font-engineering
A collection of font engineering utilities
APACHE-2.0 License
Font engineering tools
This repository contains a number of smaller scripts and libraries that I find useful in my font engineering work and don't want to keep to myself.
shape
This is a simple wrapper for hb-shape and hb-view, but I use it all the
time - like, hundreds of times a day. You will need to either be using
iTerm2 for your terminal and install the
shell integration
tools so that it can display images inline, or Kovid Goyal's TTY,
kitty. Once you've done that, use
shape <fontfile> <text> to display both the shaped glyph string and the
output image in your termal. You can also pass any command line arguments that
make sense to hb-shape or hb-view: -V to get a shaping trace,
--features=... and so on:
Top tip: if you're using zsh for your shell, add this to your .zshrc and you will be able to tab-complete font files:
compdef '_files -g "*.?tf"' shape
gimmesometext
I often need sample text in lots of different scripts and languages. I want to be able to say "give me a bunch of Tamil strings", and this does just that:
$ gimmesometext --script=Taml # ISO 15924 script code
$ gimmesometext --script=tml2 # OpenType script code
$ gimmesometext ta # ISO 639-1 language code
$ gimmesometext --script=Cyrl az # Filter by script and language
$ gimmesometext --script=Arab az # Filter by script and language
interrofont
General purpose tool for investigating OTF/TTF information from the command line. See http://www.corvelsoftware.co.uk/software/interrofont/ for usage.
otfsurgeon
General purpose tool for slicing and dicing OTF/TTF binary table data. Like
dumptable but more so:
otfsurgeon -i f.ttf dump --ttx name # Dumps name table as XML to stdout
otfsurgeon -i f.ttf dump name > name.bin # Dumps name table as binary to stdout
otfsurgeon -i f2.ttf add name < name.bin # Adds the name table from stdin
otfsurgeon -i f.ttf strip Debg DSIG # Removes the Debg and DSIG tables from the font
otfsurgeon -i f.ttf steal f2.ttf DSIG # Copies the DSIG table from f2.ttf
fontshell
Sometimes you need to poke at a font in its fontTools representation. I got very tired of typing
>>> from fontTools.ttLib import TTFont
>>> font = TTFont("whatever.otf")
every single time, so I wrote fontshell. Run fontshell whatever.otf and you are dropped into a Python REPL with the font variable set to a fontTools.ttLib.ttFont.TTFont object. Bonus: If you have IPython installed, you get tab completion of methods and pretty colours too!
dumptable
(Superceded by otfsurgeon)
This extracts a table from a sfnt file and outputs it as a raw binary string. I use this when debugging glyph naming issues and other times TTX gets too clever and doesn't tell me everything I need to know about a table. It's rare that you'll need it, but if you need it, you'll really need it.
glyphs2svg.py
Converts a .glyphs file to a directory of .svg files.
compare_shape.py
I was converting a font from one format to another, and wanted to make sure that the layout rules were equivalent between the two. This script takes two fonts and a list of words, and checks that the shaping output from Harfbuzz is equal, outputting a report of passed and failed tests, as well as any shaping differences.
shape-diff.py
So, compare_shape.py told you that a test failed and there was a difference between the shaping outputs of two fonts, but it didn't tell you why that happened. That's what shape-diff.py does. Give it two fonts and a text, and it'll report what went wrong:
$ python3 shape-diff.py NNU.ttf NNU2.ttf 'نسس'
font1⮯ ⮮font2
✔ GSUB( 5/20) Noon=0|Seen=1|SeenFin=2
✔ GSUB( 4/19) Noon=0|SeenMed=1|SeenFin=2
✔ GSUB( 3/15) BehxIni=0|OneDotAboveNS=0|SeenMed=1|SeenFin=2
✔ GSUB(26/85) BehxIni.outT2=0|OneDotAboveNS=0|SeenMed.inT2outT2=1|SeenFin=2
✔ GSUB(27/86) BehxIni.outT2tall=0|OneDotAboveNS=0|SeenMed.inT2outT2=1|SeenFin=2
✔ GSUB(28/87) BehxIni.outD1=0|OneDotAboveNS=0|SeenMed.SWinAoutT2=1|SeenFin=2
First difference appeared at GSUB lookup 158 (font1) / 37 (font2)
font1 BehxIni.outD1=0|sp0=0|sp0=0|OneDotAboveNS=0|SeenMed.SWinAoutT2=1|SeenFin=2
font2 BehxIni.outD1=0|OneDotAboveNS=0|SeenMed.SWinAoutT2=1|SeenFin=2
vharfbuzz
vharfbuzz has graduated to its own repository.
stringbrewer
StringBrewer has graduated to its own repository.
fontbakery-shaping.py
Superseded by fontbakery check-profile fontbakery.profiles.shaping
gnipahs.py
Superseded by fontbakery check-profile fontbakery.profiles.shaping
glypholympics
When composing test strings and writing OpenType rules, I often find myself wanting to know "What are the tallest glyphs in the font?" "What are the widest glyphs in the font?" And I haven't got time to open a font editor. I just want the answers.
$ ./glypholympics Hind-Regular.otf
Widest : dvK_S_P_LA(2082) dvmI.a31(2118) dvmI.a32(2169)
Narrowest : NULL(0) CR(0) space(0)
[Horizontal ink]
Fattest : itfLogo(1648) dvL_K_YA(1670) dvK_S_P_LA(2038)
Thinnest : NULL(0) CR(0) dvmU(0)
[Horizontal advance]
Tallest : dvCandrabindu(1003) dvmEcandra_Anusvara(1003) dvmOcandra_Anusvara(1003)
Shortest : dvNukta(-75) dvVirama(-68) cedilla(-51)
[yMax]
Highest : zerowidthnonjoiner(1041) zerowidthjoiner(1046) radical(1096)
Lowest : NULL(0) CR(0) space(0)
[Vertical ink]
Deepest : dvmvRR(-440) dvmvLL(-437) dvNG_R(-373)
Shallowest: dvCandrabindu(739) dvmEcandra_Anusvara(739) dvAnusvara.amI(834)
[yMin]
find-nested-components.py
ftxvalidator tells you if the font has deeply nested components (TTF components which call other TTF components). But it doesn't tell you which glyphs have the problem. This finds nested components in a TrueType font over a given depth:
$ python3 find-nested-components.py --depth=3 MyFont-Regular.ttf
Atilde has depth 3: Atilde -> tildecomb.cap -> tilde -> tildecomb
Ohungarumlaut has depth 4: Ohungarumlaut -> uni030B.cap -> uni030B -> acute -> acutecomb
...
kernmaker.py
This is a simple Python script for turning a wordlist into a list of kerning test words. Here is some sample output, given the /usr/share/dict/words wordlist which comes with Mac OS:
$ python3 kernmaker.py /usr/share/dict/words
...
q Aquarius Equisetaceae Squamata Yquem qasida quinaquina subquality nonacquiescence subheadquarters qere equidominant earthquake Iraqian triquetric alquifou umquhile unquiet qoph xyloquinone Umpqua zaqqum miqra torques sesquitertial outqueen querying bouquet cawquaw superexquisiteness triptyque q
...
As you can see, it tries to find words in the wordlist which contain Aq, qA, Bq, qB, ... cq, qc, dq, qd, and so on. You can split this into left-side and right-side pairs with the -s option:
$ python3 kernmaker.py /usr/share/dict/words
...
q+ qasida qere qintar qoph zaqqum miqra unsqueezed q+
+q Aquila Equisetum Squamipennes Yquem Guaque subquintuple acquisible headquarters equalist earthquaking bisiliquous pulque kumquat unquestionedness alloquy Umpqua hyperquadric squatina outquarters tuque cawquaw exquisitism triptyque +q
...
A key point about it is that it doesn't assume anything about the alphabet, but treats every distinctly occurring character in your wordlist as a potential letter to be kerned; this means it should work for e.g. African languages, words containing punctuation, and so on.
lookup-size.py
A tool for identifying large lookups in GSUB and GPOS tables, reporting them in order of total byte size and the location in the feature file where they were defined.
GSUB table
==========
90 = sources/build/features.fea:11214:5 (DotAvoidance_reverse_top) 183960
92 = sources/build/features.fea:25591:5 (DotAvoidance_top) 62362
56 = sources/build/features.fea:499:5 (connections) 54352
...