hrx-parser 0.1.1

#[derive(Debug)]
pub enum Error {
    UnexpectedToken {
        expected: String,
        actual: String,
        start: usize,
        end: usize,
    },
    UnexpectedEof {
        expected: String,
        start: usize,
    },
    Invalid {
        message: String,
        start: usize,
        end: usize,
    },
}

impl std::fmt::Display for Error {
    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
        match self {
            Error::UnexpectedToken {
                actual, expected, ..
            } => writeln!(f, "Expected {}, found {}", expected, actual),
            Error::UnexpectedEof { expected, .. } => {
                writeln!(f, "Expected {}, found Eof", expected)
            }
            Error::Invalid { message, .. } => writeln!(f, "{}", message),
        }
    }
}

pub struct Parser<'a> {
    source: &'a str,
    boundary_length: usize,
    pos: usize,
}

impl<'a> Parser<'a> {
    pub fn new(source: &'a str) -> Self {
        Self {
            source,
            boundary_length: 0,
            pos: 0,
        }
    }

    pub fn cur_byte(&self) -> Option<u8> {
        self.source.as_bytes().get(self.pos).map(|item| *item)
    }

    pub fn eof(&self) -> bool {
        self.pos == self.source.len()
    }

    /// ```ignore
    /// newline ::= U+000A LINE FEED
    /// ```
    pub fn is_line_feed(&self) -> bool {
        self.cur_byte() == Some(b'\n')
    }

    pub fn pos(&self) -> usize {
        self.pos
    }

    pub fn scan_boundary(&mut self) -> Result<(usize, usize), Error> {
        let start = self.pos;
        let mut eq_len = 1;
        self.expect_byte(b'<')?;
        self.expect_byte(b'=')?;
        while self.cur_byte() == Some(b'=') {
            eq_len += 1;
            self.pos += 1;
        }
        self.expect_byte(b'>')?;

        let end = self.pos;
        if self.boundary_length == 0 {
            self.boundary_length = eq_len;
        } else if self.boundary_length != eq_len {
            return Err(Error::Invalid {
                message: "Un matched boundary length".to_string(),
                start,
                end,
            });
        }
        Ok((start, end))
    }

    /// ```ignore
    /// contents ::= any sequence of characters that neither begins with boundary nor
    /// includes U+000A LINE FEED followed immediately by boundary
    /// ```
    pub fn scan_contents(&mut self) -> Result<(usize, usize), Error> {
        let start = self.pos;
        if self.cur_byte() == Some(b'<') {
            match self.scan_boundary() {
                // rewind
                Ok(_) => {
                    self.pos = start;
                    return Ok((start, start));
                }
                Err(_) => {}
            }
        }
        loop {
            if self.eof() {
                break;
            }
            if self.cur_byte() == Some(b'\n') {
                self.pos += 1;
                if self.cur_byte() == Some(b'<') {
                    let checkpoint = self.pos;
                    match self.scan_boundary() {
                        // rewind
                        Ok(_) => {
                            self.pos = checkpoint;
                            break;
                        }
                        Err(_) => {}
                    }
                }
            } else {
                self.pos += 1;
            }
        }
        let end = self.pos;
        Ok((start, end))
    }

    pub fn expect_byte(&mut self, byte: u8) -> Result<(), Error> {
        if self.cur_byte() != Some(byte) {
            if self.eof() {
                return Err(Error::UnexpectedEof {
                    expected: (byte as char).to_string(),
                    // actual: (self.cur_byte() as char).to_string(),
                    start: self.pos,
                });
            }
            return Err(Error::UnexpectedToken {
                expected: (byte as char).to_string(),
                // SAFETY: we know that we don't reach the eof yet
                actual: (self.cur_byte().unwrap() as char).to_string(),
                start: self.pos,
                end: self.pos + 1,
            });
        }
        self.pos += 1;
        Ok(())
    }

    pub fn scan_path_component(&mut self) -> Result<(usize, usize, &'a str), Error> {
        let start = self.pos;
        // SAFETY: we know that we don't reach eof yet
        while !self.eof() && self.is_path_component(self.cur_byte().unwrap()) {
            self.pos += 1;
        }
        let end = self.pos;
        let source = &self.source[start..end];
        if source.len() == 0 || source == "." || source == ".." {
            if self.eof() {
                return Err(Error::UnexpectedEof {
                    expected: "path".to_string(),
                    start: self.pos,
                });
            } else {
                return Err(Error::UnexpectedToken {
                    expected: "path".to_string(),
                    actual: source.to_string(),
                    start,
                    end,
                });
            }
        }
        Ok((start, end, source))
    }

    /// ```ignore
    /// path-character ::= any character other than U+0000 through U+001F, U+007F DELETE, U+002F
    /// SOLIDUS, U+003A COLON, or U+005C REVERSE SOLIDUS
    /// ```
    #[inline]
    fn is_path_component(&self, byte: u8) -> bool {
        !matches!(byte, 0..=0x1f | 0x7f | 0x2f | 0x3a | 0x5c)
    }
}

#[derive(Debug)]
pub enum FileOrDirectory<'a> {
    File(File<'a>),
    Directory(Directory<'a>),
}

#[derive(Debug)]
pub struct File<'a> {
    pub start: usize,
    pub end: usize,
    pub body: Option<Body<'a>>,
    pub path: Path<'a>,
}
#[derive(Debug)]
pub struct Path<'a> {
    pub start: usize,
    pub end: usize,
    pub source: &'a str,
}

#[derive(Debug)]
pub struct Directory<'a> {
    pub start: usize,
    pub end: usize,
    pub path: Path<'a>,
}

#[derive(Debug)]
pub struct Archive<'a> {
    pub start: usize,
    pub end: usize,
    pub entries: Vec<Entry<'a>>,
    pub comment: Option<Comment<'a>>,
}

#[derive(Debug)]
pub struct Entry<'a> {
    pub start: usize,
    pub end: usize,
    pub comment: Option<Comment<'a>>,
    pub body: FileOrDirectory<'a>,
}

impl<'a> Entry<'a> {
    pub fn path(&self) -> String {
        match self.body {
            FileOrDirectory::File(ref f) => f.path.source.to_string(),
            FileOrDirectory::Directory(ref d) => d.path.source.to_string() + "/",
        }
    }

    pub fn content(&self) -> Option<String> {
        match self.body {
            FileOrDirectory::File(ref f) => f.body.as_ref().map(|item| item.source.to_string()),
            FileOrDirectory::Directory(_) => None,
        }
    }
}

#[derive(Debug)]
pub struct Comment<'a> {
    pub start: usize,
    pub end: usize,
    pub source: &'a str,
    pub boundary: Boundary,
    pub body: Body<'a>,
}

#[derive(Debug, Clone)]
pub struct Body<'a> {
    pub start: usize,
    pub end: usize,
    pub source: &'a str,
}

#[derive(Debug)]
pub struct Boundary {
    pub start: usize,
    pub end: usize,
}

pub fn parse_archive<'a>(p: &mut Parser<'a>) -> Result<Archive<'a>, Error> {
    let start = p.pos;
    let mut entries = vec![];
    loop {
        let checkpoint = p.pos;
        match parse_entry(p) {
            Ok(entry) => {
                entries.push(entry);
            }
            Err(err) => {
                p.pos = checkpoint;
                match err {
                    Error::UnexpectedToken { .. } => {}
                    Error::UnexpectedEof { .. } => {}
                    Error::Invalid { .. } => {
                        return Err(err);
                    }
                }
                break;
            }
        }
    }
    dbg!(&p.source[p.pos..]);
    let comment = parse_comment(p).ok();
    dbg!(&comment);
    if !p.eof() {
        return Err(Error::UnexpectedToken {
            expected: "Eof eof".to_string(),
            // SAFETY: we know that we don't reach eof yet
            actual: p.cur_byte().unwrap().to_string(),
            start: p.pos,
            end: p.pos + 1,
        });
    }
    Ok(Archive {
        start,
        end: p.pos,
        entries,
        comment,
    })
    // let
}

pub fn parse_entry<'a>(p: &mut Parser<'a>) -> Result<Entry<'a>, Error> {
    let start = p.pos;
    let comment = parse_comment(p).ok();
    if comment.is_none() {
        p.pos = start;
    }
    parse_boundary(p)?;
    p.expect_byte(b' ')?;
    let path = parse_path(p)?;

    match p.cur_byte() {
        Some(b'/') => {
            p.expect_byte(b'/')?;
            while !p.eof() && p.is_line_feed() {
                p.pos += 1;
            }
            let end = p.pos;
            if !p.eof() && p.cur_byte() != Some(b'<') {
                return Err(Error::Invalid {
                    message: "A directory can't have text contents.".to_string(),
                    start: p.pos,
                    end: p.pos + 1,
                });
            }
            return Ok(Entry {
                start,
                end,
                comment,
                body: FileOrDirectory::Directory(Directory { start, end, path }),
            });
        }
        Some(b'\n') => {
            p.expect_byte(b'\n')?;
            let checkpoint = p.pos;

            let body = parse_body(p).ok();
            if body.is_none() {
                p.pos = checkpoint;
            }
            let end = p.pos;
            return Ok(Entry {
                start,
                end,
                comment,
                body: FileOrDirectory::File(File {
                    start,
                    end,
                    body,
                    path,
                }),
            });
        }
        _ if p.eof() => {
            return Err(Error::UnexpectedEof {
                expected: "`/` or `\n`".to_string(),
                start: p.pos,
            })
        }
        _ => {
            return Err(Error::UnexpectedToken {
                expected: "`/` or `\n`".to_string(),
                // SAFETY: we know that we don't reach the eof yet.
                actual: p.cur_byte().unwrap().to_string(),
                start,
                end: p.pos,
            });
        }
    }
}

pub fn parse_comment<'a>(p: &mut Parser<'a>) -> Result<Comment<'a>, Error> {
    let start = p.pos();
    let boundary = parse_boundary(p)?;
    p.expect_byte(b'\n')?;
    let body = parse_body(p)?;
    let end = p.pos;
    Ok(Comment {
        start,
        end,
        source: &p.source[start..end],
        boundary,
        body,
    })
}

fn parse_path<'a>(p: &mut Parser<'a>) -> Result<Path<'a>, Error> {
    let (start, mut end, _) = p.scan_path_component()?;

    while !p.eof() && p.cur_byte() == Some(b'/') {
        let checkpoint = p.pos;
        p.expect_byte(b'/')?;
        match p.scan_path_component() {
            Ok((_, e, ..)) => {
                end = e;
            }
            // rewind, maybe this is end `/` separator of directory
            Err(_) => {
                p.pos = checkpoint;
                break;
            }
        }
    }

    Ok(Path {
        start,
        end,
        source: &p.source[start..end],
    })
}

fn parse_body<'a>(p: &mut Parser<'a>) -> Result<Body<'a>, Error> {
    let (start, end) = p.scan_contents()?;
    Ok(Body {
        start,
        end,
        source: &p.source[start..end],
    })
}

pub fn parse_boundary<'a>(p: &mut Parser<'a>) -> Result<Boundary, Error> {
    let (start, end) = p.scan_boundary()?;
    Ok(Boundary { start, end })
}

pub fn parse(source: &str) -> Result<Archive, Error> {
    let mut parser = Parser::new(source);
    parse_archive(&mut parser)
}