XSD: Drops nom dependency

pull/509/head
Tpt 1 year ago committed by Thomas Tanon
parent 63945638ea
commit b2d625e10e
  1. 1
      Cargo.lock
  2. 3
      lib/oxsdatatypes/Cargo.toml
  3. 21
      lib/oxsdatatypes/src/date_time.rs
  4. 6
      lib/oxsdatatypes/src/duration.rs
  5. 872
      lib/oxsdatatypes/src/parser.rs

1
Cargo.lock generated

@ -1048,7 +1048,6 @@ name = "oxsdatatypes"
version = "0.1.2-dev" version = "0.1.2-dev"
dependencies = [ dependencies = [
"js-sys", "js-sys",
"nom",
] ]
[[package]] [[package]]

@ -13,9 +13,6 @@ An implementation of some XSD datatypes for SPARQL implementations
edition = "2021" edition = "2021"
rust-version = "1.60" rust-version = "1.60"
[dependencies]
nom = "7"
[target.'cfg(all(target_family = "wasm", target_os = "unknown"))'.dependencies] [target.'cfg(all(target_family = "wasm", target_os = "unknown"))'.dependencies]
js-sys = "0.3" js-sys = "0.3"

@ -1,8 +1,7 @@
use super::parser::{date_lexical_rep, date_time_lexical_rep, parse_value, time_lexical_rep};
use super::{DayTimeDuration, Decimal, Duration, XsdParseError, YearMonthDuration}; use super::{DayTimeDuration, Decimal, Duration, XsdParseError, YearMonthDuration};
use crate::parser::{ use crate::parser::{
g_day_lexical_rep, g_month_day_lexical_rep, g_month_lexical_rep, g_year_lexical_rep, parse_date, parse_date_time, parse_g_day, parse_g_month, parse_g_month_day, parse_g_year,
g_year_month_lexical_rep, parse_g_year_month, parse_time,
}; };
use std::cmp::{min, Ordering}; use std::cmp::{min, Ordering};
use std::error::Error; use std::error::Error;
@ -234,7 +233,7 @@ impl FromStr for DateTime {
type Err = XsdParseError; type Err = XsdParseError;
fn from_str(input: &str) -> Result<Self, XsdParseError> { fn from_str(input: &str) -> Result<Self, XsdParseError> {
parse_value(date_time_lexical_rep, input) parse_date_time(input)
} }
} }
@ -445,7 +444,7 @@ impl FromStr for Time {
type Err = XsdParseError; type Err = XsdParseError;
fn from_str(input: &str) -> Result<Self, XsdParseError> { fn from_str(input: &str) -> Result<Self, XsdParseError> {
parse_value(time_lexical_rep, input) parse_time(input)
} }
} }
@ -648,7 +647,7 @@ impl FromStr for Date {
type Err = XsdParseError; type Err = XsdParseError;
fn from_str(input: &str) -> Result<Self, XsdParseError> { fn from_str(input: &str) -> Result<Self, XsdParseError> {
parse_value(date_lexical_rep, input) parse_date(input)
} }
} }
@ -770,7 +769,7 @@ impl FromStr for GYearMonth {
type Err = XsdParseError; type Err = XsdParseError;
fn from_str(input: &str) -> Result<Self, XsdParseError> { fn from_str(input: &str) -> Result<Self, XsdParseError> {
parse_value(g_year_month_lexical_rep, input) parse_g_year_month(input)
} }
} }
@ -891,7 +890,7 @@ impl FromStr for GYear {
type Err = XsdParseError; type Err = XsdParseError;
fn from_str(input: &str) -> Result<Self, XsdParseError> { fn from_str(input: &str) -> Result<Self, XsdParseError> {
parse_value(g_year_lexical_rep, input) parse_g_year(input)
} }
} }
@ -1013,7 +1012,7 @@ impl FromStr for GMonthDay {
type Err = XsdParseError; type Err = XsdParseError;
fn from_str(input: &str) -> Result<Self, XsdParseError> { fn from_str(input: &str) -> Result<Self, XsdParseError> {
parse_value(g_month_day_lexical_rep, input) parse_g_month_day(input)
} }
} }
@ -1139,7 +1138,7 @@ impl FromStr for GMonth {
type Err = XsdParseError; type Err = XsdParseError;
fn from_str(input: &str) -> Result<Self, XsdParseError> { fn from_str(input: &str) -> Result<Self, XsdParseError> {
parse_value(g_month_lexical_rep, input) parse_g_month(input)
} }
} }
@ -1256,7 +1255,7 @@ impl FromStr for GDay {
type Err = XsdParseError; type Err = XsdParseError;
fn from_str(input: &str) -> Result<Self, XsdParseError> { fn from_str(input: &str) -> Result<Self, XsdParseError> {
parse_value(g_day_lexical_rep, input) parse_g_day(input)
} }
} }

@ -135,7 +135,7 @@ impl FromStr for Duration {
type Err = XsdParseError; type Err = XsdParseError;
fn from_str(input: &str) -> Result<Self, XsdParseError> { fn from_str(input: &str) -> Result<Self, XsdParseError> {
parse_value(duration_lexical_rep, input) parse_duration(input)
} }
} }
@ -350,7 +350,7 @@ impl FromStr for YearMonthDuration {
type Err = XsdParseError; type Err = XsdParseError;
fn from_str(input: &str) -> Result<Self, XsdParseError> { fn from_str(input: &str) -> Result<Self, XsdParseError> {
parse_value(year_month_duration_lexical_rep, input) parse_year_month_duration(input)
} }
} }
@ -537,7 +537,7 @@ impl FromStr for DayTimeDuration {
type Err = XsdParseError; type Err = XsdParseError;
fn from_str(input: &str) -> Result<Self, XsdParseError> { fn from_str(input: &str) -> Result<Self, XsdParseError> {
parse_value(day_time_duration_lexical_rep, input) parse_day_time_duration(input)
} }
} }

@ -2,15 +2,6 @@ use super::date_time::{DateTimeError, GDay, GMonth, GMonthDay, GYear, GYearMonth
use super::decimal::ParseDecimalError; use super::decimal::ParseDecimalError;
use super::duration::{DayTimeDuration, YearMonthDuration}; use super::duration::{DayTimeDuration, YearMonthDuration};
use super::*; use super::*;
use nom::branch::alt;
use nom::bytes::complete::{tag, take_while, take_while_m_n};
use nom::character::complete::{char, digit0, digit1};
use nom::combinator::{map, opt, recognize};
use nom::error::{ErrorKind, ParseError};
use nom::multi::many1;
use nom::sequence::{preceded, terminated, tuple};
use nom::Err;
use nom::{IResult, Needed};
use std::error::Error; use std::error::Error;
use std::fmt; use std::fmt;
use std::num::ParseIntError; use std::num::ParseIntError;
@ -24,46 +15,35 @@ pub struct XsdParseError {
#[derive(Debug, Clone)] #[derive(Debug, Clone)]
enum XsdParseErrorKind { enum XsdParseErrorKind {
NomKind(ErrorKind),
NomChar(char),
MissingData(Needed),
TooMuchData { count: usize },
Overflow,
ParseInt(ParseIntError), ParseInt(ParseIntError),
ParseDecimal(ParseDecimalError), ParseDecimal(ParseDecimalError),
OutOfIntegerRange { value: u8, min: u8, max: u8 },
DateTime(DateTimeError), DateTime(DateTimeError),
Message(&'static str),
} }
const OVERFLOW_ERROR: XsdParseError = XsdParseError {
kind: XsdParseErrorKind::Message("Overflow error"),
};
impl fmt::Display for XsdParseError { impl fmt::Display for XsdParseError {
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
match &self.kind { match &self.kind {
XsdParseErrorKind::NomKind(kind) => {
write!(f, "Invalid XML Schema value: {}", kind.description())
}
XsdParseErrorKind::NomChar(c) => {
write!(f, "Unexpected character in XML Schema value: '{c}'")
}
XsdParseErrorKind::MissingData(Needed::Unknown) => {
write!(f, "Too small XML Schema value")
}
XsdParseErrorKind::MissingData(Needed::Size(size)) => {
write!(f, "Too small XML Schema value: missing {size} chars")
}
XsdParseErrorKind::TooMuchData { count } => {
write!(f, "Too long XML Schema value: {count} extra chars")
}
XsdParseErrorKind::Overflow => write!(f, "Computation overflow or underflow"),
XsdParseErrorKind::ParseInt(error) => { XsdParseErrorKind::ParseInt(error) => {
write!(f, "Error while parsing integer: {error}") write!(f, "Error while parsing integer: {error}")
} }
XsdParseErrorKind::ParseDecimal(error) => { XsdParseErrorKind::ParseDecimal(error) => {
write!(f, "Error while parsing decimal: {error}") write!(f, "Error while parsing decimal: {error}")
} }
XsdParseErrorKind::OutOfIntegerRange { value, min, max } => {
write!(f, "The integer {value} is not between {min} and {max}")
}
XsdParseErrorKind::DateTime(error) => error.fmt(f), XsdParseErrorKind::DateTime(error) => error.fmt(f),
XsdParseErrorKind::Message(msg) => write!(f, "{msg}"),
}
}
}
impl XsdParseError {
const fn msg(message: &'static str) -> Self {
Self {
kind: XsdParseErrorKind::Message(message),
} }
} }
} }
@ -74,31 +54,9 @@ impl Error for XsdParseError {
XsdParseErrorKind::ParseInt(error) => Some(error), XsdParseErrorKind::ParseInt(error) => Some(error),
XsdParseErrorKind::ParseDecimal(error) => Some(error), XsdParseErrorKind::ParseDecimal(error) => Some(error),
XsdParseErrorKind::DateTime(error) => Some(error), XsdParseErrorKind::DateTime(error) => Some(error),
_ => None, XsdParseErrorKind::Message(_) => None,
}
}
}
impl ParseError<&str> for XsdParseError {
fn from_error_kind(_input: &str, kind: ErrorKind) -> Self {
Self {
kind: XsdParseErrorKind::NomKind(kind),
} }
} }
fn append(_input: &str, _kind: ErrorKind, other: Self) -> Self {
other
}
fn from_char(_input: &str, c: char) -> Self {
Self {
kind: XsdParseErrorKind::NomChar(c),
}
}
fn or(self, other: Self) -> Self {
other
}
} }
impl From<ParseIntError> for XsdParseError { impl From<ParseIntError> for XsdParseError {
@ -125,412 +83,538 @@ impl From<DateTimeError> for XsdParseError {
} }
} }
impl From<Err<Self>> for XsdParseError { // [6] duYearFrag ::= unsignedNoDecimalPtNumeral 'Y'
fn from(err: Err<Self>) -> Self { // [7] duMonthFrag ::= unsignedNoDecimalPtNumeral 'M'
match err { // [8] duDayFrag ::= unsignedNoDecimalPtNumeral 'D'
Err::Incomplete(needed) => Self { // [9] duHourFrag ::= unsignedNoDecimalPtNumeral 'H'
kind: XsdParseErrorKind::MissingData(needed), // [10] duMinuteFrag ::= unsignedNoDecimalPtNumeral 'M'
}, // [11] duSecondFrag ::= (unsignedNoDecimalPtNumeral | unsignedDecimalPtNumeral) 'S'
Err::Error(e) | Err::Failure(e) => e, // [12] duYearMonthFrag ::= (duYearFrag duMonthFrag?) | duMonthFrag
// [13] duTimeFrag ::= 'T' ((duHourFrag duMinuteFrag? duSecondFrag?) | (duMinuteFrag duSecondFrag?) | duSecondFrag)
// [14] duDayTimeFrag ::= (duDayFrag duTimeFrag?) | duTimeFrag
// [15] durationLexicalRep ::= '-'? 'P' ((duYearMonthFrag duDayTimeFrag?) | duDayTimeFrag)
struct DurationParts {
year_month: Option<i64>,
day_time: Option<Decimal>,
}
fn duration_parts(input: &str) -> Result<(DurationParts, &str), XsdParseError> {
// States
const START: u32 = 0;
const AFTER_YEAR: u32 = 1;
const AFTER_MONTH: u32 = 2;
const AFTER_DAY: u32 = 3;
const AFTER_T: u32 = 4;
const AFTER_HOUR: u32 = 5;
const AFTER_MINUTE: u32 = 6;
const AFTER_SECOND: u32 = 7;
let (negative, input) = if let Some(left) = input.strip_prefix('-') {
(true, left)
} else {
(false, input)
};
let mut input = expect_char(input, 'P', "Durations must start with 'P'")?;
let mut state = START;
let mut year_month: Option<i64> = None;
let mut day_time: Option<Decimal> = None;
while !input.is_empty() {
if let Some(left) = input.strip_prefix('T') {
if state >= AFTER_T {
return Err(XsdParseError::msg("Duplicated time separator 'T'"));
}
state = AFTER_T;
input = left;
} else {
let (number_str, left) = decimal_prefix(input);
match left.chars().next() {
Some('Y') if state < AFTER_YEAR => {
year_month = Some(
year_month
.unwrap_or_default()
.checked_add(
i64::from_str(number_str)?
.checked_mul(12)
.ok_or(OVERFLOW_ERROR)?,
)
.ok_or(OVERFLOW_ERROR)?,
);
state = AFTER_YEAR;
}
Some('M') if state < AFTER_MONTH => {
year_month = Some(
year_month
.unwrap_or_default()
.checked_add(i64::from_str(number_str)?)
.ok_or(OVERFLOW_ERROR)?,
);
state = AFTER_MONTH;
}
Some('D') if state < AFTER_DAY => {
if number_str.contains('.') {
return Err(XsdParseError::msg(
"Decimal numbers are not allowed for days",
));
}
day_time = Some(
day_time
.unwrap_or_default()
.checked_add(
Decimal::from_str(number_str)?
.checked_mul(86400)
.ok_or(OVERFLOW_ERROR)?,
)
.ok_or(OVERFLOW_ERROR)?,
);
state = AFTER_DAY;
}
Some('H') if state == AFTER_T => {
if number_str.contains('.') {
return Err(XsdParseError::msg(
"Decimal numbers are not allowed for hours",
));
}
day_time = Some(
day_time
.unwrap_or_default()
.checked_add(
Decimal::from_str(number_str)?
.checked_mul(3600)
.ok_or(OVERFLOW_ERROR)?,
)
.ok_or(OVERFLOW_ERROR)?,
);
state = AFTER_HOUR;
}
Some('M') if (AFTER_T..AFTER_MINUTE).contains(&state) => {
if number_str.contains('.') {
return Err(XsdParseError::msg(
"Decimal numbers are not allowed for minutes",
));
}
day_time = Some(
day_time
.unwrap_or_default()
.checked_add(
Decimal::from_str(number_str)?
.checked_mul(60)
.ok_or(OVERFLOW_ERROR)?,
)
.ok_or(OVERFLOW_ERROR)?,
);
state = AFTER_MINUTE;
}
Some('S') if (AFTER_T..AFTER_SECOND).contains(&state) => {
day_time = Some(
day_time
.unwrap_or_default()
.checked_add(Decimal::from_str(number_str)?)
.ok_or(OVERFLOW_ERROR)?,
);
state = AFTER_SECOND;
}
Some(_) => return Err(XsdParseError::msg("Unexpected type character")),
None => {
return Err(XsdParseError::msg(
"Numbers in durations must be followed by a type character",
))
}
}
input = &left[1..];
} }
} }
}
type XsdResult<'a, T> = IResult<&'a str, T, XsdParseError>;
const OVERFLOW_ERROR: XsdParseError = XsdParseError {
kind: XsdParseErrorKind::Overflow,
};
pub fn parse_value<'a, T>( Ok((
mut f: impl FnMut(&'a str) -> XsdResult<'a, T>, DurationParts {
input: &'a str, year_month: if let Some(v) = year_month {
) -> Result<T, XsdParseError> { Some(if negative {
let (left, result) = f(input)?; v.checked_neg().ok_or(OVERFLOW_ERROR)?
if left.is_empty() {
Ok(result)
} else { } else {
Err(XsdParseError { v
kind: XsdParseErrorKind::TooMuchData { count: left.len() },
}) })
} } else {
} None
},
//TODO: check every computation day_time: if let Some(v) = day_time {
Some(if negative {
// [6] duYearFrag ::= unsignedNoDecimalPtNumeral 'Y' v.checked_neg().ok_or(OVERFLOW_ERROR)?
fn du_year_frag(input: &str) -> XsdResult<'_, i64> { } else {
terminated(unsigned_no_decimal_pt_numeral, char('Y'))(input) v
} })
} else {
// [7] duMonthFrag ::= unsignedNoDecimalPtNumeral 'M' None
fn du_month_frag(input: &str) -> XsdResult<'_, i64> { },
terminated(unsigned_no_decimal_pt_numeral, char('M'))(input) },
input,
))
} }
// [8] duDayFrag ::= unsignedNoDecimalPtNumeral 'D' pub fn parse_duration(input: &str) -> Result<Duration, XsdParseError> {
fn du_day_frag(input: &str) -> XsdResult<'_, i64> { let parts = ensure_complete(input, duration_parts)?;
terminated(unsigned_no_decimal_pt_numeral, char('D'))(input) if parts.year_month.is_none() && parts.day_time.is_none() {
return Err(XsdParseError::msg("Empty duration"));
}
Ok(Duration::new(
parts.year_month.unwrap_or(0),
parts.day_time.unwrap_or_default(),
))
}
pub fn parse_year_month_duration(input: &str) -> Result<YearMonthDuration, XsdParseError> {
let parts = ensure_complete(input, duration_parts)?;
if parts.day_time.is_some() {
return Err(XsdParseError::msg(
"There must not be any day or time component in a yearMonthDuration",
));
}
Ok(YearMonthDuration::new(parts.year_month.ok_or(
XsdParseError::msg("No year and month values found"),
)?))
} }
// [9] duHourFrag ::= unsignedNoDecimalPtNumeral 'H' pub fn parse_day_time_duration(input: &str) -> Result<DayTimeDuration, XsdParseError> {
fn du_hour_frag(input: &str) -> XsdResult<'_, i64> { let parts = ensure_complete(input, duration_parts)?;
terminated(unsigned_no_decimal_pt_numeral, char('H'))(input) if parts.year_month.is_some() {
return Err(XsdParseError::msg(
"There must not be any year or month component in a dayTimeDuration",
));
}
Ok(DayTimeDuration::new(parts.day_time.ok_or(
XsdParseError::msg("No day or time values found"),
)?))
} }
// [10] duMinuteFrag ::= unsignedNoDecimalPtNumeral 'M' // [16] dateTimeLexicalRep ::= yearFrag '-' monthFrag '-' dayFrag 'T' ((hourFrag ':' minuteFrag ':' secondFrag) | endOfDayFrag) timezoneFrag?
fn du_minute_frag(input: &str) -> XsdResult<'_, i64> { fn date_time_lexical_rep(input: &str) -> Result<(DateTime, &str), XsdParseError> {
terminated(unsigned_no_decimal_pt_numeral, char('M'))(input) let (year, input) = year_frag(input)?;
let input = expect_char(input, '-', "The year and month must be separated by '-'")?;
let (month, input) = month_frag(input)?;
let input = expect_char(input, '-', "The month and day must be separated by '-'")?;
let (day, input) = day_frag(input)?;
let input = expect_char(input, 'T', "The date and time must be separated by 'T'")?;
let (hour, input) = hour_frag(input)?;
let input = expect_char(input, ':', "The hours and minutes must be separated by ':'")?;
let (minute, input) = minute_frag(input)?;
let input = expect_char(
input,
':',
"The minutes and seconds must be separated by ':'",
)?;
let (second, input) = second_frag(input)?;
// We validate 24:00:00
if hour == 24 && minute != 0 && second != Decimal::from(0) {
return Err(XsdParseError::msg(
"Times are not allowed to be after 24:00:00",
));
}
let (timezone_offset, input) = optional_end(input, timezone_frag)?;
Ok((
DateTime::new(year, month, day, hour, minute, second, timezone_offset)?,
input,
))
} }
// [11] duSecondFrag ::= (unsignedNoDecimalPtNumeral | unsignedDecimalPtNumeral) 'S' pub fn parse_date_time(input: &str) -> Result<DateTime, XsdParseError> {
fn du_second_frag(input: &str) -> XsdResult<'_, Decimal> { ensure_complete(input, date_time_lexical_rep)
terminated(
map_res(
recognize(tuple((digit0, opt(preceded(char('.'), digit0))))),
Decimal::from_str,
),
char('S'),
)(input)
} }
// [12] duYearMonthFrag ::= (duYearFrag duMonthFrag?) | duMonthFrag // [17] timeLexicalRep ::= ((hourFrag ':' minuteFrag ':' secondFrag) | endOfDayFrag) timezoneFrag?
fn du_year_month_frag(input: &str) -> XsdResult<'_, i64> { fn time_lexical_rep(input: &str) -> Result<(Time, &str), XsdParseError> {
alt(( let (hour, input) = hour_frag(input)?;
map(tuple((du_year_frag, opt(du_month_frag))), |(y, m)| { let input = expect_char(input, ':', "The hours and minutes must be separated by ':'")?;
12 * y + m.unwrap_or(0) let (minute, input) = minute_frag(input)?;
}), let input = expect_char(
du_month_frag, input,
))(input) ':',
"The minutes and seconds must be separated by ':'",
)?;
let (second, input) = second_frag(input)?;
// We validate 24:00:00
if hour == 24 && minute != 0 && second != Decimal::from(0) {
return Err(XsdParseError::msg(
"Times are not allowed to be after 24:00:00",
));
}
let (timezone_offset, input) = optional_end(input, timezone_frag)?;
Ok((Time::new(hour, minute, second, timezone_offset)?, input))
} }
// [13] duTimeFrag ::= 'T' ((duHourFrag duMinuteFrag? duSecondFrag?) | (duMinuteFrag duSecondFrag?) | duSecondFrag) pub fn parse_time(input: &str) -> Result<Time, XsdParseError> {
fn du_time_frag(input: &str) -> XsdResult<'_, Decimal> { ensure_complete(input, time_lexical_rep)
preceded(
char('T'),
alt((
map_res(
tuple((du_hour_frag, opt(du_minute_frag), opt(du_second_frag))),
|(h, m, s)| {
Decimal::from(3600 * h + 60 * m.unwrap_or(0))
.checked_add(s.unwrap_or_default())
.ok_or(OVERFLOW_ERROR)
},
),
map_res(tuple((du_minute_frag, opt(du_second_frag))), |(m, s)| {
Decimal::from(m * 60)
.checked_add(s.unwrap_or_default())
.ok_or(OVERFLOW_ERROR)
}),
du_second_frag,
)),
)(input)
} }
// [14] duDayTimeFrag ::= (duDayFrag duTimeFrag?) | duTimeFrag // [18] dateLexicalRep ::= yearFrag '-' monthFrag '-' dayFrag timezoneFrag? Constraint: Day-of-month Representations
fn du_day_time_frag(input: &str) -> XsdResult<'_, Decimal> { fn date_lexical_rep(input: &str) -> Result<(Date, &str), XsdParseError> {
alt(( let (year, input) = year_frag(input)?;
map_res(tuple((du_day_frag, opt(du_time_frag))), |(d, t)| { let input = expect_char(input, '-', "The year and month must be separated by '-'")?;
Decimal::from(d) let (month, input) = month_frag(input)?;
.checked_mul(Decimal::from(86400)) let input = expect_char(input, '-', "The month and day must be separated by '-'")?;
.ok_or(OVERFLOW_ERROR)? let (day, input) = day_frag(input)?;
.checked_add(t.unwrap_or_default()) let (timezone_offset, input) = optional_end(input, timezone_frag)?;
.ok_or(OVERFLOW_ERROR) Ok((Date::new(year, month, day, timezone_offset)?, input))
}),
du_time_frag,
))(input)
} }
// [15] durationLexicalRep ::= '-'? 'P' ((duYearMonthFrag duDayTimeFrag?) | duDayTimeFrag) pub fn parse_date(input: &str) -> Result<Date, XsdParseError> {
pub fn duration_lexical_rep(input: &str) -> XsdResult<'_, Duration> { ensure_complete(input, date_lexical_rep)
map(
tuple((
opt(char('-')),
preceded(
char('P'),
alt((
map(
tuple((du_year_month_frag, opt(du_day_time_frag))),
|(y, d)| Duration::new(y, d.unwrap_or_default()),
),
map(du_day_time_frag, |d| Duration::new(0, d)),
)),
),
)),
|(sign, duration)| {
if sign == Some('-') {
-duration
} else {
duration
}
},
)(input)
} }
// [16] dateTimeLexicalRep ::= yearFrag '-' monthFrag '-' dayFrag 'T' ((hourFrag ':' minuteFrag ':' secondFrag) | endOfDayFrag) timezoneFrag? // [19] gYearMonthLexicalRep ::= yearFrag '-' monthFrag timezoneFrag?
pub fn date_time_lexical_rep(input: &str) -> XsdResult<'_, DateTime> { fn g_year_month_lexical_rep(input: &str) -> Result<(GYearMonth, &str), XsdParseError> {
map_res( let (year, input) = year_frag(input)?;
tuple(( let input = expect_char(input, '-', "The year and month must be separated by '-'")?;
year_frag, let (month, input) = month_frag(input)?;
char('-'), let (timezone_offset, input) = optional_end(input, timezone_frag)?;
month_frag, Ok((GYearMonth::new(year, month, timezone_offset)?, input))
char('-'),
day_frag,
char('T'),
alt((
map(
tuple((hour_frag, char(':'), minute_frag, char(':'), second_frag)),
|(h, _, m, _, s)| (h, m, s),
),
end_of_day_frag,
)),
opt(timezone_frag),
)),
|(year, _, month, _, day, _, (hours, minutes, seconds), timezone)| {
DateTime::new(year, month, day, hours, minutes, seconds, timezone)
},
)(input)
} }
// [17] timeLexicalRep ::= ((hourFrag ':' minuteFrag ':' secondFrag) | endOfDayFrag) timezoneFrag? pub fn parse_g_year_month(input: &str) -> Result<GYearMonth, XsdParseError> {
pub fn time_lexical_rep(input: &str) -> XsdResult<'_, Time> { ensure_complete(input, g_year_month_lexical_rep)
map_res(
tuple((
alt((
map(
tuple((hour_frag, char(':'), minute_frag, char(':'), second_frag)),
|(h, _, m, _, s)| (h, m, s),
),
end_of_day_frag,
)),
opt(timezone_frag),
)),
|((hours, minutes, seconds), timezone)| Time::new(hours, minutes, seconds, timezone),
)(input)
} }
// [18] dateLexicalRep ::= yearFrag '-' monthFrag '-' dayFrag timezoneFrag? Constraint: Day-of-month Representations // [20] gYearLexicalRep ::= yearFrag timezoneFrag?
pub fn date_lexical_rep(input: &str) -> XsdResult<'_, Date> { fn g_year_lexical_rep(input: &str) -> Result<(GYear, &str), XsdParseError> {
map_res( let (year, input) = year_frag(input)?;
tuple(( let (timezone_offset, input) = optional_end(input, timezone_frag)?;
year_frag, Ok((GYear::new(year, timezone_offset)?, input))
char('-'),
month_frag,
char('-'),
day_frag,
opt(timezone_frag),
)),
|(year, _, month, _, day, timezone)| Date::new(year, month, day, timezone),
)(input)
} }
// [19] gYearMonthLexicalRep ::= yearFrag '-' monthFrag timezoneFrag? pub fn parse_g_year(input: &str) -> Result<GYear, XsdParseError> {
pub fn g_year_month_lexical_rep(input: &str) -> XsdResult<'_, GYearMonth> { ensure_complete(input, g_year_lexical_rep)
map_res(
tuple((year_frag, char('-'), month_frag, opt(timezone_frag))),
|(year, _, month, timezone)| GYearMonth::new(year, month, timezone),
)(input)
} }
// [20] gYearLexicalRep ::= yearFrag timezoneFrag? // [21] gMonthDayLexicalRep ::= '--' monthFrag '-' dayFrag timezoneFrag? Constraint: Day-of-month Representations
pub fn g_year_lexical_rep(input: &str) -> XsdResult<'_, GYear> { fn g_month_day_lexical_rep(input: &str) -> Result<(GMonthDay, &str), XsdParseError> {
map_res( let input = expect_char(input, '-', "gMonthDay values must start with '--'")?;
tuple((year_frag, opt(timezone_frag))), let input = expect_char(input, '-', "gMonthDay values must start with '--'")?;
|(year, timezone)| GYear::new(year, timezone), let (month, input) = month_frag(input)?;
)(input) let input = expect_char(input, '-', "The month and day must be separated by '-'")?;
let (day, input) = day_frag(input)?;
let (timezone_offset, input) = optional_end(input, timezone_frag)?;
Ok((GMonthDay::new(month, day, timezone_offset)?, input))
} }
// [21] gMonthDayLexicalRep ::= '--' monthFrag '-' dayFrag timezoneFrag? Constraint: Day-of-month Representations pub fn parse_g_month_day(input: &str) -> Result<GMonthDay, XsdParseError> {
pub fn g_month_day_lexical_rep(input: &str) -> XsdResult<'_, GMonthDay> { ensure_complete(input, g_month_day_lexical_rep)
map_res(
tuple((
char('-'),
char('-'),
month_frag,
char('-'),
day_frag,
opt(timezone_frag),
)),
|(_, _, month, _, day, timezone)| GMonthDay::new(month, day, timezone),
)(input)
} }
// [22] gDayLexicalRep ::= '---' dayFrag timezoneFrag? // [22] gDayLexicalRep ::= '---' dayFrag timezoneFrag?
pub fn g_day_lexical_rep(input: &str) -> XsdResult<'_, GDay> { fn g_day_lexical_rep(input: &str) -> Result<(GDay, &str), XsdParseError> {
map_res( let input = expect_char(input, '-', "gDay values must start with '---'")?;
tuple(( let input = expect_char(input, '-', "gDay values must start with '---'")?;
char('-'), let input = expect_char(input, '-', "gDay values must start with '---'")?;
char('-'), let (day, input) = day_frag(input)?;
char('-'), let (timezone_offset, input) = optional_end(input, timezone_frag)?;
day_frag, Ok((GDay::new(day, timezone_offset)?, input))
opt(timezone_frag),
)),
|(_, _, _, day, timezone)| GDay::new(day, timezone),
)(input)
} }
// [23] gMonthLexicalRep ::= '--' monthFrag timezoneFrag? pub fn parse_g_day(input: &str) -> Result<GDay, XsdParseError> {
pub fn g_month_lexical_rep(input: &str) -> XsdResult<'_, GMonth> { ensure_complete(input, g_day_lexical_rep)
map_res(
tuple((char('-'), char('-'), month_frag, opt(timezone_frag))),
|(_, _, month, timezone)| GMonth::new(month, timezone),
)(input)
}
// [42] yearMonthDurationLexicalRep ::= '-'? 'P' duYearMonthFrag
pub fn year_month_duration_lexical_rep(input: &str) -> XsdResult<'_, YearMonthDuration> {
map(
tuple((opt(char('-')), preceded(char('P'), du_year_month_frag))),
|(sign, duration)| {
YearMonthDuration::new(if sign == Some('-') {
-duration
} else {
duration
})
},
)(input)
} }
// [43] dayTimeDurationLexicalRep ::= '-'? 'P' duDayTimeFrag // [23] gMonthLexicalRep ::= '--' monthFrag timezoneFrag?
pub fn day_time_duration_lexical_rep(input: &str) -> XsdResult<'_, DayTimeDuration> { fn g_month_lexical_rep(input: &str) -> Result<(GMonth, &str), XsdParseError> {
map( let input = expect_char(input, '-', "gMonth values must start with '--'")?;
tuple((opt(char('-')), preceded(char('P'), du_day_time_frag))), let input = expect_char(input, '-', "gMonth values must start with '--'")?;
|(sign, duration)| { let (month, input) = month_frag(input)?;
DayTimeDuration::new(if sign == Some('-') { let (timezone_offset, input) = optional_end(input, timezone_frag)?;
-duration Ok((GMonth::new(month, timezone_offset)?, input))
} else {
duration
})
},
)(input)
} }
// [46] unsignedNoDecimalPtNumeral ::= digit+ pub fn parse_g_month(input: &str) -> Result<GMonth, XsdParseError> {
fn unsigned_no_decimal_pt_numeral(input: &str) -> XsdResult<'_, i64> { ensure_complete(input, g_month_lexical_rep)
map_res(digit1, i64::from_str)(input)
} }
// [56] yearFrag ::= '-'? (([1-9] digit digit digit+)) | ('0' digit digit digit)) // [56] yearFrag ::= '-'? (([1-9] digit digit digit+)) | ('0' digit digit digit))
fn year_frag(input: &str) -> XsdResult<'_, i64> { fn year_frag(input: &str) -> Result<(i64, &str), XsdParseError> {
map_res( let (sign, input) = if let Some(left) = input.strip_prefix('-') {
recognize(tuple(( (-1, left)
opt(char('-')), } else {
take_while_m_n(4, usize::MAX, |c: char| c.is_ascii_digit()), (1, input)
))), };
i64::from_str, let (number_str, input) = integer_prefix(input);
)(input) let number = i64::from_str(number_str)?;
if number < 1000 && number_str.len() != 4 {
return Err(XsdParseError::msg(
"The years below 1000 must be encoded on exactly 4 digits",
));
}
Ok((sign * number, input))
} }
// [57] monthFrag ::= ('0' [1-9]) | ('1' [0-2]) // [57] monthFrag ::= ('0' [1-9]) | ('1' [0-2])
fn month_frag(input: &str) -> XsdResult<'_, u8> { fn month_frag(input: &str) -> Result<(u8, &str), XsdParseError> {
map_res(take_while_m_n(2, 2, |c: char| c.is_ascii_digit()), |v| { let (number_str, input) = integer_prefix(input);
parsed_u8_range(v, 1, 12) if number_str.len() != 2 {
})(input) return Err(XsdParseError::msg("Month must be encoded with two digits"));
}
let number = u8::from_str(number_str)?;
if !(1..=12).contains(&number) {
return Err(XsdParseError::msg("Month must be between 01 and 12"));
}
Ok((number, input))
} }
// [58] dayFrag ::= ('0' [1-9]) | ([12] digit) | ('3' [01]) // [58] dayFrag ::= ('0' [1-9]) | ([12] digit) | ('3' [01])
fn day_frag(input: &str) -> XsdResult<'_, u8> { fn day_frag(input: &str) -> Result<(u8, &str), XsdParseError> {
map_res(take_while_m_n(2, 2, |c: char| c.is_ascii_digit()), |v| { let (number_str, input) = integer_prefix(input);
parsed_u8_range(v, 1, 31) if number_str.len() != 2 {
})(input) return Err(XsdParseError::msg("Day must be encoded with two digits"));
}
let number = u8::from_str(number_str)?;
if !(1..=31).contains(&number) {
return Err(XsdParseError::msg("Day must be between 01 and 31"));
}
Ok((number, input))
} }
// [59] hourFrag ::= ([01] digit) | ('2' [0-3]) // [59] hourFrag ::= ([01] digit) | ('2' [0-3])
fn hour_frag(input: &str) -> XsdResult<'_, u8> { // We also allow 24 for ease of parsing
map_res(take_while_m_n(2, 2, |c: char| c.is_ascii_digit()), |v| { fn hour_frag(input: &str) -> Result<(u8, &str), XsdParseError> {
parsed_u8_range(v, 0, 23) let (number_str, input) = integer_prefix(input);
})(input) if number_str.len() != 2 {
return Err(XsdParseError::msg("Hours must be encoded with two digits"));
}
let number = u8::from_str(number_str)?;
if !(0..=24).contains(&number) {
return Err(XsdParseError::msg("Hours must be between 00 and 24"));
}
Ok((number, input))
} }
// [60] minuteFrag ::= [0-5] digit // [60] minuteFrag ::= [0-5] digit
fn minute_frag(input: &str) -> XsdResult<'_, u8> { fn minute_frag(input: &str) -> Result<(u8, &str), XsdParseError> {
map_res(take_while_m_n(2, 2, |c: char| c.is_ascii_digit()), |v| { let (number_str, input) = integer_prefix(input);
parsed_u8_range(v, 0, 59) if number_str.len() != 2 {
})(input) return Err(XsdParseError::msg(
"Minutes must be encoded with two digits",
));
}
let number = u8::from_str(number_str)?;
if !(0..=59).contains(&number) {
return Err(XsdParseError::msg("Minutes must be between 00 and 59"));
}
Ok((number, input))
} }
// [61] secondFrag ::= ([0-5] digit) ('.' digit+)? // [61] secondFrag ::= ([0-5] digit) ('.' digit+)?
#[allow(clippy::cast_possible_truncation, clippy::cast_sign_loss)] fn second_frag(input: &str) -> Result<(Decimal, &str), XsdParseError> {
fn second_frag(input: &str) -> XsdResult<'_, Decimal> { let (number_str, input) = decimal_prefix(input);
map_res( let (before_dot_str, _) = number_str.split_once('.').unwrap_or((number_str, ""));
recognize(tuple(( if before_dot_str.len() != 2 {
take_while_m_n(2, 2, |c: char| c.is_ascii_digit()), return Err(XsdParseError::msg(
opt(preceded( "Seconds must be encoded with two digits",
char('.'), ));
take_while(|c: char| c.is_ascii_digit()), }
)), let number = Decimal::from_str(number_str)?;
))), if number < Decimal::from(0) || number >= Decimal::from(60) {
|v| { return Err(XsdParseError::msg("Seconds must be between 00 and 60"));
let value = Decimal::from_str(v)?; }
if Decimal::from(0) <= value && value < Decimal::from(60) { if number_str.ends_with('.') {
Ok(value) return Err(XsdParseError::msg(
"Seconds are not allowed to end with a dot",
));
}
Ok((number, input))
}
// [63] timezoneFrag ::= 'Z' | ('+' | '-') (('0' digit | '1' [0-3]) ':' minuteFrag | '14:00')
fn timezone_frag(input: &str) -> Result<(TimezoneOffset, &str), XsdParseError> {
if let Some(left) = input.strip_prefix('Z') {
return Ok((TimezoneOffset::UTC, left));
}
let (sign, input) = if let Some(left) = input.strip_prefix('-') {
(-1, left)
} else if let Some(left) = input.strip_prefix('+') {
(1, left)
} else { } else {
Err(XsdParseError { (1, input)
kind: XsdParseErrorKind::OutOfIntegerRange { };
value: value.as_i128() as u8,
min: 0, let (hour_str, input) = integer_prefix(input);
max: 60, if hour_str.len() != 2 {
}, return Err(XsdParseError::msg(
}) "The timezone hours must be encoded with two digits",
));
} }
}, let hours = i16::from_str(hour_str)?;
)(input)
let input = expect_char(
input,
':',
"The timezone hours and minutes must be separated by ':'",
)?;
let (minutes, input) = minute_frag(input)?;
if hours > 13 && !(hours == 14 && minutes == 0) {
return Err(XsdParseError::msg(
"The timezone hours must be between 00 and 13",
));
}
Ok((
TimezoneOffset::new(sign * (hours * 60 + i16::from(minutes)))?,
input,
))
} }
// [62] endOfDayFrag ::= '24:00:00' ('.' '0'+)? fn ensure_complete<T>(
fn end_of_day_frag(input: &str) -> XsdResult<'_, (u8, u8, Decimal)> { input: &str,
map( parse: impl FnOnce(&str) -> Result<(T, &str), XsdParseError>,
recognize(tuple(( ) -> Result<T, XsdParseError> {
tag("24:00:00"), let (result, left) = parse(input)?;
opt(preceded(char('.'), many1(char('0')))), if !left.is_empty() {
))), return Err(XsdParseError::msg("Unrecognized value suffix"));
|_| (24, 0, 0.into()), }
)(input) Ok(result)
} }
// [63] timezoneFrag ::= 'Z' | ('+' | '-') (('0' digit | '1' [0-3]) ':' minuteFrag | '14:00') fn expect_char<'a>(
fn timezone_frag(input: &str) -> XsdResult<'_, TimezoneOffset> { input: &'a str,
alt(( constant: char,
map(char('Z'), |_| TimezoneOffset::UTC), error_message: &'static str,
map_res( ) -> Result<&'a str, XsdParseError> {
tuple(( if let Some(left) = input.strip_prefix(constant) {
alt((map(char('+'), |_| 1), map(char('-'), |_| -1))), Ok(left)
alt((
map(
tuple((
map_res(take_while_m_n(2, 2, |c: char| c.is_ascii_digit()), |v| {
parsed_u8_range(v, 0, 13)
}),
char(':'),
minute_frag,
)),
|(hours, _, minutes)| i16::from(hours) * 60 + i16::from(minutes),
),
map(tag("14:00"), |_| 14 * 60),
)),
)),
|(sign, value)| TimezoneOffset::new(sign * value),
),
))(input)
}
fn parsed_u8_range(input: &str, min: u8, max: u8) -> Result<u8, XsdParseError> {
let value = u8::from_str(input)?;
if min <= value && value <= max {
Ok(value)
} else { } else {
Err(XsdParseError { Err(XsdParseError::msg(error_message))
kind: XsdParseErrorKind::OutOfIntegerRange { value, min, max }, }
}) }
fn integer_prefix(input: &str) -> (&str, &str) {
let mut end = input.len();
for (i, c) in input.char_indices() {
if !c.is_ascii_digit() {
end = i;
break;
}
} }
input.split_at(end)
} }
fn map_res<'a, O1, O2, E2: Into<XsdParseError>>( fn decimal_prefix(input: &str) -> (&str, &str) {
mut first: impl FnMut(&'a str) -> XsdResult<'a, O1>, let mut end = input.len();
mut second: impl FnMut(O1) -> Result<O2, E2>, let mut dot_seen = false;
) -> impl FnMut(&'a str) -> XsdResult<'a, O2> { for (i, c) in input.char_indices() {
move |input| { if c.is_ascii_digit() {
let (input, o1) = first(input)?; // Ok
Ok((input, second(o1).map_err(|e| Err::Error(e.into()))?)) } else if c == '.' && !dot_seen {
dot_seen = true;
} else {
end = i;
break;
}
} }
input.split_at(end)
}
fn optional_end<T>(
input: &str,
parse: impl FnOnce(&str) -> Result<(T, &str), XsdParseError>,
) -> Result<(Option<T>, &str), XsdParseError> {
Ok(if input.is_empty() {
(None, input)
} else {
let (result, input) = parse(input)?;
(Some(result), input)
})
} }

Loading…
Cancel
Save