nom-xml 0.1.2-pre-alpha

pub mod attribute;
mod debug;
pub mod decode;
mod error;
pub mod extract;
pub mod io;
pub mod misc;
pub mod namespaces;
pub mod parse;
pub mod processing_instruction;
pub mod prolog;
pub mod reference;
pub mod tag;

use std::borrow::Cow;
use std::collections::HashMap;
use std::error::Error;

use crate::misc::MiscState;
use crate::{misc::Misc, parse::Parse};

use crate::processing_instruction::ProcessingInstruction;
use crate::prolog::doctype::DocType;
use crate::prolog::xmldecl::XmlDecl;
use crate::reference::Reference;
use crate::tag::Tag;
//use extract::Extract;
use namespaces::ParseNamespace;
use nom::{
    branch::alt,
    bytes::complete::{tag, take_till},
    combinator::{map, not, opt, verify},
    multi::{many0, many_till},
    sequence::{pair, tuple},
    IResult,
};

#[derive(Clone, PartialEq)]
pub enum Document<'a> {
    Prolog {
        xml_decl: Option<XmlDecl<'a>>,
        misc: Option<Vec<Misc<'a>>>,
        doc_type: Option<DocType<'a>>,
    },
    Element(Tag<'a>, Box<Document<'a>>, Tag<'a>),
    Content(Option<Cow<'a, str>>),
    Nested(Vec<Document<'a>>),
    Empty,
    ProcessingInstruction(ProcessingInstruction<'a>),
    Comment(Cow<'a, str>),
    CDATA(Cow<'a, str>),
}

impl<'a> Parse<'a> for Document<'a> {}

impl<'a> Document<'a> {
    //[22 prolog ::= XMLDecl? Misc* (doctypedecl Misc*)?
    pub fn parse_prolog(input: &'a str) -> IResult<&'a str, Document<'a>> {
        let (input, xml_decl) = opt(XmlDecl::parse)(input)?;
        let (input, misc_before) =
            opt(|input| Misc::parse(input, MiscState::BeforeDoctype))(input)?;
        let (input, doc_type) = opt(DocType::parse)(input)?;
        let (input, misc_after) = match &doc_type {
            Some(_) => opt(|input| Misc::parse(input, MiscState::AfterDoctype))(input)?,
            None => (input, None),
        };

        let miscs: Vec<Option<Misc<'a>>> = vec![misc_before, misc_after];
        let miscs: Vec<Misc<'a>> = miscs.into_iter().flatten().collect();
        let misc = if miscs.is_empty() { None } else { Some(miscs) };

        Ok((
            input,
            Document::Prolog {
                xml_decl,
                misc,
                doc_type,
            },
        ))
    }

    // [14] CharData ::= [^<&]* - ([^<&]* ']]>' [^<&]*)
    fn parse_char_data(input: &'a str) -> IResult<&'a str, Cow<'a, str>> {
        let (input, data) = take_till(|c: char| c == '<' || c == '&')(input)?;
        let (input, _) = not(tag("]]>"))(input)?;
        Ok((input, Cow::Borrowed(data)))
    }

    //[18] CDSect ::= CDStart CData CDEnd
    //[19] CDStart ::= '<![CDATA['
    // [20] CData ::= (Char* - (Char* ']]>' Char*))
    fn parse_cdata(input: &'a str) -> IResult<&'a str, Cow<'a, str>> {
        // Parse until "]]>" or EOF, checking that characters are valid XML characters
        let (input, (data, _)) = many_till(Self::parse_char, tag("]]>"))(input)?;

        // Convert the Vec<char> to a String
        let data: String = data.into_iter().collect();

        Ok((input, Cow::Owned(data)))
    }
    //[21] CDEnd ::= ']]>'
    fn parse_cdata_section(input: &'a str) -> IResult<&'a str, Document<'a>> {
        let (input, _) = tag("<![CDATA[")(input)?;
        let (input, cdata_content) = Self::parse_cdata(input)?;
        let cdata_string: String = cdata_content.to_string();
        Ok((input, Document::CDATA(Cow::Owned(cdata_string))))
    }

    // [39] element	::= EmptyElemTag | STag content ETag
    pub fn parse_element(input: &'a str) -> IResult<&'a str, Document<'a>> {
        let (input, doc) = alt((
            map(Tag::parse_empty_element_tag, |tag| {
                Document::Element(tag.clone(), Box::new(Document::Empty), tag.clone())
            }),
            map(
                tuple((
                    Self::parse_multispace0, // this is not adhering strictly to the spec, but handles the case where there is whitespace before the start tag for readability
                    Tag::parse_start_tag,
                    Self::parse_content,
                    Tag::parse_end_tag,
                    Self::parse_multispace0, // this is not adhering strictly to the spec, but handles the case where there is whitespace after the start tag for readability
                )),
                |(_, start_tag, content, end_tag, _)| {
                    Document::Element(start_tag, Box::new(content), end_tag)
                },
            ),
        ))(input)?;
        Ok((input, doc))
    }

    // [43] content ::= CharData? ((element | Reference | CDSect | PI | Comment) CharData?)*
    // [43] content ::= CharData? ((element | Reference | CDSect | PI | Comment) CharData?)*
    fn parse_content(input: &'a str) -> IResult<&'a str, Document<'a>> {
        let (input, (maybe_chardata, elements)) = tuple((
            opt(Self::parse_char_data),
            many0(pair(
                alt((
                    Self::parse_element,
                    map(Reference::parse, |reference| match reference {
                        Reference::EntityRef(entity) => Document::Content(Some(entity)),
                        Reference::CharRef { value, .. } => Document::Content(Some(value)),
                    }),
                    Self::parse_cdata_section,
                    map(
                        ProcessingInstruction::parse,
                        Document::ProcessingInstruction,
                    ),
                    Self::parse_comment,
                )),
                opt(Self::parse_char_data),
            )),
        ))(input)?;

        let mut content = elements
            .into_iter()
            .flat_map(|(doc, maybe_chardata)| {
                let mut vec = Vec::new();
                vec.push(doc);
                if let Some(chardata) = maybe_chardata {
                    if !chardata.is_empty() {
                        vec.push(Document::Content(Some(chardata)));
                    }
                }
                vec
            })
            .collect::<Vec<_>>();

        Ok((
            input,
            match maybe_chardata {
                Some(chardata) if !chardata.is_empty() => {
                    let mut vec = Vec::new();
                    vec.push(Document::Content(Some(chardata)));
                    vec.append(&mut content);
                    Document::Nested(vec)
                }
                _ => match &content[..] {
                    [Document::Content(content)] => Document::Content(content.clone()),
                    _ => Document::Nested(content),
                },
            },
        ))
    }

    // [15] Comment ::= '<!--' ((Char - '-') | ('-' (Char - '-')))* '-->'
    pub fn parse_comment(input: &'a str) -> IResult<&'a str, Document<'a>> {
        let (input, _) = tag("<!--")(input)?;
        let (input, (comment_content, _)) =
            many_till(verify(Self::parse_char, |&c| c != '-'), tag("-->"))(input)?;
        let comment_string: String = comment_content.into_iter().collect();
        let (input, _) = tag("-->")(input)?;

        Ok((input, Document::Comment(Cow::Owned(comment_string))))
    }

    pub fn parse_xml_str(input: &'a str) -> IResult<&'a str, Document<'a>> {
        let (input, prolog) = opt(Self::parse_prolog)(input)?;
        let (input, start_tag) = Tag::parse_start_tag(input)?;
        let (input, content) = Self::parse_content(input)?;
        let (input, end_tag) = Tag::parse_end_tag(input)?;

        Self::construct_document(input, prolog, start_tag, content, end_tag)
    }

    fn construct_document(
        input: &'a str,
        prolog: Option<Document<'a>>,
        start_tag: Tag<'a>,
        content: Document<'a>,
        end_tag: Tag<'a>,
    ) -> IResult<&'a str, Document<'a>> {
        match (&start_tag, &end_tag) {
            (
                Tag {
                    name: start_name, ..
                },
                Tag { name: end_name, .. },
            ) if start_name == end_name => {
                //let child_document = Document::Nested(content);

                // Check if a prolog exists and construct document accordingly
                let document = match prolog {
                    Some(prolog) => Document::Nested(vec![
                        prolog,
                        Document::Element(start_tag.clone(), Box::new(content), end_tag.clone()),
                    ]),
                    None => {
                        Document::Element(start_tag.clone(), Box::new(content), end_tag.clone())
                    }
                };

                Ok((input, document))
            }
            _ => Err(nom::Err::Error(nom::error::Error::new(
                input,
                nom::error::ErrorKind::Verify,
            ))),
        }
    }
}

#[derive(Clone, Hash, Eq, PartialEq)]
pub struct QualifiedName<'a> {
    pub prefix: Option<Cow<'a, str>>,
    pub local_part: Cow<'a, str>,
}
pub type Name<'a> = QualifiedName<'a>;

impl<'a> ParseNamespace<'a> for Document<'a> {}

impl<'a> Document<'a> {
    fn extract_content(
        &self,
        tag: &QualifiedName<'a>,
        hashmap: &mut HashMap<QualifiedName<'a>, Vec<Document<'a>>>,
    ) {
        match self {
            Document::Element(start_tag, inner_doc, end_tag) => {
                if &start_tag.name == tag {
                    hashmap
                        .entry(start_tag.name.clone())
                        .or_default()
                        .push(self.clone());
                }
                inner_doc.extract_content(tag, hashmap);
            }
            Document::Nested(docs) => {
                for doc in docs {
                    doc.extract_content(tag, hashmap);
                }
            }
            _ => {} // Handle other Document variants if needed
        }
    }

    pub fn extract(
        &self,
        name: &QualifiedName<'a>,
    ) -> Result<HashMap<QualifiedName<'a>, Vec<Document<'a>>>, Box<dyn Error>> {
        let mut hashmap: HashMap<QualifiedName<'a>, Vec<Document<'a>>> = HashMap::new();
        self.extract_content(name, &mut hashmap);
        Ok(hashmap)
    }

    //pub fn extract_prolog() {}
}