organic/src/parser/plain_link.rs

use nom::branch::alt;
use nom::bytes::complete::tag;
use nom::bytes::complete::tag_no_case;
use nom::character::complete::anychar;
use nom::character::complete::none_of;
use nom::character::complete::one_of;
use nom::combinator::eof;
use nom::combinator::peek;
use nom::combinator::recognize;
use nom::combinator::verify;
use nom::multi::many_till;

use super::org_source::OrgSource;
use crate::context::parser_with_context;
use crate::context::ContextElement;
use crate::context::ExitClass;
use crate::context::ExitMatcherNode;
use crate::context::RefContext;
use crate::error::CustomError;
use crate::error::MyError;
use crate::error::Res;
use crate::parser::util::exit_matcher_parser;
use crate::parser::util::get_consumed;
use crate::parser::util::WORD_CONSTITUENT_CHARACTERS;
use crate::types::PlainLink;

// TODO: Make this a user-provided variable corresponding to elisp's org-link-parameters
const ORG_LINK_PARAMETERS: [&'static str; 23] = [
    "id",
    "eww",
    "rmail",
    "mhe",
    "irc",
    "info",
    "gnus",
    "docview",
    "bibtex",
    "bbdb",
    "w3m",
    "doi",
    "file+sys",
    "file+emacs",
    "shell",
    "news",
    "mailto",
    "https",
    "http",
    "ftp",
    "help",
    "file",
    "elisp",
];

#[cfg_attr(feature = "tracing", tracing::instrument(ret, level = "debug"))]
pub fn plain_link<'b, 'g, 'r, 's>(
    context: RefContext<'b, 'g, 'r, 's>,
    input: OrgSource<'s>,
) -> Res<OrgSource<'s>, PlainLink<'s>> {
    let (remaining, _) = pre(context, input)?;
    let (remaining, proto) = protocol(context, remaining)?;
    let (remaining, _separator) = tag(":")(remaining)?;
    let (remaining, path) = path_plain(context, remaining)?;
    peek(parser_with_context!(post)(context))(remaining)?;
    let source = get_consumed(input, remaining);
    Ok((
        remaining,
        PlainLink {
            source: source.into(),
            link_type: proto.into(),
            path: path.into(),
        },
    ))
}

#[cfg_attr(feature = "tracing", tracing::instrument(ret, level = "debug"))]
fn pre<'b, 'g, 'r, 's>(
    _context: RefContext<'b, 'g, 'r, 's>,
    input: OrgSource<'s>,
) -> Res<OrgSource<'s>, ()> {
    let preceding_character = input.get_preceding_character();
    match preceding_character {
        // If None, we are at the start of the file which is fine
        None => {}
        Some(x) if !WORD_CONSTITUENT_CHARACTERS.contains(x) => {}
        Some(_) => {
            // Not at start of line, cannot be a heading
            return Err(nom::Err::Error(CustomError::MyError(MyError(
                "Not a valid pre character for plain link.".into(),
            ))));
        }
    };
    Ok((input, ()))
}

#[cfg_attr(feature = "tracing", tracing::instrument(ret, level = "debug"))]
fn post<'b, 'g, 'r, 's>(
    _context: RefContext<'b, 'g, 'r, 's>,
    input: OrgSource<'s>,
) -> Res<OrgSource<'s>, ()> {
    let (remaining, _) = alt((eof, recognize(none_of(WORD_CONSTITUENT_CHARACTERS))))(input)?;
    Ok((remaining, ()))
}

#[cfg_attr(feature = "tracing", tracing::instrument(ret, level = "debug"))]
pub fn protocol<'b, 'g, 'r, 's>(
    _context: RefContext<'b, 'g, 'r, 's>,
    input: OrgSource<'s>,
) -> Res<OrgSource<'s>, OrgSource<'s>> {
    // TODO: This should be defined by org-link-parameters
    for link_parameter in ORG_LINK_PARAMETERS {
        let result = tag_no_case::<_, _, CustomError<_>>(link_parameter)(input);
        match result {
            Ok((remaining, ent)) => {
                return Ok((remaining, ent));
            }
            Err(_) => {}
        }
    }

    Err(nom::Err::Error(CustomError::MyError(MyError(
        "NoLinkProtocol".into(),
    ))))
}

#[cfg_attr(feature = "tracing", tracing::instrument(ret, level = "debug"))]
fn path_plain<'b, 'g, 'r, 's>(
    context: RefContext<'b, 'g, 'r, 's>,
    input: OrgSource<'s>,
) -> Res<OrgSource<'s>, OrgSource<'s>> {
    // TODO: "optionally containing parenthesis-wrapped non-whitespace non-bracket substrings up to a depth of two. The string must end with either a non-punctation non-whitespace character, a forwards slash, or a parenthesis-wrapped substring"
    let parser_context = ContextElement::ExitMatcherNode(ExitMatcherNode {
        class: ExitClass::Gamma,
        exit_matcher: &path_plain_end,
    });
    let parser_context = context.with_additional_node(&parser_context);

    let exit_matcher = parser_with_context!(exit_matcher_parser)(&parser_context);

    let (remaining, path) = recognize(verify(
        many_till(anychar, peek(exit_matcher)),
        |(children, _exit_contents)| !children.is_empty(),
    ))(input)?;

    Ok((remaining, path))
}

#[cfg_attr(feature = "tracing", tracing::instrument(ret, level = "debug"))]
fn path_plain_end<'b, 'g, 'r, 's>(
    _context: RefContext<'b, 'g, 'r, 's>,
    input: OrgSource<'s>,
) -> Res<OrgSource<'s>, OrgSource<'s>> {
    recognize(many_till(
        verify(anychar, |c| {
            *c != '/' && (c.is_ascii_punctuation() || c.is_whitespace())
        }),
        one_of(" \t\r\n()[]<>"),
    ))(input)
}