469 lines
16 KiB
Rust
469 lines
16 KiB
Rust
![]() |
// Copyright 2015 The Rust Project Developers. See the COPYRIGHT
|
||
|
// file at the top-level directory of this distribution and at
|
||
|
// http://rust-lang.org/COPYRIGHT.
|
||
|
//
|
||
|
// Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
|
||
|
// http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
|
||
|
// <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
|
||
|
// option. This file may not be copied, modified, or distributed
|
||
|
// except according to those terms.
|
||
|
|
||
|
#![feature(box_syntax)]
|
||
|
#![feature(box_patterns)]
|
||
|
#![feature(rustc_private)]
|
||
|
#![feature(collections)]
|
||
|
#![feature(os)]
|
||
|
#![feature(core)]
|
||
|
#![feature(unicode)]
|
||
|
#![feature(old_path)]
|
||
|
#![feature(exit_status)]
|
||
|
|
||
|
#[macro_use]
|
||
|
extern crate log;
|
||
|
|
||
|
extern crate getopts;
|
||
|
extern crate rustc;
|
||
|
extern crate rustc_driver;
|
||
|
extern crate syntax;
|
||
|
|
||
|
use rustc::session::Session;
|
||
|
use rustc::session::config::{self, Input};
|
||
|
use rustc_driver::{driver, CompilerCalls, Compilation};
|
||
|
|
||
|
use syntax::ast;
|
||
|
use syntax::codemap::{self, CodeMap, Span, Pos};
|
||
|
use syntax::diagnostics;
|
||
|
use syntax::parse::token;
|
||
|
use syntax::print::pprust;
|
||
|
use syntax::visit;
|
||
|
|
||
|
use std::mem;
|
||
|
|
||
|
use changes::ChangeSet;
|
||
|
|
||
|
pub mod rope;
|
||
|
mod changes;
|
||
|
|
||
|
const IDEAL_WIDTH: usize = 80;
|
||
|
const MAX_WIDTH: usize = 100;
|
||
|
const MIN_STRING: usize = 10;
|
||
|
|
||
|
// Formatting which depends on the AST.
|
||
|
fn fmt_ast<'a>(krate: &ast::Crate, codemap: &'a CodeMap) -> ChangeSet<'a> {
|
||
|
let mut visitor = FmtVisitor { codemap: codemap,
|
||
|
changes: ChangeSet::from_codemap(codemap) };
|
||
|
visit::walk_crate(&mut visitor, krate);
|
||
|
|
||
|
visitor.changes
|
||
|
}
|
||
|
|
||
|
// Formatting done on a char by char basis.
|
||
|
fn fmt_lines(changes: &mut ChangeSet) {
|
||
|
// Iterate over the chars in the change set.
|
||
|
for (f, text) in changes.text() {
|
||
|
let mut trims = vec![];
|
||
|
let mut last_wspace = None;
|
||
|
let mut line_len = 0;
|
||
|
let mut cur_line = 1;
|
||
|
for (c, b) in text.chars() {
|
||
|
if c == '\n' { // TOOD test for \r too
|
||
|
// Check for (and record) trailing whitespace.
|
||
|
if let Some(lw) = last_wspace {
|
||
|
trims.push((lw, b));
|
||
|
line_len -= b - lw;
|
||
|
}
|
||
|
// Check for any line width errors we couldn't correct.
|
||
|
if line_len > MAX_WIDTH {
|
||
|
// FIXME store the error rather than reporting immediately.
|
||
|
println!("Rustfmt couldn't fix (sorry). {}:{}: line longer than {} characters",
|
||
|
f, cur_line, MAX_WIDTH);
|
||
|
}
|
||
|
line_len = 0;
|
||
|
cur_line += 1;
|
||
|
last_wspace = None;
|
||
|
} else {
|
||
|
line_len += 1;
|
||
|
if c.is_whitespace() {
|
||
|
if last_wspace.is_none() {
|
||
|
last_wspace = Some(b);
|
||
|
}
|
||
|
} else {
|
||
|
last_wspace = None;
|
||
|
}
|
||
|
}
|
||
|
}
|
||
|
|
||
|
unsafe {
|
||
|
// Invariant: we only mutate a rope after we have searched it, then
|
||
|
// we will not search it again.
|
||
|
let mut_text: &mut rope::Rope = mem::transmute(text);
|
||
|
let mut_count: &mut u64 = mem::transmute(&changes.count);
|
||
|
let mut offset = 0;
|
||
|
// Get rid of any trailing whitespace we recorded earlier.
|
||
|
for &(s, e) in trims.iter() {
|
||
|
// Note that we change the underlying ropes directly, we don't
|
||
|
// go through the changeset because our change positions are
|
||
|
// relative to the newest text, not the original.
|
||
|
debug!("Stripping trailing whitespace {}:{}-{} \"{}\"",
|
||
|
f, s, e, text.slice(s-offset..e-offset));
|
||
|
mut_text.remove(s-offset, e-offset);
|
||
|
*mut_count += 1;
|
||
|
offset += e - s;
|
||
|
}
|
||
|
}
|
||
|
}
|
||
|
}
|
||
|
|
||
|
struct FmtVisitor<'a> {
|
||
|
codemap: &'a CodeMap,
|
||
|
changes: ChangeSet<'a>,
|
||
|
}
|
||
|
|
||
|
impl<'a, 'v> visit::Visitor<'v> for FmtVisitor<'a> {
|
||
|
fn visit_expr(&mut self, ex: &'v ast::Expr) {
|
||
|
match ex.node {
|
||
|
ast::Expr_::ExprLit(ref l) => match l.node {
|
||
|
ast::Lit_::LitStr(ref is, _) => {
|
||
|
self.rewrite_string(&is, l.span);
|
||
|
}
|
||
|
_ => {}
|
||
|
},
|
||
|
_ => {}
|
||
|
}
|
||
|
|
||
|
visit::walk_expr(self, ex)
|
||
|
}
|
||
|
|
||
|
fn visit_fn(&mut self,
|
||
|
fk: visit::FnKind<'v>,
|
||
|
fd: &'v ast::FnDecl,
|
||
|
b: &'v ast::Block,
|
||
|
s: Span,
|
||
|
_: ast::NodeId) {
|
||
|
self.fix_formal_args(fd);
|
||
|
visit::walk_fn(self, fk, fd, b, s);
|
||
|
}
|
||
|
|
||
|
fn visit_item(&mut self, item: &'v ast::Item) {
|
||
|
// TODO check each item is on a new line and is correctly indented.
|
||
|
match item.node {
|
||
|
ast::Item_::ItemUse(ref vp) => {
|
||
|
match vp.node {
|
||
|
ast::ViewPath_::ViewPathList(ref path, ref path_list) => {
|
||
|
let new_str = self.fix_use_list(path, path_list, vp.span);
|
||
|
|
||
|
// TODO move these optimisations to ChangeSet
|
||
|
if new_str != self.codemap.span_to_snippet(item.span).unwrap() {
|
||
|
self.changes.change_span(item.span, new_str);
|
||
|
}
|
||
|
}
|
||
|
ast::ViewPath_::ViewPathGlob(_) => {
|
||
|
// FIXME convert to list?
|
||
|
}
|
||
|
_ => {}
|
||
|
}
|
||
|
}
|
||
|
_ => {}
|
||
|
}
|
||
|
visit::walk_item(self, item);
|
||
|
}
|
||
|
}
|
||
|
|
||
|
fn make_indent(width: usize) -> String {
|
||
|
let mut indent = String::with_capacity(width);
|
||
|
for _ in 0..width {
|
||
|
indent.push(' ')
|
||
|
}
|
||
|
indent
|
||
|
}
|
||
|
|
||
|
impl<'a> FmtVisitor<'a> {
|
||
|
// TODO NEEDS TESTS
|
||
|
fn rewrite_string(&mut self, s: &str, span: Span) {
|
||
|
// FIXME I bet this stomps unicode escapes in the source string
|
||
|
|
||
|
// Check if there is anything to fix: we always try to fixup multi-line
|
||
|
// strings, or if the string is too long for the line.
|
||
|
let l_loc = self.codemap.lookup_char_pos(span.lo);
|
||
|
let r_loc = self.codemap.lookup_char_pos(span.hi);
|
||
|
if l_loc.line == r_loc.line && r_loc.col.to_usize() <= MAX_WIDTH {
|
||
|
return;
|
||
|
}
|
||
|
|
||
|
// TODO if lo.col > IDEAL - 10, start a new line (need cur indent for that)
|
||
|
|
||
|
let s = s.escape_default();
|
||
|
|
||
|
// TODO use fixed value.
|
||
|
let l_loc = self.codemap.lookup_char_pos(span.lo);
|
||
|
let l_col = l_loc.col.to_usize();
|
||
|
|
||
|
let indent = make_indent(l_col + 1);
|
||
|
let indent = &indent;
|
||
|
|
||
|
let max_chars = MAX_WIDTH - (l_col + 1);
|
||
|
|
||
|
let mut cur_start = 0;
|
||
|
let mut result = String::new();
|
||
|
result.push('"');
|
||
|
loop {
|
||
|
let mut cur_end = cur_start + max_chars;
|
||
|
|
||
|
if cur_end >= s.len() {
|
||
|
result.push_str(&s[cur_start..]);
|
||
|
break;
|
||
|
}
|
||
|
|
||
|
// Make sure we're on a char boundary.
|
||
|
cur_end = next_char(&s, cur_end);
|
||
|
|
||
|
// Push cur_end left until we reach whitespace
|
||
|
while !s.char_at(cur_end-1).is_whitespace() {
|
||
|
cur_end = prev_char(&s, cur_end);
|
||
|
|
||
|
if cur_end - cur_start < MIN_STRING {
|
||
|
// We can't break at whitespace, fall back to splitting
|
||
|
// anywhere that doesn't break an escape sequence
|
||
|
cur_end = next_char(&s, cur_start + max_chars);
|
||
|
while s.char_at(cur_end) == '\\' {
|
||
|
cur_end = prev_char(&s, cur_end);
|
||
|
}
|
||
|
}
|
||
|
}
|
||
|
// Make sure there is no whitespace to the right of the break.
|
||
|
while cur_end < s.len() && s.char_at(cur_end).is_whitespace() {
|
||
|
cur_end = next_char(&s, cur_end+1);
|
||
|
}
|
||
|
result.push_str(&s[cur_start..cur_end]);
|
||
|
result.push_str("\\\n");
|
||
|
result.push_str(indent);
|
||
|
|
||
|
cur_start = cur_end;
|
||
|
}
|
||
|
result.push('"');
|
||
|
|
||
|
// Check that we actually changed something.
|
||
|
if result == self.codemap.span_to_snippet(span).unwrap() {
|
||
|
return;
|
||
|
}
|
||
|
|
||
|
self.changes.change_span(span, result);
|
||
|
}
|
||
|
|
||
|
// Basically just pretty prints a multi-item import.
|
||
|
fn fix_use_list(&mut self,
|
||
|
path: &ast::Path,
|
||
|
path_list: &[ast::PathListItem],
|
||
|
vp_span: Span) -> String {
|
||
|
// FIXME remove unused imports
|
||
|
|
||
|
// FIXME check indentation
|
||
|
let l_loc = self.codemap.lookup_char_pos(vp_span.lo);
|
||
|
let path_str = pprust::path_to_string(&path);
|
||
|
let indent = l_loc.col.0;
|
||
|
// After accounting for the overhead, how much space left for
|
||
|
// the item list? ( 5 = :: + { + } + ; )
|
||
|
let space = IDEAL_WIDTH - (indent + path_str.len() + 5);
|
||
|
// 4 = `use` + one space
|
||
|
// TODO might be pub use
|
||
|
let indent = make_indent(indent-4);
|
||
|
|
||
|
let mut cur_str = String::new();
|
||
|
let mut first = true;
|
||
|
// If `self` is in the list, put it first.
|
||
|
if path_list.iter().any(|vpi|
|
||
|
if let ast::PathListItem_::PathListMod{ .. } = vpi.node {
|
||
|
true
|
||
|
} else {
|
||
|
false
|
||
|
}
|
||
|
) {
|
||
|
cur_str = "self".to_string();
|
||
|
first = false;
|
||
|
}
|
||
|
|
||
|
let mut new_str = String::new();
|
||
|
for vpi in path_list.iter() {
|
||
|
match vpi.node {
|
||
|
ast::PathListItem_::PathListIdent{ name, .. } => {
|
||
|
let next_item = &token::get_ident(name);
|
||
|
if cur_str.len() + next_item.len() > space {
|
||
|
let cur_line = format!("{}use {}::{{{}}};\n", indent, path_str, cur_str);
|
||
|
new_str.push_str(&cur_line);
|
||
|
|
||
|
cur_str = String::new();
|
||
|
first = true;
|
||
|
}
|
||
|
|
||
|
if first {
|
||
|
first = false;
|
||
|
} else {
|
||
|
cur_str.push_str(", ");
|
||
|
}
|
||
|
|
||
|
cur_str.push_str(next_item);
|
||
|
}
|
||
|
ast::PathListItem_::PathListMod{ .. } => {}
|
||
|
}
|
||
|
}
|
||
|
|
||
|
assert!(!first);
|
||
|
let cur_line = format!("{}use {}::{{{}}};", indent, path_str, cur_str);
|
||
|
new_str.push_str(&cur_line);
|
||
|
|
||
|
new_str
|
||
|
}
|
||
|
|
||
|
fn fix_formal_args<'v>(&mut self, fd: &'v ast::FnDecl) {
|
||
|
// For now, just check the arguments line up and make them per-row if the line is too long.
|
||
|
let args = &fd.inputs;
|
||
|
if args.len() <= 1 {
|
||
|
return;
|
||
|
}
|
||
|
// TODO not really using the hi positions
|
||
|
let spans: Vec<_> = args.iter().map(|a| (a.pat.span.lo, a.ty.span.hi)).collect();
|
||
|
let locs: Vec<_> = spans.iter().map(|&(a, b)| (self.codemap.lookup_char_pos(a), self.codemap.lookup_char_pos(b))).collect();
|
||
|
let first_loc = &locs[0].0;
|
||
|
// TODO need to adjust for previous changes here.
|
||
|
let same_row = locs.iter().all(|&(ref l, _)| l.line == first_loc.line);
|
||
|
let same_col = locs.iter().all(|&(ref l, _)| l.col == first_loc.col);
|
||
|
|
||
|
if same_col {
|
||
|
// TODO Check one arg per line and no lines in between (except comments)
|
||
|
return;
|
||
|
}
|
||
|
|
||
|
if same_row { // TODO check line is < 100 && first_loc.line {
|
||
|
// TODO could also fix whitespace around punctuaton here
|
||
|
// TODO and could check that we're on the same line as the function call, if possible
|
||
|
return;
|
||
|
}
|
||
|
|
||
|
let col = self.changes.col(spans[0].0);
|
||
|
let mut indent = String::with_capacity(col);
|
||
|
indent.push('\n');
|
||
|
for _ in 0..col { indent.push(' '); }
|
||
|
let last_idx = spans.len() - 1;
|
||
|
for (i, s) in spans.iter().enumerate() {
|
||
|
// Take the span from lo to lo (or the last hi for the last arg),
|
||
|
// trim, push on top of indent, then replace the old lo-lo span with it.
|
||
|
let mut new_text = if i == 0 {
|
||
|
"".to_string()
|
||
|
} else {
|
||
|
indent.clone()
|
||
|
};
|
||
|
let hi = if i == last_idx {
|
||
|
s.1
|
||
|
} else {
|
||
|
spans[i+1].0
|
||
|
};
|
||
|
// TODO need a version of slice taking locs, not a span
|
||
|
let snippet = self.changes.slice_span(Span{ lo: s.0, hi: hi, expn_id: codemap::NO_EXPANSION }).to_string();
|
||
|
let snippet = snippet.trim();
|
||
|
new_text.push_str(snippet);
|
||
|
self.changes.change(&first_loc.file.name, (s.0).0 as usize, hi.0 as usize, new_text);
|
||
|
}
|
||
|
}
|
||
|
}
|
||
|
|
||
|
#[inline]
|
||
|
fn prev_char(s: &str, mut i: usize) -> usize {
|
||
|
if i == 0 { return 0; }
|
||
|
|
||
|
i -= 1;
|
||
|
while !s.is_char_boundary(i) {
|
||
|
i -= 1;
|
||
|
}
|
||
|
i
|
||
|
}
|
||
|
|
||
|
#[inline]
|
||
|
fn next_char(s: &str, mut i: usize) -> usize {
|
||
|
if i >= s.len() { return s.len(); }
|
||
|
|
||
|
while !s.is_char_boundary(i) {
|
||
|
i += 1;
|
||
|
}
|
||
|
i
|
||
|
}
|
||
|
|
||
|
struct RustFmtCalls {
|
||
|
input_path: Option<Path>,
|
||
|
}
|
||
|
|
||
|
impl<'a> CompilerCalls<'a> for RustFmtCalls {
|
||
|
fn early_callback(&mut self,
|
||
|
_: &getopts::Matches,
|
||
|
_: &diagnostics::registry::Registry)
|
||
|
-> Compilation {
|
||
|
Compilation::Continue
|
||
|
}
|
||
|
|
||
|
fn some_input(&mut self, input: Input, input_path: Option<Path>) -> (Input, Option<Path>) {
|
||
|
match input_path {
|
||
|
Some(ref ip) => self.input_path = Some(ip.clone()),
|
||
|
_ => {
|
||
|
// FIXME should handle string input and write to stdout or something
|
||
|
panic!("No input path");
|
||
|
}
|
||
|
}
|
||
|
(input, input_path)
|
||
|
}
|
||
|
|
||
|
fn no_input(&mut self,
|
||
|
_: &getopts::Matches,
|
||
|
_: &config::Options,
|
||
|
_: &Option<Path>,
|
||
|
_: &Option<Path>,
|
||
|
_: &diagnostics::registry::Registry)
|
||
|
-> Option<(Input, Option<Path>)> {
|
||
|
panic!("No input supplied to RustFmt");
|
||
|
}
|
||
|
|
||
|
fn late_callback(&mut self,
|
||
|
_: &getopts::Matches,
|
||
|
_: &Session,
|
||
|
_: &Input,
|
||
|
_: &Option<Path>,
|
||
|
_: &Option<Path>)
|
||
|
-> Compilation {
|
||
|
Compilation::Continue
|
||
|
}
|
||
|
|
||
|
fn build_controller(&mut self, _: &Session) -> driver::CompileController<'a> {
|
||
|
let mut control = driver::CompileController::basic();
|
||
|
control.after_parse.stop = Compilation::Stop;
|
||
|
control.after_parse.callback = box |state| {
|
||
|
let krate = state.krate.unwrap();
|
||
|
let codemap = state.session.codemap();
|
||
|
let mut changes = fmt_ast(krate, codemap);
|
||
|
fmt_lines(&mut changes);
|
||
|
|
||
|
println!("Making {} changes", changes.count);
|
||
|
println!("{}", changes);
|
||
|
// FIXME(#5) Should be user specified whether to show or replace.
|
||
|
};
|
||
|
|
||
|
control
|
||
|
}
|
||
|
}
|
||
|
|
||
|
fn main() {
|
||
|
let args = std::os::args();
|
||
|
let mut call_ctxt = RustFmtCalls { input_path: None };
|
||
|
rustc_driver::run_compiler(&args, &mut call_ctxt);
|
||
|
std::env::set_exit_status(0);
|
||
|
}
|
||
|
|
||
|
// FIXME comments
|
||
|
// comments aren't in the AST, which makes processing them difficult, but then
|
||
|
// comments are complicated anyway. I think I am happy putting off tackling them
|
||
|
// for now. Long term the soluton is for comments to be in the AST, but that means
|
||
|
// only the libsyntax AST, not the rustc one, which means waiting for the ASTs
|
||
|
// to diverge one day....
|
||
|
|
||
|
// Once we do have comments, we just have to implement a simple word wrapping
|
||
|
// algorithm to keep the width under IDEAL_WIDTH. We should also convert multiline
|
||
|
// /* ... */ comments to // and check doc comments are in the right place and of
|
||
|
// the right kind.
|