debuggingsplit-into-workspace

5 files changed, 629 insertions, 0 deletions

diff --git a/crates/repo/src/ebuild/meson.build b/crates/repo/src/ebuild/meson.build
new file mode 100644
index 0000000..a7331a8
--- /dev/null
+++ b/crates/repo/src/ebuild/meson.build
@@ -0,0 +1 @@
+sources += files('mod.rs', 'parsers.rs')
diff --git a/crates/repo/src/ebuild/mod.rs b/crates/repo/src/ebuild/mod.rs
new file mode 100644
index 0000000..6c547c5
--- /dev/null
+++ b/crates/repo/src/ebuild/mod.rs
@@ -0,0 +1,83 @@
+use get::Get;
+
+use std::path::PathBuf;
+
+use atom::{Atom, Name, Slot, Version};
+
+use useflag::{IUseFlag, UseFlag};
+
+mod parsers;
+
+#[derive(Clone, Debug)]
+pub enum Conditional {
+    Negative(UseFlag),
+    Positive(UseFlag),
+}
+
+#[derive(Clone, Debug)]
+pub enum Depend<T> {
+    Element(T),
+    AllOf(Vec<Self>),
+    AnyOf(Vec<Self>),
+    OneOf(Vec<Self>),
+    ConditionalGroup(Conditional, Vec<Self>),
+}
+
+#[derive(Debug, Clone)]
+pub enum UriPrefix {
+    Mirror,
+    Fetch,
+}
+
+#[derive(Debug, Clone, Get)]
+pub struct Uri {
+    #[get(kind = "deref")]
+    protocol: String,
+    #[get(kind = "deref")]
+    path: String,
+}
+
+#[derive(Debug, Clone)]
+pub enum SrcUri {
+    Filename(PathBuf),
+    Uri {
+        prefix: Option<UriPrefix>,
+        uri: Uri,
+        filename: Option<PathBuf>,
+    },
+}
+
+#[derive(Debug, Clone, Get)]
+pub struct License(#[get(method = "get", kind = "deref")] String);
+
+#[derive(Debug, Clone, Get)]
+pub struct Eapi(#[get(method = "get", kind = "deref")] String);
+
+#[derive(Debug, Clone, Get)]
+pub struct Eclass(#[get(method = "get", kind = "deref")] String);
+
+#[derive(Debug, Clone, Get)]
+pub struct Ebuild {
+    pub(super) name: Name,
+    pub(super) version: Version,
+    pub(super) slot: Option<Slot>,
+    pub(super) homepage: Option<String>,
+    #[get(kind = "deref")]
+    pub(super) src_uri: Vec<Depend<SrcUri>>,
+    pub(super) eapi: Option<Eapi>,
+    #[get(kind = "deref")]
+    pub(super) inherit: Vec<Eclass>,
+    #[get(kind = "deref")]
+    pub(super) iuse: Vec<IUseFlag>,
+    #[get(kind = "deref")]
+    pub(super) license: Vec<Depend<License>>,
+    pub(super) description: Option<String>,
+    #[get(kind = "deref")]
+    pub(super) depend: Vec<Depend<Atom>>,
+    #[get(kind = "deref")]
+    pub(super) bdepend: Vec<Depend<Atom>>,
+    #[get(kind = "deref")]
+    pub(super) rdepend: Vec<Depend<Atom>>,
+    #[get(kind = "deref")]
+    pub(super) idepend: Vec<Depend<Atom>>,
+}
diff --git a/crates/repo/src/ebuild/parsers.rs b/crates/repo/src/ebuild/parsers.rs
new file mode 100644
index 0000000..6dc3525
--- /dev/null
+++ b/crates/repo/src/ebuild/parsers.rs
@@ -0,0 +1,207 @@
+use std::path::PathBuf;
+
+use crate::ebuild::{Conditional, Depend, Eapi, Eclass, License, SrcUri, Uri, UriPrefix};
+
+use mon::{
+    Parser, ParserIter, ascii_alpha1, ascii_alphanumeric, ascii_whitespace1, r#if, one_of, tag,
+};
+
+use parseable::Parseable;
+
+use useflag::UseFlag;
+
+impl<'a> Parseable<'a, &'a str> for UriPrefix {
+    type Parser = impl Parser<&'a str, Output = Self>;
+
+    fn parser() -> Self::Parser {
+        tag("+mirror")
+            .map(|_| UriPrefix::Mirror)
+            .or(tag("+fetch").map(|_| UriPrefix::Fetch))
+    }
+}
+
+impl<'a> Parseable<'a, &'a str> for Uri {
+    type Parser = impl Parser<&'a str, Output = Self>;
+
+    fn parser() -> Self::Parser {
+        let protocol = ascii_alpha1::<&str>()
+            .followed_by(tag("://"))
+            .map(|output: &str| output.to_string());
+        let path = r#if(|c: &char| !c.is_ascii_whitespace())
+            .repeated()
+            .at_least(1)
+            .recognize()
+            .map(|output: &str| output.to_string());
+
+        protocol
+            .and(path)
+            .map(|(protocol, path)| Uri { protocol, path })
+    }
+}
+
+impl<'a> Parseable<'a, &'a str> for SrcUri {
+    type Parser = impl Parser<&'a str, Output = Self>;
+
+    fn parser() -> Self::Parser {
+        let filename = || {
+            r#if(|c: &char| !c.is_ascii_whitespace())
+                .repeated()
+                .at_least(1)
+                .recognize()
+                .map(|output: &str| PathBuf::from(output))
+        };
+
+        let uri = UriPrefix::parser()
+            .opt()
+            .and(Uri::parser())
+            .and(filename().preceded_by(tag(" -> ")).opt())
+            .map(|((prefix, uri), filename)| SrcUri::Uri {
+                prefix,
+                uri,
+                filename,
+            });
+
+        uri.or(filename().map(|path: PathBuf| SrcUri::Filename(path)))
+    }
+}
+
+impl<'a> Parseable<'a, &'a str> for License {
+    type Parser = impl Parser<&'a str, Output = Self>;
+
+    fn parser() -> Self::Parser {
+        let start = ascii_alphanumeric().or(one_of("_".chars()));
+        let rest = ascii_alphanumeric()
+            .or(one_of("+_.-".chars()))
+            .repeated()
+            .many();
+
+        start
+            .and(rest)
+            .recognize()
+            .map(|output: &str| License(output.to_string()))
+    }
+}
+
+impl<'a> Parseable<'a, &'a str> for Eapi {
+    type Parser = impl Parser<&'a str, Output = Self>;
+
+    fn parser() -> Self::Parser {
+        let start = ascii_alphanumeric().or(one_of("_".chars()));
+        let rest = ascii_alphanumeric()
+            .or(one_of("+_.-".chars()))
+            .repeated()
+            .many();
+
+        start
+            .and(rest)
+            .recognize()
+            .map(|output: &str| Eapi(output.to_string()))
+    }
+}
+
+// TODO:
+// Cant find information about eclass names in pms so we allow anything except
+// for whitespace.
+impl<'a> Parseable<'a, &'a str> for Eclass {
+    type Parser = impl Parser<&'a str, Output = Self>;
+
+    fn parser() -> Self::Parser {
+        r#if(|c: &char| !c.is_ascii_whitespace())
+            .repeated()
+            .at_least(1)
+            .recognize()
+            .map(|output: &str| Eclass(output.to_string()))
+    }
+}
+
+impl<'a, T> Parseable<'a, &'a str> for Depend<T>
+where
+    T: Parseable<'a, &'a str>,
+{
+    type Parser = impl Parser<&'a str, Output = Self>;
+
+    fn parser() -> Self::Parser {
+        |it| {
+            let exprs = || {
+                Depend::parser()
+                    .separated_by_with_trailing(ascii_whitespace1())
+                    .at_least(1)
+                    .delimited_by(tag("(").followed_by(ascii_whitespace1()), tag(")"))
+            };
+
+            let all_of_group = exprs().map(|exprs| Depend::AllOf(exprs));
+
+            let any_of_group = exprs()
+                .preceded_by(tag("||").followed_by(ascii_whitespace1()))
+                .map(|exprs| Depend::AnyOf(exprs));
+
+            let one_of_group = exprs()
+                .preceded_by(tag("^^").followed_by(ascii_whitespace1()))
+                .map(|exprs| Depend::OneOf(exprs));
+
+            let conditional_group = Conditional::parser()
+                .followed_by(ascii_whitespace1())
+                .and(exprs())
+                .map(|(conditional, exprs)| Depend::ConditionalGroup(conditional, exprs));
+
+            T::parser()
+                .map(|e| Depend::Element(e))
+                .or(conditional_group)
+                .or(any_of_group)
+                .or(all_of_group)
+                .or(one_of_group)
+                .parse(it)
+        }
+    }
+}
+
+impl<'a> Parseable<'a, &'a str> for Conditional {
+    type Parser = impl Parser<&'a str, Output = Self>;
+
+    fn parser() -> Self::Parser {
+        UseFlag::parser()
+            .preceded_by(tag("!"))
+            .followed_by(tag("?"))
+            .map(Conditional::Negative)
+            .or(UseFlag::parser()
+                .followed_by(tag("?"))
+                .map(Conditional::Positive))
+    }
+}
+
+#[cfg(test)]
+mod test {
+
+    use super::*;
+    
+    use mon::{ParserIter, input::InputIter};
+
+    use atom::Atom;
+
+    use crate::ebuild::Depend;
+
+    #[test]
+    fn test_src_uri() {
+        let tests = [
+            "https://example.com/foo/bar.tar.gz",
+            "https://example.com/foo/bar.tar.gz -> bar.tar.gz",
+        ];
+
+        for test in tests {
+            SrcUri::parser()
+                .check_finished(InputIter::new(test))
+                .unwrap();
+        }
+    }
+
+    #[test]
+    fn test_expr() {
+        let it = InputIter::new("flag? ( || ( foo/bar foo/bar ) )");
+
+        Depend::<Atom>::parser()
+            .separated_by(ascii_whitespace1())
+            .many()
+            .check_finished(it)
+            .unwrap();
+    }
+}
diff --git a/crates/repo/src/lib.rs b/crates/repo/src/lib.rs
new file mode 100644
index 0000000..df4412d
--- /dev/null
+++ b/crates/repo/src/lib.rs
@@ -0,0 +1,335 @@
+#![deny(clippy::pedantic, unused_imports)]
+#![allow(
+    dead_code,
+    unstable_name_collisions,
+    clippy::missing_errors_doc,
+    clippy::missing_panics_doc
+)]
+#![feature(impl_trait_in_assoc_type)]
+
+use std::{
+    fs, io,
+    os::unix::ffi::OsStrExt,
+    path::{Path, PathBuf},
+};
+
+use crate::ebuild::{Depend, Eapi, Ebuild, Eclass, License, SrcUri};
+
+use get::Get;
+
+use mon::{Parser, ParserIter, ascii_whitespace1, input::InputIter, tag};
+
+use parseable::Parseable;
+
+use atom::{self, Atom};
+
+use useflag::IUseFlag;
+
+pub mod ebuild;
+
+#[derive(Debug, thiserror::Error)]
+pub enum Error {
+    #[error("io error: {0}")]
+    Io(PathBuf, io::Error),
+    #[error("error while reading directory: {0:?}: {1}")]
+    ReadDir(PathBuf, io::Error),
+    #[error("failed to decode path: {0}")]
+    Unicode(PathBuf),
+    #[error("parser error: {0}")]
+    Parser(String),
+}
+
+#[derive(Debug, Clone, Get)]
+pub struct Repo {
+    #[get(kind = "deref")]
+    path: PathBuf,
+}
+
+#[derive(Debug, Clone, Get)]
+pub struct Category {
+    name: atom::Category,
+    #[get(kind = "deref")]
+    path: PathBuf,
+}
+
+#[derive(Debug)]
+pub struct Categories(PathBuf, fs::ReadDir);
+
+#[derive(Debug)]
+pub struct Ebuilds(PathBuf, fs::ReadDir);
+
+impl Repo {
+    pub fn new<P: AsRef<Path>>(path: P) -> Self {
+        Self {
+            path: path.as_ref().to_path_buf(),
+        }
+    }
+
+    pub fn categories(&self) -> Result<Categories, Error> {
+        let path = self.path.as_path().join("metadata/md5-cache");
+
+        Ok(Categories(
+            path.clone(),
+            fs::read_dir(&path).map_err(|e| Error::Io(path, e))?,
+        ))
+    }
+}
+
+impl Category {
+    pub fn ebuilds(&self) -> Result<Ebuilds, Error> {
+        Ok(Ebuilds(
+            self.path.clone(),
+            fs::read_dir(&self.path).map_err(|e| Error::Io(self.path.clone(), e))?,
+        ))
+    }
+}
+
+impl Iterator for Categories {
+    type Item = Result<Category, Error>;
+
+    fn next(&mut self) -> Option<Self::Item> {
+        loop {
+            match self.1.next()? {
+                Ok(entry) if entry.path().file_name().unwrap().as_bytes() == b"Manifest.gz" => (),
+                Ok(entry) => match read_category(entry.path()) {
+                    Ok(category) => break Some(Ok(category)),
+                    Err(e) => break Some(Err(e)),
+                },
+                Err(e) => break Some(Err(Error::ReadDir(self.0.clone(), e))),
+            }
+        }
+    }
+}
+
+impl Iterator for Ebuilds {
+    type Item = Result<Ebuild, Error>;
+
+    fn next(&mut self) -> Option<Self::Item> {
+        loop {
+            match self.1.next()? {
+                Ok(entry) if entry.path().file_name().unwrap().as_bytes() == b"Manifest.gz" => (),
+                Ok(entry) => match read_ebuild(entry.path()) {
+                    Ok(ebuild) => break Some(Ok(ebuild)),
+                    Err(e) => break Some(Err(e)),
+                },
+                Err(e) => break Some(Err(Error::ReadDir(self.0.clone(), e))),
+            }
+        }
+    }
+}
+
+fn read_category(path: PathBuf) -> Result<Category, Error> {
+    let file_name = path
+        .as_path()
+        .file_name()
+        .unwrap()
+        .to_str()
+        .ok_or(Error::Unicode(path.clone()))?;
+
+    let name = atom::Category::parser()
+        .parse_finished(InputIter::new(file_name))
+        .map_err(|_| Error::Parser(file_name.to_string()))?;
+
+    Ok(Category { name, path })
+}
+
+fn read_ebuild(path: PathBuf) -> Result<Ebuild, Error> {
+    let file_name = path
+        .as_path()
+        .file_name()
+        .unwrap()
+        .to_str()
+        .ok_or(Error::Unicode(path.clone()))?;
+
+    let (name, version) = atom::Name::parser()
+        .and(atom::Version::parser().preceded_by(tag("-")))
+        .parse_finished(InputIter::new(file_name))
+        .map_err(|_| Error::Parser(file_name.to_string()))?;
+
+    let metadata = fs::read_to_string(path.as_path()).map_err(|e| Error::Io(path, e))?;
+
+    Ok(Ebuild {
+        name,
+        version,
+        slot: match read_slot(&metadata) {
+            Some(Ok(slot)) => Some(slot),
+            Some(Err(e)) => return Err(e),
+            None => None,
+        },
+        homepage: read_homepage(&metadata),
+        src_uri: match read_src_uri(&metadata) {
+            Some(Ok(src_uri)) => src_uri,
+            Some(Err(e)) => return Err(e),
+            None => Vec::new(),
+        },
+        eapi: match read_eapi(&metadata) {
+            Some(Ok(eapi)) => Some(eapi),
+            Some(Err(e)) => return Err(e),
+            None => None,
+        },
+        inherit: match read_inherit(&metadata) {
+            Some(Ok(inherit)) => inherit,
+            Some(Err(e)) => return Err(e),
+            None => Vec::new(),
+        },
+        iuse: match read_iuse(&metadata) {
+            Some(Ok(iuse)) => iuse,
+            Some(Err(e)) => return Err(e),
+            None => Vec::new(),
+        },
+        license: match read_license(&metadata) {
+            Some(Ok(license)) => license,
+            Some(Err(e)) => return Err(e),
+            None => Vec::new(),
+        },
+        description: read_description(&metadata),
+        depend: match read_depend(&metadata) {
+            Some(Ok(depend)) => depend,
+            Some(Err(e)) => return Err(e),
+            None => Vec::new(),
+        },
+        bdepend: match read_bdepend(&metadata) {
+            Some(Ok(depend)) => depend,
+            Some(Err(e)) => return Err(e),
+            None => Vec::new(),
+        },
+        rdepend: match read_rdepend(&metadata) {
+            Some(Ok(depend)) => depend,
+            Some(Err(e)) => return Err(e),
+            None => Vec::new(),
+        },
+        idepend: match read_idepend(&metadata) {
+            Some(Ok(depend)) => depend,
+            Some(Err(e)) => return Err(e),
+            None => Vec::new(),
+        },
+    })
+}
+
+fn read_slot(input: &str) -> Option<Result<atom::Slot, Error>> {
+    let line = input.lines().find_map(|line| line.strip_prefix("SLOT="))?;
+
+    match atom::Slot::parser().parse_finished(InputIter::new(line)) {
+        Ok(slot) => Some(Ok(slot)),
+        Err(_) => Some(Err(Error::Parser(line.to_string()))),
+    }
+}
+
+fn read_homepage(input: &str) -> Option<String> {
+    input
+        .lines()
+        .find_map(|line| line.strip_prefix("HOMEPAGE=").map(str::to_string))
+}
+
+fn read_src_uri(input: &str) -> Option<Result<Vec<Depend<SrcUri>>, Error>> {
+    let line = input
+        .lines()
+        .find_map(|line| line.strip_prefix("SRC_URI="))?;
+
+    match Depend::<SrcUri>::parser()
+        .separated_by(ascii_whitespace1())
+        .many()
+        .parse_finished(InputIter::new(line))
+    {
+        Ok(slot) => Some(Ok(slot)),
+        Err(_) => Some(Err(Error::Parser(line.to_string()))),
+    }
+}
+
+fn read_eapi(input: &str) -> Option<Result<Eapi, Error>> {
+    let line = input.lines().find_map(|line| line.strip_prefix("EAPI="))?;
+
+    match Eapi::parser().parse_finished(InputIter::new(line)) {
+        Ok(slot) => Some(Ok(slot)),
+        Err(_) => Some(Err(Error::Parser(line.to_string()))),
+    }
+}
+
+fn read_inherit(input: &str) -> Option<Result<Vec<Eclass>, Error>> {
+    let line = input
+        .lines()
+        .find_map(|line| line.strip_prefix("INHERIT="))?;
+
+    match Eclass::parser()
+        .separated_by(ascii_whitespace1())
+        .many()
+        .parse_finished(InputIter::new(line))
+    {
+        Ok(inherit) => Some(Ok(inherit)),
+        Err(_) => Some(Err(Error::Parser(line.to_string()))),
+    }
+}
+
+fn read_iuse(input: &str) -> Option<Result<Vec<IUseFlag>, Error>> {
+    let line = input.lines().find_map(|line| line.strip_prefix("IUSE="))?;
+
+    match IUseFlag::parser()
+        .separated_by(ascii_whitespace1())
+        .many()
+        .parse_finished(InputIter::new(line))
+    {
+        Ok(iuse) => Some(Ok(iuse)),
+        Err(_) => Some(Err(Error::Parser(line.to_string()))),
+    }
+}
+
+fn read_license(input: &str) -> Option<Result<Vec<Depend<License>>, Error>> {
+    let line = input
+        .lines()
+        .find_map(|line| line.strip_suffix("LICENSE="))?;
+
+    match Depend::<License>::parser()
+        .separated_by(ascii_whitespace1())
+        .many()
+        .parse_finished(InputIter::new(line))
+    {
+        Ok(license) => Some(Ok(license)),
+        Err(_) => Some(Err(Error::Parser(line.to_string()))),
+    }
+}
+
+fn read_description(input: &str) -> Option<String> {
+    input
+        .lines()
+        .find_map(|line| line.strip_prefix("DESCRIPTION=").map(str::to_string))
+}
+
+fn read_depend(input: &str) -> Option<Result<Vec<Depend<Atom>>, Error>> {
+    let line = input
+        .lines()
+        .find_map(|line| line.strip_prefix("DEPEND="))?;
+
+    Some(parse_depends(line))
+}
+
+fn read_bdepend(input: &str) -> Option<Result<Vec<Depend<Atom>>, Error>> {
+    let line = input
+        .lines()
+        .find_map(|line| line.strip_prefix("BDEPEND="))?;
+
+    Some(parse_depends(line))
+}
+
+fn read_rdepend(input: &str) -> Option<Result<Vec<Depend<Atom>>, Error>> {
+    let line = input
+        .lines()
+        .find_map(|line| line.strip_prefix("RDEPEND="))?;
+
+    Some(parse_depends(line))
+}
+
+fn read_idepend(input: &str) -> Option<Result<Vec<Depend<Atom>>, Error>> {
+    let line = input
+        .lines()
+        .find_map(|line| line.strip_prefix("IDEPEND="))?;
+
+    Some(parse_depends(line))
+}
+
+fn parse_depends(line: &str) -> Result<Vec<Depend<Atom>>, Error> {
+    Depend::<Atom>::parser()
+        .separated_by(ascii_whitespace1())
+        .many()
+        .parse_finished(InputIter::new(line))
+        .map_err(|_| Error::Parser(line.to_string()))
+}
diff --git a/crates/repo/src/meson.build b/crates/repo/src/meson.build
new file mode 100644
index 0000000..c1be7a7
--- /dev/null
+++ b/crates/repo/src/meson.build
@@ -0,0 +1,3 @@
+sources += files('mod.rs')
+
+subdir('ebuild')