path: root/src/pagespec.rs

//! Parse and match a PageSpec.
//!
//! A [PageSpec](http://ikiwiki.info/ikiwiki/pagespec/) is an
//! expression that selects pages using an expression. See the ikiwiki
//! documentation for more.

use crate::site::Site;
use log::trace;
use std::path::{Path, PathBuf};

use lalrpop_util::lalrpop_mod;

lalrpop_mod!(pub pagespec);

/// A parsed PageSpec expression, ready to be matched on page paths.
#[derive(Debug)]
pub struct PageSpec {
    container: PathBuf,
    expr: Box<Expr>,
}

impl PageSpec {
    /// Create a new PageSpec.
    pub fn new(containing_page: &Path, spec: &str) -> Result<Self, PageSpecError> {
        trace!(
            "PageSpec::new: containing_page={} spec={:?}",
            containing_page.display(),
            spec
        );
        let expr = pagespec::ExprParser::new()
            .parse(spec)
            .map_err(|e| PageSpecError::Parse(format!("{}", e)))?;
        Ok(Self {
            container: containing_page.into(),
            expr,
        })
    }

    /// Match a PageSpec on a page path.
    pub fn matches(&self, site: &Site, page_path: &Path) -> bool {
        trace!(
            "PageSpec::matches: container={} page_path={}",
            self.container.display(),
            page_path.display()
        );
        assert!(page_path.is_absolute());
        if let Ok(path) = page_path.strip_prefix(&self.container) {
            let path = format!("{}", path.display());
            self.expr.matches(site, &self.container, &path)
        } else {
            false
        }
    }
}

/// Errors from PageSpec parsing.
#[derive(Debug, thiserror::Error)]
pub enum PageSpecError {
    #[error("failed to parse PageSpec expression: {0}")]
    Parse(String),
}

/// This is public only because parser generation needs it to be.
#[derive(Debug)]
pub enum Expr {
    Glob(String),
    LinksHereFunc(String),
    PageFunc(String),
    TaggedFunc(String),
    Negate(Box<Expr>),
    Op(Box<Expr>, OpCode, Box<Expr>),
}

impl Expr {
    fn matches(&self, site: &Site, container: &Path, path: &str) -> bool {
        trace!("Expr::matches: path={:?} self={:?}", path, self);
        match self {
            Self::Glob(glob) => glob_matches(glob, path),
            Self::LinksHereFunc(glob) => links_here(site, container, path, glob),
            Self::PageFunc(glob) => page_matches(site, container, glob, path), // FIXME: check its page
            Self::TaggedFunc(glob) => tagged(site, container, glob, path),
            Self::Negate(expr) => !expr.matches(site, container, path),
            Self::Op(left, op, right) => match op {
                OpCode::And => {
                    left.matches(site, container, path) && right.matches(site, container, path)
                }
                OpCode::Or => {
                    left.matches(site, container, path) || right.matches(site, container, path)
                }
            },
        }
    }
}

/// This is public only because parser generation needs it to be.
#[derive(Debug)]
pub enum OpCode {
    And,
    Or,
}

fn glob_matches(glob: &str, path: &str) -> bool {
    trace!("glob_matches: glob={:?} path={:?}", glob, path);
    let glob: Vec<char> = glob.chars().collect();
    let path: Vec<char> = path.chars().collect();
    if glob_matches_helper(&glob, &path) {
        trace!("glob_matches: match!");
        true
    } else {
        trace!("glob_matches: no match");
        false
    }
}

fn glob_matches_helper(mut glob: &[char], mut path: &[char]) -> bool {
    while !glob.is_empty() && !path.is_empty() {
        match glob[0] {
            '?' => {
                glob = &glob[1..];
                path = &path[1..];
            }
            '*' => {
                let glob_remain = &glob[1..];
                if glob_remain.is_empty() {
                    return true;
                }
                while !path.is_empty() {
                    if glob_matches_helper(&glob[1..], path) {
                        return true;
                    }
                    path = &path[1..];
                }
                return false;
            }
            _ => {
                if glob[0] == path[0] {
                    glob = &glob[1..];
                    path = &path[1..];
                } else {
                    return false;
                }
            }
        }
    }

    while let Some('*') = glob.first() {
        glob = &glob[1..];
    }

    glob.is_empty() && path.is_empty()
}

fn links_here(site: &Site, container: &Path, path: &str, _glob: &str) -> bool {
    trace!(
        "links_here: container={} path={:?}",
        container.display(),
        path
    );
    let path = Path::new(path);
    if let Some(page) = site.page(path) {
        trace!("links_here: found page {}", path.display());
        for link in page.meta().links_to() {
            trace!("links_here: page links to {}", link.display());
            if link == container {
                trace!("links to container!");
                return true;
            }
        }
    }
    trace!("links_here: does not link to container");
    false
}

fn tagged(_site: &Site, container: &Path, path: &str, _glob: &str) -> bool {
    trace!("tagged: container={} path={:?}", container.display(), path);
    false
}

fn page_matches(site: &Site, container: &Path, glob: &str, path: &str) -> bool {
    if glob_matches(glob, path) {
        let full_path = container.join(path);
        site.is_page(&full_path)
    } else {
        false
    }
}

#[cfg(test)]
mod test {
    use super::*;

    #[test]
    fn fixed_glob_matches_itself() {
        assert!(glob_matches("foo", "foo"));
    }

    #[test]
    fn fixed_glob_doesnt_match_other_string() {
        assert!(!glob_matches("foo", "bar"));
    }

    #[test]
    fn glob_question_mark_matches() {
        assert!(glob_matches("foo?", "foo0"));
    }

    #[test]
    fn glob_question_mark_doesnt_match_too_long_string() {
        assert!(!glob_matches("foo?", "foo"));
    }

    #[test]
    fn glob_star_matches_empty_string_at_beginning() {
        assert!(glob_matches("*foo", "foo"));
    }

    #[test]
    fn glob_star_matches_empty_string_in_middle() {
        assert!(glob_matches("foo*bar", "foobar"));
    }

    #[test]
    fn glob_star_matches_empty_string_at_end() {
        assert!(glob_matches("foo*", "foo"));
    }

    #[test]
    fn glob_star_matches_nonempty_string_at_beginning() {
        assert!(glob_matches("*bar", "foobar"));
    }
    #[test]
    fn glob_star_matches_nonempty_string_at_end() {
        assert!(glob_matches("foo*", "foobar"));
    }

    #[test]
    fn glob_star_matches_nonempty_string_in_middle() {
        assert!(glob_matches("foo*bar", "fooyobar"));
    }

    #[test]
    fn spec_fixed_page_name_matches_itself() {
        let site = Site::new("/src", "/dest");
        let spec = PageSpec::new(Path::new("/"), "foo").unwrap();
        assert!(spec.matches(&site, Path::new("/foo")));
    }

    #[test]
    fn spec_glob_matches_page() {
        let site = Site::new("/src", "/dest");
        let spec = PageSpec::new(Path::new("/"), "foo*").unwrap();
        assert!(spec.matches(&site, Path::new("/foobar")));
    }

    #[test]
    fn spec_negation() {
        let site = Site::new("/src", "/dest");
        let spec = PageSpec::new(Path::new("/"), "!foo*").unwrap();
        assert!(!spec.matches(&site, Path::new("/foo")));
        assert!(spec.matches(&site, Path::new("/bar")));
    }

    #[test]
    fn spec_or_matches_either() {
        let site = Site::new("/src", "/dest");
        let spec = PageSpec::new(Path::new("/"), "foo or bar").unwrap();
        assert!(spec.matches(&site, Path::new("/foo")));
        assert!(spec.matches(&site, Path::new("/bar")));
        assert!(!spec.matches(&site, Path::new("/yo")));
    }

    #[test]
    fn spec_and_matches_both() {
        let site = Site::new("/src", "/dest");
        let spec = PageSpec::new(Path::new("/"), "foo and !bar").unwrap();
        assert!(spec.matches(&site, Path::new("/foo")));
        assert!(!spec.matches(&site, Path::new("/bar")));
        assert!(!spec.matches(&site, Path::new("/yo")));
    }
}