diff options
author | Erick Tryzelaar <erick.tryzelaar@gmail.com> | 2015-08-12 22:15:58 -0700 |
---|---|---|
committer | Erick Tryzelaar <erick.tryzelaar@gmail.com> | 2015-08-12 22:15:58 -0700 |
commit | 4e0172a9a510bdd4468dc22c562d79c3c5449daa (patch) | |
tree | 79ad2f9d42851e047c67b4495216b0f103f2902c | |
parent | fac884d032526a25eb26db54047c62a21945aaf4 (diff) | |
parent | b70f6e53b20c8f90f525219b1db7e99f6ef417dc (diff) | |
download | milf-rs-4e0172a9a510bdd4468dc22c562d79c3c5449daa.tar.gz milf-rs-4e0172a9a510bdd4468dc22c562d79c3c5449daa.zip |
Merge remote-tracking branch 'remotes/origin/master' into serde
-rw-r--r-- | .travis.yml | 25 | ||||
-rw-r--r-- | Cargo.toml | 2 | ||||
-rw-r--r-- | README.md | 3 | ||||
-rw-r--r-- | examples/toml2json.rs | 57 | ||||
-rw-r--r-- | src/decoder/rustc_serialize.rs | 75 | ||||
-rw-r--r-- | src/lib.rs | 3 | ||||
-rw-r--r-- | src/parser.rs | 296 |
7 files changed, 332 insertions, 129 deletions
diff --git a/.travis.yml b/.travis.yml index 4eeadf2..417e85d 100644 --- a/.travis.yml +++ b/.travis.yml @@ -1,5 +1,11 @@ language: rust +rust: + - 1.0.0 + - beta + - nightly sudo: false +before_script: + - pip install 'travis-cargo<0.2' --user && export PATH=$HOME/.local/bin:$PATH script: - cargo build --verbose - cargo build --verbose --no-default-features @@ -7,17 +13,18 @@ script: - cargo test --verbose --features serde - rustdoc --test README.md -L target - cargo doc --no-deps -after_success: | - [ $TRAVIS_BRANCH = master ] && - [ $TRAVIS_PULL_REQUEST = false ] && - echo '<meta http-equiv=refresh content=0;url=toml/index.html>' > target/doc/index.html && - pip install ghp-import --user $USER && - $HOME/.local/bin/ghp-import -n target/doc && - git push -qf https://${TOKEN}@github.com/${TRAVIS_REPO_SLUG}.git gh-pages +after_success: + - travis-cargo --only nightly doc-upload + - travis-cargo coveralls --no-sudo env: global: - - secure: FO8GVrtHAn5GTj4LOz2BApC3tAEsMbNzvH5UVmCIeNKPuVcKcI3oWNJC/KMCvuJZhu96J3okfRLBxBJrhxsp/YT4fS4kibhZDm6AzbCqxz6AmvHJo2d0jztoRyuLwLSkhwW8vM4VeHH+Tf4PeC56YmnpUGkccHMMidxytJzx8qI= - - secure: WVCzGVsthRub6ezJU15xzo+ahlUoZEwvZDeMPmjIMf1G28ObE9Y4BeUNW0j9CxCjyQ+5S0mrp1l0TESN326XTDosigabDiGnKyr5wfncnreN3PCUi3gx7NI+bRTy9B3eV318BhuCDgLgRWLWufCyPtkgAdT6cl+u6p+bEh+vyxo= + secure: LZMkQQJT5LqLQQ8JyakjvHNqqMPy8lm/SyC+H5cKUVI/xk7xRuti4eKY937N8uSmbff2m9ZYlG6cNwIOfk/nWn8YsqxA8Wg/xugubWzqGuqu+NQ4IZVa7INT2Fiqyk5SPCh8B5fo2x7OBJ24SCkWb2p8bEWAuW8XdZZOdmi3H2I= notifications: email: on_success: never +addons: + apt: + packages: + - libcurl4-openssl-dev + - libelf-dev + - libdw-dev @@ -1,7 +1,7 @@ [package] name = "toml" -version = "0.1.20" +version = "0.1.21" authors = ["Alex Crichton <alex@alexcrichton.com>"] license = "MIT/Apache-2.0" readme = "README.md" @@ -1,8 +1,9 @@ # toml-rs [![Build Status](https://travis-ci.org/alexcrichton/toml-rs.svg?branch=master)](https://travis-ci.org/alexcrichton/toml-rs) +[![Coverage Status](https://coveralls.io/repos/alexcrichton/toml-rs/badge.svg?branch=master&service=github)](https://coveralls.io/github/alexcrichton/toml-rs?branch=master) -[Documentation](http://alexcrichton.com/toml-rs/toml/index.html) +[Documentation](http://alexcrichton.com/toml-rs) A [TOML][toml] decoder and encoder for Rust. This library is currently compliant with the v0.4.0 version of TOML. This library will also likely continue to stay up to diff --git a/examples/toml2json.rs b/examples/toml2json.rs new file mode 100644 index 0000000..0d40680 --- /dev/null +++ b/examples/toml2json.rs @@ -0,0 +1,57 @@ +#![deny(warnings)] + +extern crate toml; +extern crate rustc_serialize; + +use std::fs::File; +use std::env; +use std::io; +use std::io::prelude::*; + +use toml::Value; +use rustc_serialize::json::Json; + +fn main() { + let mut args = env::args(); + let mut input = String::new(); + let filename = if args.len() > 1 { + let name = args.nth(1).unwrap(); + File::open(&name).and_then(|mut f| { + f.read_to_string(&mut input) + }).unwrap(); + name + } else { + io::stdin().read_to_string(&mut input).unwrap(); + "<stdin>".to_string() + }; + + let mut parser = toml::Parser::new(&input); + let toml = match parser.parse() { + Some(toml) => toml, + None => { + for err in &parser.errors { + let (loline, locol) = parser.to_linecol(err.lo); + let (hiline, hicol) = parser.to_linecol(err.hi); + println!("{}:{}:{}-{}:{} error: {}", + filename, loline, locol, hiline, hicol, err.desc); + } + return + } + }; + let json = convert(Value::Table(toml)); + println!("{}", json.pretty()); +} + +fn convert(toml: Value) -> Json { + match toml { + Value::String(s) => Json::String(s), + Value::Integer(i) => Json::I64(i), + Value::Float(f) => Json::F64(f), + Value::Boolean(b) => Json::Boolean(b), + Value::Array(arr) => Json::Array(arr.into_iter().map(convert).collect()), + Value::Table(table) => Json::Object(table.into_iter().map(|(k, v)| { + (k, convert(v)) + }).collect()), + Value::Datetime(dt) => Json::String(dt), + } +} diff --git a/src/decoder/rustc_serialize.rs b/src/decoder/rustc_serialize.rs index 6e8fe59..534154d 100644 --- a/src/decoder/rustc_serialize.rs +++ b/src/decoder/rustc_serialize.rs @@ -96,15 +96,31 @@ impl rustc_serialize::Decoder for Decoder { -> Result<T, DecodeError> where F: FnMut(&mut Decoder, usize) -> Result<T, DecodeError> { - let mut first_error = None; + // When decoding enums, this crate takes the strategy of trying to + // decode the current TOML as all of the possible variants, returning + // success on the first one that succeeds. + // + // Note that fidelity of the errors returned here is a little nebulous, + // but we try to return the error that had the relevant field as the + // longest field. This way we hopefully match an error against what was + // most likely being written down without losing too much info. + let mut first_error = None::<DecodeError>; for i in 0..names.len() { let mut d = self.sub_decoder(self.toml.clone(), ""); match f(&mut d, i) { - Ok(t) => { self.toml = d.toml; return Ok(t) } + Ok(t) => { + self.toml = d.toml; + return Ok(t) + } Err(e) => { - if first_error.is_none() { - first_error = Some(e); + if let Some(ref first) = first_error { + let my_len = e.field.as_ref().map(|s| s.len()); + let first_len = first.field.as_ref().map(|s| s.len()); + if my_len <= first_len { + continue + } } + first_error = Some(e); } } } @@ -158,7 +174,7 @@ impl rustc_serialize::Decoder for Decoder { let toml = match self.toml { Some(Value::Table(ref mut table)) => { table.remove(&field) - .or_else(|| table.remove(&f_name.replace("_", "-"))) + .or_else(|| table.remove(&f_name.replace("_", "-"))) }, ref found => return Err(self.mismatch("table", found)), }; @@ -275,8 +291,8 @@ impl rustc_serialize::Decoder for Decoder { Some(Value::Table(ref table)) => { match table.iter().skip(idx).next() { Some((key, _)) => { - let val = Value::String(format!("{}", key)); - f(&mut self.sub_decoder(Some(val), &**key)) + let val = Value::String(key.to_string()); + f(&mut self.sub_decoder(Some(val), key)) } None => Err(self.err(ExpectedMapKey(idx))), } @@ -291,9 +307,9 @@ impl rustc_serialize::Decoder for Decoder { match self.toml { Some(Value::Table(ref table)) => { match table.iter().skip(idx).next() { - Some((_, value)) => { + Some((key, value)) => { // XXX: this shouldn't clone - f(&mut self.sub_decoder(Some(value.clone()), "")) + f(&mut self.sub_decoder(Some(value.clone()), key)) } None => Err(self.err(ExpectedMapElement(idx))), } @@ -309,3 +325,44 @@ impl rustc_serialize::Decoder for Decoder { } } } + +#[cfg(test)] +mod tests { + use rustc_serialize::Decodable; + use std::collections::HashMap; + + use {Parser, Decoder, Value}; + + #[test] + fn bad_enum_chooses_longest_error() { + #[derive(RustcDecodable)] + #[allow(dead_code)] + struct Foo { + wut: HashMap<String, Bar>, + } + + #[derive(RustcDecodable)] + enum Bar { + Simple(String), + Detailed(Baz), + } + + #[derive(RustcDecodable, Debug)] + struct Baz { + features: Vec<String>, + } + + let s = r#" + [wut] + a = { features = "" } + "#; + let v = Parser::new(s).parse().unwrap(); + let mut d = Decoder::new(Value::Table(v)); + let err = match Foo::decode(&mut d) { + Ok(_) => panic!("expected error"), + Err(e) => e, + }; + assert_eq!(err.field.as_ref().unwrap(), "wut.a.features"); + + } +} @@ -44,7 +44,6 @@ use std::collections::BTreeMap; use std::str::FromStr; -use std::string; pub use parser::{Parser, ParserError}; @@ -77,7 +76,7 @@ pub enum Value { pub type Array = Vec<Value>; /// Type representing a TOML table, payload of the Value::Table variant -pub type Table = BTreeMap<string::String, Value>; +pub type Table = BTreeMap<String, Value>; impl Value { /// Tests whether this and another value have the same type. diff --git a/src/parser.rs b/src/parser.rs index 9a15de8..b7a810f 100644 --- a/src/parser.rs +++ b/src/parser.rs @@ -5,13 +5,84 @@ use std::error::Error; use std::fmt; use std::str; -use Table as TomlTable; -use Value::{self, Array, Table, Float, Integer, Boolean, Datetime}; - macro_rules! try { ($e:expr) => (match $e { Some(s) => s, None => return None }) } +// We redefine Value because we need to keep track of encountered table +// definitions, eg when parsing: +// +// [a] +// [a.b] +// [a] +// +// we have to error out on redefinition of [a]. This bit of data is difficult to +// track in a side table so we just have a "stripped down" AST to work with +// which has the relevant metadata fields in it. +struct TomlTable { + values: BTreeMap<String, Value>, + defined: bool, +} + +impl TomlTable { + fn convert(self) -> super::Table { + self.values.into_iter().map(|(k,v)| (k, v.convert())).collect() + } +} + +enum Value { + String(String), + Integer(i64), + Float(f64), + Boolean(bool), + Datetime(String), + Array(Vec<Value>), + Table(TomlTable), +} + +impl Value { + fn type_str(&self) -> &'static str { + match *self { + Value::String(..) => "string", + Value::Integer(..) => "integer", + Value::Float(..) => "float", + Value::Boolean(..) => "boolean", + Value::Datetime(..) => "datetime", + Value::Array(..) => "array", + Value::Table(..) => "table", + } + } + + fn same_type(&self, other: &Value) -> bool { + match (self, other) { + (&Value::String(..), &Value::String(..)) | + (&Value::Integer(..), &Value::Integer(..)) | + (&Value::Float(..), &Value::Float(..)) | + (&Value::Boolean(..), &Value::Boolean(..)) | + (&Value::Datetime(..), &Value::Datetime(..)) | + (&Value::Array(..), &Value::Array(..)) | + (&Value::Table(..), &Value::Table(..)) => true, + + _ => false, + } + } + + fn convert(self) -> super::Value { + match self { + Value::String(x) => super::Value::String(x), + Value::Integer(x) => super::Value::Integer(x), + Value::Float(x) => super::Value::Float(x), + Value::Boolean(x) => super::Value::Boolean(x), + Value::Datetime(x) => super::Value::Datetime(x), + Value::Array(v) => + super::Value::Array( + v.into_iter().map(|x| x.convert()).collect() + ), + Value::Table(t) => super::Value::Table(t.convert()) + } + } +} + /// Parser for converting a string to a TOML `Value` instance. /// /// This parser contains the string slice that is being parsed, and exports the @@ -161,8 +232,8 @@ impl<'a> Parser<'a> { /// /// If an error occurs, the `errors` field of this parser can be consulted /// to determine the cause of the parse failure. - pub fn parse(&mut self) -> Option<TomlTable> { - let mut ret = BTreeMap::new(); + pub fn parse(&mut self) -> Option<super::Table> { + let mut ret = TomlTable { values: BTreeMap::new(), defined: false }; while self.peek(0).is_some() { self.ws(); if self.newline() { continue } @@ -175,9 +246,8 @@ impl<'a> Parser<'a> { let mut keys = Vec::new(); loop { self.ws(); - match self.key_name() { - Some(s) => keys.push(s), - None => {} + if let Some(s) = self.key_name() { + keys.push(s); } self.ws(); if self.eat(']') { @@ -189,12 +259,16 @@ impl<'a> Parser<'a> { if keys.len() == 0 { return None } // Build the section table - let mut table = BTreeMap::new(); + let mut table = TomlTable { + values: BTreeMap::new(), + defined: true, + }; if !self.values(&mut table) { return None } if array { - self.insert_array(&mut ret, &*keys, Table(table), start) + self.insert_array(&mut ret, &keys, Value::Table(table), + start) } else { - self.insert_table(&mut ret, &*keys, table, start) + self.insert_table(&mut ret, &keys, table, start) } } else { if !self.values(&mut ret) { return None } @@ -203,7 +277,7 @@ impl<'a> Parser<'a> { if self.errors.len() > 0 { None } else { - Some(ret) + Some(ret.convert()) } } @@ -214,18 +288,13 @@ impl<'a> Parser<'a> { self.finish_string(start, false) } else { let mut ret = String::new(); - loop { - match self.cur.clone().next() { - Some((_, ch)) => { - match ch { - 'a' ... 'z' | - 'A' ... 'Z' | - '0' ... '9' | - '_' | '-' => { self.cur.next(); ret.push(ch) } - _ => break, - } - } - None => break + while let Some((_, ch)) = self.cur.clone().next() { + match ch { + 'a' ... 'z' | + 'A' ... 'Z' | + '0' ... '9' | + '_' | '-' => { self.cur.next(); ret.push(ch) } + _ => break, } } Some(ret) @@ -344,9 +413,8 @@ impl<'a> Parser<'a> { return Some(ret) } Some((pos, '\\')) => { - match escape(self, pos, multiline) { - Some(c) => ret.push(c), - None => {} + if let Some(c) = escape(self, pos, multiline) { + ret.push(c); } } Some((pos, ch)) if ch < '\u{1f}' => { @@ -391,32 +459,26 @@ impl<'a> Parser<'a> { } else { "invalid" }; - match u32::from_str_radix(num, 16).ok() { - Some(n) => { - match char::from_u32(n) { - Some(c) => { - me.cur.by_ref().skip(len - 1).next(); - return Some(c) - } - None => { - me.errors.push(ParserError { - lo: pos + 1, - hi: pos + 5, - desc: format!("codepoint `{:x}` is \ - not a valid unicode \ - codepoint", n), - }) - } - } - } - None => { + if let Some(n) = u32::from_str_radix(num, 16).ok() { + if let Some(c) = char::from_u32(n) { + me.cur.by_ref().skip(len - 1).next(); + return Some(c) + } else { me.errors.push(ParserError { - lo: pos, - hi: pos + 1, - desc: format!("expected {} hex digits \ - after a `{}` escape", len, c), + lo: pos + 1, + hi: pos + 5, + desc: format!("codepoint `{:x}` is \ + not a valid unicode \ + codepoint", n), }) } + } else { + me.errors.push(ParserError { + lo: pos, + hi: pos + 1, + desc: format!("expected {} hex digits \ + after a `{}` escape", len, c), + }) } None } @@ -519,9 +581,9 @@ impl<'a> Parser<'a> { }; let input = input.trim_left_matches('+'); if is_float { - input.parse().ok().map(Float) + input.parse().ok().map(Value::Float) } else { - input.parse().ok().map(Integer) + input.parse().ok().map(Value::Integer) } }; if ret.is_none() { @@ -603,12 +665,12 @@ impl<'a> Parser<'a> { for _ in 0..4 { self.cur.next(); } - Some(Boolean(true)) + Some(Value::Boolean(true)) } else if rest.starts_with("false") { for _ in 0..5 { self.cur.next(); } - Some(Boolean(false)) + Some(Value::Boolean(false)) } else { let next = self.next_pos(); self.errors.push(ParserError { @@ -659,7 +721,7 @@ impl<'a> Parser<'a> { valid = valid && it.next().map(is_digit).unwrap_or(false); valid = valid && it.next().map(|c| c == 'Z').unwrap_or(false); if valid { - Some(Datetime(date.clone())) + Some(Value::Datetime(date.clone())) } else { self.errors.push(ParserError { lo: start, @@ -683,7 +745,7 @@ impl<'a> Parser<'a> { loop { // Break out early if we see the closing bracket consume(self); - if self.eat(']') { return Some(Array(ret)) } + if self.eat(']') { return Some(Value::Array(ret)) } // Attempt to parse a value, triggering an error if it's the wrong // type. @@ -709,14 +771,14 @@ impl<'a> Parser<'a> { } consume(self); if !self.expect(']') { return None } - return Some(Array(ret)) + return Some(Value::Array(ret)) } fn inline_table(&mut self, _start: usize) -> Option<Value> { if !self.expect('{') { return None } self.ws(); - let mut ret = BTreeMap::new(); - if self.eat('}') { return Some(Table(ret)) } + let mut ret = TomlTable { values: BTreeMap::new(), defined: true }; + if self.eat('}') { return Some(Value::Table(ret)) } loop { let lo = self.next_pos(); let key = try!(self.key_name()); @@ -729,19 +791,19 @@ impl<'a> Parser<'a> { if !self.expect(',') { return None } self.ws(); } - return Some(Table(ret)) + return Some(Value::Table(ret)) } fn insert(&mut self, into: &mut TomlTable, key: String, value: Value, key_lo: usize) { - if into.contains_key(&key) { + if into.values.contains_key(&key) { self.errors.push(ParserError { lo: key_lo, hi: key_lo + key.len(), desc: format!("duplicate key: `{}`", key), }) } else { - into.insert(key, value); + into.values.insert(key, value); } } @@ -751,15 +813,12 @@ impl<'a> Parser<'a> { for part in keys[..keys.len() - 1].iter() { let tmp = cur; - if tmp.contains_key(part) { - match *tmp.get_mut(part).unwrap() { - Table(ref mut table) => { - cur = table; - continue - } - Array(ref mut array) => { + if tmp.values.contains_key(part) { + match *tmp.values.get_mut(part).unwrap() { + Value::Table(ref mut table) => cur = table, + Value::Array(ref mut array) => { match array.last_mut() { - Some(&mut Table(ref mut table)) => cur = table, + Some(&mut Value::Table(ref mut table)) => cur = table, _ => { self.errors.push(ParserError { lo: key_lo, @@ -770,7 +829,6 @@ impl<'a> Parser<'a> { return None } } - continue } _ => { self.errors.push(ParserError { @@ -782,12 +840,16 @@ impl<'a> Parser<'a> { return None } } + continue } // Initialize an empty table as part of this sub-key - tmp.insert(part.clone(), Table(BTreeMap::new())); - match *tmp.get_mut(part).unwrap() { - Table(ref mut inner) => cur = inner, + tmp.values.insert(part.clone(), Value::Table(TomlTable { + values: BTreeMap::new(), + defined: false, + })); + match *tmp.values.get_mut(part).unwrap() { + Value::Table(ref mut inner) => cur = inner, _ => unreachable!(), } } @@ -795,45 +857,38 @@ impl<'a> Parser<'a> { } fn insert_table(&mut self, into: &mut TomlTable, keys: &[String], - value: TomlTable, key_lo: usize) { + table: TomlTable, key_lo: usize) { let (into, key) = match self.recurse(into, keys, key_lo) { Some(pair) => pair, None => return, }; - let key = format!("{}", key); - let mut added = false; - if !into.contains_key(&key) { - into.insert(key.clone(), Table(BTreeMap::new())); - added = true; + if !into.values.contains_key(key) { + into.values.insert(key.to_owned(), Value::Table(table)); + return } - match into.get_mut(&key) { - Some(&mut Table(ref mut table)) => { - let any_tables = table.values().any(|v| v.as_table().is_some()); - if !any_tables && !added { + if let Value::Table(ref mut into) = *into.values.get_mut(key).unwrap() { + if into.defined { + self.errors.push(ParserError { + lo: key_lo, + hi: key_lo + key.len(), + desc: format!("redefinition of table `{}`", key), + }); + } + for (k, v) in table.values { + if into.values.insert(k.clone(), v).is_some() { self.errors.push(ParserError { lo: key_lo, hi: key_lo + key.len(), - desc: format!("redefinition of table `{}`", key), + desc: format!("duplicate key `{}` in table", k), }); } - for (k, v) in value.into_iter() { - if table.insert(k.clone(), v).is_some() { - self.errors.push(ParserError { - lo: key_lo, - hi: key_lo + key.len(), - desc: format!("duplicate key `{}` in table", k), - }); - } - } - } - Some(_) => { - self.errors.push(ParserError { - lo: key_lo, - hi: key_lo + key.len(), - desc: format!("duplicate key `{}` in table", key), - }); } - None => {} + } else { + self.errors.push(ParserError { + lo: key_lo, + hi: key_lo + key.len(), + desc: format!("duplicate key `{}` in table", key), + }); } } @@ -843,12 +898,11 @@ impl<'a> Parser<'a> { Some(pair) => pair, None => return, }; - let key = format!("{}", key); - if !into.contains_key(&key) { - into.insert(key.clone(), Array(Vec::new())); + if !into.values.contains_key(key) { + into.values.insert(key.to_owned(), Value::Array(Vec::new())); } - match *into.get_mut(&key).unwrap() { - Array(ref mut vec) => { + match *into.values.get_mut(key).unwrap() { + Value::Array(ref mut vec) => { match vec.first() { Some(ref v) if !v.same_type(&value) => { self.errors.push(ParserError { @@ -1333,4 +1387,32 @@ trimmed in raw strings. c = 2 ", "duplicate key `c` in table"); } + + #[test] + fn bad_table_redefine() { + bad!(" + [a] + foo=\"bar\" + [a.b] + foo=\"bar\" + [a] + ", "redefinition of table `a`"); + bad!(" + [a] + foo=\"bar\" + b = { foo = \"bar\" } + [a] + ", "redefinition of table `a`"); + bad!(" + [a] + b = {} + [a.b] + ", "redefinition of table `b`"); + + bad!(" + [a] + b = {} + [a] + ", "redefinition of table `a`"); + } } |