mod error;

pub use crate::error::Error;
use bitvec::prelude::*;
use error::InvalidHeaderKind;
use std::iter::Peekable;
use std::num::ParseIntError;
use std::str::Chars;

#[derive(Debug, PartialEq)]
struct Header {
	bitorder: BitOrder,
	byteorder: ByteOrder,
	negativekind: NegativeKind,
	pad_bits: bool,
}

#[derive(Debug, PartialEq)]
enum BitOrder {
	Msb0,
	Lsb0,
}

#[derive(Debug, PartialEq)]
enum ByteOrder {
	LittleEndian,
	BigEndian,
}

#[derive(Debug, PartialEq)]
enum NegativeKind {
	TwosCompliment,
	OnesCompliment,
	SignMagnitude,
}

pub struct Hext {
	parsed: Vec<u8>,
}

impl Hext {
	pub fn new() -> Self {
		Self { parsed: vec![] }
	}

	pub fn parse<S: AsRef<str>>(mut self, raw: S) -> Result<Vec<u8>, Error> {
		let mut chars = raw.as_ref().chars().peekable();

		// Clear through any leading comments or blank lines
		Self::skip_nondata(&mut chars);

		let header: Header;
		loop {
			match chars.next() {
				Some('~') => {
					header = Self::parse_header(Self::consume_line(&mut chars))?;
					break;
				}
				Some(_) => return Err(Error::NoHeader),
				None => return Ok(self.parsed), //todo: is this an error?
			}
		}

		let mut bits: BitVec<u8, Msb0> = BitVec::new();
		let mut state = State::ReadingHex;

		loop {
			match state {
				State::ReadingHex => match chars.next_if(|&c| c != '.') {
					Some('#') => Self::skip_line(&mut chars),
					Some(c) if c.is_whitespace() => continue,

					Some(high) if high.is_ascii_hexdigit() => {
						match chars.next_if(|&c| c.is_ascii_hexdigit()) {
							Some(low) => self.parsed.push(
								((high.to_digit(16).unwrap() * 16) + low.to_digit(16).unwrap())
									as u8,
							),
							None => return Err(Error::IncompleteOctet),
						}
					}

					Some('=') => state = State::ReadingUnsizedDecimal,
					Some('i') => state = State::ReadingSignedDecimal,
					Some('u') => state = State::ReadingUnsignedDecimal,
					Some('\"') => state = State::ReadingLiteral,
					Some(c) => return Err(Error::InvalidCharacter(c)),

					None => match chars.peek() {
						Some('.') => state = State::ReadingBinary,
						Some(_) => unreachable!(),
						None => return Ok(self.parsed),
					},
				},

				State::ReadingUnsizedDecimal => {
					let decimal = Self::consume_until_whitespace(&mut chars);
					state = State::ReadingHex;

					let is_signed = if let Some(sign) = decimal.chars().next() {
						sign == '-' || sign == '+'
					} else {
						// it was a lone =. Send the maybe-decimal string even
						// though we know it's empty
						return Err(Error::InvalidDecimal(decimal));
					};

					let mut bytes = if is_signed {
						Self::signed_smallest_le_bytes(&decimal)
					} else {
						Self::unsigned_smallest_le_bytes(&decimal)
					}
					.map_err(|_e| Error::InvalidDecimal(decimal))?;

					if header.byteorder == ByteOrder::BigEndian {
						bytes.reverse();
					}

					self.parsed.extend_from_slice(&bytes);
				}

				State::ReadingSignedDecimal => {
					let signed_decimal_string = Self::consume_until_whitespace(&mut chars);
					state = State::ReadingHex;

					let splits = signed_decimal_string.split_once('=');
					match splits {
						Some((bitness, value)) => {
							let mut bytes = Self::signed_le_bytes(bitness, value)?;

							if header.byteorder == ByteOrder::BigEndian {
								bytes.reverse();
							}

							self.parsed.extend_from_slice(&bytes);
						}
						None => return Err(Error::InvalidSignedDecimal(signed_decimal_string)),
					}
				}

				State::ReadingUnsignedDecimal => {
					let signed_decimal_string = Self::consume_until_whitespace(&mut chars);
					state = State::ReadingHex;

					let splits = signed_decimal_string.split_once('=');
					match splits {
						Some((bitness, value)) => {
							let mut bytes = Self::unsigned_le_bytes(bitness, value)?;

							if header.byteorder == ByteOrder::BigEndian {
								bytes.reverse();
							}

							self.parsed.extend_from_slice(&bytes);
						}
						None => return Err(Error::InvalidDecimal(signed_decimal_string)),
					}
				}

				State::ReadingLiteral => match chars.next() {
					Some('\"') => state = State::ReadingHex,
					Some('\\') => match chars.next() {
						Some(c) => match Self::escape(c) {
							Some(c) => self.parsed.push(c as u8),
							None => return Err(Error::InvalidEscape(c)),
						},
						None => return Err(Error::UnclosedStringLiteral),
					},
					Some('\n') => return Err(Error::UnclosedStringLiteral),
					Some(c) => {
						let mut encode = vec![0; c.len_utf8()];
						c.encode_utf8(&mut encode);
						self.parsed.extend_from_slice(&encode)
					}
					None => return Err(Error::UnclosedStringLiteral),
				},

				State::ReadingBinary => match chars.next_if(|&c| c == '.') {
					Some('.') => loop {
						match chars
							.next_if(|&c| c == '1' || c == '0' || c == '#' || c.is_whitespace())
						{
							Some('0') => bits.push(false),
							Some('1') => bits.push(true),
							Some('#') => Self::skip_line(&mut chars),
							Some(c) if c.is_whitespace() => {
								Self::skip_nondata(&mut chars);
								break;
							}
							Some(_) => return Err(Error::GarbageCharacterInBitstream),
							None => break,
						}
					},
					Some(_) => unreachable!(),
					None => {
						if bits.len() % 8 != 0 {
							if !header.pad_bits {
								eprintln!("{}", bits.len());
								return Err(Error::UnalignedBits);
							} else {
								while bits.len() % 8 != 0 {
									bits.insert(0, false);
								}
							}
						}

						self.parsed.extend_from_slice(bits.as_raw_slice());
						bits = BitVec::new();

						state = State::ReadingHex;
					}
				},
			}
		}
	}

	fn parse_header<S: AsRef<str>>(string: S) -> Result<Header, Error> {
		let splits: Vec<&str> = string.as_ref().trim_end().split(' ').collect();

		let mut bitorder = None;
		let mut byteorder = None;
		let mut negativekind = None;
		let mut pad_bits = false;

		for split in splits {
			match split {
				"msb0" => {
					if bitorder.replace(BitOrder::Msb0).is_some() {
						return Err(InvalidHeaderKind::TwoBitOrder.into());
					}
				}
				"lsb0" => {
					if bitorder.replace(BitOrder::Lsb0).is_some() {
						return Err(InvalidHeaderKind::TwoBitOrder.into());
					}
				}
				"big-endian" => {
					if byteorder.replace(ByteOrder::BigEndian).is_some() {
						return Err(InvalidHeaderKind::TwoByteOrder.into());
					}
				}
				"little-endian" => {
					if byteorder.replace(ByteOrder::LittleEndian).is_some() {
						return Err(InvalidHeaderKind::TwoByteOrder.into());
					}
				}
				"twos-compliment" => {
					if negativekind.replace(NegativeKind::TwosCompliment).is_none() {
						return Err(InvalidHeaderKind::TwoNegativeKind.into());
					}
				}
				"ones-compliment" => {
					if negativekind.replace(NegativeKind::OnesCompliment).is_none() {
						return Err(InvalidHeaderKind::TwoNegativeKind.into());
					}
				}
				"sign-magnitude" => {
					if negativekind.replace(NegativeKind::SignMagnitude).is_none() {
						return Err(InvalidHeaderKind::TwoNegativeKind.into());
					}
				}
				"padbits" => pad_bits = true,
				_ => return Err(InvalidHeaderKind::InvalidProperty(split.into()).into()),
			}
		}

		if bitorder.is_none() {
			return Err(InvalidHeaderKind::NoBitOrder.into());
		} else if byteorder.is_none() {
			return Err(InvalidHeaderKind::NoByteOrder.into());
		} else {
			Ok(Header {
				bitorder: bitorder.unwrap(),
				byteorder: byteorder.unwrap(),
				negativekind: negativekind.unwrap_or(NegativeKind::TwosCompliment),
				pad_bits,
			})
		}
	}

	fn escape(c: char) -> Option<char> {
		match c {
			'\"' => Some('\"'),
			'\\' => Some('\\'),
			'n' => Some('\n'),
			'r' => Some('\r'),
			't' => Some('\t'),
			_ => None,
		}
	}

	fn skip_nondata(mut chars: &mut Peekable<Chars>) {
		loop {
			match chars.peek() {
				Some('#') => Self::skip_line(&mut chars),
				Some(c) if c.is_whitespace() => {
					chars.next();
				}
				_ => return,
			};
		}
	}

	fn skip_line(chars: &mut Peekable<Chars>) {
		chars.find(|&c| c == '\n');
	}

	fn consume_line(chars: &mut Peekable<Chars>) -> String {
		chars.take_while(|&c| c != '\n').collect()
	}

	fn consume_until_whitespace(chars: &mut Peekable<Chars>) -> String {
		chars.take_while(|&c| !c.is_whitespace()).collect()
	}

	fn signed_le_bytes<S: AsRef<str>>(bitness: S, value: S) -> Result<Vec<u8>, Error> {
		match bitness.as_ref() {
			"8" => Ok(i8::from_str_radix(value.as_ref(), 10)
				.map_err(|_| Error::InvalidSignedDecimal(value.as_ref().to_string()))?
				.to_le_bytes()
				.to_vec()),
			"16" => Ok(i16::from_str_radix(value.as_ref(), 10)
				.map_err(|_| Error::InvalidSignedDecimal(value.as_ref().to_string()))?
				.to_le_bytes()
				.to_vec()),
			"32" => Ok(i32::from_str_radix(value.as_ref(), 10)
				.map_err(|_| Error::InvalidSignedDecimal(value.as_ref().to_string()))?
				.to_le_bytes()
				.to_vec()),
			"64" => Ok(i64::from_str_radix(value.as_ref(), 10)
				.map_err(|_| Error::InvalidSignedDecimal(value.as_ref().to_string()))?
				.to_le_bytes()
				.to_vec()),
			_ => return Err(Error::InvalidBitness(bitness.as_ref().to_string())),
		}
	}

	fn signed_smallest_le_bytes<S: AsRef<str>>(string: S) -> Result<Vec<u8>, ParseIntError> {
		let large: i64 = i64::from_str_radix(string.as_ref(), 10)?;

		Ok(if large > i32::MAX as i64 || large < i32::MIN as i64 {
			large.to_le_bytes().to_vec()
		} else if large > i16::MAX as i64 || large < i16::MIN as i64 {
			(large as i32).to_le_bytes().to_vec()
		} else if large > i8::MAX as i64 || large < i8::MIN as i64 {
			(large as i16).to_le_bytes().to_vec()
		} else {
			(large as i8).to_le_bytes().to_vec()
		})
	}

	fn unsigned_le_bytes<S: AsRef<str>>(bitness: S, value: S) -> Result<Vec<u8>, Error> {
		match bitness.as_ref() {
			"8" => Ok(u8::from_str_radix(value.as_ref(), 10)
				.map_err(|_| Error::InvalidUnsignedDecimal(value.as_ref().to_string()))?
				.to_le_bytes()
				.to_vec()),
			"16" => Ok(u16::from_str_radix(value.as_ref(), 10)
				.map_err(|_| Error::InvalidUnsignedDecimal(value.as_ref().to_string()))?
				.to_le_bytes()
				.to_vec()),
			"32" => Ok(u32::from_str_radix(value.as_ref(), 10)
				.map_err(|_| Error::InvalidUnsignedDecimal(value.as_ref().to_string()))?
				.to_le_bytes()
				.to_vec()),
			"64" => Ok(u64::from_str_radix(value.as_ref(), 10)
				.map_err(|_| Error::InvalidUnsignedDecimal(value.as_ref().to_string()))?
				.to_le_bytes()
				.to_vec()),
			_ => return Err(Error::InvalidBitness(bitness.as_ref().to_string())),
		}
	}

	fn unsigned_smallest_le_bytes<S: AsRef<str>>(string: S) -> Result<Vec<u8>, ParseIntError> {
		let large: u64 = u64::from_str_radix(string.as_ref(), 10)?;

		Ok(if large > u32::MAX as u64 {
			large.to_le_bytes().to_vec()
		} else if large > u16::MAX as u64 {
			(large as u32).to_le_bytes().to_vec()
		} else if large > u8::MAX as u64 {
			(large as u16).to_le_bytes().to_vec()
		} else {
			(large as u8).to_le_bytes().to_vec()
		})
	}
}

enum State {
	ReadingHex,
	ReadingUnsizedDecimal,
	ReadingSignedDecimal,
	ReadingUnsignedDecimal,
	ReadingBinary,
	ReadingLiteral,
}

#[cfg(test)]
mod test {
	use super::*;

	#[test]
	fn pares_header_success() {
		// Recognizes the keytwords...
		assert_eq!(
			Hext::parse_header("msb0 big-endian").unwrap(),
			Header {
				byteorder: ByteOrder::BigEndian,
				bitorder: crate::BitOrder::Msb0,
				negativekind: NegativeKind::TwosCompliment,
				pad_bits: false
			}
		);

		assert_eq!(
			Hext::parse_header("lsb0 little-endian").unwrap(),
			Header {
				byteorder: ByteOrder::LittleEndian,
				bitorder: crate::BitOrder::Lsb0,
				negativekind: NegativeKind::TwosCompliment,
				pad_bits: false
			}
		);

		// ...In either order
		assert_eq!(
			Hext::parse_header("big-endian lsb0").unwrap(),
			Header {
				byteorder: ByteOrder::BigEndian,
				bitorder: crate::BitOrder::Lsb0,
				negativekind: NegativeKind::TwosCompliment,
				pad_bits: false
			}
		);
	}

	#[test]
	fn parse_header_fail_twobits() {
		assert_eq!(
			Hext::parse_header("lsb0 msb0"),
			Err(InvalidHeaderKind::TwoBitOrder.into())
		)
	}

	#[test]
	fn parse_header_fail_twobytes() {
		assert_eq!(
			Hext::parse_header("little-endian big-endian"),
			Err(InvalidHeaderKind::TwoByteOrder.into())
		)
	}

	#[test]
	fn parse_header_fail_nobits() {
		assert_eq!(
			Hext::parse_header("big-endian"),
			Err(InvalidHeaderKind::NoBitOrder.into())
		)
	}

	#[test]
	fn parse_header_fail_nobytes() {
		assert_eq!(
			Hext::parse_header("msb0"),
			Err(InvalidHeaderKind::NoByteOrder.into())
		)
	}

	#[test]
	fn parse_header_fail_invalidproperty() {
		assert_eq!(
			Hext::parse_header("lsb0 big-endian invalidproperty"),
			Err(InvalidHeaderKind::InvalidProperty("invalidproperty".into()).into())
		)
	}

	//## Bytes tests ##
	#[test]
	fn test_onebyte() {
		let test = "~little-endian msb0\n41";
		let cmp = vec![0x41];

		assert_eq!(Hext::new().parse(&test).unwrap(), cmp);
	}

	#[test]
	fn test_only_comment() {
		let test = "~little-endian msb0\n# Comment";

		assert_eq!(Hext::new().parse(&test).unwrap(), vec![]);
	}

	#[test]
	fn test_1byte_comment() {
		let test = "~little-endian msb0\n41 #A";
		let cmp = vec![0x41];

		assert_eq!(Hext::new().parse(&test).unwrap(), cmp);
	}

	#[test]
	fn test_byte_nospace_comment() {
		let test = "~little-endian msb0\n41#A";
		let cmp = vec![0x41];

		assert_eq!(Hext::new().parse(&test).unwrap(), cmp);
	}

	#[test]
	fn test_2byte_multiline() {
		let test = "~little-endian msb0\n41\n42";
		let cmp = vec![0x41, 0x42];

		assert_eq!(Hext::new().parse(&test).unwrap(), cmp);
	}

	#[test]
	fn test_2bytes_nospace() {
		let test = "~little-endian msb0\n4142";
		let cmp = vec![0x41, 0x42];

		assert_eq!(Hext::new().parse(&test).unwrap(), cmp);
	}

	//## Bit Tests ##
	#[test]
	fn test_8bits() {
		let test = "~little-endian msb0\n.01000001";
		let cmp = vec![0x41];

		assert_eq!(Hext::new().parse(&test).unwrap(), cmp);
	}

	#[test]
	fn test_8bits_hex10() {
		let test = "~little-endian msb0\n.01000001 10";
		let cmp = vec![0x41, 0x10];

		assert_eq!(Hext::new().parse(&test).unwrap(), cmp);
	}

	#[test]
	fn test_8bit_comment() {
		let test = "~little-endian msb0\n.01000001 # A";
		let cmp = vec![0x41];

		assert_eq!(Hext::new().parse(&test).unwrap(), cmp)
	}

	#[test]
	fn test_8bit_nospace_comment() {
		let test = "~little-endian msb0\n.01000001#A";
		let cmp = vec![0x41];

		assert_eq!(Hext::new().parse(&test).unwrap(), cmp);
	}

	#[test]
	fn test_1bit() {
		let test = "~little-endian msb0 padbits\n.1";
		let cmp = vec![0x01];

		assert_eq!(Hext::new().parse(&test).unwrap(), cmp);
	}

	#[test]
	fn test_8bits_halved_space() {
		let test_space = "~little-endian msb0\n.0100 .0010";
		let cmp = vec![0x42];

		assert_eq!(Hext::new().parse(&test_space).unwrap(), cmp);
	}

	#[test]
	fn test_8bits_halved_line() {
		let test_line = "~little-endian msb0\n.0100\n.0010";
		let cmp = vec![0x42];

		assert_eq!(Hext::new().parse(&test_line).unwrap(), cmp);
	}

	#[test]
	fn test_8bits_halved_line_comments() {
		let test_line_comments = "~little-endian msb0\n.0100#Half of capital letter\n.0010 # B";
		let cmp = vec![0x42];

		assert_eq!(Hext::new().parse(&test_line_comments).unwrap(), cmp);
	}

	#[test]
	fn test_1bit_then_byte() {
		let test = "~little-endian msb0 padbits\n.1 41";
		let cmp = vec![0x01, 0x41];

		assert_eq!(Hext::new().parse(&test).unwrap(), cmp);
	}

	//## Literal Tests ##
	#[test]
	fn literal_multibyte() {
		let test = "~big-endian lsb0\n\"🥺\"";
		let cmp = vec![0xf0, 0x9f, 0xa5, 0xba];

		assert_eq!(Hext::new().parse(&test).unwrap(), cmp);
	}

	//## Decimal Tests ##
	#[test]
	fn decimal_unsized_u8() {
		let test = "~big-endian lsb0\n=200";
		let cmp = vec![200];

		assert_eq!(Hext::new().parse(&test).unwrap(), cmp);
	}

	#[test]
	fn decimal_unsized_i8() {
		let test = "~big-endian lsb0\n=-127";
		let cmp = (-127i8).to_be_bytes().to_vec();

		assert_eq!(Hext::new().parse(&test).unwrap(), cmp);
	}

	#[test]
	fn decimal_unsized_u32() {
		let test = "~little-endian lsb0\n=65536";
		let cmp = 65536u32.to_le_bytes().to_vec();

		assert_eq!(Hext::new().parse(&test).unwrap(), cmp);
	}

	#[test]
	fn decimal_unsized_i32() {
		let test = "~little-endian lsb0\n=-40000";
		let cmp = (-40000i32).to_le_bytes().to_vec();

		assert_eq!(Hext::new().parse(&test).unwrap(), cmp);
	}

	#[test]
	fn decimal_sized_u16() {
		let test = "~little-endian lsb0\nu16=65534";
		let cmp = 65534u16.to_le_bytes().to_vec();

		assert_eq!(Hext::new().parse(&test).unwrap(), cmp);
	}

	#[test]
	fn decimal_overflow_sized_u16() {
		let test = "~little-endian lsb0\nu16=65536";

		assert_eq!(
			Hext::new().parse(&test).unwrap_err(),
			Error::InvalidUnsignedDecimal("65536".into())
		);
	}

	//## Everything ##
	#[test]
	fn everything() {
		let to_parse = std::fs::read_to_string("tests/everything.hxt").unwrap();
		let cmp = std::fs::read_to_string("tests/everything.correct")
			.unwrap()
			.into_bytes();

		assert_eq!(Hext::new().parse(&to_parse).unwrap(), cmp)
	}

	//## Failing Tests ##
	#[test]
	fn ftest_incompleteoctet() {
		let test = "~little-endian msb0\n4";

		assert_eq!(
			Hext::new().parse(&test).unwrap_err(),
			Error::IncompleteOctet
		);
	}

	#[test]
	fn ftest_invalidcharacter() {
		let test = "~little-endian msb0\nG";

		assert_eq!(
			Hext::new().parse(&test).unwrap_err(),
			Error::InvalidCharacter('G')
		);
	}

	#[test]
	fn ftest_unaligned_bit() {
		let test = "~little-endian msb0\n.1";
		let cmp = Error::UnalignedBits;

		assert_eq!(Hext::new().parse(&test).unwrap_err(), cmp);
	}

	#[test]
	fn ftest_unaligned_bit_then_byte() {
		let test = "~little-endian msb0\n.1 41";
		let cmp = Error::UnalignedBits;

		assert_eq!(Hext::new().parse(&test).unwrap_err(), cmp);
	}
}