src/tools/rustfmt/src/formatting.rs

   1 // High level formatting functions.
   2
   3 use std::collections::HashMap;
   4 use std::io::{self, Write};
   5 use std::time::{Duration, Instant};
   6
   7 use rustc_ast::ast;
   8 use rustc_span::Span;
   9
  10 use self::newline_style::apply_newline_style;
  11 use crate::comment::{CharClasses, FullCodeCharKind};
  12 use crate::config::{Config, FileName, Verbosity};
  13 use crate::formatting::generated::is_generated_file;
  14 use crate::issues::BadIssueSeeker;
  15 use crate::modules::Module;
  16 use crate::parse::parser::{DirectoryOwnership, Parser, ParserError};
  17 use crate::parse::session::ParseSess;
  18 use crate::utils::{contains_skip, count_newlines};
  19 use crate::visitor::FmtVisitor;
  20 use crate::{modules, source_file, ErrorKind, FormatReport, Input, Session};
  21
  22 mod generated;
  23 mod newline_style;
  24
  25 // A map of the files of a crate, with their new content
  26 pub(crate) type SourceFile = Vec<FileRecord>;
  27 pub(crate) type FileRecord = (FileName, String);
  28
  29 impl<'b, T: Write + 'b> Session<'b, T> {
  30     pub(crate) fn format_input_inner(
  31         &mut self,
  32         input: Input,
  33         is_macro_def: bool,
  34     ) -> Result<FormatReport, ErrorKind> {
  35         if !self.config.version_meets_requirement() {
  36             return Err(ErrorKind::VersionMismatch);
  37         }
  38
  39         rustc_span::create_session_if_not_set_then(self.config.edition().into(), |_| {
  40             if self.config.disable_all_formatting() {
  41                 // When the input is from stdin, echo back the input.
  42                 if let Input::Text(ref buf) = input {
  43                     if let Err(e) = io::stdout().write_all(buf.as_bytes()) {
  44                         return Err(From::from(e));
  45                     }
  46                 }
  47                 return Ok(FormatReport::new());
  48             }
  49
  50             let config = &self.config.clone();
  51             let format_result = format_project(input, config, self, is_macro_def);
  52
  53             format_result.map(|report| {
  54                 self.errors.add(&report.internal.borrow().1);
  55                 report
  56             })
  57         })
  58     }
  59 }
  60
  61 /// Determine if a module should be skipped. True if the module should be skipped, false otherwise.
  62 fn should_skip_module<T: FormatHandler>(
  63     config: &Config,
  64     context: &FormatContext<'_, T>,
  65     input_is_stdin: bool,
  66     main_file: &FileName,
  67     path: &FileName,
  68     module: &Module<'_>,
  69 ) -> bool {
  70     if contains_skip(module.attrs()) {
  71         return true;
  72     }
  73
  74     if config.skip_children() && path != main_file {
  75         return true;
  76     }
  77
  78     if !input_is_stdin && context.ignore_file(path) {
  79         return true;
  80     }
  81
  82     // FIXME(calebcartwright) - we need to determine how we'll handle the
  83     // `format_generated_files` option with stdin based input.
  84     if !input_is_stdin && !config.format_generated_files() {
  85         let source_file = context.parse_session.span_to_file_contents(module.span);
  86         let src = source_file.src.as_ref().expect("SourceFile without src");
  87
  88         if is_generated_file(src) {
  89             return true;
  90         }
  91     }
  92
  93     false
  94 }
  95
  96 // Format an entire crate (or subset of the module tree).
  97 fn format_project<T: FormatHandler>(
  98     input: Input,
  99     config: &Config,
 100     handler: &mut T,
 101     is_macro_def: bool,
 102 ) -> Result<FormatReport, ErrorKind> {
 103     let mut timer = Timer::start();
 104
 105     let main_file = input.file_name();
 106     let input_is_stdin = main_file == FileName::Stdin;
 107
 108     let parse_session = ParseSess::new(config)?;
 109     if config.skip_children() && parse_session.ignore_file(&main_file) {
 110         return Ok(FormatReport::new());
 111     }
 112
 113     // Parse the crate.
 114     let mut report = FormatReport::new();
 115     let directory_ownership = input.to_directory_ownership();
 116     let krate = match Parser::parse_crate(input, &parse_session) {
 117         Ok(krate) => krate,
 118         // Surface parse error via Session (errors are merged there from report)
 119         Err(e) => {
 120             let forbid_verbose = input_is_stdin || e != ParserError::ParsePanicError;
 121             should_emit_verbose(forbid_verbose, config, || {
 122                 eprintln!("The Rust parser panicked");
 123             });
 124             report.add_parsing_error();
 125             return Ok(report);
 126         }
 127     };
 128
 129     let mut context = FormatContext::new(&krate, report, parse_session, config, handler);
 130     let files = modules::ModResolver::new(
 131         &context.parse_session,
 132         directory_ownership.unwrap_or(DirectoryOwnership::UnownedViaBlock),
 133         !input_is_stdin && !config.skip_children(),
 134     )
 135     .visit_crate(&krate)?
 136     .into_iter()
 137     .filter(|(path, module)| {
 138         !should_skip_module(config, &context, input_is_stdin, &main_file, path, module)
 139     })
 140     .collect::<Vec<_>>();
 141
 142     timer = timer.done_parsing();
 143
 144     // Suppress error output if we have to do any further parsing.
 145     context.parse_session.set_silent_emitter();
 146
 147     for (path, module) in files {
 148         should_emit_verbose(input_is_stdin, config, || println!("Formatting {}", path));
 149         context.format_file(path, &module, is_macro_def)?;
 150     }
 151     timer = timer.done_formatting();
 152
 153     should_emit_verbose(input_is_stdin, config, || {
 154         println!(
 155             "Spent {0:.3} secs in the parsing phase, and {1:.3} secs in the formatting phase",
 156             timer.get_parse_time(),
 157             timer.get_format_time(),
 158         )
 159     });
 160
 161     Ok(context.report)
 162 }
 163
 164 // Used for formatting files.
 165 #[derive(new)]
 166 struct FormatContext<'a, T: FormatHandler> {
 167     krate: &'a ast::Crate,
 168     report: FormatReport,
 169     parse_session: ParseSess,
 170     config: &'a Config,
 171     handler: &'a mut T,
 172 }
 173
 174 impl<'a, T: FormatHandler + 'a> FormatContext<'a, T> {
 175     fn ignore_file(&self, path: &FileName) -> bool {
 176         self.parse_session.ignore_file(path)
 177     }
 178
 179     // Formats a single file/module.
 180     fn format_file(
 181         &mut self,
 182         path: FileName,
 183         module: &Module<'_>,
 184         is_macro_def: bool,
 185     ) -> Result<(), ErrorKind> {
 186         let snippet_provider = self.parse_session.snippet_provider(module.span);
 187         let mut visitor = FmtVisitor::from_parse_sess(
 188             &self.parse_session,
 189             self.config,
 190             &snippet_provider,
 191             self.report.clone(),
 192         );
 193         visitor.skip_context.update_with_attrs(&self.krate.attrs);
 194         visitor.is_macro_def = is_macro_def;
 195         visitor.last_pos = snippet_provider.start_pos();
 196         visitor.skip_empty_lines(snippet_provider.end_pos());
 197         visitor.format_separate_mod(module, snippet_provider.end_pos());
 198
 199         debug_assert_eq!(
 200             visitor.line_number,
 201             count_newlines(&visitor.buffer),
 202             "failed in format_file visitor.buffer:\n {:?}",
 203             &visitor.buffer
 204         );
 205
 206         // For some reason, the source_map does not include terminating
 207         // newlines so we must add one on for each file. This is sad.
 208         source_file::append_newline(&mut visitor.buffer);
 209
 210         format_lines(
 211             &mut visitor.buffer,
 212             &path,
 213             &visitor.skipped_range.borrow(),
 214             self.config,
 215             &self.report,
 216         );
 217
 218         apply_newline_style(
 219             self.config.newline_style(),
 220             &mut visitor.buffer,
 221             snippet_provider.entire_snippet(),
 222         );
 223
 224         if visitor.macro_rewrite_failure {
 225             self.report.add_macro_format_failure();
 226         }
 227         self.report
 228             .add_non_formatted_ranges(visitor.skipped_range.borrow().clone());
 229
 230         self.handler.handle_formatted_file(
 231             &self.parse_session,
 232             path,
 233             visitor.buffer.to_owned(),
 234             &mut self.report,
 235         )
 236     }
 237 }
 238
 239 // Handle the results of formatting.
 240 trait FormatHandler {
 241     fn handle_formatted_file(
 242         &mut self,
 243         parse_session: &ParseSess,
 244         path: FileName,
 245         result: String,
 246         report: &mut FormatReport,
 247     ) -> Result<(), ErrorKind>;
 248 }
 249
 250 impl<'b, T: Write + 'b> FormatHandler for Session<'b, T> {
 251     // Called for each formatted file.
 252     fn handle_formatted_file(
 253         &mut self,
 254         parse_session: &ParseSess,
 255         path: FileName,
 256         result: String,
 257         report: &mut FormatReport,
 258     ) -> Result<(), ErrorKind> {
 259         if let Some(ref mut out) = self.out {
 260             match source_file::write_file(
 261                 Some(parse_session),
 262                 &path,
 263                 &result,
 264                 out,
 265                 &mut *self.emitter,
 266                 self.config.newline_style(),
 267             ) {
 268                 Ok(ref result) if result.has_diff => report.add_diff(),
 269                 Err(e) => {
 270                     // Create a new error with path_str to help users see which files failed
 271                     let err_msg = format!("{}: {}", path, e);
 272                     return Err(io::Error::new(e.kind(), err_msg).into());
 273                 }
 274                 _ => {}
 275             }
 276         }
 277
 278         self.source_file.push((path, result));
 279         Ok(())
 280     }
 281 }
 282
 283 pub(crate) struct FormattingError {
 284     pub(crate) line: usize,
 285     pub(crate) kind: ErrorKind,
 286     is_comment: bool,
 287     is_string: bool,
 288     pub(crate) line_buffer: String,
 289 }
 290
 291 impl FormattingError {
 292     pub(crate) fn from_span(
 293         span: Span,
 294         parse_sess: &ParseSess,
 295         kind: ErrorKind,
 296     ) -> FormattingError {
 297         FormattingError {
 298             line: parse_sess.line_of_byte_pos(span.lo()),
 299             is_comment: kind.is_comment(),
 300             kind,
 301             is_string: false,
 302             line_buffer: parse_sess.span_to_first_line_string(span),
 303         }
 304     }
 305
 306     pub(crate) fn is_internal(&self) -> bool {
 307         match self.kind {
 308             ErrorKind::LineOverflow(..)
 309             | ErrorKind::TrailingWhitespace
 310             | ErrorKind::IoError(_)
 311             | ErrorKind::ParseError
 312             | ErrorKind::LostComment => true,
 313             _ => false,
 314         }
 315     }
 316
 317     pub(crate) fn msg_suffix(&self) -> &str {
 318         if self.is_comment || self.is_string {
 319             "set `error_on_unformatted = false` to suppress \
 320              the warning against comments or string literals\n"
 321         } else {
 322             ""
 323         }
 324     }
 325
 326     // (space, target)
 327     pub(crate) fn format_len(&self) -> (usize, usize) {
 328         match self.kind {
 329             ErrorKind::LineOverflow(found, max) => (max, found - max),
 330             ErrorKind::TrailingWhitespace
 331             | ErrorKind::DeprecatedAttr
 332             | ErrorKind::BadIssue(_)
 333             | ErrorKind::BadAttr
 334             | ErrorKind::LostComment
 335             | ErrorKind::LicenseCheck => {
 336                 let trailing_ws_start = self
 337                     .line_buffer
 338                     .rfind(|c: char| !c.is_whitespace())
 339                     .map(|pos| pos + 1)
 340                     .unwrap_or(0);
 341                 (
 342                     trailing_ws_start,
 343                     self.line_buffer.len() - trailing_ws_start,
 344                 )
 345             }
 346             _ => unreachable!(),
 347         }
 348     }
 349 }
 350
 351 pub(crate) type FormatErrorMap = HashMap<FileName, Vec<FormattingError>>;
 352
 353 #[derive(Default, Debug, PartialEq)]
 354 pub(crate) struct ReportedErrors {
 355     // Encountered e.g., an IO error.
 356     pub(crate) has_operational_errors: bool,
 357
 358     // Failed to reformat code because of parsing errors.
 359     pub(crate) has_parsing_errors: bool,
 360
 361     // Code is valid, but it is impossible to format it properly.
 362     pub(crate) has_formatting_errors: bool,
 363
 364     // Code contains macro call that was unable to format.
 365     pub(crate) has_macro_format_failure: bool,
 366
 367     // Failed a check, such as the license check or other opt-in checking.
 368     pub(crate) has_check_errors: bool,
 369
 370     /// Formatted code differs from existing code (--check only).
 371     pub(crate) has_diff: bool,
 372
 373     /// Formatted code missed something, like lost comments or extra trailing space
 374     pub(crate) has_unformatted_code_errors: bool,
 375 }
 376
 377 impl ReportedErrors {
 378     /// Combine two summaries together.
 379     pub(crate) fn add(&mut self, other: &ReportedErrors) {
 380         self.has_operational_errors |= other.has_operational_errors;
 381         self.has_parsing_errors |= other.has_parsing_errors;
 382         self.has_formatting_errors |= other.has_formatting_errors;
 383         self.has_macro_format_failure |= other.has_macro_format_failure;
 384         self.has_check_errors |= other.has_check_errors;
 385         self.has_diff |= other.has_diff;
 386         self.has_unformatted_code_errors |= other.has_unformatted_code_errors;
 387     }
 388 }
 389
 390 #[derive(Clone, Copy, Debug)]
 391 enum Timer {
 392     Disabled,
 393     Initialized(Instant),
 394     DoneParsing(Instant, Instant),
 395     DoneFormatting(Instant, Instant, Instant),
 396 }
 397
 398 impl Timer {
 399     fn start() -> Timer {
 400         if cfg!(target_arch = "wasm32") {
 401             Timer::Disabled
 402         } else {
 403             Timer::Initialized(Instant::now())
 404         }
 405     }
 406     fn done_parsing(self) -> Self {
 407         match self {
 408             Timer::Disabled => Timer::Disabled,
 409             Timer::Initialized(init_time) => Timer::DoneParsing(init_time, Instant::now()),
 410             _ => panic!("Timer can only transition to DoneParsing from Initialized state"),
 411         }
 412     }
 413
 414     fn done_formatting(self) -> Self {
 415         match self {
 416             Timer::Disabled => Timer::Disabled,
 417             Timer::DoneParsing(init_time, parse_time) => {
 418                 Timer::DoneFormatting(init_time, parse_time, Instant::now())
 419             }
 420             _ => panic!("Timer can only transition to DoneFormatting from DoneParsing state"),
 421         }
 422     }
 423
 424     /// Returns the time it took to parse the source files in seconds.
 425     fn get_parse_time(&self) -> f32 {
 426         match *self {
 427             Timer::Disabled => panic!("this platform cannot time execution"),
 428             Timer::DoneParsing(init, parse_time) | Timer::DoneFormatting(init, parse_time, _) => {
 429                 // This should never underflow since `Instant::now()` guarantees monotonicity.
 430                 Self::duration_to_f32(parse_time.duration_since(init))
 431             }
 432             Timer::Initialized(..) => unreachable!(),
 433         }
 434     }
 435
 436     /// Returns the time it took to go from the parsed AST to the formatted output. Parsing time is
 437     /// not included.
 438     fn get_format_time(&self) -> f32 {
 439         match *self {
 440             Timer::Disabled => panic!("this platform cannot time execution"),
 441             Timer::DoneFormatting(_init, parse_time, format_time) => {
 442                 Self::duration_to_f32(format_time.duration_since(parse_time))
 443             }
 444             Timer::DoneParsing(..) | Timer::Initialized(..) => unreachable!(),
 445         }
 446     }
 447
 448     fn duration_to_f32(d: Duration) -> f32 {
 449         d.as_secs() as f32 + d.subsec_nanos() as f32 / 1_000_000_000f32
 450     }
 451 }
 452
 453 // Formatting done on a char by char or line by line basis.
 454 // FIXME(#20): other stuff for parity with make tidy.
 455 fn format_lines(
 456     text: &mut String,
 457     name: &FileName,
 458     skipped_range: &[(usize, usize)],
 459     config: &Config,
 460     report: &FormatReport,
 461 ) {
 462     let mut formatter = FormatLines::new(name, skipped_range, config);
 463     formatter.check_license(text);
 464     formatter.iterate(text);
 465
 466     if formatter.newline_count > 1 {
 467         debug!("track truncate: {} {}", text.len(), formatter.newline_count);
 468         let line = text.len() - formatter.newline_count + 1;
 469         text.truncate(line);
 470     }
 471
 472     report.append(name.clone(), formatter.errors);
 473 }
 474
 475 struct FormatLines<'a> {
 476     name: &'a FileName,
 477     skipped_range: &'a [(usize, usize)],
 478     last_was_space: bool,
 479     line_len: usize,
 480     cur_line: usize,
 481     newline_count: usize,
 482     errors: Vec<FormattingError>,
 483     issue_seeker: BadIssueSeeker,
 484     line_buffer: String,
 485     current_line_contains_string_literal: bool,
 486     format_line: bool,
 487     allow_issue_seek: bool,
 488     config: &'a Config,
 489 }
 490
 491 impl<'a> FormatLines<'a> {
 492     fn new(
 493         name: &'a FileName,
 494         skipped_range: &'a [(usize, usize)],
 495         config: &'a Config,
 496     ) -> FormatLines<'a> {
 497         let issue_seeker = BadIssueSeeker::new(config.report_todo(), config.report_fixme());
 498         FormatLines {
 499             name,
 500             skipped_range,
 501             last_was_space: false,
 502             line_len: 0,
 503             cur_line: 1,
 504             newline_count: 0,
 505             errors: vec![],
 506             allow_issue_seek: !issue_seeker.is_disabled(),
 507             issue_seeker,
 508             line_buffer: String::with_capacity(config.max_width() * 2),
 509             current_line_contains_string_literal: false,
 510             format_line: config.file_lines().contains_line(name, 1),
 511             config,
 512         }
 513     }
 514
 515     fn check_license(&mut self, text: &mut String) {
 516         if let Some(ref license_template) = self.config.license_template {
 517             if !license_template.is_match(text) {
 518                 self.errors.push(FormattingError {
 519                     line: self.cur_line,
 520                     kind: ErrorKind::LicenseCheck,
 521                     is_comment: false,
 522                     is_string: false,
 523                     line_buffer: String::new(),
 524                 });
 525             }
 526         }
 527     }
 528
 529     // Iterate over the chars in the file map.
 530     fn iterate(&mut self, text: &mut String) {
 531         for (kind, c) in CharClasses::new(text.chars()) {
 532             if c == '\r' {
 533                 continue;
 534             }
 535
 536             if self.allow_issue_seek && self.format_line {
 537                 // Add warnings for bad todos/ fixmes
 538                 if let Some(issue) = self.issue_seeker.inspect(c) {
 539                     self.push_err(ErrorKind::BadIssue(issue), false, false);
 540                 }
 541             }
 542
 543             if c == '\n' {
 544                 self.new_line(kind);
 545             } else {
 546                 self.char(c, kind);
 547             }
 548         }
 549     }
 550
 551     fn new_line(&mut self, kind: FullCodeCharKind) {
 552         if self.format_line {
 553             // Check for (and record) trailing whitespace.
 554             if self.last_was_space {
 555                 if self.should_report_error(kind, &ErrorKind::TrailingWhitespace)
 556                     && !self.is_skipped_line()
 557                 {
 558                     self.push_err(
 559                         ErrorKind::TrailingWhitespace,
 560                         kind.is_comment(),
 561                         kind.is_string(),
 562                     );
 563                 }
 564                 self.line_len -= 1;
 565             }
 566
 567             // Check for any line width errors we couldn't correct.
 568             let error_kind = ErrorKind::LineOverflow(self.line_len, self.config.max_width());
 569             if self.line_len > self.config.max_width()
 570                 && !self.is_skipped_line()
 571                 && self.should_report_error(kind, &error_kind)
 572             {
 573                 let is_string = self.current_line_contains_string_literal;
 574                 self.push_err(error_kind, kind.is_comment(), is_string);
 575             }
 576         }
 577
 578         self.line_len = 0;
 579         self.cur_line += 1;
 580         self.format_line = self
 581             .config
 582             .file_lines()
 583             .contains_line(self.name, self.cur_line);
 584         self.newline_count += 1;
 585         self.last_was_space = false;
 586         self.line_buffer.clear();
 587         self.current_line_contains_string_literal = false;
 588     }
 589
 590     fn char(&mut self, c: char, kind: FullCodeCharKind) {
 591         self.newline_count = 0;
 592         self.line_len += if c == '\t' {
 593             self.config.tab_spaces()
 594         } else {
 595             1
 596         };
 597         self.last_was_space = c.is_whitespace();
 598         self.line_buffer.push(c);
 599         if kind.is_string() {
 600             self.current_line_contains_string_literal = true;
 601         }
 602     }
 603
 604     fn push_err(&mut self, kind: ErrorKind, is_comment: bool, is_string: bool) {
 605         self.errors.push(FormattingError {
 606             line: self.cur_line,
 607             kind,
 608             is_comment,
 609             is_string,
 610             line_buffer: self.line_buffer.clone(),
 611         });
 612     }
 613
 614     fn should_report_error(&self, char_kind: FullCodeCharKind, error_kind: &ErrorKind) -> bool {
 615         let allow_error_report = if char_kind.is_comment()
 616             || self.current_line_contains_string_literal
 617             || error_kind.is_comment()
 618         {
 619             self.config.error_on_unformatted()
 620         } else {
 621             true
 622         };
 623
 624         match error_kind {
 625             ErrorKind::LineOverflow(..) => {
 626                 self.config.error_on_line_overflow() && allow_error_report
 627             }
 628             ErrorKind::TrailingWhitespace | ErrorKind::LostComment => allow_error_report,
 629             _ => true,
 630         }
 631     }
 632
 633     /// Returns `true` if the line with the given line number was skipped by `#[rustfmt::skip]`.
 634     fn is_skipped_line(&self) -> bool {
 635         self.skipped_range
 636             .iter()
 637             .any(|&(lo, hi)| lo <= self.cur_line && self.cur_line <= hi)
 638     }
 639 }
 640
 641 fn should_emit_verbose<F>(forbid_verbose_output: bool, config: &Config, f: F)
 642 where
 643     F: Fn(),
 644 {
 645     if config.verbose() == Verbosity::Verbose && !forbid_verbose_output {
 646         f();
 647     }
 648 }