YAMLToArray

constant kMultiLineModeNone = 0
constant kMultiLineModeLiteral = 1
constant kMultiLineModeFolded = 2

command YAMLToArray pYaml
   local tInDocument = true
   local tPath
   local tArray
   local tPathLists
   local tReferences
   local tMultiLineMode
   local tCharNo

   put kMultiLineModeNone into tMultiLineMode

   local tMultiLine
   repeat for each line tLine in pYaml
      -- ignore directives
      if tLine begins with "%" then
         next repeat
      end if

      if tLine begins with "---" then
         put true into tInDocument
         next repeat
      end if

      if not tInDocument then
         next repeat
      end if

      if tLine begins with "..." then
         if tMultiLineMode is not kMultiLineModeNone then
            __ClearQuotes tMultiLine
            put tMultiLine into tArray[tPath]
            put kMultiLineModeNone into tMultiLineMode
         end if
         put false into tInDocument
         next repeat
      end if

      local tPathElement, tListItem
      put __PathElement(tLine, tListItem, tPathLists) into tPathElement

      -- comment lines
      put __unquotedCharOffset("#", tLine) into tCharNo
      if tCharNo > 0 then
         delete char tCharNo to -1 of tLine
      end if

      -- remain in multiline until indent is lower
      if tMultiLineMode is not kMultiLineModeNone then
         if tPathElement < the number of elements of tPath + 1 and tLine is not empty then
            __ClearQuotes tMultiLine
            put tMultiLine into tArray[tPath]
            put kMultiLineModeNone into tMultiLineMode
         else
            -- add back any stripped literal indents
            if tPathElement > the number of elements of tPath + 1 then
               repeat for tPathElement - (the number of elements of tPath + 1)
                  put "    " before tLine
               end repeat
            end if

            -- empty lines and indented lines are literal even in folded mode
            if tMultiLineMode is kMultiLineModeLiteral then
               if tMultiLine is not empty then
                  put return after tMultiLine
               end if
               put tLine after tMultiLine
            else
               if tLine is empty then
                  put return after tMultiLine
               else if char 1 of tLine is space or \
                     the last line of tMultiLine is empty or \
                     the last line of tMultiLine begins with space then
                  if tMultiLine is not empty then
                     put return after tMultiLine
                  end if
                  put tLine after tMultiLine
               else
                  if tMultiLine is not empty then
                     put space after tMultiLine
                  end if
                  put tLine after tMultiLine
               end if
            end if

            next repeat
         end if
      end if

      -- empty lines
      if tLine is empty then
         next repeat
      end if

      local tElements
      put the number of elements of tPath into tElements
      if tPathLists[tPathElement] and \
            tListItem and \
            tPath[tPathElement] is an integer then
         add 1 to tPath[tPathElement]
         repeat with tIndex = tElements down to tPathElement + 1
            delete variable tPath[tIndex]
            delete variable tPathLists[tIndex]
         end repeat
         add 1 to tPathElement
      else
         if tPathElement < tElements then
            repeat with tIndex = tElements down to tPathElement + 1
               delete variable tPath[tIndex]
               delete variable tPathLists[tIndex]
            end repeat
         else if tPathElement > tElements + 1 then
            return "invalid YAML" for error
         end if

         put tListItem into tPathLists[tPathElement]
         if tListItem then
            put 1 into tPath[tPathElement]
            add 1 to tPathElement
         end if
      end if

      -- handle line with just list marker
      if the number of words of tLine is 0 then
         next repeat
      end if

      -- map
      put __unquotedCharOffset(":", tLine) into tCharNo
      if tCharNo > 0 then
         local tKey

         put char 1 to tCharNo-1 of tLine into tKey
         put char tCharNo+1 to -1 of tLine into tLine
         __ClearQuotes tKey

         if tKey is empty then
            return "invalid YAML" for error
         end if

         put false into tPathLists[tPathElement]
         put tKey into tPath[tPathElement]

         -- clean whitespace
         put word 1 to -1 of tLine into tLine
         
         -- referenced element
         local tRef
         if tLine begins with "&" then
            put word 1 of tLine into tRef
            put "*" into char 1 of tRef
            delete word 1 of tLine
         else
            put empty into tRef
         end if

         -- check for referenced element
         if word 1 of tLine is among the keys of tReferences then
            put tArray[tReferences[word 1 of tLine]] into tLine
         end if

         -- store referenced element
         if tRef is not empty then
            put tPath into tReferences[tRef]
         end if

         -- ignore explicit typing
         if tLine begins with "!" then
            delete word 1 of tLine
         end if

         if tLine is "|" then
            put kMultiLineModeLiteral into tMultiLineMode
            put empty into tMultiLine
         else if tLine is ">" then
            put kMultiLineModeFolded into tMultiLineMode
            put empty into tMultiLine
         else
            __AddToArray tLine, tArray[tPath]
         end if
      else
         __AddToArray tLine, tArray[tPath]
      end if
   end repeat

   if tMultiLineMode is not kMultiLineModeNone then
      __ClearQuotes tMultiLine
      put tMultiLine into tArray[tPath]
   end if

   return tArray for value
end YAMLToArray


private function __unquotedCharOffset pChar, pLine
  local tOffset = 0
  local tCharOffset
  local tInQuote = "false"

  repeat for each char tChar in pLine
    add 1 to tOffset
    if tChar is quote then
      put not tInQuote into tInQuote
    else if tChar is pChar and not tInQuote then
      return tOffset
    end if
  end repeat

  return 0
end __unquotedCharOffset

private command __AddToArray pLine, @xElement
   local tElements, tKey
   -- trim whitespace
   if pLine is not an array then
      put word 1 to -1 of pLine into pLine
      if pLine begins with "[" and pLine ends with "]" then
         -- support single line flow sequence
         put char 2 to -2 of pLine into pLine
         split pLine by comma
         put the number of elements of pLine into tElements
         repeat with tIndex = 1 to tElements
            put word 1 to -1 of pLine[tIndex] into pLine[tIndex]
            __ClearQuotes pLine[tIndex]
         end repeat
      else if pLine begins with "{" and pLine ends with "}" then
         -- support single line flow sequence
         put char 2 to -2 of pLine into pLine
         split pLine by comma
         set the itemDelimiter to ":"
         
         local tLine
         put the number of elements of pLine into tElements
         repeat with tIndex = 1 to tElements
            put word 1 to -1 of pLine[tIndex] into pLine[tIndex]
            
            local tValue
            put item 1 of pLine[tIndex] into tKey
            __ClearQuotes tKey
            put item 2 to -1 of pLine[tIndex] into tValue
            __ClearQuotes tValue
            
            put tValue into tLine[tKey]
         end repeat
         put tLine into pLine
      else
         __ClearQuotes pLine
      end if
   end if
   
   put pLine into xElement
end __AddToArray

private command __ClearQuotes @xLine
   if xLine begins with quote and xLine ends with quote then
      put format( char 2 to -2 of xLine) into xLine
   else if xLine begins with "'" and xLine ends with "'" then
      put char 2 to -2 of xLine into xLine
   end if
end __ClearQuotes

private function __PathElement @xLine, @rListItem, pPathLists
   local tPathElement = 1
   
   put false into rListItem
   repeat
      if xLine begins with "  " then
         add 1 to tPathElement
      else if xLine begins with "- " then
         -- list
         put true into rListItem
      else
         return tPathElement
      end if
      delete char 1 to 2 of xLine
   end repeat
end __PathElement