dependency on the image/draw package. That package will
give the right answer for arbitrary source images, including those of
type plan9Image, but doing the conversion directly avoids bouncing uint8
or color.Alpha values through the general-purpose draw.Image,
image.Image and color.Color interfaces.
It is possible to optimize this even further, but this will do for now.
benchmark old ns/op new ns/op delta
BenchmarkParseSubfont-8 2298492 492443 -78.58%
Change-Id: Iea9436bffa097a1ba0052dbabf21516bce8b61e0
Reviewed-on: https://go-review.googlesource.com/21693
Reviewed-by: Nigel Tao <nigeltao@golang.org>
image.Alpha is fast-pathed in draw.Draw, plan9Image is not.
Change-Id: I5ed8fc4d310bb5c6ec2cffdd0ba4295dae2274cf
Reviewed-on: https://go-review.googlesource.com/21453
Reviewed-by: Nigel Tao <nigeltao@golang.org>
The key feature is that its data is entirely self-contained and does not
require loading from separate files.
Change-Id: I4ef72b52cde93597be89dcfd55659f418c6cab23
Reviewed-on: https://go-review.googlesource.com/14486
Reviewed-by: Rob Pike <r@golang.org>
