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