git.delta.rocks / jrsonnet / refs/commits / ff3e2c836938

difftreelog

source

crates/jrsonnet-evaluator/src/stdlib/format.rs16.0 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], &str[i + 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	#[allow(clippy::bool_to_int_with_if)]314	let zp = padding.saturating_sub(if neg || blank || sign { 1 } else { 0 });315	let zp2 = zp316		.max(precision)317		.saturating_sub(prefix.len() + digits.len());318319	if neg {320		out.push('-');321	} else if sign {322		out.push('+');323	} else if blank {324		out.push(' ');325	}326327	out.reserve(zp2);328	for _ in 0..zp2 {329		out.push('0');330	}331	out.push_str(prefix);332333	for digit in digits.into_iter().rev() {334		let ch = NUMBERS[digit as usize] as char;335		out.push(if caps { ch.to_ascii_uppercase() } else { ch });336	}337}338339pub fn render_decimal(340	out: &mut String,341	iv: f64,342	padding: usize,343	precision: usize,344	blank: bool,345	sign: bool,346) {347	render_integer(out, iv, padding, precision, blank, sign, 10, "", false);348}349pub fn render_octal(350	out: &mut String,351	iv: f64,352	padding: usize,353	precision: usize,354	alt: bool,355	blank: bool,356	sign: bool,357) {358	render_integer(359		out,360		iv,361		padding,362		precision,363		blank,364		sign,365		8,366		if alt && iv != 0.0 { "0" } else { "" },367		false,368	);369}370371#[allow(clippy::fn_params_excessive_bools)]372pub fn render_hexadecimal(373	out: &mut String,374	iv: f64,375	padding: usize,376	precision: usize,377	alt: bool,378	blank: bool,379	sign: bool,380	caps: bool,381) {382	render_integer(383		out,384		iv,385		padding,386		precision,387		blank,388		sign,389		16,390		match (alt, caps) {391			(true, true) => "0X",392			(true, false) => "0x",393			(false, _) => "",394		},395		caps,396	);397}398399#[allow(clippy::fn_params_excessive_bools)]400pub fn render_float(401	out: &mut String,402	n: f64,403	mut padding: usize,404	precision: usize,405	blank: bool,406	sign: bool,407	ensure_pt: bool,408	trailing: bool,409) {410	#[allow(clippy::bool_to_int_with_if)]411	let dot_size = if precision == 0 && !ensure_pt { 0 } else { 1 };412	padding = padding.saturating_sub(dot_size + precision);413	render_decimal(out, n.floor(), padding, 0, blank, sign);414	if precision == 0 {415		if ensure_pt {416			out.push('.');417		}418		return;419	}420	let frac = n421		.fract()422		.mul_add(10.0_f64.powf(precision as f64), 0.5)423		.floor();424	if trailing || frac > 0.0 {425		out.push('.');426		let mut frac_str = String::new();427		render_decimal(&mut frac_str, frac, precision, 0, false, false);428		let mut trim = frac_str.len();429		if !trailing {430			for b in frac_str.as_bytes().iter().rev() {431				if *b == b'0' {432					trim -= 1;433				}434			}435		}436		out.push_str(&frac_str[..trim]);437	} else if ensure_pt {438		out.push('.');439	}440}441442#[allow(clippy::fn_params_excessive_bools)]443pub fn render_float_sci(444	out: &mut String,445	n: f64,446	mut padding: usize,447	precision: usize,448	blank: bool,449	sign: bool,450	ensure_pt: bool,451	trailing: bool,452	caps: bool,453) {454	let exponent = n.log10().floor();455	let mantissa = if exponent as i16 == -324 {456		n * 10.0 / 10.0_f64.powf(exponent + 1.0)457	} else {458		n / 10.0_f64.powf(exponent)459	};460	let mut exponent_str = String::new();461	render_decimal(&mut exponent_str, exponent, 3, 0, false, true);462463	// +1 for e464	padding = padding.saturating_sub(exponent_str.len() + 1);465466	render_float(467		out, mantissa, padding, precision, blank, sign, ensure_pt, trailing,468	);469	out.push(if caps { 'E' } else { 'e' });470	out.push_str(&exponent_str);471}472473#[allow(clippy::too_many_lines)]474pub fn format_code(475	s: State,476	out: &mut String,477	value: &Val,478	code: &Code<'_>,479	width: usize,480	precision: Option<usize>,481) -> Result<()> {482	let clfags = &code.cflags;483	let (fpprec, iprec) = precision.map_or((6, 0), |v| (v, v));484	let padding = if clfags.zero && !clfags.left {485		width486	} else {487		0488	};489490	// TODO: If left padded, can optimize by writing directly to out491	let mut tmp_out = String::new();492493	match code.convtype {494		ConvTypeV::String => tmp_out.push_str(&value.clone().to_string(s)?),495		ConvTypeV::Decimal => {496			let value = f64::from_untyped(value.clone(), s)?;497			render_decimal(498				&mut tmp_out,499				value,500				padding,501				iprec,502				clfags.blank,503				clfags.sign,504			);505		}506		ConvTypeV::Octal => {507			let value = f64::from_untyped(value.clone(), s)?;508			render_octal(509				&mut tmp_out,510				value,511				padding,512				iprec,513				clfags.alt,514				clfags.blank,515				clfags.sign,516			);517		}518		ConvTypeV::Hexadecimal => {519			let value = f64::from_untyped(value.clone(), s)?;520			render_hexadecimal(521				&mut tmp_out,522				value,523				padding,524				iprec,525				clfags.alt,526				clfags.blank,527				clfags.sign,528				code.caps,529			);530		}531		ConvTypeV::Scientific => {532			let value = f64::from_untyped(value.clone(), s)?;533			render_float_sci(534				&mut tmp_out,535				value,536				padding,537				fpprec,538				clfags.blank,539				clfags.sign,540				clfags.alt,541				true,542				code.caps,543			);544		}545		ConvTypeV::Float => {546			let value = f64::from_untyped(value.clone(), s)?;547			render_float(548				&mut tmp_out,549				value,550				padding,551				fpprec,552				clfags.blank,553				clfags.sign,554				clfags.alt,555				true,556			);557		}558		ConvTypeV::Shorter => {559			let value = f64::from_untyped(value.clone(), s)?;560			let exponent = value.log10().floor();561			if exponent < -4.0 || exponent >= fpprec as f64 {562				render_float_sci(563					&mut tmp_out,564					value,565					padding,566					fpprec - 1,567					clfags.blank,568					clfags.sign,569					clfags.alt,570					clfags.alt,571					code.caps,572				);573			} else {574				let digits_before_pt = 1.max(exponent as usize + 1);575				render_float(576					&mut tmp_out,577					value,578					padding,579					fpprec - digits_before_pt,580					clfags.blank,581					clfags.sign,582					clfags.alt,583					clfags.alt,584				);585			}586		}587		ConvTypeV::Char => match value.clone() {588			Val::Num(n) => tmp_out.push(589				std::char::from_u32(n as u32)590					.ok_or_else(|| InvalidUnicodeCodepointGot(n as u32))?,591			),592			Val::Str(s) => {593				if s.chars().count() != 1 {594					throw!(RuntimeError(595						format!("%c expected 1 char string, got {}", s.chars().count()).into(),596					));597				}598				tmp_out.push_str(&s);599			}600			_ => {601				throw!(TypeMismatch(602					"%c requires number/string",603					vec![ValType::Num, ValType::Str],604					value.value_type(),605				));606			}607		},608		ConvTypeV::Percent => tmp_out.push('%'),609	};610611	let padding = width.saturating_sub(tmp_out.len());612613	if !clfags.left {614		for _ in 0..padding {615			out.push(' ');616		}617	}618	out.push_str(&tmp_out);619	if clfags.left {620		for _ in 0..padding {621			out.push(' ');622		}623	}624625	Ok(())626}627628pub fn format_arr(s: State, str: &str, mut values: &[Val]) -> Result<String> {629	let codes = parse_codes(str)?;630	let mut out = String::new();631632	for code in codes {633		match code {634			Element::String(s) => {635				out.push_str(s);636			}637			Element::Code(c) => {638				let width = match c.width {639					Width::Star => {640						if values.is_empty() {641							throw!(NotEnoughValues);642						}643						let value = &values[0];644						values = &values[1..];645						usize::from_untyped(value.clone(), s.clone())?646					}647					Width::Fixed(n) => n,648				};649				let precision = match c.precision {650					Some(Width::Star) => {651						if values.is_empty() {652							throw!(NotEnoughValues);653						}654						let value = &values[0];655						values = &values[1..];656						Some(usize::from_untyped(value.clone(), s.clone())?)657					}658					Some(Width::Fixed(n)) => Some(n),659					None => None,660				};661662				// %% should not consume a value663				let value = if c.convtype == ConvTypeV::Percent {664					&Val::Null665				} else {666					if values.is_empty() {667						throw!(NotEnoughValues);668					}669					let value = &values[0];670					values = &values[1..];671					value672				};673674				format_code(s.clone(), &mut out, value, &c, width, precision)?;675			}676		}677	}678679	Ok(out)680}681682pub fn format_obj(s: State, str: &str, values: &ObjValue) -> Result<String> {683	let codes = parse_codes(str)?;684	let mut out = String::new();685686	for code in codes {687		match code {688			Element::String(s) => {689				out.push_str(s);690			}691			Element::Code(c) => {692				// TODO: Operate on ref693				let f: IStr = c.mkey.into();694				let width = match c.width {695					Width::Star => {696						throw!(CannotUseStarWidthWithObject);697					}698					Width::Fixed(n) => n,699				};700				let precision = match c.precision {701					Some(Width::Star) => {702						throw!(CannotUseStarWidthWithObject);703					}704					Some(Width::Fixed(n)) => Some(n),705					None => None,706				};707708				let value = if c.convtype == ConvTypeV::Percent {709					Val::Null710				} else {711					if f.is_empty() {712						throw!(MappingKeysRequired);713					}714					if let Some(v) = values.get(s.clone(), f.clone())? {715						v716					} else {717						throw!(NoSuchFormatField(f));718					}719				};720721				format_code(s.clone(), &mut out, &value, &c, width, precision)?;722			}723		}724	}725726	Ok(out)727}728729#[cfg(test)]730pub mod test_format {731	use super::*;732733	#[test]734	fn parse() {735		assert_eq!(736			parse_codes(737				"How much error budget is left looking at our %.3f%% availability gurantees?"738			)739			.unwrap()740			.len(),741			4742		);743	}744745	#[test]746	fn octals() {747		let s = State::default();748		assert_eq!(749			format_arr(s.clone(), "%#o", &[Val::Num(8.0)]).unwrap(),750			"010"751		);752		assert_eq!(753			format_arr(s.clone(), "%#4o", &[Val::Num(8.0)]).unwrap(),754			" 010"755		);756		assert_eq!(757			format_arr(s.clone(), "%4o", &[Val::Num(8.0)]).unwrap(),758			"  10"759		);760		assert_eq!(761			format_arr(s.clone(), "%04o", &[Val::Num(8.0)]).unwrap(),762			"0010"763		);764		assert_eq!(765			format_arr(s.clone(), "%+4o", &[Val::Num(8.0)]).unwrap(),766			" +10"767		);768		assert_eq!(769			format_arr(s.clone(), "%+04o", &[Val::Num(8.0)]).unwrap(),770			"+010"771		);772		assert_eq!(773			format_arr(s.clone(), "%-4o", &[Val::Num(8.0)]).unwrap(),774			"10  "775		);776		assert_eq!(777			format_arr(s.clone(), "%+-4o", &[Val::Num(8.0)]).unwrap(),778			"+10 "779		);780		assert_eq!(format_arr(s, "%+-04o", &[Val::Num(8.0)]).unwrap(), "+10 ");781	}782783	#[test]784	fn percent_doesnt_consumes_values() {785		let s = State::default();786		assert_eq!(787			format_arr(788				s,789				"How much error budget is left looking at our %.3f%% availability gurantees?",790				&[Val::Num(4.0)]791			)792			.unwrap(),793			"How much error budget is left looking at our 4.000% availability gurantees?"794		);795	}796}