Serialize/load trie data as JSON

.gitignore

@@ -1,5 +1,5 @@
 data.json
-data.trie
+trie.json
 .DS_Store
 node_modules/
 *.js

generate.coffee

@@ -4,7 +4,7 @@ UnicodeTrieBuilder = require 'unicode-trie/builder'
 
 log2 = Math.log2 or (n) ->
   Math.log(n) / Math.LN2
-  
+
 bits = (n) ->
   (log2(n) + 1) | 0
 
@@ -18,7 +18,7 @@ scriptCount = 0
 eawCount = 0
 
 for codePoint in codePoints when codePoint?
-  categories[codePoint.category] ?= categoryCount++  
+  categories[codePoint.category] ?= categoryCount++
   combiningClasses[codePoint.combiningClassName] ?= combiningClassCount++
   scripts[codePoint.script] ?= scriptCount++
   eaws[codePoint.eastAsianWidth] ?= eawCount++
@@ -58,27 +58,31 @@ numericValue = (numeric) ->
         while (mant % 60) is 0
           mant /= 60
           ++exp
-      
+
         ((mant + 0xbf) << 2) + (exp - 1)
   else
     0
 
 trie = new UnicodeTrieBuilder
-for codePoint in codePoints when codePoint?  
+for codePoint in codePoints when codePoint?
   category = categories[codePoint.category]
   combiningClass = combiningClasses[codePoint.combiningClassName] or 0
   script = scripts[codePoint.script] or 0
   eaw = eaws[codePoint.eastAsianWidth] or 0
-  
+
   val = (category << categoryShift) |
         (combiningClass << combiningShift) |
         (script << scriptShift) |
         (eaw << eawShift) |
         numericValue(codePoint.numeric)
-  
+
   trie.set codePoint.code, val
 
-fs.writeFileSync 'data.trie', trie.toBuffer()
+
+# Trie is serialized suboptimally as JSON so it can be loaded via require,
+# allowing unicode-properties to work in the browser
+fs.writeFileSync 'trie.json', JSON.stringify trie.toBuffer()
+
 fs.writeFileSync 'data.json', JSON.stringify
   categories: Object.keys(categories)
   combiningClasses: Object.keys(combiningClasses)

index.coffee

@@ -1,11 +1,15 @@
 UnicodeTrie = require 'unicode-trie'
 data = require './data.json'
-fs = require 'fs'
-trie = new UnicodeTrie fs.readFileSync __dirname + '/data.trie'
+
+# Trie is serialized as a Buffer in node, but here
+# we may be running in a browser so we make an Uint8Array
+trieBuffer = require './trie.json'
+trieData = new Uint8Array trieBuffer.data
+trie = new UnicodeTrie trieData
 
 log2 = Math.log2 or (n) ->
   Math.log(n) / Math.LN2
-  
+
 bits = (n) ->
   (log2(n) + 1) | 0
 
@@ -31,23 +35,23 @@ NUMBER_MASK = (1 << NUMBER_BITS) - 1
 exports.getCategory = (codePoint) ->
   val = trie.get codePoint
   data.categories[(val >> CATEGORY_SHIFT) & CATEGORY_MASK]
-  
+
 exports.getCombiningClass = (codePoint) ->
   val = trie.get codePoint
   data.combiningClasses[(val >> COMBINING_SHIFT) & COMBINING_MASK]
-  
+
 exports.getScript = (codePoint) ->
   val = trie.get codePoint
   data.scripts[(val >> SCRIPT_SHIFT) & SCRIPT_MASK]
-  
+
 exports.getEastAsianWidth = (codePoint) ->
   val = trie.get codePoint
   data.eaw[(val >> EAW_SHIFT) & EAW_MASK]
-  
+
 exports.getNumericValue = (codePoint) ->
   val = trie.get codePoint
   num = val & NUMBER_MASK
-    
+
   if num is 0
     return null
   else if num <= 50
@@ -61,46 +65,46 @@ exports.getNumericValue = (codePoint) ->
     # base 10
     val = (num >> 5) - 14
     exp = (num & 0x1f) + 2
-    
+
     while exp > 0
       val *= 10
       exp--
-      
+
     return val
   else
     # base 60
     val = (num >> 2) - 0xbf
     exp = (num & 3) + 1
-    
+
     while exp > 0
       val *= 60
       exp--
-    
+
     return val
-  
+
 exports.isAlphabetic = (codePoint) ->
   exports.getCategory(codePoint) in ['Lu', 'Ll', 'Lt', 'Lm', 'Lo', 'Nl']
-  
+
 exports.isDigit = (codePoint) ->
   exports.getCategory(codePoint) is 'Nd'
-  
+
 exports.isPunctuation = (codePoint) ->
   exports.getCategory(codePoint) in ['Pc', 'Pd', 'Pe', 'Pf', 'Pi', 'Po', 'Ps']
-  
+
 exports.isLowerCase = (codePoint) ->
   exports.getCategory(codePoint) is 'Ll'
-  
+
 exports.isUpperCase = (codePoint) ->
   exports.getCategory(codePoint) is 'Lu'
-  
+
 exports.isTitleCase = (codePoint) ->
   exports.getCategory(codePoint) is 'Lt'
-  
+
 exports.isWhiteSpace = (codePoint) ->
   exports.getCategory(codePoint) in ['Zs', 'Zl', 'Zp']
-    
+
 exports.isBaseForm = (codePoint) ->
   exports.getCategory(codePoint) in ['Nd', 'No', 'Nl', 'Lu', 'Ll', 'Lt', 'Lm', 'Lo', 'Me', 'Mc']
-  
+
 exports.isMark = (codePoint) ->
   exports.getCategory(codePoint) in ['Mn', 'Me', 'Mc']