use nom::{self, IResult}; use std::str; #[derive(Debug)] pub enum Expr<'a> { NumLit(&'a str), StrLit(&'a str), Var(&'a str), Attr(Box>, &'a str), Filter(&'a str, Vec>), BinOp(&'a str, Box>, Box>), Group(Box>), Call(Box>, &'a str, Vec>), } #[derive(Debug)] pub enum Target<'a> { Name(&'a str), } #[derive(Clone, Copy, Debug)] pub struct WS(pub bool, pub bool); #[derive(Debug)] pub enum Node<'a> { Lit(&'a str, &'a str, &'a str), Comment(), Expr(WS, Expr<'a>), Cond(Vec<(WS, Option>, Vec>)>, WS), Loop(WS, Target<'a>, Expr<'a>, Vec>, WS), Extends(Expr<'a>), BlockDef(WS, &'a str, Vec>, WS), Block(WS, &'a str, WS), } pub type Cond<'a> = (WS, Option>, Vec>); fn split_ws_parts(s: &[u8]) -> Node { if s.is_empty() { let rs = str::from_utf8(s).unwrap(); return Node::Lit(rs, rs, rs); } let is_ws = |c: &u8| { *c != b' ' && *c != b'\t' && *c != b'\r' && *c != b'\n' }; let start = s.iter().position(&is_ws); let res = if start.is_none() { (s, &s[0..0], &s[0..0]) } else { let start = start.unwrap(); let end = s.iter().rposition(&is_ws); if end.is_none() { (&s[..start], &s[start..], &s[0..0]) } else { let end = end.unwrap(); (&s[..start], &s[start..end + 1], &s[end + 1..]) } }; Node::Lit(str::from_utf8(res.0).unwrap(), str::from_utf8(res.1).unwrap(), str::from_utf8(res.2).unwrap()) } enum ContentState { Any, Brace(usize), End(usize), } fn take_content(i: &[u8]) -> IResult<&[u8], Node> { use parser::ContentState::*; let mut state = Any; for (idx, c) in i.iter().enumerate() { state = match (state, *c) { (Any, b'{') => Brace(idx), (Any, _) => Any, (Brace(start), b'{') | (Brace(start), b'%') | (Brace(start), b'#') => End(start), (Brace(_), _) => Any, (End(_), _) => panic!("cannot happen"), }; if let End(_) = state { break; } } match state { Any | Brace(_) => IResult::Done(&i[..0], split_ws_parts(i)), End(0) => IResult::Error(nom::ErrorKind::Custom(0)), End(start) => IResult::Done(&i[start..], split_ws_parts(&i[..start])), } } fn identifier(input: &[u8]) -> IResult<&[u8], &str> { if !nom::is_alphabetic(input[0]) && input[0] != b'_' { return IResult::Error(nom::ErrorKind::Custom(0)); } for (i, ch) in input.iter().enumerate() { if i == 0 || nom::is_alphanumeric(*ch) || *ch == b'_' { continue; } return IResult::Done(&input[i..], str::from_utf8(&input[..i]).unwrap()); } IResult::Done(&input[1..], str::from_utf8(&input[..1]).unwrap()) } named!(expr_num_lit, map!(nom::digit, |s| Expr::NumLit(str::from_utf8(s).unwrap()) )); named!(expr_str_lit, map!( delimited!(char!('"'), is_not!("\""), char!('"')), |s| Expr::StrLit(str::from_utf8(s).unwrap()) )); named!(expr_var, map!(identifier, |s| Expr::Var(s)) ); named!(target_single, map!(identifier, |s| Target::Name(s) )); named!(arguments>>, opt!( do_parse!( tag_s!("(") >> arg0: ws!(opt!(expr_any)) >> args: many0!(do_parse!( tag_s!(",") >> argn: ws!(expr_any) >> (argn) )) >> tag_s!(")") >> ({ let mut res = Vec::new(); if arg0.is_some() { res.push(arg0.unwrap()); } res.extend(args); res }) ) )); named!(expr_group, map!( delimited!(char!('('), expr_any, char!(')')), |s| Expr::Group(Box::new(s)) )); named!(expr_single, alt!( expr_num_lit | expr_str_lit | expr_var | expr_group )); named!(expr_attr, alt!( do_parse!( obj: expr_single >> tag_s!(".") >> attr: identifier >> args: arguments >> (if args.is_some() { Expr::Call(Box::new(obj), attr, args.unwrap()) } else { Expr::Attr(Box::new(obj), attr) }) ) | expr_single )); named!(filter<(&str, Option>)>, do_parse!( tag_s!("|") >> fname: identifier >> args: arguments >> (fname, args) )); named!(expr_filtered, do_parse!( obj: expr_attr >> filters: many0!(filter) >> ({ let mut res = obj; for (fname, args) in filters { res = Expr::Filter(fname, { let mut args = match args { Some(inner) => inner, None => Vec::new(), }; args.insert(0, res); args }); } res }) )); macro_rules! expr_prec_layer { ( $name:ident, $inner:ident, $( $op:expr ),* ) => { named!($name, alt!( do_parse!( left: $inner >> op: ws!(alt!($( tag_s!($op) )|*)) >> right: $inner >> (Expr::BinOp(str::from_utf8(op).unwrap(), Box::new(left), Box::new(right))) ) | $inner )); } } expr_prec_layer!(expr_muldivmod, expr_filtered, "*", "/", "%"); expr_prec_layer!(expr_addsub, expr_muldivmod, "+", "-"); expr_prec_layer!(expr_shifts, expr_addsub, ">>", "<<"); expr_prec_layer!(expr_band, expr_shifts, "&"); expr_prec_layer!(expr_bxor, expr_band, "^"); expr_prec_layer!(expr_bor, expr_bxor, "|"); expr_prec_layer!(expr_compare, expr_bor, "==", "!=", ">=", ">", "<=", "<" ); expr_prec_layer!(expr_and, expr_compare, "&&"); expr_prec_layer!(expr_any, expr_and, "||"); named!(expr_node, do_parse!( tag_s!("{{") >> pws: opt!(tag_s!("-")) >> expr: ws!(expr_any) >> nws: opt!(tag_s!("-")) >> tag_s!("}}") >> (Node::Expr(WS(pws.is_some(), nws.is_some()), expr)) )); named!(cond_if, do_parse!( ws!(tag_s!("if")) >> cond: ws!(expr_any) >> (cond) )); named!(cond_block, do_parse!( tag_s!("{%") >> pws: opt!(tag_s!("-")) >> ws!(tag_s!("else")) >> cond: opt!(cond_if) >> nws: opt!(tag_s!("-")) >> tag_s!("%}") >> block: parse_template >> (WS(pws.is_some(), nws.is_some()), cond, block) )); named!(block_if, do_parse!( tag_s!("{%") >> pws1: opt!(tag_s!("-")) >> cond: ws!(cond_if) >> nws1: opt!(tag_s!("-")) >> tag_s!("%}") >> block: parse_template >> elifs: many0!(cond_block) >> tag_s!("{%") >> pws2: opt!(tag_s!("-")) >> ws!(tag_s!("endif")) >> nws2: opt!(tag_s!("-")) >> tag_s!("%}") >> ({ let mut res = Vec::new(); res.push((WS(pws1.is_some(), nws1.is_some()), Some(cond), block)); res.extend(elifs); Node::Cond(res, WS(pws2.is_some(), nws2.is_some())) }) )); named!(block_for, do_parse!( tag_s!("{%") >> pws1: opt!(tag_s!("-")) >> ws!(tag_s!("for")) >> var: ws!(target_single) >> ws!(tag_s!("in")) >> iter: ws!(expr_any) >> nws1: opt!(tag_s!("-")) >> tag_s!("%}") >> block: parse_template >> tag_s!("{%") >> pws2: opt!(tag_s!("-")) >> ws!(tag_s!("endfor")) >> nws2: opt!(tag_s!("-")) >> tag_s!("%}") >> (Node::Loop(WS(pws1.is_some(), nws1.is_some()), var, iter, block, WS(pws2.is_some(), pws2.is_some()))) )); named!(block_extends, do_parse!( tag_s!("{%") >> ws!(tag_s!("extends")) >> name: ws!(expr_str_lit) >> tag_s!("%}") >> (Node::Extends(name)) )); named!(block_block, do_parse!( tag_s!("{%") >> pws1: opt!(tag_s!("-")) >> ws!(tag_s!("block")) >> name: ws!(identifier) >> nws1: opt!(tag_s!("-")) >> tag_s!("%}") >> contents: parse_template >> tag_s!("{%") >> pws2: opt!(tag_s!("-")) >> ws!(tag_s!("endblock")) >> nws2: opt!(tag_s!("-")) >> tag_s!("%}") >> (Node::BlockDef(WS(pws1.is_some(), nws1.is_some()), name, contents, WS(pws2.is_some(), pws2.is_some()))) )); named!(block_comment, do_parse!( tag_s!("{#") >> take_until_s!("#}") >> tag_s!("#}") >> (Node::Comment()) )); named!(parse_template>>, many0!(alt!( take_content | block_comment | expr_node | block_if | block_for | block_extends | block_block ))); pub fn parse(src: &str) -> Vec { match parse_template(src.as_bytes()) { IResult::Done(_, res) => res, IResult::Error(err) => panic!("problems parsing template source: {}", err), IResult::Incomplete(_) => panic!("parsing incomplete"), } } #[cfg(test)] mod tests { fn check_ws_split(s: &str, res: &(&str, &str, &str)) { let node = super::split_ws_parts(s.as_bytes()); match node { super::Node::Lit(lws, s, rws) => { assert_eq!(lws, res.0); assert_eq!(s, res.1); assert_eq!(rws, res.2); }, _ => { panic!("fail"); }, } } #[test] fn test_ws_splitter() { check_ws_split("", &("", "", "")); check_ws_split("a", &("", "a", "")); check_ws_split("\ta", &("\t", "a", "")); check_ws_split("b\n", &("", "b", "\n")); check_ws_split(" \t\r\n", &(" \t\r\n", "", "")); } }