From b615cb5c8e638cffe77bbe5cb86c9362e2b2fc18 Mon Sep 17 00:00:00 2001
From: Bruce Hill <bruce@bruce-hill.com>
Date: Wed, 3 Oct 2018 16:14:17 -0700
Subject: [PATCH] Fixed up some edge cases with as_lua_id and from_lua_id that
 were producing bad results.

---
 string2.lua  | 25 +++++++++++++++++++++----
 string2.moon | 13 +++++++------
 2 files changed, 28 insertions(+), 10 deletions(-)

diff --git a/string2.lua b/string2.lua
index 1f15fed..ae0e49d 100644
--- a/string2.lua
+++ b/string2.lua
@@ -143,8 +143,6 @@ local string2 = {
     return '"' .. escaped .. '"'
   end,
   as_lua_id = function(str)
-    local orig = str
-    str = gsub(str, "^ *$", "%1 ")
     str = gsub(str, "x([0-9A-F][0-9A-F])", "x78%1")
     str = gsub(str, "%W", function(c)
       if c == ' ' then
@@ -159,18 +157,37 @@ local string2 = {
     return str
   end,
   from_lua_id = function(str)
-    if not (is_lua_id("^_+(.*)$")) then
+    if not (is_lua_id(str:match("^_*(.*)$"))) then
       str = str:sub(2, -1)
     end
     str = gsub(str, "_", " ")
     str = gsub(str, "x([0-9A-F][0-9A-F])", function(hex)
       return char(tonumber(hex, 16))
     end)
-    str = gsub(str, "^ ([ ]*)$", "%1")
     return str
   end
 }
 for k, v in pairs(string) do
   string2[k] = string2[k] or v
 end
+local _list_0 = {
+  "",
+  "_",
+  " ",
+  "return",
+  "asdf",
+  "one two",
+  "one_two",
+  "Hex2Dec",
+  "He-ec",
+  "\3"
+}
+for _index_0 = 1, #_list_0 do
+  local test = _list_0[_index_0]
+  local lua_id = string2.as_lua_id(test)
+  assert(is_lua_id(lua_id), "failed to convert '" .. tostring(test) .. "' to a valid Lua identifier (got '" .. tostring(lua_id) .. "')")
+  local roundtrip = string2.from_lua_id(lua_id)
+  assert(roundtrip == test, "Failed lua_id roundtrip: '" .. tostring(test) .. "' -> " .. tostring(lua_id) .. " -> " .. tostring(roundtrip))
+end
+assert(string2.as_lua_id('') == '_')
 return string2
diff --git a/string2.moon b/string2.moon
index 140c586..8d2f3f9 100644
--- a/string2.moon
+++ b/string2.moon
@@ -67,10 +67,6 @@ string2 = {
     -- but not idempotent. In logic terms: (x != y) => (as_lua_id(x) != as_lua_id(y)),
     -- but not (as_lua_id(a) == b) => (as_lua_id(b) == b).
     as_lua_id: (str)->
-        orig = str
-        -- Empty strings are not valid lua identifiers, so treat them like " ",
-        -- and treat " " as "  ", etc. to preserve injectivity.
-        str = gsub str, "^ *$", "%1 "
         -- Escape 'x' (\x78) when it precedes something that looks like an uppercase hex sequence.
         -- This way, all Lua IDs can be unambiguously reverse-engineered, but normal usage
         -- of 'x' won't produce ugly Lua IDs.
@@ -88,13 +84,18 @@ string2 = {
     -- from_lua_id(as_lua_id(str)) == str, but behavior is unspecified for inputs that
     -- did not come from as_lua_id()
     from_lua_id: (str)->
-        unless is_lua_id("^_+(.*)$")
+        unless is_lua_id(str\match("^_*(.*)$"))
             str = str\sub(2,-1)
         str = gsub(str, "_", " ")
         str = gsub(str, "x([0-9A-F][0-9A-F])", (hex)-> char(tonumber(hex, 16)))
-        str = gsub(str, "^ ([ ]*)$", "%1")
         return str
 }
 for k,v in pairs(string) do string2[k] or= v
 
+for test in *{"", "_", " ", "return", "asdf", "one two", "one_two", "Hex2Dec", "He-ec", "\3"}
+    lua_id = string2.as_lua_id(test)
+    assert is_lua_id(lua_id), "failed to convert '#{test}' to a valid Lua identifier (got '#{lua_id}')"
+    roundtrip = string2.from_lua_id(lua_id)
+    assert roundtrip == test, "Failed lua_id roundtrip: '#{test}' -> #{lua_id} -> #{roundtrip}"
+
 return string2