// rustfmt doesn't do a very good job on nom parser invocations. #![cfg_attr(rustfmt, rustfmt_skip)] use nom; use std::str; use askama_shared::Syntax; #[derive(Debug)] pub enum Expr<'a> { NumLit(&'a str), StrLit(&'a str), Var(&'a str), Path(Vec<&'a str>), Array(Vec>), Attr(Box>, &'a str), Index(Box>, Box>), Filter(&'a str, Vec>), Unary(&'a str, Box>), BinOp(&'a str, Box>, Box>), Range(&'a str, Option>>, Option>>), Group(Box>), MethodCall(Box>, &'a str, Vec>), RustMacro(&'a str, Vec>), } #[derive(Debug)] pub enum MatchVariant<'a> { Path(Vec<&'a str>), Name(&'a str), NumLit(&'a str), StrLit(&'a str), } #[derive(Debug)] pub enum MatchParameter<'a> { Name(&'a str), NumLit(&'a str), StrLit(&'a str), } #[derive(Debug)] pub enum Target<'a> { Name(&'a str), } #[derive(Clone, Copy, Debug)] pub struct WS(pub bool, pub bool); #[derive(Debug)] pub struct Macro<'a> { pub ws1: WS, pub args: Vec<&'a str>, pub nodes: Vec>, pub ws2: WS, } #[derive(Debug)] pub enum Node<'a> { Lit(&'a str, &'a str, &'a str), Comment(WS), Expr(WS, Expr<'a>), Call(WS, Option<&'a str>, &'a str, Vec>), LetDecl(WS, Target<'a>), Let(WS, Target<'a>, Expr<'a>), Cond(Vec<(WS, Option>, Vec>)>, WS), Match(WS, Expr<'a>, Option<&'a str>, Vec>, WS), Loop(WS, Target<'a>, Expr<'a>, Vec>, WS), Extends(Expr<'a>), BlockDef(WS, &'a str, Vec>, WS), Include(WS, &'a str), Import(WS, &'a str, &'a str), Macro(&'a str, Macro<'a>), } pub type Cond<'a> = (WS, Option>, Vec>); pub type When<'a> = ( WS, Option>, Vec>, Vec>, ); type Input<'a> = nom::types::CompleteByteSlice<'a>; #[allow(non_snake_case)] fn Input(input: &[u8]) -> Input { nom::types::CompleteByteSlice(input) } 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<'a>( i: Input<'a>, s: &'a Syntax<'a>, ) -> Result<(Input<'a>, Node<'a>), nom::Err>> { use crate::parser::ContentState::*; let bs = s.block_start.as_bytes()[0]; let be = s.block_start.as_bytes()[1]; let cs = s.comment_start.as_bytes()[0]; let ce = s.comment_start.as_bytes()[1]; let es = s.expr_start.as_bytes()[0]; let ee = s.expr_start.as_bytes()[1]; let mut state = Any; for (idx, c) in i.iter().enumerate() { state = match state { Any => { if *c == bs || *c == es || *c == cs { Brace(idx) } else { Any } } Brace(start) => { if *c == be || *c == ee || *c == ce { End(start) } else { Any } } End(_) => panic!("cannot happen"), }; if let End(_) = state { break; } } match state { Any | Brace(_) => Ok((Input(&i[..0]), split_ws_parts(i.0))), End(0) => Err(nom::Err::Error(error_position!( i, nom::ErrorKind::Custom(0) ))), End(start) => Ok((Input(&i[start..]), split_ws_parts(&i[..start]))), } } fn identifier(input: Input) -> Result<(Input, &str), nom::Err> { if !nom::is_alphabetic(input[0]) && input[0] != b'_' { return Err(nom::Err::Error(error_position!( input, nom::ErrorKind::Custom(0) ))); } for (i, ch) in input.iter().enumerate() { if i == 0 || nom::is_alphanumeric(*ch) || *ch == b'_' { continue; } return Ok((Input(&input[i..]), str::from_utf8(&input[..i]).unwrap())); } Ok((Input(&input[1..]), str::from_utf8(&input[..1]).unwrap())) } named!(num_lit, map!(nom::digit, |s| str::from_utf8(s.0).unwrap() )); named!(expr_num_lit, map!(num_lit, |s| Expr::NumLit(s) )); named!(expr_array_lit, do_parse!( ws!(tag!("[")) >> first: expr_any >> rest: many0!(do_parse!( ws!(tag!(",")) >> part: expr_any >> (part) )) >> ws!(tag!("]")) >> ({ let mut elements = vec![first]; elements.extend(rest); Expr::Array(elements) }) )); named!(variant_num_lit, map!(num_lit, |s| MatchVariant::NumLit(s) )); named!(param_num_lit, map!(num_lit, |s| MatchParameter::NumLit(s) )); named!(expr_str_lit, map!( delimited!(char!('"'), take_until!("\""), char!('"')), |s| Expr::StrLit(str::from_utf8(&s).unwrap()) )); named!(variant_str_lit, map!( delimited!(char!('"'), is_not!("\""), char!('"')), |s| MatchVariant::StrLit(str::from_utf8(&s).unwrap()) )); named!(param_str_lit, map!( delimited!(char!('"'), is_not!("\""), char!('"')), |s| MatchParameter::StrLit(str::from_utf8(&s).unwrap()) )); named!(expr_var, map!(identifier, |s| Expr::Var(s)) ); named!(expr_path, do_parse!( start: call!(identifier) >> rest: many1!(do_parse!( tag!("::") >> part: identifier >> (part) )) >> ({ let mut path = vec![start]; path.extend(rest); Expr::Path(path) }) )); named!(variant_path, do_parse!( start: call!(identifier) >> rest: many1!(do_parse!( tag!("::") >> part: identifier >> (part) )) >> ({ let mut path = vec![start]; path.extend(rest); MatchVariant::Path(path) }) )); named!(target_single, map!(identifier, |s| Target::Name(s) )); named!(variant_name, map!(identifier, |s| MatchVariant::Name(s) )); named!(param_name, map!(identifier, |s| MatchParameter::Name(s) )); named!(arguments>, do_parse!( tag!("(") >> args: opt!(do_parse!( arg0: ws!(expr_any) >> args: many0!(do_parse!( tag!(",") >> argn: ws!(expr_any) >> (argn) )) >> ({ let mut res = vec![arg0]; res.extend(args); res }) )) >> tag!(")") >> (args.unwrap_or_default()) )); named!(parameters>, do_parse!( tag!("(") >> vals: opt!(do_parse!( arg0: ws!(identifier) >> args: many0!(do_parse!( tag!(",") >> argn: ws!(identifier) >> (argn) )) >> ({ let mut res = vec![arg0]; res.extend(args); res }) )) >> tag!(")") >> (vals.unwrap_or_default()) )); named!(with_parameters>, do_parse!( tag!("with") >> ws!(tag!("(")) >> vals: opt!(do_parse!( arg0: ws!(match_parameter) >> args: many0!(do_parse!( tag!(",") >> argn: ws!(match_parameter) >> (argn) )) >> ({ let mut res = vec![arg0]; res.extend(args); res }) )) >> tag!(")") >> (vals.unwrap_or_default()) )); 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_path | expr_array_lit | expr_var | expr_group )); named!(match_variant, alt!( variant_path | variant_name | variant_num_lit | variant_str_lit )); named!(match_parameter, alt!( param_name | param_num_lit | param_str_lit )); named!(attr>)>, do_parse!( tag!(".") >> attr: alt!(num_lit | identifier) >> args: opt!(arguments) >> (attr, args) )); named!(expr_attr, do_parse!( obj: expr_single >> attrs: many0!(attr) >> ({ let mut res = obj; for (aname, args) in attrs { res = if args.is_some() { Expr::MethodCall(Box::new(res), aname, args.unwrap()) } else { Expr::Attr(Box::new(res), aname) }; } res }) )); named!(expr_index, do_parse!( obj: expr_attr >> key: opt!(do_parse!( ws!(tag!("[")) >> key: expr_any >> ws!(tag!("]")) >> (key) )) >> (match key { Some(key) => Expr::Index(Box::new(obj), Box::new(key)), None => obj, }) )); named!(filter>)>, do_parse!( tag!("|") >> fname: identifier >> args: opt!(arguments) >> (fname, args) )); named!(expr_filtered, do_parse!( obj: expr_index >> 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 }) )); named!(expr_unary, do_parse!( op: opt!(alt!(tag!("!") | tag!("-"))) >> expr: expr_filtered >> (match op { Some(op) => Expr::Unary(str::from_utf8(op.0).unwrap(), Box::new(expr)), None => expr, }) )); named!(rust_macro, do_parse!( mname: identifier >> tag!("!") >> args: arguments >> (Expr::RustMacro(mname, args)) )); macro_rules! expr_prec_layer { ( $name:ident, $inner:ident, $( $op:expr ),* ) => { named!($name, do_parse!( left: $inner >> op_and_right: opt!(pair!(ws!(alt!($( tag!($op) )|*)), expr_any)) >> (match op_and_right { Some((op, right)) => Expr::BinOp( str::from_utf8(op.0).unwrap(), Box::new(left), Box::new(right) ), None => left, }) )); } } expr_prec_layer!(expr_muldivmod, expr_unary, "*", "/", "%"); 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_or, expr_and, "||"); named!(range_right, do_parse!( ws!(tag!("..")) >> incl: opt!(ws!(tag!("="))) >> right: opt!(expr_or) >> (Expr::Range(if incl.is_some() { "..=" } else { ".." }, None, right.map(Box::new))) )); named!(expr_any, alt!( range_right | rust_macro | do_parse!( left: expr_or >> rest: range_right >> (match rest { Expr::Range(op, _, right) => Expr::Range(op, Some(Box::new(left)), right), _ => unreachable!(), }) ) | expr_or )); named_args!(expr_node<'a>(s: &'a Syntax<'a>) , Node<'a>>, do_parse!( call!(tag_expr_start, s) >> pws: opt!(tag!("-")) >> expr: ws!(expr_any) >> nws: opt!(tag!("-")) >> call!(tag_expr_end, s) >> (Node::Expr(WS(pws.is_some(), nws.is_some()), expr)) )); named!(block_call, do_parse!( pws: opt!(tag!("-")) >> ws!(tag!("call")) >> scope: opt!(do_parse!( scope: ws!(identifier) >> ws!(tag!("::")) >> (scope) )) >> name: ws!(identifier) >> args: ws!(arguments) >> nws: opt!(tag!("-")) >> (Node::Call(WS(pws.is_some(), nws.is_some()), scope, name, args)) )); named!(cond_if, do_parse!( ws!(tag!("if")) >> cond: ws!(expr_any) >> (cond) )); named_args!(cond_block<'a>(s: &'a Syntax<'a>) , Cond<'a>>, do_parse!( call!(tag_block_start, s) >> pws: opt!(tag!("-")) >> ws!(tag!("else")) >> cond: opt!(cond_if) >> nws: opt!(tag!("-")) >> call!(tag_block_end, s) >> block: call!(parse_template, s) >> (WS(pws.is_some(), nws.is_some()), cond, block) )); named_args!(block_if<'a>(s: &'a Syntax<'a>) , Node<'a>>, do_parse!( pws1: opt!(tag!("-")) >> cond: ws!(cond_if) >> nws1: opt!(tag!("-")) >> call!(tag_block_end, s) >> block: call!(parse_template, s) >> elifs: many0!(call!(cond_block, s)) >> call!(tag_block_start, s) >> pws2: opt!(tag!("-")) >> ws!(tag!("endif")) >> nws2: opt!(tag!("-")) >> ({ 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_args!(match_else_block<'a>(s: &'a Syntax<'a>) , When<'a>>, do_parse!( call!(tag_block_start, s) >> pws: opt!(tag!("-")) >> ws!(tag!("else")) >> nws: opt!(tag!("-")) >> call!(tag_block_end, s) >> block: call!(parse_template, s) >> (WS(pws.is_some(), nws.is_some()), None, vec![], block) )); named_args!(when_block<'a>(s: &'a Syntax<'a>) , When<'a>>, do_parse!( call!(tag_block_start, s) >> pws: opt!(tag!("-")) >> ws!(tag!("when")) >> variant: ws!(match_variant) >> params: opt!(ws!(with_parameters)) >> nws: opt!(tag!("-")) >> call!(tag_block_end, s) >> block: call!(parse_template, s) >> (WS(pws.is_some(), nws.is_some()), Some(variant), params.unwrap_or_default(), block) )); named_args!(block_match<'a>(s: &'a Syntax<'a>) , Node<'a>>, do_parse!( pws1: opt!(tag!("-")) >> ws!(tag!("match")) >> expr: ws!(expr_any) >> nws1: opt!(tag!("-")) >> call!(tag_block_end, s) >> inter: opt!(call!(take_content, s)) >> arms: many1!(call!(when_block, s)) >> else_arm: opt!(call!(match_else_block, s)) >> ws!(call!(tag_block_start, s)) >> pws2: opt!(tag!("-")) >> ws!(tag!("endmatch")) >> nws2: opt!(tag!("-")) >> ({ let mut arms = arms; if let Some(arm) = else_arm { arms.push(arm); } let inter = match inter { Some(Node::Lit(lws, val, rws)) => { assert!(val.is_empty(), "only whitespace allowed between match and first when, found {}", val); assert!(rws.is_empty(), "only whitespace allowed between match and first when, found {}", rws); Some(lws) }, None => None, _ => panic!("only literals allowed between match and first when"), }; Node::Match( WS(pws1.is_some(), nws1.is_some()), expr, inter, arms, WS(pws2.is_some(), nws2.is_some()), ) }) )); named!(block_let, do_parse!( pws: opt!(tag!("-")) >> ws!(tag!("let")) >> var: ws!(target_single) >> val: opt!(do_parse!( ws!(tag!("=")) >> val: ws!(expr_any) >> (val) )) >> nws: opt!(tag!("-")) >> (if val.is_some() { Node::Let(WS(pws.is_some(), nws.is_some()), var, val.unwrap()) } else { Node::LetDecl(WS(pws.is_some(), nws.is_some()), var) }) )); named_args!(block_for<'a>(s: &'a Syntax<'a>) , Node<'a>>, do_parse!( pws1: opt!(tag!("-")) >> ws!(tag!("for")) >> var: ws!(target_single) >> ws!(tag!("in")) >> iter: ws!(expr_any) >> nws1: opt!(tag!("-")) >> call!(tag_block_end, s) >> block: call!(parse_template, s) >> call!(tag_block_start, s) >> pws2: opt!(tag!("-")) >> ws!(tag!("endfor")) >> nws2: opt!(tag!("-")) >> (Node::Loop(WS(pws1.is_some(), nws1.is_some()), var, iter, block, WS(pws2.is_some(), nws2.is_some()))) )); named!(block_extends, do_parse!( ws!(tag!("extends")) >> name: ws!(expr_str_lit) >> (Node::Extends(name)) )); named_args!(block_block<'a>(s: &'a Syntax<'a>) , Node<'a>>, do_parse!( pws1: opt!(tag!("-")) >> ws!(tag!("block")) >> name: ws!(identifier) >> nws1: opt!(tag!("-")) >> call!(tag_block_end, s) >> contents: call!(parse_template, s) >> call!(tag_block_start, s) >> pws2: opt!(tag!("-")) >> ws!(tag!("endblock")) >> opt!(ws!(tag!(name))) >> nws2: opt!(tag!("-")) >> (Node::BlockDef(WS(pws1.is_some(), nws1.is_some()), name, contents, WS(pws2.is_some(), nws2.is_some()))) )); named!(block_include, do_parse!( pws: opt!(tag!("-")) >> ws!(tag!("include")) >> name: ws!(expr_str_lit) >> nws: opt!(tag!("-")) >> (Node::Include(WS(pws.is_some(), nws.is_some()), match name { Expr::StrLit(s) => s, _ => panic!("include path must be a string literal"), })) )); named!(block_import, do_parse!( pws: opt!(tag!("-")) >> ws!(tag!("import")) >> name: ws!(expr_str_lit) >> ws!(tag!("as")) >> scope: ws!(identifier) >> nws: opt!(tag!("-")) >> (Node::Import(WS(pws.is_some(), nws.is_some()), match name { Expr::StrLit(s) => s, _ => panic!("import path must be a string literal"), }, scope)) )); named_args!(block_macro<'a>(s: &'a Syntax<'a>) , Node<'a>>, do_parse!( pws1: opt!(tag!("-")) >> ws!(tag!("macro")) >> name: ws!(identifier) >> params: ws!(parameters) >> nws1: opt!(tag!("-")) >> call!(tag_block_end, s) >> contents: call!(parse_template, s) >> call!(tag_block_start, s) >> pws2: opt!(tag!("-")) >> ws!(tag!("endmacro")) >> nws2: opt!(tag!("-")) >> ({ if name == "super" { panic!("invalid macro name 'super'"); } Node::Macro( name, Macro { ws1: WS(pws1.is_some(), nws1.is_some()), args: params, nodes: contents, ws2: WS(pws2.is_some(), nws2.is_some()) } ) }) )); named_args!(block_node<'a>(s: &'a Syntax<'a>) , Node<'a>>, do_parse!( call!(tag_block_start, s) >> contents: alt!( block_call | block_let | call!(block_if, s) | call!(block_for, s) | call!(block_match, s) | block_extends | block_include | block_import | call!(block_block, s) | call!(block_macro, s) ) >> call!(tag_block_end, s) >> (contents) )); named_args!(block_comment<'a>(s: &'a Syntax<'a>) , Node<'a>>, do_parse!( call!(tag_comment_start, s) >> pws: opt!(tag!("-")) >> inner: take_until_s!(s.comment_end) >> call!(tag_comment_end, s) >> (Node::Comment(WS(pws.is_some(), inner.len() > 1 && inner[inner.len() - 1] == b'-'))) )); named_args!(parse_template<'a>(s: &'a Syntax<'a>), Vec>>, many0!(alt!( call!(take_content, s) | call!(block_comment, s) | call!(expr_node, s) | call!(block_node, s) ))); named_args!(tag_block_start<'a>(s: &'a Syntax<'a>) , Input<'a>>, tag!(s.block_start)); named_args!(tag_block_end<'a>(s: &'a Syntax<'a>) , Input<'a>>, tag!(s.block_end)); named_args!(tag_comment_start<'a>(s: &'a Syntax<'a>) , Input<'a>>, tag!(s.comment_start)); named_args!(tag_comment_end<'a>(s: &'a Syntax<'a>) , Input<'a>>, tag!(s.comment_end)); named_args!(tag_expr_start<'a>(s: &'a Syntax<'a>) , Input<'a>>, tag!(s.expr_start)); named_args!(tag_expr_end<'a>(s: &'a Syntax<'a>) , Input<'a>>, tag!(s.expr_end)); pub fn parse<'a>(src: &'a str, syntax: &'a Syntax<'a>) -> Vec> { match parse_template(Input(src.as_bytes()), syntax) { Ok((left, res)) => { if !left.is_empty() { let s = str::from_utf8(left.0).unwrap(); panic!("unable to parse template:\n\n{:?}", s); } else { res } } Err(nom::Err::Error(err)) => panic!("problems parsing template source: {:?}", err), Err(nom::Err::Failure(err)) => panic!("problems parsing template source: {:?}", err), Err(nom::Err::Incomplete(_)) => panic!("parsing incomplete"), } } #[cfg(test)] mod tests { use askama_shared::Syntax; 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", "", "")); } #[test] #[should_panic] fn test_invalid_block() { super::parse("{% extend \"blah\" %}", &Syntax::default()); } #[test] fn test_parse_filter() { super::parse("{{ strvar|e }}", &Syntax::default()); } #[test] fn change_delimiters_parse_filter() { let syntax = Syntax { expr_start: "{~", expr_end: "~}", ..Syntax::default() }; super::parse("{~ strvar|e ~}", &syntax); } }