adding unicode friendly clue parsing

src/Global.-1.ttslua

@@ -232,6 +232,76 @@ analytics =
   sessions  = {}
 }
 
+----------[ Character sets ]----------
+
+digits_table = {
+[0]={48,1632,1776,1984,2406,2534,2662,2790,2918,3046,3174,3302,3430,3558,3664,3792,3872,4160,4240,6112,6160,6470,6608,6784,6800,6992,7088,7232,7248,42528,43216,43264,43472,43504,43600,44016,65296},
+[1]={49,1633,1777,1985,2407,2535,2663,2791,2919,3047,3175,3303,3431,3559,3665,3793,3873,4161,4241,6113,6161,6471,6609,6785,6801,6993,7089,7233,7249,42529,43217,43265,43473,43505,43601,44017,65297},
+[2]={50,1634,1778,1986,2408,2536,2664,2792,2920,3048,3176,3304,3432,3560,3666,3794,3874,4162,4242,6114,6162,6472,6610,6786,6802,6994,7090,7234,7250,42530,43218,43266,43474,43506,43602,44018,65298},
+[3]={51,1635,1779,1987,2409,2537,2665,2793,2921,3049,3177,3305,3433,3561,3667,3795,3875,4163,4243,6115,6163,6473,6611,6787,6803,6995,7091,7235,7251,42531,43219,43267,43475,43507,43603,44019,65299},
+[4]={52,1636,1780,1988,2410,2538,2666,2794,2922,3050,3178,3306,3434,3562,3668,3796,3876,4164,4244,6116,6164,6474,6612,6788,6804,6996,7092,7236,7252,42532,43220,43268,43476,43508,43604,44020,65300},
+[5]={53,1637,1781,1989,2411,2539,2667,2795,2923,3051,3179,3307,3435,3563,3669,3797,3877,4165,4245,6117,6165,6475,6613,6789,6805,6997,7093,7237,7253,42533,43221,43269,43477,43509,43605,44021,65301},
+[6]={54,1638,1782,1990,2412,2540,2668,2796,2924,3052,3180,3308,3436,3564,3670,3798,3878,4166,4246,6118,6166,6476,6614,6790,6806,6998,7094,7238,7254,42534,43222,43270,43478,43510,43606,44022,65302},
+[7]={55,1639,1783,1991,2413,2541,2669,2797,2925,3053,3181,3309,3437,3565,3671,3799,3879,4167,4247,6119,6167,6477,6615,6791,6807,6999,7095,7239,7255,42535,43223,43271,43479,43511,43607,44023,65303},
+[8]={56,1640,1784,1992,2414,2542,2670,2798,2926,3054,3182,3310,3438,3566,3672,3800,3880,4168,4248,6120,6168,6478,6616,6792,6808,7000,7096,7240,7256,42536,43224,43272,43480,43512,43608,44024,65304},
+[9]={57,1641,1785,1993,2415,2543,2671,2799,2927,3055,3183,3311,3439,3567,3673,3801,3881,4169,4249,6121,6169,6479,6617,6793,6809,7001,7097,7241,7257,42537,43225,43273,43481,43513,43609,44025,65305}
+}
+
+spaceCharacters = {}
+for _, code in pairs({32,160,5760,6158,8192,8193,8194,8195,8196,8197,8198,8199,8200,8201,8202,8203,8239,8287,12288,65279}) do
+  spaceCharacters[code] = true
+end
+
+hyphenCharacters = {}
+for _, code in pairs({45,8208,8209,8210,8211,8212,8213,11834,11835,65112,65123,65293}) do
+  hyphenCharacters[code] = true
+end
+
+digitCharacters = {}
+for digit, arr in pairs(digits_table) do
+  for _, code in pairs(arr) do
+    digitCharacters[code] = tostring(digit) -- strings are interpreted as 'true' for boolean expressions, even empty ones
+  end
+end
+
+whitespaceCharacters = {}
+for _, code in pairs({9,10,11,12,13,28,29,30,31,133,8232,8233}) do
+  whitespaceCharacters[code] = true
+end
+for code, _ in pairs(spaceCharacters) do
+  whitespaceCharacters[code] = true
+end
+
+illegalCharacters   = {}
+-- put several character ranges into the illegalCharacters table
+local illegalCharacterRanges = 
+          {
+            {0, 64},    -- before uppercase letters
+            {91, 96},   -- before lowercase letters
+            {123, 191}, -- before accented letters
+            {215, 215}, {247, 247}, -- multiplication and division sign
+            {448, 451}, -- symbols
+            {688, 767}, -- symbols
+          }
+for _, range in pairs(illegalCharacterRanges) do
+  local code_st, code_end = table.unpack(range)
+  for code = code_st, code_end do 
+    illegalCharacters[code] = true
+  end
+end                                                 
+-- include hyphens
+for code, _ in pairs(hyphenCharacters) do
+  illegalCharacters[code] = true
+end
+-- include whitespace characters
+for code, _ in pairs(whitespaceCharacters) do
+  illegalCharacters[code] = true
+end
+-- include digit characters
+for code, _ in pairs(digitCharacters) do
+  illegalCharacters[code] = true
+end
+
 function onload(saveState)
 
   -- Codenames script version
@@ -263,6 +333,9 @@ function onload(saveState)
   redToken.interactable       = false
   blueToken.interactable      = false
 
+  -- Create variable for storing typed clues
+  currentEnteredClue = {}
+
   -- Get the list of decks
   api_getDecks()
 
@@ -498,8 +571,25 @@ function searchDecks(searchTerm)
 end
 
 function clueEntered(player, value)
-  if value:match("\n") then
-    local color = player.color
+  -- Because of how MoonSharp handles its matching expressions,
+  -- absolutely no expression matching from the string library can be used for unicode clues.
+  -- It converts all characters to values 0..255 before running the expression on it.
+  -- eg. %s will match all characters with codes 0x0A, 0x20, 0x10A, 0x400A, 0x7020
+  -- string.lower upper find etc. work fine and properly even for other character sets
+
+  local color = player.color
+  local newLineInd = string.find(value, "\n")
+  if newLineInd == nil then
+    -- Save the clue for when the user presses enter
+    currentEnteredClue[color] = value
+  else
+    -- Get the clue from before the user pressed enter
+    if currentEnteredClue[color] then
+      value = currentEnteredClue[color]
+      currentEnteredClue[color] = nil
+    else
+      value = ""
+    end
 
     -- Reset the text box
     local resetInput = {
@@ -508,6 +598,11 @@ function clueEntered(player, value)
     }
     UI.setAttributes(color:lower() .. "ClueText", resetInput)
 
+    --if the clue is empty then do nothing
+    if #value == 0 then
+      return
+    end
+
     -- if the game hasn't been started, a clue cannot be entered
     if gameState.status ~= 1 then
       Player[color].broadcast("[a020f0]» [da1918]ERROR: [ffffff]You must start a game to enter a clue! [a020f0]«")
@@ -520,8 +615,11 @@ function clueEntered(player, value)
       return
     end
 
-    -- Remove the newline, trim the clue, and convert to lowercase
-    value = value:gsub("\n", ""):match("%s*(.-)%s*$"):lower()
+    -- if the clue is long, dont bother processing it
+    if #value > 50 then
+      Player[color].broadcast("[a020f0]» [da1918]ERROR: [ffffff]Invalid clue. Please enter a valid clue and push ENTER! [a020f0]«")
+      return
+    end
 
     -- Parse the entered clue into its respective parts
     local clue, number, error = getClueDetails(value)
@@ -541,6 +639,9 @@ function clueEntered(player, value)
       return
     end
 
+    -- Standardize clue to lowercase
+    clue = clue:lower()
+
     -- Don't allow a clue that isn't covered
     for cardIndex, cardData in ipairs(cards) do
       if not cardData.covered and cardData.value:lower() == clue:lower() then
@@ -552,10 +653,10 @@ function clueEntered(player, value)
     -- Track remaining clues
     if number == "inf" then
       gameState.guessesLeft = -1
-    elseif number == "0" then
+    elseif tonumber(number) == 0 then
       gameState.guessesLeft = -1
     else
-      gameState.guessesLeft = tostring(number) + 1
+      gameState.guessesLeft = tonumber(number) + 1
     end
 
     -- Encode the finished clue
@@ -564,99 +665,143 @@ function clueEntered(player, value)
     -- Enable voting for the current team
     gameState.canVote = true
 
-    -- Send analytics data for the a new clue
+    -- Send analytics data for the new clue
     api_newClue(clue, (number == "inf" and -1 or number), Player[color].steam_id)
   end
 end
 
 function getClueDetails(processedClue)
-  -- How many hyphens are there?
-  local clue, number
-  local _, hyphenCount = string.gsub(processedClue, "%-", "")
-  local _, spaceCount = string.gsub(processedClue, "%s", "")
-
-  if hyphenCount == 0 then
-    -- Single word with space (or no space) as delimiter
-    if spaceCount > 1 then
-      return nil, nil, true
-    end
-
-    local checks = {
-      "^(%a+)(%s*)(%d+)$",
-      "^(%a+)(%s+)(inf)$"
-    }
-
-    for _, check in ipairs(checks) do
-      local status, clue, _, number = pcall(function() return string.match(processedClue, check) end)
-      if status then
-        -- Parsing successful - check for nil values just in case
-        if clue != nil and number != nil then
-          -- Return the clue and number
-          return clue, number, false
-        end
+  local clueState = {}
+  clueState.PRE_WHITESPACE = 1
+  clueState.INF_N = 2
+  clueState.INF_I = 3
+  clueState.NUMBER = 4
+  clueState.INF_WHITESPACE = 5 -- inf must have 1 space before clue
+  clueState.PRE_CLUE_WHITESPACE = 6 -- this state allows for 1 hyphen
+  clueState.CLUE = 7 -- the clue allows 1 hyphen
+  clueState.CLUE_ON_HYPHEN = 8
+  clueState.POST_WHITESPACE = 9
+
+  local clue = ""
+  local number = ""
+
+  local invalid = false
+  local state = clueState.PRE_WHITESPACE
+  local hyphenCount = 0
+  -- process clue backwards as it is easier
+  for ind = #processedClue, 1, -1 do
+    -- This state machine will detect an invalid clue and stop processing if it
+    -- finds an invalid character before it reaches the beginning of the input
+    local ch = string.sub(processedClue,ind,ind)
+    local code = string.unicode(ch)
+    if state == clueState.PRE_WHITESPACE then
+      if string.lower(ch) == "f" then
+        state = clueState.INF_N
+        number = "inf"
+      elseif digitCharacters[code] then
+        state = clueState.NUMBER
+        number = digitCharacters[code]..number
+      elseif whitespaceCharacters[code] then
+        -- continue
+      else
+        invalid = true
       end
-    end
-
-    -- No valid clues detected
-    return nil, nil, true
-
-  elseif hyphenCount == 1 then
-    -- Either a hypenated word with a space (or no space) as delimiter
-    -- or a single word with a hyphen (and possibly spaces) as delimiter
-    if spaceCount > 2 then
-      return nil, nil, true
-    end
-
-    local checks = {
-      "^(%a+%-%a+)(%s*)(%d+)$",
-      "^(%a+%-%a+)(%s+)(inf)$",
-      "^(%a+)(%s*%-%s*)(%d+)$",
-      "^(%a+)(%s*%-%s*)(inf)$"
-    }
-
-    for _, check in ipairs(checks) do
-      local status, clue, _, number = pcall(function() return string.match(processedClue, check) end)
-      if status then
-        -- Parsing successful - check for nil values just in case
-        if clue != nil and number != nil then
-          -- Return the clue and number
-          return clue, number, false
+    elseif state == clueState.INF_N then
+      if string.lower(ch) == "n" then
+        state = clueState.INF_I
+      else
+        invalid = true
+      end
+    elseif state == clueState.INF_I then
+      if string.lower(ch) == "i" then
+        state = clueState.INF_WHITESPACE
+      else
+        invalid = true
+      end
+    elseif state == clueState.NUMBER then
+      if digitCharacters[code] then
+        number = digitCharacters[code]..number
+      elseif whitespaceCharacters[code] then
+        state = clueState.PRE_CLUE_WHITESPACE
+      elseif hyphenCharacters[code] then
+        state = clueState.PRE_CLUE_WHITESPACE
+        hyphenCount = hyphenCount + 1
+      elseif illegalCharacters[code] then
+        invalid = true
+      else
+        state = clueState.CLUE
+        hyphenCount = 0
+        clue = ch..clue
+      end
+    elseif state == clueState.INF_WHITESPACE then
+      if whitespaceCharacters[code] then
+        state = clueState.PRE_CLUE_WHITESPACE
+      elseif hyphenCharacters[code] then
+        state = clueState.PRE_CLUE_WHITESPACE
+        hyphenCount = hyphenCount + 1
+      else
+        invalid = true
+      end
+    elseif state == clueState.PRE_CLUE_WHITESPACE then
+      if whitespaceCharacters[code] then
+        -- continue
+      elseif hyphenCharacters[code] and hyphenCount < 1 then
+        hyphenCount = hyphenCount + 1
+      elseif illegalCharacters[code] then
+        invalid = true
+      else
+        state = clueState.CLUE
+        hyphenCount = 0
+        clue = ch..clue
+      end
+    elseif state == clueState.CLUE or state == clueState.CLUE_ON_HYPHEN then
+      if whitespaceCharacters[code] then
+        if state == clueState.CLUE_ON_HYPHEN then
+          invalid = true
+        end
+        state = clueState.POST_WHITESPACE
+      elseif hyphenCharacters[code] then
+        state = clueState.CLUE_ON_HYPHEN
+        hyphenCount = hyphenCount + 1
+        if hyphenCount > 1 then
+          invalid = true
         end
+        clue = ch..clue
+      elseif illegalCharacters[code] then
+        invalid = true
+      else
+        state = clueState.CLUE
+        clue = ch..clue
+      end
+    elseif state == clueState.POST_WHITESPACE then
+      if whitespaceCharacters[code] then
+        -- continue
+      else
+        invalid = true
       end
+    else
+      invalid = true -- we should never reach here
     end
 
-    -- No valid clues detected
-    return nil, nil, true
-
-  elseif hyphenCount == 2 then
-
-    if spaceCount > 2 then
+    if invalid then
+      -- This is an invalid clue
+      -- print("c:", clue, " n:", number, " i:", ind, " ch:", ch, " code:", code, " state:", state)
       return nil, nil, true
     end
+  end
 
-    local checks = {
-      "^(%a+%-%a+)(%s*%-%s*)(%d+)$",
-      "^(%a+%-%a+)(%s*%-%s*)(inf)$"
-    }
-
-    for _, check in ipairs(checks) do
-      local status, clue, _, number = pcall(function() return string.match(processedClue, check) end)
-      if status then
-        -- Parsing successful - check for nil values just in case
-        if clue != nil and number != nil then
-          -- Return the clue and number
-          return clue, number, false
-        end
-      end
-    end
-
-    -- No valid clues detected
+  if not (state == clueState.CLUE or state == clueState.POST_WHITESPACE) or clue == "" then
+    -- We either ended on a hyphen or did not receive a parsable clue
     return nil, nil, true
+  end
 
-  else
-    -- Clue has too many hyphens
-    return nil, nil, true
+  -- Clean number value
+  if number != "inf" then
+    number = tostring(tonumber(number))
   end
+
+  -- Return the clue and number
+  return clue, number, false
 end
 
 function rotateclues()
@@ -1621,7 +1766,7 @@ function processChat(message)
     end
   end
 
-  return command, args:gsub("^%s*(.-)%s*$", "%1")
+  return command, args:gsub("^%s*(.*%S+)%s*$", "%1"):gsub("^%s+$","")
 end
 
 function toggleTurns()