xso: allow FromXml implementations to provide matching hints

That way, callers can put multiple candidate implementations in, for
example, a sorted vector and more efficiently select candidates to try
when looking at a new element.
This commit is contained in:
Jonas Schäfer 2025-05-01 13:30:44 +02:00 committed by Link Mauve
commit 126588b47d
3 changed files with 110 additions and 22 deletions

View file

@ -54,6 +54,9 @@ Version NEXT:
`AsXmlText` and `FromXmlText` implementations based on the standard
library's `Display` and `FromStr` traits.
- `AsXmlDyn`, a dyn-compatible variant of `AsXml` (!573).
- `FromXml::xml_name_matcher()` and related types, which allow `FromXml`
implementors to provide cachable hints to callers about which XML
elements are valid candidates for parsing (!573).
* Changes
- Generated AsXml iterator and FromXml builder types are now
doc(hidden), to not clutter hand-written documentation with auto

View file

@ -17,6 +17,68 @@ use alloc::boxed::Box;
use crate::error::{Error, FromEventsError};
use crate::{FromEventsBuilder, FromXml};
/// Match an XML element qualified name.
#[derive(Debug, Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Hash)]
pub enum XmlNameMatcher<'x> {
/// Match any XML element
Any,
/// Match any XML element in the given namespace.
InNamespace(&'x str),
/// Match any XML element with the exact namespace/name combination.
Specific(&'x str, &'x str),
}
impl<'x> XmlNameMatcher<'x> {
/// Return the superset of two `XmlNameMatcher` instances.
pub const fn superset(self, other: Self) -> Self {
match self {
Self::Any => Self::Any,
Self::InNamespace(my_namespace) => match other {
Self::Any => Self::Any,
Self::InNamespace(other_namespace) | Self::Specific(other_namespace, _) => {
if crate::util::const_str_eq(my_namespace, other_namespace) {
Self::InNamespace(my_namespace)
} else {
Self::Any
}
}
},
Self::Specific(my_namespace, my_name) => match other {
Self::Any => Self::Any,
Self::InNamespace(other_namespace) => {
if crate::util::const_str_eq(my_namespace, other_namespace) {
Self::InNamespace(my_namespace)
} else {
Self::Any
}
}
Self::Specific(other_namespace, other_name) => {
if crate::util::const_str_eq(my_namespace, other_namespace) {
if crate::util::const_str_eq(my_name, other_name) {
Self::Specific(my_name, other_name)
} else {
Self::InNamespace(my_namespace)
}
} else {
Self::Any
}
}
},
}
}
/// Return true if the given `qname` matches this matcher.
pub fn matches(&self, qname: &rxml::QName) -> bool {
match self {
Self::Any => true,
Self::InNamespace(ns) => qname.0.as_str() == *ns,
Self::Specific(ns, name) => qname.0.as_str() == *ns && qname.1.as_str() == *name,
}
}
}
/// # Parsing context for [`FromEventsBuilder`]
///
/// For the most part, [`FromEventsBuilder`] implementations can work with

View file

@ -38,6 +38,34 @@ pub mod minidom_compat;
mod rxml_util;
pub mod text;
// This is a hack to not make `const_str_eq` publicly available, except
// through the `exports` module if the `macros` feature is enabled, but have
// it available internally in all cases.
mod util {
/// Compile-time comparison of two strings.
///
/// Used by macro-generated code.
///
/// This is necessary because `<str as PartialEq>::eq` is not `const`.
pub const fn const_str_eq(a: &str, b: &str) -> bool {
let a = a.as_bytes();
let b = b.as_bytes();
if a.len() != b.len() {
return false;
}
let mut i = 0;
while i < a.len() {
if a[i] != b[i] {
return false;
}
i += 1;
}
true
}
}
#[doc(hidden)]
pub mod exports {
#[cfg(all(feature = "minidom", feature = "macros"))]
@ -75,35 +103,15 @@ pub mod exports {
#[cfg(feature = "macros")]
pub type CoreU8 = u8;
/// Compile-time comparison of two strings.
///
/// Used by macro-generated code.
///
/// This is necessary because `<str as PartialEq>::eq` is not `const`.
#[cfg(feature = "macros")]
pub const fn const_str_eq(a: &'static str, b: &'static str) -> bool {
let a = a.as_bytes();
let b = b.as_bytes();
if a.len() != b.len() {
return false;
}
let mut i = 0;
while i < a.len() {
if a[i] != b[i] {
return false;
}
i += 1;
}
true
}
pub use super::util::const_str_eq;
}
use alloc::{borrow::Cow, boxed::Box, string::String, vec::Vec};
#[doc(inline)]
pub use fromxml::Context;
use fromxml::XmlNameMatcher;
pub use text::TextCodec;
@ -235,6 +243,21 @@ pub trait FromXml {
attrs: rxml::AttrMap,
ctx: &Context<'_>,
) -> Result<Self::Builder, self::error::FromEventsError>;
/// Return a predicate which determines if `Self` *may* be parsed from
/// a given XML element.
///
/// The returned matcher **must** match all elements from which `Self`
/// can be parsed, but it may also match elements from which `Self`
/// cannot be parsed.
///
/// This is an optimisation utility for code locations which have to
/// disambiguate between many `FromXml` implementations. The provided
/// implementation returns a matcher which matches all elements, which is
/// correct, but also very inefficient.
fn xml_name_matcher() -> XmlNameMatcher<'static> {
XmlNameMatcher::Any
}
}
/// Trait allowing to convert XML text to a value.