git.delta.rocks / jrsonnet / refs/commits / 46fae7b561fd

1//! faster std.format impl2#![allow(clippy::too_many_arguments)]34use crate::{error::Error::*, throw, LocError, ObjValue, Result, Val, ValType};56#[derive(Debug, Clone)]7pub enum FormatError {8	TruncatedFormatCode,9	UnrecognizedConversionType(char),1011	NotEnoughValues,1213	CannotUseStarWidthWithObject,14	MappingKeysRequired,15	NoSuchFormatField(Rc<str>),16}1718impl From<FormatError> for LocError {19	fn from(e: FormatError) -> Self {20		Self::new(Format(e))21	}22}2324use std::rc::Rc;25use FormatError::*;2627type ParseResult<'t, T> = std::result::Result<(T, &'t str), FormatError>;2829pub fn try_parse_mapping_key(str: &str) -> ParseResult<&str> {30	if str.is_empty() {31		return Err(TruncatedFormatCode);32	}33	let bytes = str.as_bytes();34	if bytes[0] == b'(' {35		let mut i = 1;36		while i < bytes.len() {37			if bytes[i] == b')' {38				return Ok((&str[1..i as usize], &str[i as usize + 1..]));39			}40			i += 1;41		}42		Err(TruncatedFormatCode)43	} else {44		Ok(("", str))45	}46}4748#[cfg(test)]49pub mod tests_key {50	use super::*;5152	#[test]53	fn parse_key() {54		assert_eq!(55			try_parse_mapping_key("(hello ) world").unwrap(),56			("hello ", " world")57		);58		assert_eq!(try_parse_mapping_key("() world").unwrap(), ("", " world"));59		assert_eq!(try_parse_mapping_key(" world").unwrap(), ("", " world"));60		assert_eq!(61			try_parse_mapping_key(" () world").unwrap(),62			("", " () world")63		);64	}6566	#[test]67	#[should_panic]68	fn parse_key_missing_start() {69		try_parse_mapping_key("").unwrap();70	}7172	#[test]73	#[should_panic]74	fn parse_key_missing_end() {75		try_parse_mapping_key("(   ").unwrap();76	}77}7879#[derive(Default, Debug)]80pub struct CFlags {81	pub alt: bool,82	pub zero: bool,83	pub left: bool,84	pub blank: bool,85	pub sign: bool,86}8788pub fn try_parse_cflags(str: &str) -> ParseResult<CFlags> {89	if str.is_empty() {90		return Err(TruncatedFormatCode);91	}92	let bytes = str.as_bytes();93	let mut i = 0;94	let mut out = CFlags::default();95	loop {96		if bytes.len() == i {97			return Err(TruncatedFormatCode);98		}99		match bytes[i] {100			b'#' => out.alt = true,101			b'0' => out.zero = true,102			b'-' => out.left = true,103			b' ' => out.blank = true,104			b'+' => out.sign = true,105			_ => break,106		}107		i += 1;108	}109	Ok((out, &str[i..]))110}111112#[derive(Debug, PartialEq)]113pub enum Width {114	Star,115	Fixed(usize),116}117pub fn try_parse_field_width(str: &str) -> ParseResult<Width> {118	if str.is_empty() {119		return Err(TruncatedFormatCode);120	}121	let bytes = str.as_bytes();122	if bytes[0] == b'*' {123		return Ok((Width::Star, &str[1..]));124	}125	let mut out: usize = 0;126	let mut digits = 0;127	while let Some(digit) = (bytes[digits] as char).to_digit(10) {128		out *= 10;129		out += digit as usize;130		digits += 1;131		if digits == bytes.len() {132			return Err(TruncatedFormatCode);133		}134	}135	Ok((Width::Fixed(out), &str[digits..]))136}137138pub fn try_parse_precision(str: &str) -> ParseResult<Option<Width>> {139	if str.is_empty() {140		return Err(TruncatedFormatCode);141	}142	let bytes = str.as_bytes();143	if bytes[0] == b'.' {144		try_parse_field_width(&str[1..]).map(|(r, s)| (Some(r), s))145	} else {146		Ok((None, str))147	}148}149150// Only skips151pub fn try_parse_length_modifier(str: &str) -> ParseResult<()> {152	if str.is_empty() {153		return Err(TruncatedFormatCode);154	}155	let bytes = str.as_bytes();156	let mut idx = 0;157	while bytes[idx] == b'h' || bytes[idx] == b'l' || bytes[idx] == b'L' {158		idx += 1;159		if bytes.len() == idx {160			return Err(TruncatedFormatCode);161		}162	}163	Ok(((), &str[idx..]))164}165166#[derive(Debug, PartialEq)]167pub enum ConvTypeV {168	Decimal,169	Octal,170	Hexadecimal,171	Scientific,172	Float,173	Shorter,174	Char,175	String,176	Percent,177}178pub struct ConvType {179	v: ConvTypeV,180	caps: bool,181}182183pub fn parse_conversion_type(str: &str) -> ParseResult<ConvType> {184	if str.is_empty() {185		return Err(TruncatedFormatCode);186	}187188	let code = str.as_bytes()[0];189	let v: (ConvTypeV, bool) = match code {190		b'd' | b'i' | b'u' => (ConvTypeV::Decimal, false),191		b'o' => (ConvTypeV::Octal, false),192		b'x' => (ConvTypeV::Hexadecimal, false),193		b'X' => (ConvTypeV::Hexadecimal, true),194		b'e' => (ConvTypeV::Scientific, false),195		b'E' => (ConvTypeV::Scientific, true),196		b'f' => (ConvTypeV::Float, false),197		b'F' => (ConvTypeV::Float, true),198		b'g' => (ConvTypeV::Shorter, false),199		b'G' => (ConvTypeV::Shorter, true),200		b'c' => (ConvTypeV::Char, false),201		b's' => (ConvTypeV::String, false),202		b'%' => (ConvTypeV::Percent, false),203		c => return Err(UnrecognizedConversionType(c as char)),204	};205206	Ok((ConvType { v: v.0, caps: v.1 }, &str[1..]))207}208209#[derive(Debug)]210pub struct Code<'s> {211	mkey: &'s str,212	cflags: CFlags,213	width: Width,214	precision: Option<Width>,215	convtype: ConvTypeV,216	caps: bool,217}218pub fn parse_code(str: &str) -> ParseResult<Code> {219	if str.is_empty() {220		return Err(TruncatedFormatCode);221	}222	let (mkey, str) = try_parse_mapping_key(str)?;223	let (cflags, str) = try_parse_cflags(str)?;224	let (width, str) = try_parse_field_width(str)?;225	let (precision, str) = try_parse_precision(str)?;226	let (_, str) = try_parse_length_modifier(str)?;227	let (convtype, str) = parse_conversion_type(str)?;228229	Ok((230		Code {231			mkey,232			cflags,233			width,234			precision,235			convtype: convtype.v,236			caps: convtype.caps,237		},238		str,239	))240}241242#[derive(Debug)]243pub enum Element<'s> {244	String(&'s str),245	Code(Code<'s>),246}247pub fn parse_codes(mut str: &str) -> Result<Vec<Element>> {248	let mut bytes = str.as_bytes();249	let mut out = vec![];250	let mut offset = 0;251252	loop {253		while offset != bytes.len() && bytes[offset] != b'%' {254			offset += 1;255		}256		if offset != 0 {257			out.push(Element::String(&str[0..offset]));258		}259		if offset == bytes.len() {260			return Ok(out);261		}262		str = &str[offset + 1..];263		let (code, nstr) = parse_code(str)?;264		str = nstr;265		bytes = str.as_bytes();266		offset = 0;267268		out.push(Element::Code(code))269	}270}271272const NUMBERS: &[u8] = b"0123456789abcdefghijklmnopqrstuvwxyz";273274#[inline]275pub fn render_integer(276	out: &mut String,277	iv: i64,278	padding: usize,279	precision: usize,280	blank: bool,281	sign: bool,282	radix: i64,283	prefix: &str,284	caps: bool,285) {286	// Digit char indexes in reverse order, i.e287	// for radix = 16 and n = 12f: [15, 2, 1]288	let digits = if iv == 0 {289		vec![0u8]290	} else {291		let mut v = iv.abs();292		let mut nums = Vec::with_capacity(1);293		while v > 0 {294			nums.push((v % radix) as u8);295			v /= radix;296		}297		nums298	};299	let neg = iv < 0;300	let zp = padding.saturating_sub(if neg || blank || sign { 1 } else { 0 });301	let zp2 = zp302		.max(precision)303		.saturating_sub(prefix.len() + digits.len());304305	if neg {306		out.push('-')307	} else if sign {308		out.push('+');309	} else if blank {310		out.push(' ');311	}312313	out.reserve(zp2);314	for _ in 0..zp2 {315		out.push('0');316	}317	out.push_str(&prefix);318319	for digit in digits.into_iter().rev() {320		let ch = NUMBERS[digit as usize] as char;321		out.push(if caps { ch.to_ascii_uppercase() } else { ch });322	}323}324325pub fn render_decimal(326	out: &mut String,327	iv: i64,328	padding: usize,329	precision: usize,330	blank: bool,331	sign: bool,332) {333	render_integer(out, iv, padding, precision, blank, sign, 10, "", false)334}335pub fn render_octal(336	out: &mut String,337	iv: i64,338	padding: usize,339	precision: usize,340	alt: bool,341	blank: bool,342	sign: bool,343) {344	render_integer(345		out,346		iv,347		padding,348		precision,349		blank,350		sign,351		8,352		if alt && iv != 0 { "0" } else { "" },353		false,354	)355}356pub fn render_hexadecimal(357	out: &mut String,358	iv: i64,359	padding: usize,360	precision: usize,361	alt: bool,362	blank: bool,363	sign: bool,364	caps: bool,365) {366	render_integer(367		out,368		iv,369		padding,370		precision,371		blank,372		sign,373		16,374		match (alt, caps) {375			(true, true) => "0X",376			(true, false) => "0x",377			(false, _) => "",378		},379		caps,380	)381}382383pub fn render_float(384	out: &mut String,385	n: f64,386	mut padding: usize,387	precision: usize,388	blank: bool,389	sign: bool,390	ensure_pt: bool,391	trailing: bool,392) {393	let dot_size = if precision == 0 && !ensure_pt { 0 } else { 1 };394	padding = padding.saturating_sub(dot_size + precision);395	render_decimal(out, n.floor() as i64, padding, 0, blank, sign);396	if precision == 0 {397		if ensure_pt {398			out.push('.');399		}400		return;401	}402	let frac = (n.fract() * 10.0_f64.powf(precision as f64) + 0.5).floor();403	if trailing || frac > 0.0 {404		out.push('.');405		let mut frac_str = String::new();406		render_decimal(&mut frac_str, frac as i64, precision, 0, false, false);407		let mut trim = frac_str.len();408		if !trailing {409			for b in frac_str.as_bytes().iter().rev() {410				if *b == b'0' {411					trim -= 1;412				}413			}414		}415		out.push_str(&frac_str[..trim]);416	} else if ensure_pt {417		out.push('.');418	}419}420421pub fn render_float_sci(422	out: &mut String,423	n: f64,424	mut padding: usize,425	precision: usize,426	blank: bool,427	sign: bool,428	ensure_pt: bool,429	trailing: bool,430	caps: bool,431) {432	let exponent = n.log10().floor();433	let mantissa = if exponent as i16 == -324 {434		n * 10.0 / 10.0_f64.powf(exponent + 1.0)435	} else {436		n / 10.0_f64.powf(exponent)437	};438	let mut exponent_str = String::new();439	render_decimal(&mut exponent_str, exponent as i64, 3, 0, false, true);440441	// +1 for e442	padding = padding.saturating_sub(exponent_str.len() + 1);443444	render_float(445		out, mantissa, padding, precision, blank, sign, ensure_pt, trailing,446	);447	out.push(if caps { 'E' } else { 'e' });448	out.push_str(&exponent_str);449}450451pub fn format_code(452	out: &mut String,453	value: &Val,454	code: &Code,455	width: usize,456	precision: Option<usize>,457) -> Result<()> {458	let clfags = &code.cflags;459	let (fpprec, iprec) = match precision {460		Some(v) => (v, v),461		None => (6, 0),462	};463	let padding = if clfags.zero && !clfags.left {464		width465	} else {466		0467	};468469	// TODO: If left padded, can optimize by writing directly to out470	let mut tmp_out = String::new();471472	match code.convtype {473		ConvTypeV::String => tmp_out.push_str(&value.clone().to_string()?),474		ConvTypeV::Decimal => {475			let value = value.clone().try_cast_num("%d/%u/%i requires number")?;476			render_decimal(477				&mut tmp_out,478				value as i64,479				padding,480				iprec,481				clfags.blank,482				clfags.sign,483			);484		}485		ConvTypeV::Octal => {486			let value = value.clone().try_cast_num("%o requires number")?;487			render_octal(488				&mut tmp_out,489				value as i64,490				padding,491				iprec,492				clfags.alt,493				clfags.blank,494				clfags.sign,495			);496		}497		ConvTypeV::Hexadecimal => {498			let value = value.clone().try_cast_num("%x/%X requires number")?;499			render_hexadecimal(500				&mut tmp_out,501				value as i64,502				padding,503				iprec,504				clfags.alt,505				clfags.blank,506				clfags.sign,507				code.caps,508			);509		}510		ConvTypeV::Scientific => {511			let value = value.clone().try_cast_num("%e/%E requires number")?;512			render_float_sci(513				&mut tmp_out,514				value,515				padding,516				fpprec,517				clfags.blank,518				clfags.sign,519				clfags.alt,520				true,521				code.caps,522			);523		}524		ConvTypeV::Float => {525			let value = value.clone().try_cast_num("%e/%E requires number")?;526			render_float(527				&mut tmp_out,528				value,529				padding,530				fpprec,531				clfags.blank,532				clfags.sign,533				clfags.alt,534				true,535			);536		}537		ConvTypeV::Shorter => {538			let value = value.clone().try_cast_num("%g/%G requires number")?;539			let exponent = value.log10().floor();540			if exponent < -4.0 || exponent >= fpprec as f64 {541				render_float_sci(542					&mut tmp_out,543					value,544					padding,545					fpprec - 1,546					clfags.blank,547					clfags.sign,548					clfags.alt,549					clfags.alt,550					code.caps,551				);552			} else {553				let digits_before_pt = 1.max(exponent as usize + 1);554				render_float(555					&mut tmp_out,556					value,557					padding,558					fpprec - digits_before_pt,559					clfags.blank,560					clfags.sign,561					clfags.alt,562					clfags.alt,563				);564			}565		}566		ConvTypeV::Char => match value.clone().unwrap_if_lazy()? {567			Val::Num(n) => tmp_out.push(568				std::char::from_u32(n as u32)569					.ok_or_else(|| InvalidUnicodeCodepointGot(n as u32))?,570			),571			Val::Str(s) => {572				if s.chars().count() != 1 {573					throw!(RuntimeError(574						format!("%c expected 1 char string, got {}", s.chars().count()).into(),575					));576				}577				tmp_out.push_str(&s);578			}579			_ => {580				throw!(TypeMismatch(581					"%c requires number/string",582					vec![ValType::Num, ValType::Str],583					value.value_type()?,584				));585			}586		},587		ConvTypeV::Percent => tmp_out.push('%'),588	};589590	let padding = width.saturating_sub(tmp_out.len());591592	if !clfags.left {593		for _ in 0..padding {594			out.push(' ');595		}596	}597	out.push_str(&tmp_out);598	if clfags.left {599		for _ in 0..padding {600			out.push(' ');601		}602	}603604	Ok(())605}606607pub fn format_arr(str: &str, mut values: &[Val]) -> Result<String> {608	let codes = parse_codes(&str)?;609	let mut out = String::new();610611	for code in codes {612		match code {613			Element::String(s) => {614				out.push_str(s);615			}616			Element::Code(c) => {617				let width = match c.width {618					Width::Star => {619						if values.is_empty() {620							throw!(NotEnoughValues);621						}622						let value = &values[0];623						values = &values[1..];624						value.clone().try_cast_num("field width")? as usize625					}626					Width::Fixed(n) => n,627				};628				let precision = match c.precision {629					Some(Width::Star) => {630						if values.is_empty() {631							throw!(NotEnoughValues);632						}633						let value = &values[0];634						values = &values[1..];635						Some(value.clone().try_cast_num("field precision")? as usize)636					}637					Some(Width::Fixed(n)) => Some(n),638					None => None,639				};640641				// %% should not consume a value642				let value = if c.convtype == ConvTypeV::Percent {643					&Val::Null644				} else {645					if values.is_empty() {646						throw!(NotEnoughValues);647					}648					let value = &values[0];649					values = &values[1..];650					value651				};652653				format_code(&mut out, value, &c, width, precision)?;654			}655		}656	}657658	Ok(out)659}660661pub fn format_obj(str: &str, values: &ObjValue) -> Result<String> {662	let codes = parse_codes(&str)?;663	let mut out = String::new();664665	for code in codes {666		match code {667			Element::String(s) => {668				out.push_str(s);669			}670			Element::Code(c) => {671				// TODO: Operate on ref672				let f: Rc<str> = c.mkey.into();673				let width = match c.width {674					Width::Star => {675						throw!(CannotUseStarWidthWithObject);676					}677					Width::Fixed(n) => n,678				};679				let precision = match c.precision {680					Some(Width::Star) => {681						throw!(CannotUseStarWidthWithObject);682					}683					Some(Width::Fixed(n)) => Some(n),684					None => None,685				};686687				let value = if c.convtype == ConvTypeV::Percent {688					Val::Null689				} else {690					if f.is_empty() {691						throw!(MappingKeysRequired);692					}693					let value = if let Some(v) = values.get(f.clone())? {694						v695					} else {696						throw!(NoSuchFormatField(f));697					};698					value699				};700701				format_code(&mut out, &value, &c, width, precision)?;702			}703		}704	}705706	Ok(out)707}708709#[cfg(test)]710pub mod test_format {711	use super::*;712713	#[test]714	fn parse() {715		assert_eq!(716			parse_codes(717				"How much error budget is left looking at our %.3f%% availability gurantees?"718			)719			.unwrap()720			.len(),721			4722		);723	}724725	#[test]726	fn octals() {727		assert_eq!(format_arr("%#o", &[Val::Num(8.0)]).unwrap(), "010");728		assert_eq!(format_arr("%#4o", &[Val::Num(8.0)]).unwrap(), " 010");729		assert_eq!(format_arr("%4o", &[Val::Num(8.0)]).unwrap(), "  10");730		assert_eq!(format_arr("%04o", &[Val::Num(8.0)]).unwrap(), "0010");731		assert_eq!(format_arr("%+4o", &[Val::Num(8.0)]).unwrap(), " +10");732		assert_eq!(format_arr("%+04o", &[Val::Num(8.0)]).unwrap(), "+010");733		assert_eq!(format_arr("%-4o", &[Val::Num(8.0)]).unwrap(), "10  ");734		assert_eq!(format_arr("%+-4o", &[Val::Num(8.0)]).unwrap(), "+10 ");735		assert_eq!(format_arr("%+-04o", &[Val::Num(8.0)]).unwrap(), "+10 ");736	}737738	#[test]739	fn percent_doesnt_consumes_values() {740		assert_eq!(741			format_arr(742				"How much error budget is left looking at our %.3f%% availability gurantees?",743				&[Val::Num(4.0)]744			)745			.unwrap(),746			"How much error budget is left looking at our 4.000% availability gurantees?"747		);748	}749}