xrust 2.0.3

/*! # Parse XPath expressions

An XPath expression parser using the xrust parser combinator that produces a xrust transformation.

```rust
use xrust::parser::xpath::parse;
# use xrust::item::Node;
# fn do_parse<N: Node>() {
let t = parse::<N>("/child::A/child::B/child::C", None, None).expect("unable to parse XPath expression");
# }
```

"t" now contains a [Transform] that will return "C" elements that have a "B" parent and an "A" grandparent in the source document.

To evaluate the transformation we need a Context with a source document as its current item.

```rust
# use std::rc::Rc;
# use xrust::xdmerror::{Error, ErrorKind};
use xrust::item::{Sequence, SequenceTrait, Item, Node, NodeType};
use xrust::trees::smite::RNode;
use xrust::parser::ParseError;
use xrust::parser::xml::parse as xmlparse;
use xrust::parser::xpath::parse;
use xrust::transform::context::{Context, ContextBuilder, StaticContext, StaticContextBuilder};

let t = parse("/child::A/child::B/child::C", None, None)
    .expect("unable to parse XPath expression");

let source = RNode::new_document();
xmlparse(source.clone(), "<A><B><C/></B><B><C/></B></A>", Some(|_: &_| Err(ParseError::MissingNameSpace)))
    .expect("unable to parse XML");
let mut static_context = StaticContextBuilder::new()
    .message(|_| Ok(()))
    .fetcher(|_| Ok(String::new()))
    .parser(|_| Err(Error::new(ErrorKind::NotImplemented, "not implemented")))
    .build();
let context = ContextBuilder::new()
    .context(vec![Item::Node(source)])
    .build();
let sequence = context.dispatch(&mut static_context, &t)
    .expect("evaluation failed");
assert_eq!(sequence.len(), 2);
assert_eq!(sequence.to_xml(), "<C/><C/>")
```
*/

mod compare;
mod context;
mod expressions;
mod flwr;
mod functions;
pub(crate) mod literals;
mod logic;
mod nodes;
pub(crate) mod nodetests;
mod numbers;
pub(crate) mod predicates;
mod strings;
pub(crate) mod support;
mod types;
pub(crate) mod variables;

use crate::parser::combinators::alt::alt4;
//use crate::parser::combinators::debug::inspect;
use crate::parser::combinators::list::separated_list1;
use crate::parser::combinators::map::map;
use crate::parser::combinators::tag::tag;
use crate::parser::combinators::tuple::tuple3;
use crate::parser::combinators::whitespace::xpwhitespace;
use crate::parser::xpath::flwr::{for_expr, if_expr, let_expr};
use crate::parser::xpath::logic::or_expr;
use crate::parser::xpath::support::noop;
use crate::parser::{
    ParseError, ParseInput, ParserState, ParserStateBuilder, StaticState, StaticStateBuilder,
};

use crate::item::Node;
use crate::transform::Transform;
use crate::xdmerror::{Error, ErrorKind};
use qualname::{NamespaceMap, NamespacePrefix, NamespaceUri};

/// Parse an XPath expression to produce a [Transform]. The optional [Node] or [NamespaceMap] may be used to resolve XML Namespaces (The [Node] will be searched first).
pub fn parse<N: Node>(
    input: &str,
    n: Option<N>,
    nmap: Option<NamespaceMap>,
) -> Result<Transform<N>, Error> {
    // Shortcut for empty
    if input.is_empty() {
        return Ok(Transform::Empty);
    }

    let state = n.clone().map_or_else(
        || ParserState::new(),
        |m| ParserStateBuilder::new().doc(m.clone()).current(m).build(),
    );
    let result = if let Some(m) = n.clone() {
        let mut static_state = StaticStateBuilder::new()
            .namespace(move |pre| {
                m.namespace_iter()
                    .find(|nsd| nsd.as_namespace_prefix().unwrap().is_some_and(|p| p == pre))
                    .map_or_else(
                        || Err(ParseError::MissingNameSpace),
                        |nsd| Ok(nsd.as_namespace_uri().unwrap().clone()),
                    )
            })
            .build();
        xpath_expr((input, state), &mut static_state)
    } else if let Some(nm) = nmap {
        let mut static_state = StaticStateBuilder::new()
            .namespace(move |pre: &NamespacePrefix| {
                nm.namespace_uri(&Some(pre.clone()))
                    .ok_or(ParseError::MissingNameSpace)
            })
            .build();
        xpath_expr((input, state), &mut static_state)
    } else {
        let mut static_state = StaticStateBuilder::new()
            .namespace(|_| Err(ParseError::MissingNameSpace))
            .build();
        xpath_expr((input, state), &mut static_state)
    };
    match result {
        Ok((_, x)) => Ok(x),
        Err(err) => match err {
            ParseError::Combinator(f) => Err(Error::new(
                ErrorKind::ParseError,
                format!(
                    "Unrecoverable parser error ({}) while parsing XPath expression \"{}\"",
                    f, input
                ),
            )),
            ParseError::NotWellFormed(e) => Err(Error::new(
                ErrorKind::ParseError,
                format!("Unrecognised extra characters: \"{}\"", e),
            )),
            ParseError::MissingNameSpace => Err(Error::new(
                ErrorKind::ParseError,
                "Missing namespace declaration.".to_string(),
            )),
            ParseError::Notimplemented => Err(Error::new(
                ErrorKind::ParseError,
                "Unimplemented feature.".to_string(),
            )),
            _ => Err(Error::new(ErrorKind::Unknown, "Unknown error".to_string())),
        },
    }
}

fn xpath_expr<'a, N: Node + 'a, L>(
    input: ParseInput<'a, N>,
    ss: &mut StaticState<L>,
) -> Result<(ParseInput<'a, N>, Transform<N>), ParseError>
where
    L: FnMut(&NamespacePrefix) -> Result<NamespaceUri, ParseError> + 'a,
{
    match expr::<N, L>()(input, ss) {
        Err(err) => Err(err),
        Ok(((input1, state1), e)) => {
            //Check nothing remaining in iterator, nothing after the end of the root node.
            if input1.is_empty() {
                Ok(((input1, state1), e))
            } else {
                Err(ParseError::NotWellFormed(format!(
                    "Unrecognised extra characters: \"{}\"",
                    input1
                )))
            }
        }
    }
}
// Implementation note: cannot use opaque type because XPath expressions are recursive, and Rust *really* doesn't like recursive opaque types. Dynamic trait objects aren't ideal, but compiling XPath expressions is a one-off operation so that shouldn't cause a major performance issue.
// Implementation note 2: since XPath is recursive, must lazily evaluate arguments to avoid stack overflow.
pub fn expr<'a, N: Node + 'a, L>() -> Box<
    dyn Fn(
            ParseInput<'a, N>,
            &mut StaticState<L>,
        ) -> Result<(ParseInput<'a, N>, Transform<N>), ParseError>
        + 'a,
>
where
    L: FnMut(&NamespacePrefix) -> Result<NamespaceUri, ParseError> + 'a,
{
    Box::new(map(
        separated_list1(
            map(tuple3(xpwhitespace(), tag(","), xpwhitespace()), |_| ()),
            expr_single::<N, L>(),
        ),
        |mut v| {
            if v.len() == 1 {
                v.pop().unwrap()
            } else {
                Transform::SequenceItems(v)
            }
        },
    ))
}

pub(crate) fn expr_wrapper<'a, N: Node + 'a, L>(
    b: bool,
) -> Box<
    dyn Fn(
        ParseInput<'a, N>,
        &mut StaticState<L>,
    ) -> Result<(ParseInput<'a, N>, Transform<N>), ParseError>,
>
where
    L: FnMut(&NamespacePrefix) -> Result<NamespaceUri, ParseError> + 'a,
{
    Box::new(move |input, ss| {
        if b {
            expr::<N, L>()(input, ss)
        } else {
            noop::<N, L>()(input, ss)
        }
    })
}

// ExprSingle ::= ForExpr | LetExpr | QuantifiedExpr | IfExpr | OrExpr
fn expr_single<'a, N: Node + 'a, L>() -> Box<
    dyn Fn(
            ParseInput<'a, N>,
            &mut StaticState<L>,
        ) -> Result<(ParseInput<'a, N>, Transform<N>), ParseError>
        + 'a,
>
where
    L: FnMut(&NamespacePrefix) -> Result<NamespaceUri, ParseError> + 'a,
{
    Box::new(alt4(let_expr(), for_expr(), if_expr(), or_expr()))
}

pub(crate) fn expr_single_wrapper<'a, N: Node + 'a, L>(
    b: bool,
) -> Box<
    dyn Fn(
        ParseInput<'a, N>,
        &mut StaticState<L>,
    ) -> Result<(ParseInput<'a, N>, Transform<N>), ParseError>,
>
where
    L: FnMut(&NamespacePrefix) -> Result<NamespaceUri, ParseError> + 'a,
{
    Box::new(move |input, ss| {
        if b {
            expr_single::<N, L>()(input, ss)
        } else {
            noop::<N, L>()(input, ss)
        }
    })
}