git.delta.rocks / jrsonnet / refs/commits / e8ba76475de8

difftreelog

source

crates/jrsonnet-evaluator/src/stdlib/format.rs15.9 KiBsourcehistory
1//! faster std.format impl2#![allow(clippy::too_many_arguments)]34use jrsonnet_gcmodule::Trace;5use jrsonnet_interner::IStr;6use jrsonnet_types::ValType;7use thiserror::Error;89use crate::{error::Error::*, throw, typed::Typed, LocError, ObjValue, Result, State, Val};1011#[derive(Debug, Clone, Error, Trace)]12pub enum FormatError {13	#[error("truncated format code")]14	TruncatedFormatCode,15	#[error("unrecognized conversion type: {0}")]16	UnrecognizedConversionType(char),1718	#[error("not enough values")]19	NotEnoughValues,2021	#[error("cannot use * width with object")]22	CannotUseStarWidthWithObject,23	#[error("mapping keys required")]24	MappingKeysRequired,25	#[error("no such format field: {0}")]26	NoSuchFormatField(IStr),27}2829impl From<FormatError> for LocError {30	fn from(e: FormatError) -> Self {31		Self::new(Format(e))32	}33}3435use FormatError::*;3637type ParseResult<'t, T> = std::result::Result<(T, &'t str), FormatError>;3839pub fn try_parse_mapping_key(str: &str) -> ParseResult<&str> {40	if str.is_empty() {41		return Err(TruncatedFormatCode);42	}43	let bytes = str.as_bytes();44	if bytes[0] == b'(' {45		let mut i = 1;46		while i < bytes.len() {47			if bytes[i] == b')' {48				return Ok((&str[1..i as usize], &str[i as usize + 1..]));49			}50			i += 1;51		}52		Err(TruncatedFormatCode)53	} else {54		Ok(("", str))55	}56}5758#[cfg(test)]59pub mod tests_key {60	use super::*;6162	#[test]63	fn parse_key() {64		assert_eq!(65			try_parse_mapping_key("(hello ) world").unwrap(),66			("hello ", " world")67		);68		assert_eq!(try_parse_mapping_key("() world").unwrap(), ("", " world"));69		assert_eq!(try_parse_mapping_key(" world").unwrap(), ("", " world"));70		assert_eq!(71			try_parse_mapping_key(" () world").unwrap(),72			("", " () world")73		);74	}7576	#[test]77	#[should_panic]78	fn parse_key_missing_start() {79		try_parse_mapping_key("").unwrap();80	}8182	#[test]83	#[should_panic]84	fn parse_key_missing_end() {85		try_parse_mapping_key("(   ").unwrap();86	}87}8889#[allow(clippy::struct_excessive_bools)]90#[derive(Default, Debug)]91pub struct CFlags {92	pub alt: bool,93	pub zero: bool,94	pub left: bool,95	pub blank: bool,96	pub sign: bool,97}9899pub fn try_parse_cflags(str: &str) -> ParseResult<CFlags> {100	if str.is_empty() {101		return Err(TruncatedFormatCode);102	}103	let bytes = str.as_bytes();104	let mut i = 0;105	let mut out = CFlags::default();106	loop {107		if bytes.len() == i {108			return Err(TruncatedFormatCode);109		}110		match bytes[i] {111			b'#' => out.alt = true,112			b'0' => out.zero = true,113			b'-' => out.left = true,114			b' ' => out.blank = true,115			b'+' => out.sign = true,116			_ => break,117		}118		i += 1;119	}120	Ok((out, &str[i..]))121}122123#[derive(Debug, PartialEq, Eq)]124pub enum Width {125	Star,126	Fixed(usize),127}128pub fn try_parse_field_width(str: &str) -> ParseResult<Width> {129	if str.is_empty() {130		return Err(TruncatedFormatCode);131	}132	let bytes = str.as_bytes();133	if bytes[0] == b'*' {134		return Ok((Width::Star, &str[1..]));135	}136	let mut out: usize = 0;137	let mut digits = 0;138	while let Some(digit) = (bytes[digits] as char).to_digit(10) {139		out *= 10;140		out += digit as usize;141		digits += 1;142		if digits == bytes.len() {143			return Err(TruncatedFormatCode);144		}145	}146	Ok((Width::Fixed(out), &str[digits..]))147}148149pub fn try_parse_precision(str: &str) -> ParseResult<Option<Width>> {150	if str.is_empty() {151		return Err(TruncatedFormatCode);152	}153	let bytes = str.as_bytes();154	if bytes[0] == b'.' {155		try_parse_field_width(&str[1..]).map(|(r, s)| (Some(r), s))156	} else {157		Ok((None, str))158	}159}160161// Only skips162pub fn try_parse_length_modifier(str: &str) -> ParseResult<()> {163	if str.is_empty() {164		return Err(TruncatedFormatCode);165	}166	let bytes = str.as_bytes();167	let mut idx = 0;168	while bytes[idx] == b'h' || bytes[idx] == b'l' || bytes[idx] == b'L' {169		idx += 1;170		if bytes.len() == idx {171			return Err(TruncatedFormatCode);172		}173	}174	Ok(((), &str[idx..]))175}176177#[derive(Debug, PartialEq, Eq)]178pub enum ConvTypeV {179	Decimal,180	Octal,181	Hexadecimal,182	Scientific,183	Float,184	Shorter,185	Char,186	String,187	Percent,188}189pub struct ConvType {190	v: ConvTypeV,191	caps: bool,192}193194pub fn parse_conversion_type(str: &str) -> ParseResult<ConvType> {195	if str.is_empty() {196		return Err(TruncatedFormatCode);197	}198199	let code = str.as_bytes()[0];200	let v: (ConvTypeV, bool) = match code {201		b'd' | b'i' | b'u' => (ConvTypeV::Decimal, false),202		b'o' => (ConvTypeV::Octal, false),203		b'x' => (ConvTypeV::Hexadecimal, false),204		b'X' => (ConvTypeV::Hexadecimal, true),205		b'e' => (ConvTypeV::Scientific, false),206		b'E' => (ConvTypeV::Scientific, true),207		b'f' => (ConvTypeV::Float, false),208		b'F' => (ConvTypeV::Float, true),209		b'g' => (ConvTypeV::Shorter, false),210		b'G' => (ConvTypeV::Shorter, true),211		b'c' => (ConvTypeV::Char, false),212		b's' => (ConvTypeV::String, false),213		b'%' => (ConvTypeV::Percent, false),214		c => return Err(UnrecognizedConversionType(c as char)),215	};216217	Ok((ConvType { v: v.0, caps: v.1 }, &str[1..]))218}219220#[derive(Debug)]221pub struct Code<'s> {222	mkey: &'s str,223	cflags: CFlags,224	width: Width,225	precision: Option<Width>,226	convtype: ConvTypeV,227	caps: bool,228}229pub fn parse_code(str: &str) -> ParseResult<Code> {230	if str.is_empty() {231		return Err(TruncatedFormatCode);232	}233	let (mkey, str) = try_parse_mapping_key(str)?;234	let (cflags, str) = try_parse_cflags(str)?;235	let (width, str) = try_parse_field_width(str)?;236	let (precision, str) = try_parse_precision(str)?;237	let (_, str) = try_parse_length_modifier(str)?;238	let (convtype, str) = parse_conversion_type(str)?;239240	Ok((241		Code {242			mkey,243			cflags,244			width,245			precision,246			convtype: convtype.v,247			caps: convtype.caps,248		},249		str,250	))251}252253#[derive(Debug)]254pub enum Element<'s> {255	String(&'s str),256	Code(Code<'s>),257}258pub fn parse_codes(mut str: &str) -> Result<Vec<Element>> {259	let mut bytes = str.as_bytes();260	let mut out = vec![];261	let mut offset = 0;262263	loop {264		while offset != bytes.len() && bytes[offset] != b'%' {265			offset += 1;266		}267		if offset != 0 {268			out.push(Element::String(&str[0..offset]));269		}270		if offset == bytes.len() {271			return Ok(out);272		}273		str = &str[offset + 1..];274		let code;275		(code, str) = parse_code(str)?;276		bytes = str.as_bytes();277		offset = 0;278279		out.push(Element::Code(code));280	}281}282283const NUMBERS: &[u8] = b"0123456789abcdefghijklmnopqrstuvwxyz";284285#[inline]286pub fn render_integer(287	out: &mut String,288	iv: f64,289	padding: usize,290	precision: usize,291	blank: bool,292	sign: bool,293	radix: i64,294	prefix: &str,295	caps: bool,296) {297	let radix = radix as f64;298	let iv = iv.floor();299	// Digit char indexes in reverse order, i.e300	// for radix = 16 and n = 12f: [15, 2, 1]301	let digits = if iv == 0.0 {302		vec![0u8]303	} else {304		let mut v = iv.abs();305		let mut nums = Vec::with_capacity(1);306		while v != 0.0 {307			nums.push((v % radix) as u8);308			v = (v / radix).floor();309		}310		nums311	};312	let neg = iv < 0.0;313	let zp = padding.saturating_sub(if neg || blank || sign { 1 } else { 0 });314	let zp2 = zp315		.max(precision)316		.saturating_sub(prefix.len() + digits.len());317318	if neg {319		out.push('-');320	} else if sign {321		out.push('+');322	} else if blank {323		out.push(' ');324	}325326	out.reserve(zp2);327	for _ in 0..zp2 {328		out.push('0');329	}330	out.push_str(prefix);331332	for digit in digits.into_iter().rev() {333		let ch = NUMBERS[digit as usize] as char;334		out.push(if caps { ch.to_ascii_uppercase() } else { ch });335	}336}337338pub fn render_decimal(339	out: &mut String,340	iv: f64,341	padding: usize,342	precision: usize,343	blank: bool,344	sign: bool,345) {346	render_integer(out, iv, padding, precision, blank, sign, 10, "", false);347}348pub fn render_octal(349	out: &mut String,350	iv: f64,351	padding: usize,352	precision: usize,353	alt: bool,354	blank: bool,355	sign: bool,356) {357	render_integer(358		out,359		iv,360		padding,361		precision,362		blank,363		sign,364		8,365		if alt && iv != 0.0 { "0" } else { "" },366		false,367	);368}369370#[allow(clippy::fn_params_excessive_bools)]371pub fn render_hexadecimal(372	out: &mut String,373	iv: f64,374	padding: usize,375	precision: usize,376	alt: bool,377	blank: bool,378	sign: bool,379	caps: bool,380) {381	render_integer(382		out,383		iv,384		padding,385		precision,386		blank,387		sign,388		16,389		match (alt, caps) {390			(true, true) => "0X",391			(true, false) => "0x",392			(false, _) => "",393		},394		caps,395	);396}397398#[allow(clippy::fn_params_excessive_bools)]399pub fn render_float(400	out: &mut String,401	n: f64,402	mut padding: usize,403	precision: usize,404	blank: bool,405	sign: bool,406	ensure_pt: bool,407	trailing: bool,408) {409	let dot_size = if precision == 0 && !ensure_pt { 0 } else { 1 };410	padding = padding.saturating_sub(dot_size + precision);411	render_decimal(out, n.floor(), padding, 0, blank, sign);412	if precision == 0 {413		if ensure_pt {414			out.push('.');415		}416		return;417	}418	let frac = n419		.fract()420		.mul_add(10.0_f64.powf(precision as f64), 0.5)421		.floor();422	if trailing || frac > 0.0 {423		out.push('.');424		let mut frac_str = String::new();425		render_decimal(&mut frac_str, frac, precision, 0, false, false);426		let mut trim = frac_str.len();427		if !trailing {428			for b in frac_str.as_bytes().iter().rev() {429				if *b == b'0' {430					trim -= 1;431				}432			}433		}434		out.push_str(&frac_str[..trim]);435	} else if ensure_pt {436		out.push('.');437	}438}439440#[allow(clippy::fn_params_excessive_bools)]441pub fn render_float_sci(442	out: &mut String,443	n: f64,444	mut padding: usize,445	precision: usize,446	blank: bool,447	sign: bool,448	ensure_pt: bool,449	trailing: bool,450	caps: bool,451) {452	let exponent = n.log10().floor();453	let mantissa = if exponent as i16 == -324 {454		n * 10.0 / 10.0_f64.powf(exponent + 1.0)455	} else {456		n / 10.0_f64.powf(exponent)457	};458	let mut exponent_str = String::new();459	render_decimal(&mut exponent_str, exponent, 3, 0, false, true);460461	// +1 for e462	padding = padding.saturating_sub(exponent_str.len() + 1);463464	render_float(465		out, mantissa, padding, precision, blank, sign, ensure_pt, trailing,466	);467	out.push(if caps { 'E' } else { 'e' });468	out.push_str(&exponent_str);469}470471#[allow(clippy::too_many_lines)]472pub fn format_code(473	s: State,474	out: &mut String,475	value: &Val,476	code: &Code,477	width: usize,478	precision: Option<usize>,479) -> Result<()> {480	let clfags = &code.cflags;481	let (fpprec, iprec) = match precision {482		Some(v) => (v, v),483		None => (6, 0),484	};485	let padding = if clfags.zero && !clfags.left {486		width487	} else {488		0489	};490491	// TODO: If left padded, can optimize by writing directly to out492	let mut tmp_out = String::new();493494	match code.convtype {495		ConvTypeV::String => tmp_out.push_str(&value.clone().to_string(s)?),496		ConvTypeV::Decimal => {497			let value = f64::from_untyped(value.clone(), s)?;498			render_decimal(499				&mut tmp_out,500				value,501				padding,502				iprec,503				clfags.blank,504				clfags.sign,505			);506		}507		ConvTypeV::Octal => {508			let value = f64::from_untyped(value.clone(), s)?;509			render_octal(510				&mut tmp_out,511				value,512				padding,513				iprec,514				clfags.alt,515				clfags.blank,516				clfags.sign,517			);518		}519		ConvTypeV::Hexadecimal => {520			let value = f64::from_untyped(value.clone(), s)?;521			render_hexadecimal(522				&mut tmp_out,523				value,524				padding,525				iprec,526				clfags.alt,527				clfags.blank,528				clfags.sign,529				code.caps,530			);531		}532		ConvTypeV::Scientific => {533			let value = f64::from_untyped(value.clone(), s)?;534			render_float_sci(535				&mut tmp_out,536				value,537				padding,538				fpprec,539				clfags.blank,540				clfags.sign,541				clfags.alt,542				true,543				code.caps,544			);545		}546		ConvTypeV::Float => {547			let value = f64::from_untyped(value.clone(), s)?;548			render_float(549				&mut tmp_out,550				value,551				padding,552				fpprec,553				clfags.blank,554				clfags.sign,555				clfags.alt,556				true,557			);558		}559		ConvTypeV::Shorter => {560			let value = f64::from_untyped(value.clone(), s)?;561			let exponent = value.log10().floor();562			if exponent < -4.0 || exponent >= fpprec as f64 {563				render_float_sci(564					&mut tmp_out,565					value,566					padding,567					fpprec - 1,568					clfags.blank,569					clfags.sign,570					clfags.alt,571					clfags.alt,572					code.caps,573				);574			} else {575				let digits_before_pt = 1.max(exponent as usize + 1);576				render_float(577					&mut tmp_out,578					value,579					padding,580					fpprec - digits_before_pt,581					clfags.blank,582					clfags.sign,583					clfags.alt,584					clfags.alt,585				);586			}587		}588		ConvTypeV::Char => match value.clone() {589			Val::Num(n) => tmp_out590				.push(std::char::from_u32(n as u32).ok_or(InvalidUnicodeCodepointGot(n as u32))?),591			Val::Str(s) => {592				if s.chars().count() != 1 {593					throw!(RuntimeError(594						format!("%c expected 1 char string, got {}", s.chars().count()).into(),595					));596				}597				tmp_out.push_str(&s);598			}599			_ => {600				throw!(TypeMismatch(601					"%c requires number/string",602					vec![ValType::Num, ValType::Str],603					value.value_type(),604				));605			}606		},607		ConvTypeV::Percent => tmp_out.push('%'),608	};609610	let padding = width.saturating_sub(tmp_out.len());611612	if !clfags.left {613		for _ in 0..padding {614			out.push(' ');615		}616	}617	out.push_str(&tmp_out);618	if clfags.left {619		for _ in 0..padding {620			out.push(' ');621		}622	}623624	Ok(())625}626627pub fn format_arr(s: State, str: &str, mut values: &[Val]) -> Result<String> {628	let codes = parse_codes(str)?;629	let mut out = String::new();630631	for code in codes {632		match code {633			Element::String(s) => {634				out.push_str(s);635			}636			Element::Code(c) => {637				let width = match c.width {638					Width::Star => {639						if values.is_empty() {640							throw!(NotEnoughValues);641						}642						let value = &values[0];643						values = &values[1..];644						usize::from_untyped(value.clone(), s.clone())?645					}646					Width::Fixed(n) => n,647				};648				let precision = match c.precision {649					Some(Width::Star) => {650						if values.is_empty() {651							throw!(NotEnoughValues);652						}653						let value = &values[0];654						values = &values[1..];655						Some(usize::from_untyped(value.clone(), s.clone())?)656					}657					Some(Width::Fixed(n)) => Some(n),658					None => None,659				};660661				// %% should not consume a value662				let value = if c.convtype == ConvTypeV::Percent {663					&Val::Null664				} else {665					if values.is_empty() {666						throw!(NotEnoughValues);667					}668					let value = &values[0];669					values = &values[1..];670					value671				};672673				format_code(s.clone(), &mut out, value, &c, width, precision)?;674			}675		}676	}677678	Ok(out)679}680681pub fn format_obj(s: State, str: &str, values: &ObjValue) -> Result<String> {682	let codes = parse_codes(str)?;683	let mut out = String::new();684685	for code in codes {686		match code {687			Element::String(s) => {688				out.push_str(s);689			}690			Element::Code(c) => {691				// TODO: Operate on ref692				let f: IStr = c.mkey.into();693				let width = match c.width {694					Width::Star => {695						throw!(CannotUseStarWidthWithObject);696					}697					Width::Fixed(n) => n,698				};699				let precision = match c.precision {700					Some(Width::Star) => {701						throw!(CannotUseStarWidthWithObject);702					}703					Some(Width::Fixed(n)) => Some(n),704					None => None,705				};706707				let value = if c.convtype == ConvTypeV::Percent {708					Val::Null709				} else {710					if f.is_empty() {711						throw!(MappingKeysRequired);712					}713					if let Some(v) = values.get(s.clone(), f.clone())? {714						v715					} else {716						throw!(NoSuchFormatField(f));717					}718				};719720				format_code(s.clone(), &mut out, &value, &c, width, precision)?;721			}722		}723	}724725	Ok(out)726}727728#[cfg(test)]729pub mod test_format {730	use super::*;731732	#[test]733	fn parse() {734		assert_eq!(735			parse_codes(736				"How much error budget is left looking at our %.3f%% availability gurantees?"737			)738			.unwrap()739			.len(),740			4741		);742	}743744	#[test]745	fn octals() {746		let s = State::default();747		assert_eq!(748			format_arr(s.clone(), "%#o", &[Val::Num(8.0)]).unwrap(),749			"010"750		);751		assert_eq!(752			format_arr(s.clone(), "%#4o", &[Val::Num(8.0)]).unwrap(),753			" 010"754		);755		assert_eq!(756			format_arr(s.clone(), "%4o", &[Val::Num(8.0)]).unwrap(),757			"  10"758		);759		assert_eq!(760			format_arr(s.clone(), "%04o", &[Val::Num(8.0)]).unwrap(),761			"0010"762		);763		assert_eq!(764			format_arr(s.clone(), "%+4o", &[Val::Num(8.0)]).unwrap(),765			" +10"766		);767		assert_eq!(768			format_arr(s.clone(), "%+04o", &[Val::Num(8.0)]).unwrap(),769			"+010"770		);771		assert_eq!(772			format_arr(s.clone(), "%-4o", &[Val::Num(8.0)]).unwrap(),773			"10  "774		);775		assert_eq!(776			format_arr(s.clone(), "%+-4o", &[Val::Num(8.0)]).unwrap(),777			"+10 "778		);779		assert_eq!(format_arr(s, "%+-04o", &[Val::Num(8.0)]).unwrap(), "+10 ");780	}781782	#[test]783	fn percent_doesnt_consumes_values() {784		let s = State::default();785		assert_eq!(786			format_arr(787				s,788				"How much error budget is left looking at our %.3f%% availability gurantees?",789				&[Val::Num(4.0)]790			)791			.unwrap(),792			"How much error budget is left looking at our 4.000% availability gurantees?"793		);794	}795}