Add the ability to extract metrics from ttfs

Summary:
Upon switching over from ttf metrics to TeX metrics, we lost metrics for a
couple of the characters that are dynamically generated by TeX. Thus TeX doesn't
have metrics for them, but our fonts do have the characters because the MathJax
scripts also dynamically build them. This adds the ability to extract metrics
from the generated font files so that we can use the generated characters
correctly. A better solution would be to dynamically generate the characters
ourselves, but that is much harder, and will be left to a future time.

Test Plan:
 - Make sure typing "\neq \cong \text{ }" produces no warnings in the console.
 - Make sure huxley screenshots look the same

Reviewers: alpert

Reviewed By: alpert

Differential Revision: http://phabricator.khanacademy.org/D13107

Makefile

@@ -31,7 +31,7 @@ test:
 	./node_modules/.bin/jasmine-node test/katex-spec.js
 
 metrics:
-	cd metrics && ./mapping.pl | ./extract_tfms.py | ./replace_line.py
+	cd metrics && ./mapping.pl | ./extract_tfms.py | ./extract_ttfs.py | ./replace_line.py
 
 clean:
 	rm -rf build/*

metrics/README.md

@@ -3,12 +3,15 @@
 
 There are several requirements for generating the metrics used by KaTeX.
 
-- You need to have an installation of TeX which supports kpathsea (you can check
-this by running `tex --version`, and seeing if it has a line that looks like >
-kpathsea version 6.2.0
+- You need to have an installation of TeX which supports kpathsea. You can check
+  this by running `tex --version`, and seeing if it has a line that looks like
+  > kpathsea version 6.2.0
 
 - You need the JSON module for perl. You can install this either from CPAN or with
-your package manager.
+  your package manager.
+
+- You need the python fontforge module. This is probably either installed with
+  fontforge or can be installed from your package manager.
 
 Once you have these things, run
 

metrics/extract_ttfs.py

@@ -0,0 +1,47 @@
+#!/usr/bin/env python
+
+import fontforge
+import sys
+import json
+
+metrics_to_extract = {
+    "Main-Regular": [
+        u"\u2260",
+        u"\u2245",
+        u"\u0020",
+        u"\u00a0",
+    ]
+}
+
+
+def main():
+    start_json = json.load(sys.stdin)
+
+    for font, chars in metrics_to_extract.iteritems():
+        fontInfo = fontforge.open("../static/fonts/KaTeX_" + font + ".ttf")
+
+        for glyph in fontInfo.glyphs():
+            try:
+                char = unichr(glyph.unicode)
+            except ValueError:
+                continue
+
+            if char in chars:
+                _, depth, _, height = glyph.boundingBox()
+
+                depth = -depth
+
+                # TODO(emily): Figure out a real way to calculate this
+                italic = 0
+
+                start_json[font][ord(char)] = {
+                    height: height / fontInfo.em,
+                    depth: depth / fontInfo.em,
+                    italic: italic / fontInfo.em,
+                }
+
+    sys.stdout.write(
+        json.dumps(start_json, separators=(',', ':'), sort_keys=True))
+
+if __name__ == "__main__":
+    main()