updated documentation and readme

README.md

@@ -2,7 +2,7 @@
 [![docs](https://docs.rs/antlr-rust/badge.svg)](https://docs.rs/antlr-rust)
 [![Crate](https://img.shields.io/crates/v/antlr_rust.svg)](https://crates.io/crates/antlr_rust)
 
-ANTLR4 runtime for Rust programming language 
+[ANTLR4](https://github.com/antlr/antlr4) runtime for Rust programming language.
 
 Tool(generator) part is currently located in rust-target branch of my antlr4 fork [rrevenantt/antlr4/tree/rust-target](https://github.com/rrevenantt/antlr4/tree/rust-target)
 Latest version is automatically built to [releases](https://github.com/rrevenantt/antlr4rust/releases) on this repository.
@@ -13,9 +13,6 @@ and [tests/my_tests.rs](tests/my_test.rs) for actual usage examples
 
 ### Implementation status
 
-Everything is implemented, "business" logic is quite stable and well tested, but user facing 
-API is not very robust yet and very likely will have some changes.
-
 For now development is going on in this repository 
 but eventually it will be merged to main ANTLR4 repo
 
@@ -40,7 +37,7 @@ Can be done after merge:
  - run rustfmt on generated parser
 ###### Long term improvements
  - generate enum for labeled alternatives without redundant `Error` option
- - option to generate fields instead of getters by default
+ - option to generate fields instead of getters by default and make visiting based on fields
  - make tree generic over pointer type and allow tree nodes to arena.
  (requires GAT, otherwise it would be a problem for users that want ownership for parse tree)
  - support stable rust
@@ -84,12 +81,6 @@ I.e. for `MultContext` struct will contain `a` and `b` fields containing child s
 `op` field with `TerminalNode` type which corresponds to individual `Token`.
 It also is possible to disable generic parse tree creation to keep only selected children via
 `parser.build_parse_trees = false`.
-
-### Key properties
- - Supports full zero-copy parsing including byte parsers
-    (you should be able to write zero-copy serde deserializers).
- - Supports downcasting in places where type is not known statically(trait objects and embedded action)
- - Listener and 
 
 ### Differences with Java
 Although Rust runtime API has been made as close as possible to Java, 
@@ -106,11 +97,12 @@ there are quite some differences because Rust is not an OOP language and is much
  If you need exactly the same behavior, use `[u32]` based `InputStream`, or implement custom `CharStream`.
  - In actions you have to escape `'` in rust lifetimes with `\ ` because ANTLR considers them as strings, e.g. `Struct<\'lifetime>`
  - To make custom tokens you should use `@tokenfactory` custom action, instead of usual `TokenLabelType` parser option.
- In Rust target TokenFactory is main customisation interface that allows to specify input type of token type. 
+ ANTLR parser options can accept only single identifiers while Rust target needs know about lifetime as well. 
+ Also in Rust target `TokenFactory` is the way to specify token type. As example you can see [CSV](grammars/CSV.g4) test grammar.
  - All rule context variables (rule argument or rule return) should implement `Default + Clone`.
 
 ### Unsafe
-Currently, unsafe is used only to cast from trait object back to original type 
+Currently, unsafe is used only for downcasting (through another crate) 
 and to update data inside Rc via `get_mut_unchecked`(returned mutable reference is used immediately and not stored anywhere)
 
 ### Versioning

src/input_stream.rs

@@ -51,11 +51,13 @@ pub type CodePoint8BitCharStream<'a> = InputStream<&'a [u8]>;
 pub type CodePoint16BitCharStream<'a> = InputStream<&'a [u16]>;
 pub type CodePoint32BitCharStream<'a> = InputStream<&'a [u32]>;
 
-impl<'a, T> CharStream<&'a [T]> for InputStream<&'a [T]>
+impl<'a, T> CharStream<Cow<'a, [T]>> for InputStream<&'a [T]>
 where
     [T]: InputData,
 {
-    fn get_text(&self, a: isize, b: isize) -> &'a [T] { self.get_text_inner(a, b).into() }
+    fn get_text(&self, a: isize, b: isize) -> Cow<'a, [T]> {
+        Cow::Borrowed(self.get_text_inner(a, b))
+    }
 }
 
 impl<'a, T> CharStream<String> for InputStream<&'a [T]>

src/lib.rs

@@ -16,26 +16,73 @@
 #![warn(trivial_numeric_casts)]
 //! # Antlr4 runtime
 //!
+//! **This is pre-release version.**
+//! **Some small breaking changes are still possible, although none is currently planned**
+//!
 //! This is a Rust runtime for [ANTLR4] parser generator.
 //! It is required to use parsers and lexers generated by [ANTLR4] parser generator
 //!
 //! This documentation refers to particular api used by generated parsers,lexers and syntax trees.
 //!
-//! For info on how to generate parser please refer to:
+//! For info on what is [ANTLR4] and how to generate parser please refer to:
 //!  - [ANTLR4] main repository
-//!  - [README](https://github.com/rrevenantt/antlr4rust/blob/master/README.md) for Rust target
+//!  - [README] for Rust target
 //!
 //! [ANTLR4]: https://github.com/antlr/antlr4
+//! [README]: https://github.com/rrevenantt/antlr4rust/blob/master/README.md
+//!
+//! ### Customization
+//!
+//! All input and output can be customized and optimized for particular usecase by implementing
+//! related trait. Each of them already has different implementations that should be enough for most cases.
+//! For more details see docs for corresponding trait and containing module.
+//!
+//! Currently available are:
+//!  - [`CharStream`] - Lexer input, stream of char values with slicing support
+//!  - [`TokenFactory`] - How lexer creates tokens.
+//!  - [`Token`] - Element of [`TokenStream`]
+//!  - [`TokenStream`] - Parser input, created from lexer or other token source.
+//!  - [`ParserRuleContext`] - Node of created syntax tree.
 //!
 //! ### Zero-copy and lifetimes
 //!
 //! This library supports full zero-copy parsing. To allow this
-//! `'input` lifetime is used everywhere inside.
+//! `'input` lifetime is used everywhere inside to refer to data borrowed by parser.
+//! Besides reference to input it also can be [`TokenFactory`] if it returns references to tokens.
+//! See [`ArenaFactory`] as an example of such behavior. It allocates tokens in [`Arena`](typed_arena::Arena) and return references.
+//!
+//! Using generated parse tree you should be careful to not require longer lifetime after the parsing.
+//! If that's the case you will likely get "does not live long enough" error on the input string,
+//! despite actual lifetime conflict is happening much later
 //!
-//! Rust infers lifetimes from the end. It means that if something requires longer lifetime
-//! when you are using generated tree, then you will get error
+//! If you need to generate owned versions of parse tree or you want simpler usage,
+//! you can opt out zero-copy by requiring `'input` to be static. In this case it is easier to also use
+//! types that contains "owned" in their name or constructor function like `OwningTokenFactory`
+//! or `InputStream::new_owned()`
 //!
+//! ### Visitors and Listeners
 //!
+//! Currently visitors and listeners must outlive `'input`.
+//! In general this means that visitor has either `'static` or `'input` lifetime.
+//! Thus you can retrieve references to parsed data from syntax tree to listener/visitor. (as example you can see visitor test)
+//!
+//! You can try to give visitor outside references but in this case
+//! if those references do not outlive `'input` you will get very confusing error messages,
+//! so this is not recommended.
+//!
+//! ### Downcasting
+//!
+//! Rule context trait object support downcasting even for zero-copy case.
+//! Also generic types(currently these are `H:ErrorStrategy` and `I:`[`TokenStream`]) that you can
+//! access in generated parser from embedded actions also can be downcasted to concrete types.
+//! To do it `TidExt::downcast_*` extension methods should be used.
+//!
+//! [`CharStream`]: crate::char_stream::CharStream
+//! [`TokenFactory`]: crate::token_factory::TokenFactory
+//! [`ArenaFactory`]: crate::token_factory::ArenaFactory
+//! [`Token`]: crate::token::Token
+//! [`TokenStream`]: crate::token_stream::TokenStream
+//! [`ParserRuleContext`]: crate::parser_rule_context::ParserRuleContext
 
 #[macro_use]
 extern crate lazy_static;
@@ -112,8 +159,7 @@ pub mod token;
 pub mod trees;
 mod utils;
 //pub mod tokenstream_rewriter_test;
-#[doc(hidden)]
-pub mod atn_type;
+mod atn_type;
 pub mod rule_context;
 pub mod vocabulary;
 

src/parser_rule_context.rs

@@ -1,3 +1,7 @@
+//!
+//!
+//!
+//!
 use std::any::{type_name, Any};
 use std::borrow::{Borrow, BorrowMut};
 use std::cell::{Ref, RefCell, RefMut};
@@ -20,8 +24,6 @@ use crate::tree::{
 };
 use better_any::{Tid, TidAble, TidExt};
 
-// use crate::utils::IndexIter;
-
 pub trait ParserRuleContext<'input>:
     ParseTree<'input> + RuleContext<'input> + Debug + Tid<'input>
 {

src/token_factory.rs

@@ -192,6 +192,15 @@ pub type ArenaCommonFactory<'a> = ArenaFactory<'a, CommonTokenFactory, CommonTok
 
 /// This is a wrapper for Token factory that allows to allocate tokens in separate arena.
 /// It can allow to significantly improve performance by passing Tokens by references everywhere.
+///
+/// Requires `&'a Tok: Default` bound to produce invalid tokens, which can be easily implemented
+/// like this:
+/// ```text
+/// lazy_static!{ static ref INVALID_TOKEN:CustomToken = ... }
+/// impl Default for &'_ CustomToken {
+///     fn default() -> Self { &**INVALID_TOKEN }
+/// }
+/// ```
 // Box is used here because it is almost always should be used for token factory
 #[derive(Tid)]
 pub struct ArenaFactory<'input, TF, T>

tests/gen/labelsparser.rs

@@ -1031,6 +1031,7 @@ where
                         &mut recog.base,
                     )))?,
                 }
+
                 let tmp = recog.input.lt(-1).cloned();
                 recog.ctx.as_ref().unwrap().set_stop(tmp);
                 recog.base.set_state(36);

tests/gen/simplelrparser.rs

@@ -404,6 +404,7 @@ where
                     recog.base.set_state(7);
                     recog.base.match_token(ID, &mut recog.err_handler)?;
                 }
+
                 let tmp = recog.input.lt(-1).cloned();
                 recog.ctx.as_ref().unwrap().set_stop(tmp);
                 recog.base.set_state(13);