pretty/pretty.lua

-- Ensure loading library, if it exists, no matter where pretty.lua was loaded from.

-- Load the library component
local LIBRARY
do
    local thispath = ... and select('1', ...):match('.+%.') or ''
    local was_loaded, library  =  pcall(require, thispath..'library')
    LIBRARY  =  was_loaded and library or {}
end
--

local ERROR_UNKNOWN_TYPE = [[
[pretty]: Attempting to format unsupported value of type "%s".
    A native formatting of the value is: %s
    This is a bug, and should be reported.
]]

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

local SINGLE_LINE_TABLE_TYPES = {
    [TABLE_TYPE_SEQUENCE]   = true,
    [TABLE_TYPE_PURE_MAP]   = true,
    [TABLE_TYPE_STRING_MAP] = true,
}

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 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 contains_only_nice_number_indexes (t)
    -- A "nice" index is here defined as one which would be visited when using
    -- ipairs: An integer larger than 1 and less than #t

    local max_index  =  #t
    for k, v in pairs(t) do
        if type(k) ~= 'number' or k < 1 or max_index < k or k ~= math.floor(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

local function get_function_info (f)
    -- NOTE: Functions best in LuaJIT or Lua 5.2+

    -- Regarding get-info:
    -- *  No need to includ 'f'. Function is already known
    -- *  No need to include 'L' (active lines) option. Ignored
    -- *  No need to include 'n' (name and namewhat). Won't work.
    local info = debug.getinfo(f, 'Su')
          info.params = {}
          info.ups = {}
          info.env = debug.getfenv and debug.getfenv(f)
          info.builtin = info.source == '=[C]'
    for i = 1, info.nparams or 0 do  info.params[i] = debug.getlocal(f, i)  end
    if info.isvararg or not info.nparams then  info.params[#info.params+1] = '...'  end
    -- Get upvalues
    for i = 1, info.nups do
        local k, v = debug.getupvalue(f, i)
        if k == '_ENV' and not debug.getfenv then
            info.env = v
        else
            info.ups[k] = v
        end
    end

    if     info.source:sub(1,1) == '=' then  info.defined_how = 'C'
    elseif info.source:sub(1,1) == '@' then  info.defined_how = 'file'
    else                                     info.defined_how = 'string'
    end

    if info.builtin and LIBRARY[f] then
        info.name       = LIBRARY[f].name
        info.params[1]  = LIBRARY[f].para
        info.doc        = LIBRARY[f].docs
    end

    return info
end

local function get_line_index (str, line_nr)
    local index = 0
    for _ = 2, line_nr do
        index = str:find('\n', index, true)
        if not index then  return #str  end
        index = index + 1
    end
    return index
end

local function get_full_function_str (str, start_line, end_line)
    local start_line_index  =  get_line_index(str, start_line)
    local end_line_index    =  get_line_index(str, end_line + 1)

    local matched_function  =  str:sub(start_line_index, end_line_index):match('function.+end')
    return matched_function
end

local function width_of_strings_in_l (l, start_i, end_i)
    local  width = 0
    for i = start_i or 1, (end_i or #l) do
        width = width + #l[i]
    end
    return width
end

local function ignore_alignment_info (l, start_i, stop_i)
    for i = start_i or 1, stop_i or #l do
        if type(l[i]) == 'table' then
            l[i] = ''
        end
    end
end

local function fix_alignment (l, start_i, stop_i)
    -- Find maximums
    local max  =  {}
    for i = start_i or 1, stop_i or #l do
        if type(l[i]) == 'table' then
            max[ l[i][2] ]  =  math.max( l[i][1], max[ l[i][2] ] or 0 )
        end
    end

    -- Insert the proper whitespace
    for i = start_i, stop_i do
        if type(l[i]) == 'table' then
            l[i] = string.rep(' ', max[ l[i][2] ] - l[i][1])
        end
    end
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 is_sequence      =  contains_only_nice_number_indexes(value)
    local only_string_keys =  contains_only_nice_string_keys(value)
    local is_pure_map      =  (#value == 0)

    -- Return type
    if     is_sequence      then  return TABLE_TYPE_SEQUENCE
    elseif only_string_keys then  return TABLE_TYPE_STRING_MAP
    elseif is_pure_map      then  return TABLE_TYPE_PURE_MAP
    else                          return TABLE_TYPE_MIXED
    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 SINGLE_LINE_TABLE_TYPES[table_type]
       and #value <= SINGLE_LINE_SEQ_MAX_ELEMENTS
       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, 'key' }
    l[#l+1] = ' = '
    return format_value(value, options, depth, l)
end

local function format_key_and_value_arbitr_map (l, key, value, options, depth)
    local index_before_key = #l+1
    l[#l+1] = '['
    format_value(key, options, 'max', l)
    l[#l+1] = ']'
    l[#l+1] = { width_of_strings_in_l(l, index_before_key), 'key' }
    l[#l+1] = ' = '
    return format_value(value, options, depth, l)
end

local function format_key_and_value_sequence (l, key, value, options, depth)
    return format_value(value, options, depth, l)
end

local TABLE_TYPE_TO_PAIR_FORMAT = {
    [TABLE_TYPE_SEQUENCE]   = format_key_and_value_sequence,
    [TABLE_TYPE_STRING_MAP] = format_key_and_value_string_map,
    [TABLE_TYPE_PURE_MAP]   = format_key_and_value_arbitr_map,
    [TABLE_TYPE_MIXED]      = format_key_and_value_arbitr_map,
}

-- Formatting tables

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

    local key_value_pairs  = get_key_value_pairs_in_proper_order(t)
    local table_type       = get_table_type(t)
    local pair_format_func = TABLE_TYPE_TO_PAIR_FORMAT[table_type]

    l[#l+1] = '{ '
    local top_before = #l

    for _, pair in ipairs(key_value_pairs) do
        pair_format_func(l, pair[1], pair[2], options, 'max')
        l[#l+1] = ', '
    end

    -- Ignore the "width of key"-shit
    ignore_alignment_info(l, top_before, #l)

    if l[#l] == ', ' then  l[#l] = nil  end
    l[#l+1] = ' }'
end

local function format_map (t, options, depth, l)
    -- TODO: l.

    local key_value_pairs  = get_key_value_pairs_in_proper_order(t)
    local table_type       = get_table_type(t)
    local pair_format_func = TABLE_TYPE_TO_PAIR_FORMAT[table_type]

    l[#l+1] = '{\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

    -- Fix whitespace alignment
    fix_alignment(l, top_before, #l)

    -- Fix and cleanup
    if l[#l] == ',\n' then  l[#l] = nil  end
    l[#l+1]   = '\n'
    l[#l+1] = options.indent:rep(depth)
    l[#l+1] = '}'

end

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

    -- Empty or exteeding max-depth?
    if table_type == TABLE_TYPE_EMPTY then  l[#l+1] = '{}'; return
    elseif depth ~= 'max' and depth >= options.max_depth then  l[#l+1] = '{...}'; return
    end

    -- Single line?
    if is_single_line_table(t)  then  format_single_line_map(t, options, l); return  end

    if depth == 'max' then  l[#l+1] = '{...}'; return  end

    -- Normal table
    format_map(t, options, depth, l)
end

local function format_string (str, options, depth, l)
    -- 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

    l[#l+1] = left
    l[#l+1] = str
    l[#l+1] = right
end

local function format_number (value, options, depth, l)
    local shorthand = options.math_shorthand
    if     value ~= value then  l[#l+1]  =  shorthand and  'nan' or  '0/0'
    elseif value ==  1/0  then  l[#l+1]  =  shorthand and  'inf' or  '1/0'
    elseif value == -1/0  then  l[#l+1]  =  shorthand and '-inf' or '-1/0'
    else                        l[#l+1]  =  tostring(value)
    end
end

local function format_coroutine (value, options, depth, l)
    l[#l+1] = coroutine.status(value)
    l[#l+1] = ' coroutine: '
    l[#l+1] = tostring(value):sub(9)
end

local function format_primitive (value, options, depth, l)
    l[#l+1] = tostring(value)
end

local function format_string_defined_function (value, options, depth, l)
    local info = get_function_info(value)
    l[#l+1] = get_full_function_str(info.source, info.linedefined, info.lastlinedefined)
end

local function format_function (value, options, depth, l)
    local info = get_function_info(value)

    if info.defined_how == 'string' then
        return format_string_defined_function(value, options, depth, l)
    end

    -- Include function modifier, and alignment info.
    l[#l+1] = info.builtin and 'builtin ' or ''
    l[#l+1] = { #l[#l], 'func_mod'}

    -- Build rest of function signature
    l[#l+1] = 'function ('
    local top_before = #l
    for _, param in ipairs(info.params) do  l[#l+1], l[#l+2] = param, ', '  end
    if l[#l] == ', ' then  l[#l] = nil  end
    l[#l+1] = ')'
    l[#l+1] = { width_of_strings_in_l(l, top_before), 'func_def' }

    -- Cleanup and finish
    if not options.more_function_info or depth ~= 0  then
        l[#l+1] = ' ... end'
    elseif options._all_function_info then
        -- NOTE: This is for testing/debugging/experimentation purposes.

        local file = io.open(info.short_src, 'r')
        local function_str = get_full_function_str(file:read('*all'), info.linedefined, info.lastlinedefined)
        file:close()

        l[#l+1] = '\n\t--[[ Function Body\n\t'
        l[#l+1] = function_str
        l[#l+1] = '\n\t--]]'

        l[#l+1] = '\n\t--[[\n\tNative repr:'
        l[#l+1] = tostring(value)
        l[#l+1] = '\n\t'
        format_value(info, options, depth + 1, l)
        l[#l+1] = '--]]'
    else
        -- More info! --

        -- Name
        if info.name then
            l[#l+1] = '\n'
            l[#l+1] = options.indent
            l[#l+1] = '-- '
            l[#l+1] = info.name
        end

        -- Doc
        if info.doc then
            for doc_line in info.doc:gmatch('[^\n]+') do
                l[#l+1] = '\n'
                l[#l+1] = options.indent
                l[#l+1] = '-- '

                l[#l+1] = doc_line
            end
        end

        -- source
        if not info.builtin then
            l[#l+1] = '\n'
            l[#l+1] = options.indent
            l[#l+1] = '-- source_file: \''
            l[#l+1] = info.short_src
            l[#l+1] = '\' [Line'
            if info.linedefined == info.lastlinedefined then
                l[#l+1] = ': '
                l[#l+1] = tostring(info.linedefined)
            else
                l[#l+1] = 's: '
                l[#l+1] = tostring(info.linedefined)
                l[#l+1] = ' - '
                l[#l+1] = tostring(info.lastlinedefined)
            end
            l[#l+1] = ']'
        end

        -- upvalues
        if info.nups > 0 and not info.builtin then
            l[#l+1] = '\n'
            l[#l+1] = options.indent
            l[#l+1] = '-- up_values:   '
            format_value(info.ups, options, depth + 1, l)
        end

        l[#l+1] = '\n\n'
        l[#l+1] = options.indent
        l[#l+1] = '...\nend'
    end
end

local TYPE_TO_FORMAT_FUNC = {
    ['nil']      =  format_primitive,
    ['boolean']  =  format_primitive,
    ['number']   =  format_number,
    ['string']   =  format_string,
    ['thread']   =  format_coroutine,
    ['table']    =  format_table,

    -- TODO
    ['function'] =  format_function,
    ['userdata'] =  format_primitive,
    ['cdata']    =  format_primitive,  -- Luajit exclusive ?
}

function format_value (value, options, depth, l)
    local format_func = TYPE_TO_FORMAT_FUNC[type(value)]
    if format_func then
        format_func(value, options, depth, l)
    else
        error(ERROR_UNKNOWN_TYPE:format(type(value), tostring(value)), 2)
    end
end

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

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

    -- If any alignment info still exists, ignore it
    ignore_alignment_info(l)

    return table.concat(l, '')
end

return pretty_format