lua-dog/src/dog.lua

-- Variable for storing all the functions
local dog = {}

-- Functions are divided according to the standard library to extend
dog.io = {}
dog.os = {}
dog.string = {}
dog.utf8 = {}

-- Imports dog functions to _G
-- Allows using the other functions without a prefix, i.e.:
-- dog.os.uname() => os.uname()
function dog.import()
  local msg = "[WARNING] [DOG] "
  msg = msg .. "Ignoring import: '@1' already exists; use '@2' instead"
  for libkey, dogfns in pairs(dog) do
    local lib = _G[libkey]
    if type(dogfns) == "table" then
      for fnkey, fn in pairs(dogfns) do
        if not lib[fnkey] then
          lib[fnkey] = fn
        else
          local name = libkey .. "." .. fnkey .. "()"
          msg = msg:gsub("@1", name):gsub("@2", "dog." .. name)
          print(msg)
        end
      end
    end
  end
end

-- Tries popen
-- @param ... string: Chunks for popen
-- @return boolean, string: Status and output of popen
function dog.io.try(...)
  local cmd = table.concat({ ... }, " ") .. " 2>&1"
  local handle = io.popen(cmd)
  local output = handle:read("*a")
  local status = (handle:close() ~= nil and true or false)
  return status, output
end

-- Gets OS short name
-- It is just intented to know if it is Linux, macOS or Windows.
-- @return string: "linux" or "bsd" or "macos" or "windows"
function dog.os.uname()
  local status, output = dog.io.try("uname")
  if status then
    output = output:gsub(" .*", ""):lower()
    if output ~= "linux" and output:match("bsd") ~= nil then
      return "bsd"
    elseif output ~= "linux" then
      return "macos"
    end
    return "linux"
  end
  return "windows"
end

-- Checks if OS is windows
-- @return boolean: Windows or not
function dog.os.iswin() return dog.os.uname() == "windows" end

-- Checks if OS is Unix
-- @return boolean: Unix or not
function dog.os.isunix() return dog.os.uname() ~= "windows" end

-- Gets OS language
-- @return string, string, string: Language, locale and encoding
function dog.os.lang()
  local lang = os.getenv("LANG")
  if lang ~= nil then return lang:match("(%w%w)_?(%w?%w?)%.?(.*)") end
  return "en", "US", "UTF-8"
end

-- Checks if string is empty
-- @return boolean: Empty or not
function dog.string.isempty(str) return str == "" end

-- Changes newlines so everything is in one line
-- @return string: String with formatted newlines as literal "\n"
function dog.string.linearize(str) return str:gsub("\n", "\\n") end

-- Normalizes string
-- @return string: Normalized string
function dog.string.normalize(str)
  str = str:lower()
  for newchar, chars in pairs(dog.utf8.table()) do
    for _, oldchar in ipairs(chars) do
      str = str:gsub(oldchar, newchar)
    end
  end
  return str
end

-- Adds indent
-- @param num number: Indent size, 2 by default
-- @param char string: Indent character, space by default
-- @return strin: Indented string
function dog.string.indent(str, num, char)
  num = num or 2
  char = char or " "
  char = string.rep(char, num)
  return char .. str:gsub("\n", "\n" .. char)
end

-- Splits the string
-- Note: only support splitting by one character.
-- Could be solved by slicing with find.
-- @param sep string: String separator, space by default
-- @return table: String matches
function dog.string.split(str, sep)
  sep = sep or "%s+"
  local parts = {}
  for part in str:gmatch("([^" .. sep .. "]+)") do
    table.insert(parts, part)
  end
  return parts
end

-- Removes spaces at the beginning of the string
-- @return string: Left stripped string
function dog.string.lstrip(str) return str:gsub("^%s+", "") end

-- Removes spaces at the end of the string
-- @return string: Right stripped string
function dog.string.rstrip(str) return str:gsub("%s+$", "") end

-- Removes spaces at the beginning and at the end of the string
-- @return string: Stripped string
function dog.string.strip(str)
  str = dog.string.lstrip(str)
  return dog.string.rstrip(str)
end

-- Alias of strip
function dog.string.trim(str) return dog.string.strip(str) end

-- The following are heavily influenced by Python pathlib
-- Check: https://docs.python.org/3/library/pathlib.html

-- Checks if string is a file or a directory
-- @return boolean: Exists or not
function dog.string.exists(str) return os.rename(str, str) ~= nil end

-- Checks if string is a file
-- @return boolean: File or not
function dog.string.isfile(str)
  if dog.string.exists(str) then return io.open(str, "a+") ~= nil end
  return false
end

-- Checks if string is a directory
-- @return boolean: Directory or not
function dog.string.isdir(str)
  if dog.string.exists(str) then
    return io.open(str, "a+") == nil
  elseif str == "." or str == ".." then
    return true
  end
  return false
end

-- Reads file content as string
-- @return string or nil: File as string or nil
function dog.string.readtext(str)
  if dog.string.exists(str) then return io.open(str):read("*a") end
end

-- Read file content as lines
-- @return table: Table of file lines or nil
function dog.string.readlines(str)
  local lines = {}
  if dog.string.exists(str) then
    for line in io.open(str):lines() do
      table.insert(lines, line)
    end
  end
  return lines
end

-- Gets file without suffix
-- @return string: File wihtout suffix
function dog.string.stem(str) return str:gsub("%.%a+$", "") end

-- Gets file extensions
-- @return table: List of file extensions
function dog.string.suffixes(str)
  local suffixes = {}
  for suffix in str:gmatch("%.%a+") do
    table.insert(suffixes, suffix)
  end
  return suffixes
end

-- Gets file final extension
-- @return string: Final file extension
function dog.string.suffix(str)
  local suffixes = str:suffixes()
  if suffixes[#suffixes] then return suffixes[#suffixes] end
  return ""
end

-- Gets an equivalency table between ASCII and Unicode
-- Note: filled on demand.
-- @return table: ASCII-Unicode table equivalency
function dog.utf8.table()
  return {
    ["a"] = { "á", "à", "ä" },
    ["e"] = { "é", "è", "ë" },
    ["i"] = { "í", "ì", "ï" },
    ["o"] = { "ó", "ò", "ö" },
    ["u"] = { "ú", "ù", "ü" },
    ["n"] = { "ñ" },
  }
end

-- Extends Pandoc Library
-- Check: https://pandoc.org/lua-filters.html#module-pandoc

if pandoc ~= nil then
  -- Gets file extension namespace
  -- Check: https://pandoc.org/MANUAL.html#general-options
  -- If suffix is 'md' or empty, the default is markdown.
  -- For the rest, the format is the suffix.
  -- @param file string: File name
  -- @return string: File extension according to Pandoc format namespaces
  function pandoc.getext(file)
    local ext = dog.string.suffix(file):gsub("^.", "")
    return ((ext == "md" or dog.string.isempty(ext)) and "markdown" or ext)
  end

  -- Pandoc converter
  -- Converts input file name into output format name.
  -- If ofile is not nil, writes output file name.
  -- If iformat is nil, defaults to ifile extension name.
  -- @param ifile string: Input file name
  -- @param oformat string: Output format name
  -- @param ofile string or nil: Output file name
  -- @param iformat string or nil: Input format name
  -- @return string: Output file content
  function pandoc.convert(ifile, oformat, ofile, iformat)
    iformat = (iformat == nil and pandoc.getext(ifile) or iformat)
    local itext = pandoc.read(dog.string.readtext(ifile), iformat)
    local doc = pandoc.write(itext, oformat)
    if ofile ~= nil then
      local eol = (dog.os:isunix() and "\n" or "\r\n")
      io.open(ofile, "w"):write(doc, eol):close()
    end
    return doc
  end

  -- Stringifies Pandoc content
  -- Avoids undesired behavios of pandoc.utils.stringify, such as quotes and
  -- backslashes conversions.
  -- @param content pandoc.MetaValue: Pandoc content value
  -- @return string: Pandoc stringified value
  function pandoc.utils.rawstringify(content)
    return pandoc.utils.stringify(content:walk({
      Plain = function(plain)
        table.insert(plain.content, pandoc.Space())
        return plain
      end,
      Quoted = function(quoted)
        local q = (quoted.quotetype == SingleQuoted and '"' or "'")
        return pandoc.Str(q .. pandoc.utils.stringify(quoted.content) .. q)
      end,
      RawInline = function(rawinline)
        return pandoc.Str(rawinline.text:gsub("\\n", "\n"):gsub("\\t", "\t"))
      end,
      SoftBreak = function(_) return pandoc.Str("\n") end,
      Inline = function(inline) return pandoc.utils.stringify(inline) end,
    }))
  end

  -- Converts pandoc.Meta to table
  -- Note that stringified numbers, such "1", are converted to numbers because
  -- Pandoc doesn't make distinctions between strings and numbers.
  -- Check: https://pandoc.org/lua-filters.html#type-metavalue
  -- @param meta pandoc.Meta: Pandoc MetaValues collection
  -- @return table: Pandoc metadata as table
  function pandoc.metatotable(meta)
    local newmeta = {}
    for k, v in pairs(meta) do
      if pandoc.utils.type(v) == "table" then
        newmeta[k] = pandoc.metatotable(v)
      else
        if type(v) == "boolean" then
          newmeta[k] = v
        elseif tonumber(v) then
          newmeta[k] = tonumber(v)
        else
          newmeta[k] = pandoc.utils.rawstringify(v)
        end
      end
    end
    return newmeta
  end
end

return dog