From 1267e966f47893bdaa9ab0f3f15d32110eddfd6e Mon Sep 17 00:00:00 2001 From: David Tolnay Date: Thu, 12 Apr 2018 11:42:36 -0700 Subject: Isolate proc-macro dependency to the proc macro The dependency of askama on askama_shared on syn on libproc_macro used to mean libproc_macro would be dynamically linked into any crate using askama. We want only askama_derive to have a runtime dependency on proc macro. This commit moves all proc macro code from askama_shared into askama_derive so that the askama crate no longer dynamically links libproc_macro. --- askama_derive/src/parser.rs | 694 ++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 694 insertions(+) create mode 100644 askama_derive/src/parser.rs (limited to 'askama_derive/src/parser.rs') diff --git a/askama_derive/src/parser.rs b/askama_derive/src/parser.rs new file mode 100644 index 0000000..402801f --- /dev/null +++ b/askama_derive/src/parser.rs @@ -0,0 +1,694 @@ +// rustfmt doesn't do a very good job on nom parser invocations. +#![cfg_attr(rustfmt, rustfmt_skip)] + +use nom::{self, IResult}; +use std::str; + +#[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), + Filter(&'a str, Vec>), + BinOp(&'a str, Box>, Box>), + Group(Box>), + MethodCall(Box>, &'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(), + 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>); + +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), + (Brace(start), b'{') | + (Brace(start), b'%') | + (Brace(start), b'#') => End(start), + (Any, _) | + (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!(num_lit<&str>, map!(nom::digit, + |s| str::from_utf8(s).unwrap() +)); + +named!(expr_num_lit, map!(num_lit, + |s| Expr::NumLit(s) +)); + +named!(expr_array_lit, do_parse!( + ws!(tag_s!("[")) >> + first: expr_any >> + rest: many0!(do_parse!( + ws!(tag_s!(",")) >> + part: expr_any >> + (part) + )) >> + ws!(tag_s!("]")) >> + ({ + 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_s!("::") >> + 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_s!("::") >> + 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_s!("(") >> + args: opt!(do_parse!( + arg0: ws!(expr_any) >> + args: many0!(do_parse!( + tag_s!(",") >> + argn: ws!(expr_any) >> + (argn) + )) >> + ({ + let mut res = vec![arg0]; + res.extend(args); + res + }) + )) >> + tag_s!(")") >> + (args.unwrap_or_default()) +)); + +named!(parameters>, do_parse!( + tag_s!("(") >> + vals: opt!(do_parse!( + arg0: ws!(identifier) >> + args: many0!(do_parse!( + tag_s!(",") >> + argn: ws!(identifier) >> + (argn) + )) >> + ({ + let mut res = vec![arg0]; + res.extend(args); + res + }) + )) >> + tag_s!(")") >> + (vals.unwrap_or_default()) +)); + +named!(with_parameters>, do_parse!( + tag_s!("with") >> + ws!(tag_s!("(")) >> + vals: opt!(do_parse!( + arg0: ws!(match_parameter) >> + args: many0!(do_parse!( + tag_s!(",") >> + argn: ws!(match_parameter) >> + (argn) + )) >> + ({ + let mut res = vec![arg0]; + res.extend(args); + res + }) + )) >> + tag_s!(")") >> + (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<(&str, Option>)>, do_parse!( + tag_s!(".") >> + 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!(filter<(&str, Option>)>, do_parse!( + tag_s!("|") >> + fname: identifier >> + args: opt!(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, do_parse!( + left: $inner >> + op_and_right: opt!(pair!(ws!(alt!($( tag_s!($op) )|*)), expr_any)) >> + (match op_and_right { + Some((op, right)) => Expr::BinOp( + str::from_utf8(op).unwrap(), Box::new(left), Box::new(right) + ), + None => left, + }) + )); + } +} + +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!(block_call, do_parse!( + pws: opt!(tag_s!("-")) >> + ws!(tag_s!("call")) >> + scope: opt!(do_parse!( + scope: ws!(identifier) >> + ws!(tag_s!("::")) >> + (scope) + )) >> + name: ws!(identifier) >> + args: ws!(arguments) >> + nws: opt!(tag_s!("-")) >> + (Node::Call(WS(pws.is_some(), nws.is_some()), scope, name, args)) +)); + +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!( + 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!("-")) >> + ({ + 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!(match_else_block, do_parse!( + tag_s!("{%") >> + pws: opt!(tag_s!("-")) >> + ws!(tag_s!("else")) >> + nws: opt!(tag_s!("-")) >> + tag_s!("%}") >> + block: parse_template >> + (WS(pws.is_some(), nws.is_some()), None, vec![], block) +)); + +named!(when_block, do_parse!( + tag_s!("{%") >> + pws: opt!(tag_s!("-")) >> + ws!(tag_s!("when")) >> + variant: ws!(match_variant) >> + params: opt!(ws!(with_parameters)) >> + nws: opt!(tag_s!("-")) >> + tag_s!("%}") >> + block: parse_template >> + (WS(pws.is_some(), nws.is_some()), Some(variant), params.unwrap_or_default(), block) +)); + +named!(block_match, do_parse!( + pws1: opt!(tag_s!("-")) >> + ws!(tag_s!("match")) >> + expr: ws!(expr_any) >> + nws1: opt!(tag_s!("-")) >> + tag_s!("%}") >> + inter: opt!(take_content) >> + arms: many1!(when_block) >> + else_arm: opt!(match_else_block) >> + ws!(tag_s!("{%")) >> + pws2: opt!(tag_s!("-")) >> + ws!(tag_s!("endmatch")) >> + nws2: opt!(tag_s!("-")) >> + ({ + 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_s!("-")) >> + ws!(tag_s!("let")) >> + var: ws!(target_single) >> + val: opt!(do_parse!( + ws!(tag_s!("=")) >> + val: ws!(expr_any) >> + (val) + )) >> + nws: opt!(tag_s!("-")) >> + (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!(block_for, do_parse!( + 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!("-")) >> + (Node::Loop(WS(pws1.is_some(), nws1.is_some()), + var, iter, block, + WS(pws2.is_some(), pws2.is_some()))) +)); + +named!(block_extends, do_parse!( + ws!(tag_s!("extends")) >> + name: ws!(expr_str_lit) >> + (Node::Extends(name)) +)); + +named!(block_block, do_parse!( + 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")) >> + opt!(ws!(tag_s!(name))) >> + nws2: opt!(tag_s!("-")) >> + (Node::BlockDef(WS(pws1.is_some(), nws1.is_some()), + name, contents, + WS(pws2.is_some(), pws2.is_some()))) +)); + +named!(block_include, do_parse!( + pws: opt!(tag_s!("-")) >> + ws!(tag_s!("include")) >> + name: ws!(expr_str_lit) >> + nws: opt!(tag_s!("-")) >> + (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_s!("-")) >> + ws!(tag_s!("import")) >> + name: ws!(expr_str_lit) >> + ws!(tag_s!("as")) >> + scope: ws!(identifier) >> + nws: opt!(tag_s!("-")) >> + (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!(block_macro, do_parse!( + pws1: opt!(tag_s!("-")) >> + ws!(tag_s!("macro")) >> + name: ws!(identifier) >> + params: ws!(parameters) >> + nws1: opt!(tag_s!("-")) >> + tag_s!("%}") >> + contents: parse_template >> + tag_s!("{%") >> + pws2: opt!(tag_s!("-")) >> + ws!(tag_s!("endmacro")) >> + nws2: opt!(tag_s!("-")) >> + (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!(block_node, do_parse!( + tag_s!("{%") >> + contents: alt!( + block_call | + block_let | + block_if | + block_for | + block_match | + block_extends | + block_include | + block_import | + block_block | + block_macro + ) >> + tag_s!("%}") >> + (contents) +)); + +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_node +))); + +pub fn parse(src: &str) -> Vec { + match parse_template(src.as_bytes()) { + IResult::Done(left, res) => { + if !left.is_empty() { + let s = str::from_utf8(left).unwrap(); + panic!("unable to parse template:\n\n{:?}", s); + } else { + 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", "", "")); + } + #[test] + #[should_panic] + fn test_invalid_block() { + super::parse("{% extend \"blah\" %}"); + } +} -- cgit