-
-
Notifications
You must be signed in to change notification settings - Fork 2k
/
hiargs.rs
1471 lines (1402 loc) · 55.6 KB
/
hiargs.rs
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
/*!
Provides the definition of high level arguments from CLI flags.
*/
use std::{
collections::HashSet,
path::{Path, PathBuf},
};
use {
bstr::BString,
grep::printer::{ColorSpecs, SummaryKind},
};
use crate::{
flags::lowargs::{
BinaryMode, BoundaryMode, BufferMode, CaseMode, ColorChoice,
ContextMode, ContextSeparator, EncodingMode, EngineChoice,
FieldContextSeparator, FieldMatchSeparator, LowArgs, MmapMode, Mode,
PatternSource, SearchMode, SortMode, SortModeKind, TypeChange,
},
haystack::{Haystack, HaystackBuilder},
search::{PatternMatcher, Printer, SearchWorker, SearchWorkerBuilder},
};
/// A high level representation of CLI arguments.
///
/// The distinction between low and high level arguments is somewhat arbitrary
/// and wishy washy. The main idea here is that high level arguments generally
/// require all of CLI parsing to be finished. For example, one cannot
/// construct a glob matcher until all of the glob patterns are known.
///
/// So while low level arguments are collected during parsing itself, high
/// level arguments aren't created until parsing has completely finished.
#[derive(Debug)]
pub(crate) struct HiArgs {
binary: BinaryDetection,
boundary: Option<BoundaryMode>,
buffer: BufferMode,
byte_offset: bool,
case: CaseMode,
color: ColorChoice,
colors: grep::printer::ColorSpecs,
column: bool,
context: ContextMode,
context_separator: ContextSeparator,
crlf: bool,
dfa_size_limit: Option<usize>,
encoding: EncodingMode,
engine: EngineChoice,
field_context_separator: FieldContextSeparator,
field_match_separator: FieldMatchSeparator,
file_separator: Option<Vec<u8>>,
fixed_strings: bool,
follow: bool,
globs: ignore::overrides::Override,
heading: bool,
hidden: bool,
hyperlink_config: grep::printer::HyperlinkConfig,
ignore_file_case_insensitive: bool,
ignore_file: Vec<PathBuf>,
include_zero: bool,
invert_match: bool,
is_terminal_stdout: bool,
line_number: bool,
max_columns: Option<u64>,
max_columns_preview: bool,
max_count: Option<u64>,
max_depth: Option<usize>,
max_filesize: Option<u64>,
mmap_choice: grep::searcher::MmapChoice,
mode: Mode,
multiline: bool,
multiline_dotall: bool,
no_ignore_dot: bool,
no_ignore_exclude: bool,
no_ignore_files: bool,
no_ignore_global: bool,
no_ignore_parent: bool,
no_ignore_vcs: bool,
no_require_git: bool,
no_unicode: bool,
null_data: bool,
one_file_system: bool,
only_matching: bool,
path_separator: Option<u8>,
paths: Paths,
path_terminator: Option<u8>,
patterns: Patterns,
pre: Option<PathBuf>,
pre_globs: ignore::overrides::Override,
quiet: bool,
quit_after_match: bool,
regex_size_limit: Option<usize>,
replace: Option<BString>,
search_zip: bool,
sort: Option<SortMode>,
stats: Option<grep::printer::Stats>,
stop_on_nonmatch: bool,
threads: usize,
trim: bool,
types: ignore::types::Types,
vimgrep: bool,
with_filename: bool,
}
impl HiArgs {
/// Convert low level arguments into high level arguments.
///
/// This process can fail for a variety of reasons. For example, invalid
/// globs or some kind of environment issue.
pub(crate) fn from_low_args(mut low: LowArgs) -> anyhow::Result<HiArgs> {
// Callers should not be trying to convert low-level arguments when
// a short-circuiting special mode is present.
assert_eq!(None, low.special, "special mode demands short-circuiting");
// If the sorting mode isn't supported, then we bail loudly. I'm not
// sure if this is the right thing to do. We could silently "not sort"
// as well. If we wanted to go that route, then we could just set
// `low.sort = None` if `supported()` returns an error.
if let Some(ref sort) = low.sort {
sort.supported()?;
}
// We modify the mode in-place on `low` so that subsequent conversions
// see the correct mode.
match low.mode {
Mode::Search(ref mut mode) => match *mode {
// treat `-v --count-matches` as `-v --count`
SearchMode::CountMatches if low.invert_match => {
*mode = SearchMode::Count;
}
// treat `-o --count` as `--count-matches`
SearchMode::Count if low.only_matching => {
*mode = SearchMode::CountMatches;
}
_ => {}
},
_ => {}
}
let mut state = State::new()?;
let patterns = Patterns::from_low_args(&mut state, &mut low)?;
let paths = Paths::from_low_args(&mut state, &patterns, &mut low)?;
let binary = BinaryDetection::from_low_args(&state, &low);
let colors = take_color_specs(&mut state, &mut low);
let hyperlink_config = take_hyperlink_config(&mut state, &mut low)?;
let stats = stats(&low);
let types = types(&low)?;
let globs = globs(&state, &low)?;
let pre_globs = preprocessor_globs(&state, &low)?;
let color = match low.color {
ColorChoice::Auto if !state.is_terminal_stdout => {
ColorChoice::Never
}
_ => low.color,
};
let column = low.column.unwrap_or(low.vimgrep);
let heading = match low.heading {
None => !low.vimgrep && state.is_terminal_stdout,
Some(false) => false,
Some(true) => !low.vimgrep,
};
let path_terminator = if low.null { Some(b'\x00') } else { None };
let quit_after_match = stats.is_none() && low.quiet;
let threads = if low.sort.is_some() || paths.is_one_file {
1
} else if let Some(threads) = low.threads {
threads
} else {
std::thread::available_parallelism().map_or(1, |n| n.get()).min(12)
};
log::debug!("using {threads} thread(s)");
let with_filename = low
.with_filename
.unwrap_or_else(|| low.vimgrep || !paths.is_one_file);
let file_separator = match low.mode {
Mode::Search(SearchMode::Standard) => {
if heading {
Some(b"".to_vec())
} else if let ContextMode::Limited(ref limited) = low.context {
let (before, after) = limited.get();
if before > 0 || after > 0 {
low.context_separator.clone().into_bytes()
} else {
None
}
} else {
None
}
}
_ => None,
};
let line_number = low.line_number.unwrap_or_else(|| {
if low.quiet {
return false;
}
let Mode::Search(ref search_mode) = low.mode else { return false };
match *search_mode {
SearchMode::FilesWithMatches
| SearchMode::FilesWithoutMatch
| SearchMode::Count
| SearchMode::CountMatches => return false,
SearchMode::JSON => return true,
SearchMode::Standard => {
// A few things can imply counting line numbers. In
// particular, we generally want to show line numbers by
// default when printing to a tty for human consumption,
// except for one interesting case: when we're only
// searching stdin. This makes pipelines work as expected.
(state.is_terminal_stdout && !paths.is_only_stdin())
|| column
|| low.vimgrep
}
}
});
let mmap_choice = {
// SAFETY: Memory maps are difficult to impossible to encapsulate
// safely in a portable way that doesn't simultaneously negate some
// of the benfits of using memory maps. For ripgrep's use, we never
// mutate a memory map and generally never store the contents of
// memory map in a data structure that depends on immutability.
// Generally speaking, the worst thing that can happen is a SIGBUS
// (if the underlying file is truncated while reading it), which
// will cause ripgrep to abort. This reasoning should be treated as
// suspect.
let maybe = unsafe { grep::searcher::MmapChoice::auto() };
let never = grep::searcher::MmapChoice::never();
match low.mmap {
MmapMode::Auto => {
if paths.paths.len() <= 10
&& paths.paths.iter().all(|p| p.is_file())
{
// If we're only searching a few paths and all of them
// are files, then memory maps are probably faster.
maybe
} else {
never
}
}
MmapMode::AlwaysTryMmap => maybe,
MmapMode::Never => never,
}
};
Ok(HiArgs {
mode: low.mode,
patterns,
paths,
binary,
boundary: low.boundary,
buffer: low.buffer,
byte_offset: low.byte_offset,
case: low.case,
color,
colors,
column,
context: low.context,
context_separator: low.context_separator,
crlf: low.crlf,
dfa_size_limit: low.dfa_size_limit,
encoding: low.encoding,
engine: low.engine,
field_context_separator: low.field_context_separator,
field_match_separator: low.field_match_separator,
file_separator,
fixed_strings: low.fixed_strings,
follow: low.follow,
heading,
hidden: low.hidden,
hyperlink_config,
ignore_file: low.ignore_file,
ignore_file_case_insensitive: low.ignore_file_case_insensitive,
include_zero: low.include_zero,
invert_match: low.invert_match,
is_terminal_stdout: state.is_terminal_stdout,
line_number,
max_columns: low.max_columns,
max_columns_preview: low.max_columns_preview,
max_count: low.max_count,
max_depth: low.max_depth,
max_filesize: low.max_filesize,
mmap_choice,
multiline: low.multiline,
multiline_dotall: low.multiline_dotall,
no_ignore_dot: low.no_ignore_dot,
no_ignore_exclude: low.no_ignore_exclude,
no_ignore_files: low.no_ignore_files,
no_ignore_global: low.no_ignore_global,
no_ignore_parent: low.no_ignore_parent,
no_ignore_vcs: low.no_ignore_vcs,
no_require_git: low.no_require_git,
no_unicode: low.no_unicode,
null_data: low.null_data,
one_file_system: low.one_file_system,
only_matching: low.only_matching,
globs,
path_separator: low.path_separator,
path_terminator,
pre: low.pre,
pre_globs,
quiet: low.quiet,
quit_after_match,
regex_size_limit: low.regex_size_limit,
replace: low.replace,
search_zip: low.search_zip,
sort: low.sort,
stats,
stop_on_nonmatch: low.stop_on_nonmatch,
threads,
trim: low.trim,
types,
vimgrep: low.vimgrep,
with_filename,
})
}
/// Returns a writer for printing buffers to stdout.
///
/// This is intended to be used from multiple threads. Namely, a buffer
/// writer can create new buffers that are sent to threads. Threads can
/// then independently write to the buffers. Once a unit of work is
/// complete, a buffer can be given to the buffer writer to write to
/// stdout.
pub(crate) fn buffer_writer(&self) -> termcolor::BufferWriter {
let mut wtr =
termcolor::BufferWriter::stdout(self.color.to_termcolor());
wtr.separator(self.file_separator.clone());
wtr
}
/// Returns true when ripgrep had to guess to search the current working
/// directory. That is, it's true when ripgrep is called without any file
/// paths or directories to search.
///
/// Other than changing how file paths are printed (i.e., without the
/// leading `./`), it's also useful to know for diagnostic reasons. For
/// example, ripgrep will print an error message when nothing is searched
/// since it's possible the ignore rules in play are too aggressive. But
/// this warning is only emitted when ripgrep was called without any
/// explicit file paths since otherwise the warning would likely be too
/// aggressive.
pub(crate) fn has_implicit_path(&self) -> bool {
self.paths.has_implicit_path
}
/// Return a properly configured builder for constructing haystacks.
///
/// The builder can be used to turn a directory entry (from the `ignore`
/// crate) into something that can be searched.
pub(crate) fn haystack_builder(&self) -> HaystackBuilder {
let mut builder = HaystackBuilder::new();
builder.strip_dot_prefix(self.paths.has_implicit_path);
builder
}
/// Return the matcher that should be used for searching using the engine
/// choice made by the user.
///
/// If there was a problem building the matcher (e.g., a syntax error),
/// then this returns an error.
pub(crate) fn matcher(&self) -> anyhow::Result<PatternMatcher> {
match self.engine {
EngineChoice::Default => match self.matcher_rust() {
Ok(m) => Ok(m),
Err(err) => {
anyhow::bail!(suggest_other_engine(err.to_string()));
}
},
EngineChoice::PCRE2 => Ok(self.matcher_pcre2()?),
EngineChoice::Auto => {
let rust_err = match self.matcher_rust() {
Ok(m) => return Ok(m),
Err(err) => err,
};
log::debug!(
"error building Rust regex in hybrid mode:\n{rust_err}",
);
let pcre_err = match self.matcher_pcre2() {
Ok(m) => return Ok(m),
Err(err) => err,
};
let divider = "~".repeat(79);
anyhow::bail!(
"regex could not be compiled with either the default \
regex engine or with PCRE2.\n\n\
default regex engine error:\n\
{divider}\n\
{rust_err}\n\
{divider}\n\n\
PCRE2 regex engine error:\n{pcre_err}",
);
}
}
}
/// Build a matcher using PCRE2.
///
/// If there was a problem building the matcher (such as a regex syntax
/// error), then an error is returned.
///
/// If the `pcre2` feature is not enabled then this always returns an
/// error.
fn matcher_pcre2(&self) -> anyhow::Result<PatternMatcher> {
#[cfg(feature = "pcre2")]
{
let mut builder = grep::pcre2::RegexMatcherBuilder::new();
builder.multi_line(true).fixed_strings(self.fixed_strings);
match self.case {
CaseMode::Sensitive => builder.caseless(false),
CaseMode::Insensitive => builder.caseless(true),
CaseMode::Smart => builder.case_smart(true),
};
if let Some(ref boundary) = self.boundary {
match *boundary {
BoundaryMode::Line => builder.whole_line(true),
BoundaryMode::Word => builder.word(true),
};
}
// For whatever reason, the JIT craps out during regex compilation with
// a "no more memory" error on 32 bit systems. So don't use it there.
if cfg!(target_pointer_width = "64") {
builder
.jit_if_available(true)
// The PCRE2 docs say that 32KB is the default, and that 1MB
// should be big enough for anything. But let's crank it to
// 10MB.
.max_jit_stack_size(Some(10 * (1 << 20)));
}
if !self.no_unicode {
builder.utf(true).ucp(true);
}
if self.multiline {
builder.dotall(self.multiline_dotall);
}
if self.crlf {
builder.crlf(true);
}
let m = builder.build_many(&self.patterns.patterns)?;
Ok(PatternMatcher::PCRE2(m))
}
#[cfg(not(feature = "pcre2"))]
{
Err(anyhow::anyhow!(
"PCRE2 is not available in this build of ripgrep"
))
}
}
/// Build a matcher using Rust's regex engine.
///
/// If there was a problem building the matcher (such as a regex syntax
/// error), then an error is returned.
fn matcher_rust(&self) -> anyhow::Result<PatternMatcher> {
let mut builder = grep::regex::RegexMatcherBuilder::new();
builder
.multi_line(true)
.unicode(!self.no_unicode)
.octal(false)
.fixed_strings(self.fixed_strings);
match self.case {
CaseMode::Sensitive => builder.case_insensitive(false),
CaseMode::Insensitive => builder.case_insensitive(true),
CaseMode::Smart => builder.case_smart(true),
};
if let Some(ref boundary) = self.boundary {
match *boundary {
BoundaryMode::Line => builder.whole_line(true),
BoundaryMode::Word => builder.word(true),
};
}
if self.multiline {
builder.dot_matches_new_line(self.multiline_dotall);
if self.crlf {
builder.crlf(true).line_terminator(None);
}
} else {
builder.line_terminator(Some(b'\n')).dot_matches_new_line(false);
if self.crlf {
builder.crlf(true);
}
// We don't need to set this in multiline mode since multiline
// matchers don't use optimizations related to line terminators.
// Moreover, a multiline regex used with --null-data should
// be allowed to match NUL bytes explicitly, which this would
// otherwise forbid.
if self.null_data {
builder.line_terminator(Some(b'\x00'));
}
}
if let Some(limit) = self.regex_size_limit {
builder.size_limit(limit);
}
if let Some(limit) = self.dfa_size_limit {
builder.dfa_size_limit(limit);
}
if !self.binary.is_none() {
builder.ban_byte(Some(b'\x00'));
}
let m = match builder.build_many(&self.patterns.patterns) {
Ok(m) => m,
Err(err) => {
anyhow::bail!(suggest_text(suggest_multiline(err.to_string())))
}
};
Ok(PatternMatcher::RustRegex(m))
}
/// Returns true if some non-zero number of matches is believed to be
/// possible.
///
/// When this returns false, it is impossible for ripgrep to ever report
/// a match.
pub(crate) fn matches_possible(&self) -> bool {
if self.patterns.patterns.is_empty() {
return false;
}
if self.max_count == Some(0) {
return false;
}
true
}
/// Returns the "mode" that ripgrep should operate in.
///
/// This is generally useful for determining what action ripgrep should
/// take. The main mode is of course to "search," but there are other
/// non-search modes such as `--type-list` and `--files`.
pub(crate) fn mode(&self) -> Mode {
self.mode
}
/// Returns a builder for constructing a "path printer."
///
/// This is useful for the `--files` mode in ripgrep, where the printer
/// just needs to emit paths and not need to worry about the functionality
/// of searching.
pub(crate) fn path_printer_builder(
&self,
) -> grep::printer::PathPrinterBuilder {
let mut builder = grep::printer::PathPrinterBuilder::new();
builder
.color_specs(self.colors.clone())
.hyperlink(self.hyperlink_config.clone())
.separator(self.path_separator.clone())
.terminator(self.path_terminator.unwrap_or(b'\n'));
builder
}
/// Returns a printer for the given search mode.
///
/// This chooses which printer to build (JSON, summary or standard) based
/// on the search mode given.
pub(crate) fn printer<W: termcolor::WriteColor>(
&self,
search_mode: SearchMode,
wtr: W,
) -> Printer<W> {
let summary_kind = if self.quiet {
SummaryKind::Quiet
} else {
match search_mode {
SearchMode::FilesWithMatches => SummaryKind::PathWithMatch,
SearchMode::FilesWithoutMatch => SummaryKind::PathWithoutMatch,
SearchMode::Count => SummaryKind::Count,
SearchMode::CountMatches => SummaryKind::CountMatches,
SearchMode::JSON => {
return Printer::JSON(self.printer_json(wtr))
}
SearchMode::Standard => {
return Printer::Standard(self.printer_standard(wtr))
}
}
};
Printer::Summary(self.printer_summary(wtr, summary_kind))
}
/// Builds a JSON printer.
fn printer_json<W: std::io::Write>(
&self,
wtr: W,
) -> grep::printer::JSON<W> {
grep::printer::JSONBuilder::new()
.pretty(false)
.max_matches(self.max_count)
.always_begin_end(false)
.build(wtr)
}
/// Builds a "standard" grep printer where matches are printed as plain
/// text lines.
fn printer_standard<W: termcolor::WriteColor>(
&self,
wtr: W,
) -> grep::printer::Standard<W> {
let mut builder = grep::printer::StandardBuilder::new();
builder
.byte_offset(self.byte_offset)
.color_specs(self.colors.clone())
.column(self.column)
.heading(self.heading)
.hyperlink(self.hyperlink_config.clone())
.max_columns_preview(self.max_columns_preview)
.max_columns(self.max_columns)
.max_matches(self.max_count)
.only_matching(self.only_matching)
.path(self.with_filename)
.path_terminator(self.path_terminator.clone())
.per_match_one_line(true)
.per_match(self.vimgrep)
.replacement(self.replace.clone().map(|r| r.into()))
.separator_context(self.context_separator.clone().into_bytes())
.separator_field_context(
self.field_context_separator.clone().into_bytes(),
)
.separator_field_match(
self.field_match_separator.clone().into_bytes(),
)
.separator_path(self.path_separator.clone())
.stats(self.stats.is_some())
.trim_ascii(self.trim);
// When doing multi-threaded searching, the buffer writer is
// responsible for writing separators since it is the only thing that
// knows whether something has been printed or not. But for the single
// threaded case, we don't use a buffer writer and thus can let the
// printer own this.
if self.threads == 1 {
builder.separator_search(self.file_separator.clone());
}
builder.build(wtr)
}
/// Builds a "summary" printer where search results are aggregated on a
/// file-by-file basis.
fn printer_summary<W: termcolor::WriteColor>(
&self,
wtr: W,
kind: SummaryKind,
) -> grep::printer::Summary<W> {
grep::printer::SummaryBuilder::new()
.color_specs(self.colors.clone())
.exclude_zero(!self.include_zero)
.hyperlink(self.hyperlink_config.clone())
.kind(kind)
.max_matches(self.max_count)
.path(self.with_filename)
.path_terminator(self.path_terminator.clone())
.separator_field(b":".to_vec())
.separator_path(self.path_separator.clone())
.stats(self.stats.is_some())
.build(wtr)
}
/// Returns true if ripgrep should operate in "quiet" mode.
///
/// Generally speaking, quiet mode means that ripgrep should not print
/// anything to stdout. There are some exceptions. For example, when the
/// user has provided `--stats`, then ripgrep will print statistics to
/// stdout.
pub(crate) fn quiet(&self) -> bool {
self.quiet
}
/// Returns true when ripgrep should stop searching after a single match is
/// found.
///
/// This is useful for example when quiet mode is enabled. In that case,
/// users generally can't tell the difference in behavior between a search
/// that finds all matches and a search that only finds one of them. (An
/// exception here is if `--stats` is given, then `quit_after_match` will
/// always return false since the user expects ripgrep to find everything.)
pub(crate) fn quit_after_match(&self) -> bool {
self.quit_after_match
}
/// Build a worker for executing searches.
///
/// Search results are found using the given matcher and written to the
/// given printer.
pub(crate) fn search_worker<W: termcolor::WriteColor>(
&self,
matcher: PatternMatcher,
searcher: grep::searcher::Searcher,
printer: Printer<W>,
) -> anyhow::Result<SearchWorker<W>> {
let mut builder = SearchWorkerBuilder::new();
builder
.preprocessor(self.pre.clone())?
.preprocessor_globs(self.pre_globs.clone())
.search_zip(self.search_zip)
.binary_detection_explicit(self.binary.explicit.clone())
.binary_detection_implicit(self.binary.implicit.clone());
Ok(builder.build(matcher, searcher, printer))
}
/// Build a searcher from the command line parameters.
pub(crate) fn searcher(&self) -> anyhow::Result<grep::searcher::Searcher> {
let line_term = if self.crlf {
grep::matcher::LineTerminator::crlf()
} else if self.null_data {
grep::matcher::LineTerminator::byte(b'\x00')
} else {
grep::matcher::LineTerminator::byte(b'\n')
};
let mut builder = grep::searcher::SearcherBuilder::new();
builder
.line_terminator(line_term)
.invert_match(self.invert_match)
.line_number(self.line_number)
.multi_line(self.multiline)
.memory_map(self.mmap_choice.clone())
.stop_on_nonmatch(self.stop_on_nonmatch);
match self.context {
ContextMode::Passthru => {
builder.passthru(true);
}
ContextMode::Limited(ref limited) => {
let (before, after) = limited.get();
builder.before_context(before);
builder.after_context(after);
}
}
match self.encoding {
EncodingMode::Auto => {} // default for the searcher
EncodingMode::Some(ref enc) => {
builder.encoding(Some(enc.clone()));
}
EncodingMode::Disabled => {
builder.bom_sniffing(false);
}
}
Ok(builder.build())
}
/// Given an iterator of haystacks, sort them if necessary.
///
/// When sorting is necessary, this will collect the entire iterator into
/// memory, sort them and then return a new iterator. When sorting is not
/// necessary, then the iterator given is returned as is without collecting
/// it into memory.
///
/// Once special case is when sorting by path in ascending order has been
/// requested. In this case, the iterator given is returned as is without
/// any additional sorting. This is done because `walk_builder()` will sort
/// the iterator it yields during directory traversal, so no additional
/// sorting is needed.
pub(crate) fn sort<'a, I>(
&self,
haystacks: I,
) -> Box<dyn Iterator<Item = Haystack> + 'a>
where
I: Iterator<Item = Haystack> + 'a,
{
use std::{cmp::Ordering, fs::Metadata, io, time::SystemTime};
fn attach_timestamps(
haystacks: impl Iterator<Item = Haystack>,
get: impl Fn(&Metadata) -> io::Result<SystemTime>,
) -> impl Iterator<Item = (Haystack, Option<SystemTime>)> {
haystacks.map(move |s| {
let time = s.path().metadata().and_then(|m| get(&m)).ok();
(s, time)
})
}
let Some(ref sort) = self.sort else { return Box::new(haystacks) };
let mut with_timestamps: Vec<_> = match sort.kind {
SortModeKind::Path if !sort.reverse => return Box::new(haystacks),
SortModeKind::Path => {
let mut haystacks = haystacks.collect::<Vec<Haystack>>();
haystacks.sort_by(|ref h1, ref h2| {
h1.path().cmp(h2.path()).reverse()
});
return Box::new(haystacks.into_iter());
}
SortModeKind::LastModified => {
attach_timestamps(haystacks, |md| md.modified()).collect()
}
SortModeKind::LastAccessed => {
attach_timestamps(haystacks, |md| md.accessed()).collect()
}
SortModeKind::Created => {
attach_timestamps(haystacks, |md| md.created()).collect()
}
};
with_timestamps.sort_by(|(_, ref t1), (_, ref t2)| {
let ordering = match (*t1, *t2) {
// Both have metadata, do the obvious thing.
(Some(t1), Some(t2)) => t1.cmp(&t2),
// Things that error should appear later (when ascending).
(Some(_), None) => Ordering::Less,
// Things that error should appear later (when ascending).
(None, Some(_)) => Ordering::Greater,
// When both error, we can't distinguish, so treat as equal.
(None, None) => Ordering::Equal,
};
if sort.reverse {
ordering.reverse()
} else {
ordering
}
});
Box::new(with_timestamps.into_iter().map(|(s, _)| s))
}
/// Returns a stats object if the user requested that ripgrep keep track
/// of various metrics during a search.
///
/// When this returns `None`, then callers may assume that the user did
/// not request statistics.
pub(crate) fn stats(&self) -> Option<grep::printer::Stats> {
self.stats.clone()
}
/// Returns a color-enabled writer for stdout.
///
/// The writer returned is also configured to do either line or block
/// buffering, based on either explicit configuration from the user via CLI
/// flags, or automatically based on whether stdout is connected to a tty.
pub(crate) fn stdout(&self) -> grep::cli::StandardStream {
let color = self.color.to_termcolor();
match self.buffer {
BufferMode::Auto => {
if self.is_terminal_stdout {
grep::cli::stdout_buffered_line(color)
} else {
grep::cli::stdout_buffered_block(color)
}
}
BufferMode::Line => grep::cli::stdout_buffered_line(color),
BufferMode::Block => grep::cli::stdout_buffered_block(color),
}
}
/// Returns the total number of threads ripgrep should use to execute a
/// search.
///
/// This number is the result of reasoning about both heuristics (like
/// the available number of cores) and whether ripgrep's mode supports
/// parallelism. It is intended that this number be used to directly
/// determine how many threads to spawn.
pub(crate) fn threads(&self) -> usize {
self.threads
}
/// Returns the file type matcher that was built.
///
/// The matcher includes both the default rules and any rules added by the
/// user for this specific invocation.
pub(crate) fn types(&self) -> &ignore::types::Types {
&self.types
}
/// Create a new builder for recursive directory traversal.
///
/// The builder returned can be used to start a single threaded or multi
/// threaded directory traversal. For multi threaded traversal, the number
/// of threads configured is equivalent to `HiArgs::threads`.
///
/// If `HiArgs::threads` is equal to `1`, then callers should generally
/// choose to explicitly use single threaded traversal since it won't have
/// the unnecessary overhead of synchronization.
pub(crate) fn walk_builder(&self) -> anyhow::Result<ignore::WalkBuilder> {
let mut builder = ignore::WalkBuilder::new(&self.paths.paths[0]);
for path in self.paths.paths.iter().skip(1) {
builder.add(path);
}
if !self.no_ignore_files {
for path in self.ignore_file.iter() {
if let Some(err) = builder.add_ignore(path) {
ignore_message!("{err}");
}
}
}
builder
.max_depth(self.max_depth)
.follow_links(self.follow)
.max_filesize(self.max_filesize)
.threads(self.threads)
.same_file_system(self.one_file_system)
.skip_stdout(matches!(self.mode, Mode::Search(_)))
.overrides(self.globs.clone())
.types(self.types.clone())
.hidden(!self.hidden)
.parents(!self.no_ignore_parent)
.ignore(!self.no_ignore_dot)
.git_global(!self.no_ignore_vcs && !self.no_ignore_global)
.git_ignore(!self.no_ignore_vcs)
.git_exclude(!self.no_ignore_vcs && !self.no_ignore_exclude)
.require_git(!self.no_require_git)
.ignore_case_insensitive(self.ignore_file_case_insensitive);
if !self.no_ignore_dot {
builder.add_custom_ignore_filename(".rgignore");
}
// When we want to sort paths lexicographically in ascending order,
// then we can actually do this during directory traversal itself.
// Otherwise, sorting is done by collecting all paths, sorting them and
// then searching them.
if let Some(ref sort) = self.sort {
assert_eq!(1, self.threads, "sorting implies single threaded");
if !sort.reverse && matches!(sort.kind, SortModeKind::Path) {
builder.sort_by_file_name(|a, b| a.cmp(b));
}
}
Ok(builder)
}
}
/// State that only needs to be computed once during argument parsing.
///
/// This state is meant to be somewhat generic and shared across multiple
/// low->high argument conversions. The state can even be mutated by various
/// conversions as a way to communicate changes to other conversions. For
/// example, reading patterns might consume from stdin. If we know stdin
/// has been consumed and no other file paths have been given, then we know
/// for sure that we should search the CWD. In this way, a state change
/// when reading the patterns can impact how the file paths are ultimately
/// generated.
#[derive(Debug)]
struct State {
/// Whether it's believed that tty is connected to stdout. Note that on
/// unix systems, this is always correct. On Windows, heuristics are used
/// by Rust's standard library, particularly for cygwin/MSYS environments.
is_terminal_stdout: bool,
/// Whether stdin has already been consumed. This is useful to know and for
/// providing good error messages when the user has tried to read from stdin
/// in two different places. For example, `rg -f - -`.
stdin_consumed: bool,
/// The current working directory.
cwd: PathBuf,
}
impl State {
/// Initialize state to some sensible defaults.
///
/// Note that the state values may change throughout the lifetime of
/// argument parsing.
fn new() -> anyhow::Result<State> {
use std::io::IsTerminal;
Ok(State {
is_terminal_stdout: std::io::stdout().is_terminal(),
stdin_consumed: false,
cwd: current_dir()?,
})
}
}
/// The disjunction of patterns to search for.
///
/// The number of patterns can be empty, e.g., via `-f /dev/null`.
#[derive(Debug)]
struct Patterns {
/// The actual patterns to match.
patterns: Vec<String>,
}
impl Patterns {
/// Pulls the patterns out of the low arguments.
///
/// This includes collecting patterns from -e/--regexp and -f/--file.
///
/// If the invocation implies that the first positional argument is a
/// pattern (the common case), then the first positional argument is
/// extracted as well.
fn from_low_args(
state: &mut State,
low: &mut LowArgs,
) -> anyhow::Result<Patterns> {
// The first positional is only a pattern when ripgrep is instructed to
// search and neither -e/--regexp nor -f/--file is given. Basically,
// the first positional is a pattern only when a pattern hasn't been
// given in some other way.
// No search means no patterns. Even if -e/--regexp or -f/--file is
// given, we know we won't use them so don't bother collecting them.
if !matches!(low.mode, Mode::Search(_)) {
return Ok(Patterns { patterns: vec![] });
}
// If we got nothing from -e/--regexp and -f/--file, then the first
// positional is a pattern.
if low.patterns.is_empty() {
anyhow::ensure!(
!low.positional.is_empty(),
"ripgrep requires at least one pattern to execute a search"
);
let ospat = low.positional.remove(0);
let Ok(pat) = ospat.into_string() else {
anyhow::bail!("pattern given is not valid UTF-8")
};
return Ok(Patterns { patterns: vec![pat] });
}
// Otherwise, we need to slurp up our patterns from -e/--regexp and
// -f/--file. We de-duplicate as we go. If we don't de-duplicate,
// then it can actually lead to major slow downs for sloppy inputs.