pretty/pretty.lua

local TABLE_TYPE_SEQUENCE    = 'SEQUENCE'
local TABLE_TYPE_PURE_MAP    = 'PURE MAP'
local TABLE_TYPE_MIXED       = 'MIXED TABLE'
local TABLE_TYPE_EMPTY       = 'EMPTY TABLE'

local SINGLE_LINE_SEQ_MAX_ELEMENTS = 10
local SINGLE_LINE_MAP_MAX_ELEMENTS = 5
local NR_CHARS_IN_LONG_STRING      = 40

local TYPE_SORT_ORDER = {
    ['nil']      = 0,
    ['boolean']  = 1,
    ['number']   = 2,
    ['string']   = 3,
    ['table']    = 4,
    ['userdata'] = 5,
    ['thread']   = 6,
    ['function'] = 7,
}

local RESERVED_LUA_WORDS = {
    ['and']      = true,
    ['break']    = true,
    ['do']       = true,
    ['else']     = true,
    ['elseif']   = true,
    ['end']      = true,
    ['false']    = true,
    ['for']      = true,
    ['function'] = true,
    ['if']       = true,
    ['in']       = true,
    ['local']    = true,
    ['nil']      = true,
    ['not']      = true,
    ['or']       = true,
    ['repeat']   = true,
    ['return']   = true,
    ['then']     = true,
    ['true']     = true,
    ['until']    = true,
    ['while']    = true,
}

local CHAR_TO_STR_REPR = {}

do
    for i = 00, 031 do  CHAR_TO_STR_REPR[i] = '\\0'..(i < 10 and '0' or '')..i  end
    for i = 32, 255 do  CHAR_TO_STR_REPR[i] = string.char(i)  end
    CHAR_TO_STR_REPR[7]   = '\\a'
    CHAR_TO_STR_REPR[8]   = '\\b'
    CHAR_TO_STR_REPR[9]   = '\t'
    CHAR_TO_STR_REPR[10]  = '\n'
    CHAR_TO_STR_REPR[11]  = '\\v'
    CHAR_TO_STR_REPR[12]  = '\\f'
    CHAR_TO_STR_REPR[13]  = '\\r'
    CHAR_TO_STR_REPR[92]  = '\\\\'
    CHAR_TO_STR_REPR[127] = '\\127'
end

--------------------------------------------------------------------------------
-- Util

local function padnum(d)
    local dec, n = string.match(d, "(%.?)0*(.+)")
    return #dec > 0 and ("%.12f"):format(d) or ("%s%03d%s"):format(dec, #n, n)
end

local function alphanum_compare_strings (a, b)
    return tostring(a):gsub("%.?%d+", padnum)..("%3d"):format(#b)
         < tostring(b):gsub("%.?%d+", padnum)..("%3d"):format(#a)
end

local function count_occurances_of_substring_in_string (str, substr)
    local _, count = string.gsub(str, substr, '')
    return count
end

local function smallest_secure_longform_string_level (str)
    -- Determines the level a longform string needs to use, to avoid "code"
    -- injection. For example, if we want to use longform on the string
    -- 'Hello ]] World', we cannot use level-0 as this would result in
    -- '[[Hello ]] World]]', which could be an issue in certain applications.

    local levels = { [1] = 1 }
    str:gsub('%]=*%]', function (m) levels[m:len()] = true end)
    return #levels - 1
end

local function compare_key_value_pairs (a, b)
    -- Get types
    local type_key_a,   type_key_b   = type(a[1]), type(b[1])
    local type_value_a, type_value_b = type(a[2]), type(b[2])

    -- Tons of compare
    if (type_key_a ~= 'string' or type_key_b ~= 'string') then
        return TYPE_SORT_ORDER[type_key_a] < TYPE_SORT_ORDER[type_key_b]
    elseif (type_value_a == type_value_b) then
        return alphanum_compare_strings(a[1], b[1])
    else
        return TYPE_SORT_ORDER[type_value_a] < TYPE_SORT_ORDER[type_value_b]
    end
end

local function get_key_value_pairs_in_proper_order (t)
    -- Generates a sequence of key value pairs, in proper order.
    -- Proper order is:
    -- 1. By value type: as defined by the TYPE_SORT_ORDER in the top.
    -- 2. By key type:  TODO: Implement this.
    -- 2.1.  Numbers
    -- 2.2.  Strings in alphanumeric order
    -- 2.3.  Other wierdness.

    local key_value_pairs = {}

    for key, value in pairs(t) do
        key_value_pairs[#key_value_pairs+1] = { key, value }
    end

    table.sort(key_value_pairs, compare_key_value_pairs)

    return key_value_pairs
end

local function nr_elements_in_map (t)
    local k, count = nil, -1
    repeat
        k, count = next(t, k), count + 1
    until not k
    return count
end

local function is_identifier(str)
    -- An identier is defined in the lua reference guide

    return str:match('^[_%a][_%w]*$') and not RESERVED_LUA_WORDS[str]
end

local function contains_only_nice_string_keys (t)
    -- A "nice" string is here defined is one following the rules of lua
    -- identifiers.

    for k, _ in pairs(t) do
        if type(k) ~= 'string' or not is_identifier(k) then
            return false
        end
    end
    return true
end

local function escape_string (str)
    local l = {}
    for i = 1, #str do
        l[#l+1] = CHAR_TO_STR_REPR[str:byte(i)]
    end
    return table.concat(l, '')
end

--------------------------------------------------------------------------------
-- Identifyer stuff

local SIMPLE_VALUE_TYPES = {
    ['nil']      = true,
    ['boolean']  = true,
    ['number']   = true,
    ['string']   = true,
}

local function is_empty_table (value)
    assert( type(value) == 'table', '[is_empty_table]: Only tables allowed!' )
    return next(value) == nil
end

local function is_short_table (value)
    -- In this context, a short table is either an empty table, or one with a
    -- single element.

    assert( type(value) == 'table', '[is_short_table]: Only tables allowed!' )

    local first_key  = next(value)
    return (not first_key or SIMPLE_VALUE_TYPES[type(value[first_key])])
       and (next(value, first_key) == nil)
end

local function is_simple_value (value)
    -- In this context, a simple value is a either nil, a boolean, a number,
    -- a string or a short table.
    -- TODO: Add clause about long strings. (Maybe >7 chars?)

    --if type(value) == 'table' then print(value, is_short_table(value))  end
    return SIMPLE_VALUE_TYPES[ type(value) ]
        or type(value) == 'table' and is_short_table(value)
end

local function contains_non_simple_key_or_value (t)
    for k, v in pairs(t) do
        if not is_simple_value(k) or not is_simple_value(v) then
            return true
        end
    end
    return false
end

local function get_table_type (value)
    -- Determines table type:
    -- * Sequence:  All keys are integer in the range 1..#value
    -- * Pure Map:  #value == 0
    -- * Mixed:     Any other

    if is_empty_table(value) then  return TABLE_TYPE_EMPTY  end

    local not_sequence  =  false
    local not_pure_map  =  (#value ~= 0)
    local max_index     =  #value

    -- Determine if there exist some non-index
    for k, v in pairs(value) do
        if type(k) ~= 'number' or k < 1 or max_index < k or k ~= math.floor(k) then
            not_sequence = true
            break
        end
    end

    -- Return type
    if not not_sequence then
        return TABLE_TYPE_SEQUENCE
    elseif not_pure_map then
        return TABLE_TYPE_MIXED
    else
        return TABLE_TYPE_PURE_MAP
    end
end

local function is_single_line_table (value)
    -- In this context, a single-line table, is:
    -- A)  Either a sequence or a pure map.
    -- B)  Has no non-simple keys or values
    -- C 1)  If sequence, has at most SINGLE_LINE_SEQ_MAX_ELEMENTS elements.
    -- C 2)  If map, has at most SINGLE_LINE_MAP_MAX_ELEMENTS elements.

    local table_type = get_table_type(value)

    return not contains_non_simple_key_or_value(value)
       and(table_type == TABLE_TYPE_SEQUENCE and #value <= SINGLE_LINE_SEQ_MAX_ELEMENTS
        or table_type == TABLE_TYPE_PURE_MAP and nr_elements_in_map(value) <= SINGLE_LINE_MAP_MAX_ELEMENTS)
end

--------------------------------------------------------------------------------
-- Formatting stuff

local format_table, format_value

-- Ways to format keys

local function format_key_and_value_string_map (l, key, value, options, depth)
    l[#l+1] = key
    l[#l+1] = #key
    l[#l+1] = ' = '
    l[#l+1] = format_value(value, options, depth)
end

local function format_key_and_value_arbitr_map (l, key, value, options, depth)
    l[#l+1] = '['
    l[#l+1] = format_value(key, options, 'max')
    l[#l+1] = ']'
    l[#l+1] = #l[#l-1] + 2
    l[#l+1] = ' = '
    l[#l+1] = format_value(value, options, depth)
end

-- Formatting tables

local function format_single_line_sequence (t, options)
    -- NOTE: Assumes that the input table was pre-checked with `is_single_line_table()`

    local l = {}
    for i = 1, #t do  l[i] = format_value(t[i], options)  end
    return '{ ' .. table.concat(l, ', ') .. ' }'
end

local function format_single_line_map (t, options)
    -- NOTE: Assumes that the input table was pre-checked with `is_single_line_table()`

    local pair_format_func = contains_only_nice_string_keys(t) and format_key_and_value_string_map or format_key_and_value_arbitr_map

    local key_value_pairs = get_key_value_pairs_in_proper_order(t)
    local l = {'{ '}
    for _, pair in ipairs(key_value_pairs) do
        local top_before = #l
        pair_format_func(l, pair[1], pair[2], options, 'max')
        l[#l+1] = ', '
        -- Ignore the "width of key"-shit
        for i = top_before, #l do  if type(l[i]) == 'number' then l[i] = '' end  end
    end
    if l[#l] == ', ' then  l[#l] = nil  end
    l[#l+1] = ' }'
    return table.concat(l, '')
end

local function format_sequence (t, options, depth)
    if depth ~= 'max' and depth >= options.max_depth    then  return '{...}'
    elseif is_single_line_table(t)                      then  return format_single_line_sequence(t, options)
    elseif depth == 'max'                               then  return '{...}'
    end

    local l = {'{\n'}
    for index, value in ipairs(t) do
        l[#l+1] = options.indent:rep(depth + 1)
        l[#l+1] = format_value(value, options, depth + 1)
        l[#l+1] = ',\n'
    end
    l[#l]   = '\n'
    l[#l+1] = options.indent:rep(depth)
    l[#l+1] = '}'
    return table.concat(l, '')
end

local function format_map (t, options, depth)
    if depth ~= 'max' and depth >= options.max_depth    then  return '{...}'
    elseif is_single_line_table(t)                      then  return format_single_line_map(t, options)
    elseif depth == 'max'                               then  return '{...}'
    end

    local key_value_pairs  =  get_key_value_pairs_in_proper_order(t)
    local pair_format_func  =  contains_only_nice_string_keys(t) and format_key_and_value_string_map or format_key_and_value_arbitr_map

    -- Figure out the max key length
    local l = {'{\n'}
    local top_before = #l
    for _, pair in pairs(key_value_pairs) do
        l[#l+1] = options.indent:rep(depth + 1)
        pair_format_func(l, pair[1], pair[2], options, depth + 1)
        l[#l+1] = ',\n'
    end
    -- Figure out max key len
    local max_key_len      =  0
    for i = top_before, #l do
        if type(l[i]) == 'number' and l[i] > max_key_len then
            max_key_len = l[i]
        end
    end
    -- Replace in the proper whitespace
    for i = top_before, #l do
        if type(l[i]) == 'number' then
            l[i] = string.rep(' ', max_key_len - l[i])
        end
    end

    l[#l]   = '\n'
    l[#l+1] = options.indent:rep(depth)
    l[#l+1] = '}'
    return table.concat(l, '')
end

function format_table (t, options, depth)
    local table_type = get_table_type(t)

    if     table_type == TABLE_TYPE_EMPTY    then   return '{}'
    elseif table_type == TABLE_TYPE_SEQUENCE then   return format_sequence(t, options, depth)
    else                                            return format_map(t, options, depth)
    end
end

local function format_string (str, options)
    -- TODO: Add option for escaping unicode characters.

    local is_long_string         =  (str:len() >= NR_CHARS_IN_LONG_STRING)
    local newline_or_tab_index   =  str:find('[\n\t]')
    local single_quote_index     =  str:find('\'')
    local double_quote_index     =  str:find('\"')


    local chance_of_longform     =  is_long_string and (newline_or_tab_index <= NR_CHARS_IN_LONG_STRING) or double_quote_index and single_quote_index


    local cut_string_index       =  options.cut_strings and (is_long_string or chance_of_longform)
                                                        and math.min(NR_CHARS_IN_LONG_STRING - 3, newline_or_tab_index or 1/0, double_quote_index or 1/0, single_quote_index or 1/0)

    local longform   = chance_of_longform and ((not cut_string_index) or cut_string_index < math.min(newline_or_tab_index or 1/0, double_quote_index or 1/0, single_quote_index or 1/0))

    local escape_newline_and_tab  =  not longform and newline_or_tab_index

    -- Determine string delimiters
    local left, right
    if longform then
        local level = smallest_secure_longform_string_level(str)
        left, right = '['..string.rep('=', level)..'[', ']'..string.rep('=', level)..']'
        if newline_or_tab_index then  str = '\n' .. str  end
    elseif not single_quote_index then
        left, right = '\'', '\''
    else
        left, right = '\"', '\"'
    end

    -- Cut string
    if cut_string_index then        str = str:sub(1, cut_string_index)  end
    str = escape_string(str)
    -- Escape newline and tab
    if escape_newline_and_tab then  str = str:gsub('\n', '\\n'):gsub('\t', '\\t')  end

    return left .. str .. right
end

local function format_number (value, shorthand)
    if     value ~= value then  return shorthand and  'nan' or  '0/0'
    elseif value ==  1/0  then  return shorthand and  'inf' or  '1/0'
    elseif value == -1/0  then  return shorthand and '-inf' or '-1/0'
    else                        return tostring(value)
    end
end

function format_value (value, options, depth)
    local type = type(value)

    if     type == 'table'  then  return format_table(value, options, depth or 'max')
    elseif type == 'string' then  return format_string(value, options)
    elseif type == 'number' then  return format_number(value, options.math_shorthand)
    else                          return tostring(value)
    end
end

--------------------------------------------------------------------------------

local function pretty_format (value, options)
    local options = options or {}
          options.max_depth = options.max_depth or math.huge
          options.indent    = options.indent    or '\t'
    return format_value(value, options, 0)
end

return pretty_format