complex string splitting

lib/underscore.lua

@@ -747,17 +747,39 @@ function _.tap(value, func)
   return value
 end
 
-function _.split(value)
-  if _.isString(value) then
-    local values = {}
-    string.gsub(value, "[^%s+]", function(c)
-      table.insert(values, c)
-    end)
-
-    return values
+function splitIterator(value, pattern, start)
+  if pattern then
+    return string.find(value, pattern, start)
   else
-    return {}
+    if start > string.len(value) then
+      return nil
+    else
+      return start+1, start
+    end
+  end
+end
+
+
+function _.split(value, pattern)
+  if not _.isString(value) then return {} end
+  local values = {}
+  local start = 1
+  local start_pattern, end_pattern = splitIterator(value, pattern, start)
+
+  while start_pattern do
+    table.insert(
+      values,
+      string.sub(value, start, start_pattern - 1)
+    )
+    start = end_pattern + 1
+    start_pattern, end_pattern = splitIterator(value, pattern, start)
+  end
+
+  if start <= string.len(value) then
+    table.insert(values, string.sub(value, start))
   end
+
+  return values
 end
 
 function _.chain(value)

spec/string_spec.lua

@@ -13,5 +13,22 @@ describe("string functions", function()
       assert.same(_.split(123),{})
       assert.same(_.split(function() end),{})
     end)
+
+    describe("with a specific split string", function()
+      it("can split a specific characters", function()
+        assert.same(_.split("a|b|c|d", "|"), {"a","b","c","d"})
+        assert.same(_.split("a 1 c 2", " "), {"a","1","c","2"})
+        assert.same(_.split("a\tb\tc\td", "\t"), {"a","b","c","d"})
+      end)
+
+      it("can split using Lua pattern matchers", function()
+        assert.same(_.split("a   b c     d", "%s+"), {"a","b","c","d"})
+        assert.same(_.split("a1b2c3d4e", "%d"), {"a","b","c","d","e"})
+      end)
+
+      it("can split on whole words", function()
+        assert.same(_.split("arabbitbrabbitc", "rabbit"), {"a","b","c"})
+      end)
+    end)
   end)
 end)