From dd8f7cd126403955295c8b0cdbccc5ca5cbef763 Mon Sep 17 00:00:00 2001 From: Inokentiy Babushkin Date: Sat, 10 Jun 2017 16:09:51 +0200 Subject: Moved FileMap construction to it's own constructor. The rationale is that BOM stripping is needed for lazy source loading for external crates, and duplication can be avoided by moving the corresponding functionality to libsyntax_pos. --- src/libsyntax_pos/Cargo.toml | 1 + src/libsyntax_pos/lib.rs | 37 +++++++++++++++++++++++++++++++++++++ 2 files changed, 38 insertions(+) (limited to 'src/libsyntax_pos') diff --git a/src/libsyntax_pos/Cargo.toml b/src/libsyntax_pos/Cargo.toml index 760aaa8a957..dd8129bab51 100644 --- a/src/libsyntax_pos/Cargo.toml +++ b/src/libsyntax_pos/Cargo.toml @@ -10,3 +10,4 @@ crate-type = ["dylib"] [dependencies] serialize = { path = "../libserialize" } +rustc_data_structures = { path = "../librustc_data_structures" } diff --git a/src/libsyntax_pos/lib.rs b/src/libsyntax_pos/lib.rs index caea1497181..75ce8b675f3 100644 --- a/src/libsyntax_pos/lib.rs +++ b/src/libsyntax_pos/lib.rs @@ -38,6 +38,11 @@ use std::ops::{Add, Sub}; use std::rc::Rc; use std::cmp; use std::fmt; +use std::hash::Hasher; + +use rustc_data_structures::stable_hasher::StableHasher; + +extern crate rustc_data_structures; use serialize::{Encodable, Decodable, Encoder, Decoder}; @@ -522,6 +527,31 @@ impl fmt::Debug for FileMap { } impl FileMap { + pub fn new(name: FileName, + name_was_remapped: bool, + mut src: String, + start_pos: BytePos) -> FileMap { + remove_bom(&mut src); + + let mut hasher: StableHasher = StableHasher::new(); + hasher.write(src.as_bytes()); + let src_hash = hasher.finish(); + + let end_pos = start_pos.to_usize() + src.len(); + + FileMap { + name: name, + name_was_remapped: name_was_remapped, + crate_of_origin: 0, + src: Some(Rc::new(src)), + src_hash: src_hash, + start_pos: start_pos, + end_pos: Pos::from_usize(end_pos), + lines: RefCell::new(Vec::new()), + multibyte_chars: RefCell::new(Vec::new()), + } + } + /// EFFECT: register a start-of-line offset in the /// table of line-beginnings. /// UNCHECKED INVARIANT: these offsets must be added in the right @@ -621,6 +651,13 @@ impl FileMap { } } +/// Remove utf-8 BOM if any. +fn remove_bom(src: &mut String) { + if src.starts_with("\u{feff}") { + src.drain(..3); + } +} + // _____________________________________________________________________________ // Pos, BytePos, CharPos // -- cgit 1.4.1-3-g733a5