yerba 0.8.0 → 0.9.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
data/rust/src/ffi.rs CHANGED
@@ -485,25 +485,34 @@ pub unsafe extern "C" fn yerba_document_find(document: *const Document, path: *c
485
485
  CString::new(json).unwrap_or_default().into_raw()
486
486
  }
487
487
 
488
+ unsafe fn borrow_value<'a>(value: *const c_char, length: usize) -> &'a str {
489
+ if value.is_null() {
490
+ return "";
491
+ }
492
+
493
+ std::str::from_utf8(std::slice::from_raw_parts(value as *const u8, length)).unwrap_or("")
494
+ }
495
+
488
496
  #[no_mangle]
489
497
  pub unsafe extern "C" fn yerba_document_set(
490
498
  document: *mut Document,
491
499
  path: *const c_char,
492
500
  value: *const c_char,
501
+ value_length: usize,
493
502
  value_type: YerbaValueType,
503
+ plain: bool,
494
504
  all: bool,
495
505
  ) -> YerbaResult {
496
506
  let document = &mut *document;
497
507
  let selector_string = CStr::from_ptr(path).to_str().unwrap_or("");
498
- let value_string = CStr::from_ptr(value).to_str().unwrap_or("");
499
-
500
- let result = if all {
501
- document.set_all(selector_string, value_string)
502
- } else {
503
- match value_type {
504
- YerbaValueType::String => document.set(selector_string, value_string),
505
- _ => document.set_plain(selector_string, value_string),
506
- }
508
+ let value_string = borrow_value(value, value_length);
509
+ let verbatim = plain || value_type != YerbaValueType::String;
510
+
511
+ let result = match (all, verbatim) {
512
+ (true, false) => document.set_all(selector_string, value_string),
513
+ (true, true) => document.set_all_plain(selector_string, value_string),
514
+ (false, false) => document.set(selector_string, value_string),
515
+ (false, true) => document.set_plain(selector_string, value_string),
507
516
  };
508
517
 
509
518
  match result {
@@ -517,17 +526,25 @@ pub unsafe extern "C" fn yerba_document_insert(
517
526
  document: *mut Document,
518
527
  path: *const c_char,
519
528
  value: *const c_char,
529
+ value_length: usize,
520
530
  value_type: YerbaValueType,
531
+ plain: bool,
532
+ style: *const c_char,
521
533
  before: *const c_char,
522
534
  after: *const c_char,
523
535
  at: i64,
524
536
  ) -> YerbaResult {
525
537
  let document = &mut *document;
526
538
  let selector_string = CStr::from_ptr(path).to_str().unwrap_or("");
527
- let raw_value = CStr::from_ptr(value).to_str().unwrap_or("");
539
+ let raw_value = borrow_value(value, value_length);
528
540
 
529
541
  let value_string = match value_type {
530
- YerbaValueType::String => crate::syntax::quote_if_needed(raw_value),
542
+ YerbaValueType::String if !plain => {
543
+ let style = if style.is_null() { None } else { CStr::from_ptr(style).to_str().ok() };
544
+
545
+ crate::syntax::quote_scalar_styled(raw_value, style)
546
+ }
547
+
531
548
  _ => raw_value.to_string(),
532
549
  };
533
550
 
@@ -543,7 +560,13 @@ pub unsafe extern "C" fn yerba_document_insert(
543
560
  InsertPosition::Last
544
561
  };
545
562
 
546
- match document.insert_into(selector_string, &value_string, position) {
563
+ let result = if plain {
564
+ document.insert_fragment_into(selector_string, &value_string, position)
565
+ } else {
566
+ document.insert_into(selector_string, &value_string, position)
567
+ };
568
+
569
+ match result {
547
570
  Ok(()) => YerbaResult::ok(),
548
571
  Err(e) => YerbaResult::err(&e.to_string()),
549
572
  }
@@ -746,6 +769,7 @@ pub unsafe extern "C" fn yerba_document_blank_lines(document: *mut Document, pat
746
769
  }
747
770
 
748
771
  /// Caller must free with yerba_string_free.
772
+ #[cfg(feature = "schema")]
749
773
  #[no_mangle]
750
774
  pub unsafe extern "C" fn yerba_document_validate_schema(document: *const Document, schema_json: *const c_char, selector: *const c_char) -> *mut c_char {
751
775
  let document = &*document;
@@ -784,6 +808,7 @@ pub unsafe extern "C" fn yerba_document_validate_schema(document: *const Documen
784
808
  }
785
809
 
786
810
  /// Caller must free with yerba_string_free.
811
+ #[cfg(feature = "cli")]
787
812
  #[no_mangle]
788
813
  pub unsafe extern "C" fn yerba_yerbafile_find(directory: *const c_char) -> *mut c_char {
789
814
  let start = if directory.is_null() {
@@ -803,6 +828,7 @@ pub unsafe extern "C" fn yerba_yerbafile_find(directory: *const c_char) -> *mut
803
828
  }
804
829
  }
805
830
 
831
+ #[cfg(feature = "cli")]
806
832
  #[no_mangle]
807
833
  pub unsafe extern "C" fn yerba_document_apply_yerbafile(document: *mut Document, file_path: *const c_char, yerbafile_path: *const c_char) -> YerbaResult {
808
834
  let document = &mut *document;
@@ -840,6 +866,21 @@ pub unsafe extern "C" fn yerba_document_to_string(document: *const Document) ->
840
866
  CString::new(content).unwrap_or_default().into_raw()
841
867
  }
842
868
 
869
+ #[no_mangle]
870
+ pub unsafe extern "C" fn yerba_quote_scalar(value: *const c_char, length: usize, style: *const c_char) -> *mut c_char {
871
+ let bytes = std::slice::from_raw_parts(value as *const u8, length);
872
+
873
+ let Ok(value) = std::str::from_utf8(bytes) else {
874
+ return std::ptr::null_mut();
875
+ };
876
+
877
+ let style = if style.is_null() { None } else { CStr::from_ptr(style).to_str().ok() };
878
+
879
+ CString::new(crate::syntax::quote_scalar_styled(value, style))
880
+ .map(|string| string.into_raw())
881
+ .unwrap_or(std::ptr::null_mut())
882
+ }
883
+
843
884
  #[no_mangle]
844
885
  pub unsafe extern "C" fn yerba_string_free(s: *mut c_char) {
845
886
  if !s.is_null() {
@@ -924,6 +965,7 @@ fn located_nodes_to_list(nodes: &[crate::LocatedNode]) -> YerbaTypedList {
924
965
  }
925
966
  }
926
967
 
968
+ #[cfg(feature = "glob")]
927
969
  #[no_mangle]
928
970
  pub unsafe extern "C" fn yerba_glob_get(glob_pattern: *const c_char, path: *const c_char) -> YerbaTypedList {
929
971
  let pattern = CStr::from_ptr(glob_pattern).to_str().unwrap_or("");
@@ -940,6 +982,7 @@ pub unsafe extern "C" fn yerba_document_get_all(document: *const Document, path:
940
982
  located_nodes_to_list(&document.get_all_located(selector_string))
941
983
  }
942
984
 
985
+ #[cfg(feature = "glob")]
943
986
  #[no_mangle]
944
987
  pub unsafe extern "C" fn yerba_glob_find(glob_pattern: *const c_char, path: *const c_char, condition: *const c_char, select: *const c_char) -> YerbaTypedList {
945
988
  let pattern = CStr::from_ptr(glob_pattern).to_str().unwrap_or("");
data/rust/src/lib.rs CHANGED
@@ -4,21 +4,29 @@ pub mod error;
4
4
  pub mod ffi;
5
5
  pub mod json;
6
6
  mod quote_style;
7
+ #[cfg(feature = "schema")]
7
8
  pub mod schema;
8
9
  pub mod selector;
9
10
  mod syntax;
11
+ mod validation;
10
12
  mod yaml_writer;
13
+ #[cfg(feature = "cli")]
11
14
  pub mod yerbafile;
12
15
 
13
- pub use document::style::StyleEnforcement;
16
+ pub use document::style::{BlankLineOptions, StyleEnforcement};
14
17
  pub use document::{
15
18
  collect_selectors, validate_condition, validate_item_condition, Document, DuplicateInfo, InsertPosition, LocatedNode, Location, NodeInfo, NodeType, SortField,
16
19
  };
17
20
  pub use error::YerbaError;
18
21
  pub use quote_style::{KeyStyle, QuoteStyle};
19
22
  pub use selector::Selector;
20
- pub use syntax::{detect_yaml_type, ScalarValue, YerbaValueType};
23
+ pub use syntax::{
24
+ detect_yaml_type, escape_double_quoted, is_control_character, is_flow_collection, is_inline_scalar_safe, is_plain_safe, is_plain_safe_in_flow,
25
+ is_quoted_scalar, is_single_quotable, needs_quoting, needs_quoting_in_flow, quote_scalar, quote_scalar_styled, ScalarValue, YerbaValueType,
26
+ };
27
+ pub use validation::{find_control_characters, ControlCharacter};
21
28
  pub use yaml_writer::json_to_yaml_text;
29
+ #[cfg(feature = "cli")]
22
30
  pub use yerbafile::Yerbafile;
23
31
 
24
32
  pub fn version() -> &'static str {
@@ -33,6 +41,7 @@ pub fn parse_file(path: impl AsRef<std::path::Path>) -> Result<Document, YerbaEr
33
41
  Document::parse_file(path)
34
42
  }
35
43
 
44
+ #[cfg(feature = "glob")]
36
45
  pub fn glob_get(pattern: &str, selector: &str) -> Vec<document::LocatedNode> {
37
46
  use rayon::prelude::*;
38
47
 
@@ -55,6 +64,7 @@ pub fn glob_get(pattern: &str, selector: &str) -> Vec<document::LocatedNode> {
55
64
  .collect()
56
65
  }
57
66
 
67
+ #[cfg(feature = "glob")]
58
68
  pub fn glob_find(pattern: &str, selector: &str, condition: Option<&str>, select: Option<&str>) -> Result<Vec<serde_json::Value>, YerbaError> {
59
69
  use rayon::prelude::*;
60
70
 
@@ -1,7 +1,7 @@
1
- use clap::ValueEnum;
2
1
  use yaml_parser::SyntaxKind;
3
2
 
4
- #[derive(Debug, Clone, PartialEq, ValueEnum)]
3
+ #[derive(Debug, Clone, PartialEq)]
4
+ #[cfg_attr(feature = "cli", derive(clap::ValueEnum))]
5
5
  pub enum KeyStyle {
6
6
  /// Unquoted key (host:)
7
7
  Plain,
@@ -34,33 +34,34 @@ impl KeyStyle {
34
34
  }
35
35
  }
36
36
 
37
- #[derive(Debug, Clone, PartialEq, ValueEnum)]
37
+ #[derive(Debug, Clone, PartialEq)]
38
+ #[cfg_attr(feature = "cli", derive(clap::ValueEnum))]
38
39
  pub enum QuoteStyle {
39
40
  /// Unquoted value (host: localhost)
40
41
  Plain,
41
42
  /// Single-quoted value (host: 'localhost')
42
- #[value(alias = "single-quoted")]
43
+ #[cfg_attr(feature = "cli", value(alias = "single-quoted"))]
43
44
  Single,
44
45
  /// Double-quoted value (host: "localhost")
45
- #[value(alias = "double-quoted")]
46
+ #[cfg_attr(feature = "cli", value(alias = "double-quoted"))]
46
47
  Double,
47
48
  /// Literal block scalar, strip trailing newline (|-)
48
- #[value(alias = "block-literal", alias = "|-")]
49
+ #[cfg_attr(feature = "cli", value(alias = "block-literal", alias = "|-"))]
49
50
  Literal,
50
51
  /// Literal block scalar, keep one trailing newline (|)
51
- #[value(alias = "|")]
52
+ #[cfg_attr(feature = "cli", value(alias = "|"))]
52
53
  LiteralClip,
53
54
  /// Literal block scalar, keep all trailing newlines (|+)
54
- #[value(alias = "|+")]
55
+ #[cfg_attr(feature = "cli", value(alias = "|+"))]
55
56
  LiteralKeep,
56
57
  /// Folded block scalar, strip trailing newline (>-)
57
- #[value(alias = "block-folded", alias = ">-")]
58
+ #[cfg_attr(feature = "cli", value(alias = "block-folded", alias = ">-"))]
58
59
  Folded,
59
60
  /// Folded block scalar, keep one trailing newline (>)
60
- #[value(alias = ">")]
61
+ #[cfg_attr(feature = "cli", value(alias = ">"))]
61
62
  FoldedClip,
62
63
  /// Folded block scalar, keep all trailing newlines (>+)
63
- #[value(alias = ">+")]
64
+ #[cfg_attr(feature = "cli", value(alias = ">+"))]
64
65
  FoldedKeep,
65
66
  }
66
67
 
data/rust/src/syntax.rs CHANGED
@@ -33,7 +33,7 @@ pub fn detect_yaml_type(scalar: &ScalarValue) -> YerbaValueType {
33
33
 
34
34
  pub fn raw_scalar_value(token: &SyntaxToken) -> Option<String> {
35
35
  match token.kind() {
36
- SyntaxKind::PLAIN_SCALAR => Some(token.text().to_string()),
36
+ SyntaxKind::PLAIN_SCALAR => Some(fold_flow_scalar(token.text())),
37
37
 
38
38
  SyntaxKind::DOUBLE_QUOTED_SCALAR => {
39
39
  let text = token.text();
@@ -65,8 +65,16 @@ pub fn extract_scalar(node: &SyntaxNode) -> Option<ScalarValue> {
65
65
  .filter_map(|element| element.into_token())
66
66
  .find(|token| token.kind() == SyntaxKind::BLOCK_SCALAR_TEXT)?;
67
67
 
68
+ let dedented = dedent_block_scalar(block_token.text());
69
+
70
+ let folded = node
71
+ .descendants()
72
+ .find(|descendant| descendant.kind() == SyntaxKind::BLOCK_SCALAR)
73
+ .map(|scalar| scalar.text().to_string().trim_start().starts_with('>'))
74
+ .unwrap_or(false);
75
+
68
76
  Some(ScalarValue {
69
- text: dedent_block_scalar(block_token.text()),
77
+ text: if folded { fold_flow_scalar(&dedented) } else { dedented },
70
78
  kind: SyntaxKind::BLOCK_SCALAR_TEXT,
71
79
  file_path: None,
72
80
  selector: None,
@@ -168,10 +176,7 @@ pub fn find_scalar_token(node: &SyntaxNode) -> Option<SyntaxToken> {
168
176
 
169
177
  pub fn format_scalar_value(value: &str, kind: SyntaxKind) -> String {
170
178
  match kind {
171
- SyntaxKind::DOUBLE_QUOTED_SCALAR => {
172
- let escaped = value.replace('\\', "\\\\").replace('"', "\\\"");
173
- format!("\"{}\"", escaped)
174
- }
179
+ SyntaxKind::DOUBLE_QUOTED_SCALAR => format!("\"{}\"", escape_double_quoted(value)),
175
180
 
176
181
  SyntaxKind::SINGLE_QUOTED_SCALAR => {
177
182
  let escaped = value.replace('\'', "''");
@@ -182,14 +187,168 @@ pub fn format_scalar_value(value: &str, kind: SyntaxKind) -> String {
182
187
  }
183
188
  }
184
189
 
185
- pub fn quote_if_needed(value: &str) -> String {
186
- if is_yaml_non_string(value) {
190
+ pub fn is_control_character(character: char) -> bool {
191
+ matches!(character as u32, 0x00..=0x08 | 0x0b | 0x0c | 0x0e..=0x1f | 0x7f | 0x80..=0x84 | 0x86..=0x9f)
192
+ }
193
+
194
+ pub fn is_single_quotable(value: &str) -> bool {
195
+ !value.contains('\n') && !value.contains('\r') && !value.chars().any(is_control_character)
196
+ }
197
+
198
+ pub fn escape_double_quoted(value: &str) -> String {
199
+ let mut result = String::with_capacity(value.len());
200
+
201
+ for character in value.chars() {
202
+ match character {
203
+ '\\' => result.push_str("\\\\"),
204
+ '"' => result.push_str("\\\""),
205
+ '\n' => result.push_str("\\n"),
206
+ '\r' => result.push_str("\\r"),
207
+ '\t' => result.push_str("\\t"),
208
+ _ if is_control_character(character) => result.push_str(&format!("\\x{:02x}", character as u32)),
209
+ _ => result.push(character),
210
+ }
211
+ }
212
+
213
+ result
214
+ }
215
+
216
+ const LEADING_INDICATORS: [char; 16] = ['#', '&', '*', '!', '|', '>', '\'', '"', '%', '@', '`', ',', '[', ']', '{', '}'];
217
+ const FLOW_INDICATORS: [char; 5] = [',', '[', ']', '{', '}'];
218
+
219
+ pub fn is_plain_safe(value: &str) -> bool {
220
+ if value.is_empty() || value != value.trim() {
221
+ return false;
222
+ }
223
+
224
+ if value.contains('\n') || value.contains('\t') {
225
+ return false;
226
+ }
227
+
228
+ if value.chars().any(is_control_character) {
229
+ return false;
230
+ }
231
+
232
+ if value.contains(": ") || value.ends_with(':') {
233
+ return false;
234
+ }
235
+
236
+ if value.contains(" #") {
237
+ return false;
238
+ }
239
+
240
+ if value.starts_with("---") || value.starts_with("...") {
241
+ return false;
242
+ }
243
+
244
+ let mut characters = value.chars();
245
+ let first = characters.next().expect("value is non-empty");
246
+
247
+ if LEADING_INDICATORS.contains(&first) {
248
+ return false;
249
+ }
250
+
251
+ if matches!(first, '-' | '?' | ':') {
252
+ return matches!(characters.next(), Some(next) if next != ' ');
253
+ }
254
+
255
+ true
256
+ }
257
+
258
+ pub fn is_plain_safe_in_flow(value: &str) -> bool {
259
+ is_plain_safe(value) && !value.contains(FLOW_INDICATORS)
260
+ }
261
+
262
+ pub fn is_quoted_scalar(value: &str) -> bool {
263
+ let bytes = value.as_bytes();
264
+
265
+ if bytes.len() < 2 {
266
+ return false;
267
+ }
268
+
269
+ match (bytes[0], bytes[bytes.len() - 1]) {
270
+ (b'"', b'"') => {
271
+ let interior = &value[1..value.len() - 1];
272
+ let mut escaped = false;
273
+
274
+ for character in interior.chars() {
275
+ if escaped {
276
+ escaped = false;
277
+ continue;
278
+ }
279
+
280
+ match character {
281
+ '\\' => escaped = true,
282
+ '"' => return false,
283
+ _ => {}
284
+ }
285
+ }
286
+
287
+ !escaped
288
+ }
289
+
290
+ (b'\'', b'\'') => {
291
+ let interior = &value[1..value.len() - 1];
292
+ let mut characters = interior.chars().peekable();
293
+
294
+ while let Some(character) = characters.next() {
295
+ if character == '\'' {
296
+ if characters.peek() == Some(&'\'') {
297
+ characters.next();
298
+ } else {
299
+ return false;
300
+ }
301
+ }
302
+ }
303
+
304
+ true
305
+ }
306
+
307
+ _ => false,
308
+ }
309
+ }
310
+
311
+ pub fn is_flow_collection(value: &str) -> bool {
312
+ (value.starts_with('[') && value.ends_with(']')) || (value.starts_with('{') && value.ends_with('}'))
313
+ }
314
+
315
+ pub fn is_inline_scalar_safe(value: &str) -> bool {
316
+ if value.is_empty() {
317
+ return true;
318
+ }
319
+
320
+ if value.contains('\n') || value.starts_with("- ") {
321
+ return false;
322
+ }
323
+
324
+ is_quoted_scalar(value) || is_flow_collection(value) || is_plain_safe(value)
325
+ }
326
+
327
+ pub fn needs_quoting(value: &str) -> bool {
328
+ is_yaml_non_string(value) || !is_plain_safe(value)
329
+ }
330
+
331
+ pub fn needs_quoting_in_flow(value: &str) -> bool {
332
+ is_yaml_non_string(value) || !is_plain_safe_in_flow(value)
333
+ }
334
+
335
+ pub fn quote_scalar(value: &str) -> String {
336
+ if needs_quoting(value) {
187
337
  format_scalar_value(value, SyntaxKind::DOUBLE_QUOTED_SCALAR)
188
338
  } else {
189
339
  value.to_string()
190
340
  }
191
341
  }
192
342
 
343
+ pub fn quote_scalar_styled(value: &str, style: Option<&str>) -> String {
344
+ match style {
345
+ Some("double") => format_scalar_value(value, SyntaxKind::DOUBLE_QUOTED_SCALAR),
346
+ Some("single") if is_single_quotable(value) => format_scalar_value(value, SyntaxKind::SINGLE_QUOTED_SCALAR),
347
+
348
+ _ => quote_scalar(value),
349
+ }
350
+ }
351
+
193
352
  pub fn extract_scalar_text(node: &SyntaxNode) -> Option<String> {
194
353
  extract_scalar(node).map(|scalar| scalar.text)
195
354
  }
@@ -212,11 +371,114 @@ pub fn dedent_block_scalar(text: &str) -> String {
212
371
  dedented.trim().to_string()
213
372
  }
214
373
 
374
+ fn push_hex_escape(result: &mut String, characters: &mut Scan<'_>, digits: usize) {
375
+ let escape: String = characters.clone().take(digits).collect();
376
+
377
+ let decoded = (escape.len() == digits && escape.chars().all(|character| character.is_ascii_hexdigit()))
378
+ .then(|| u32::from_str_radix(&escape, 16).ok().and_then(char::from_u32))
379
+ .flatten();
380
+
381
+ match decoded {
382
+ Some(character) => {
383
+ for _ in 0..digits {
384
+ characters.next();
385
+ }
386
+
387
+ result.push(character);
388
+ }
389
+
390
+ None => {
391
+ result.push('\\');
392
+ result.push(match digits {
393
+ 2 => 'x',
394
+ 4 => 'u',
395
+ _ => 'U',
396
+ });
397
+ }
398
+ }
399
+ }
400
+
401
+ type Scan<'a> = std::iter::Peekable<std::str::Chars<'a>>;
402
+
403
+ fn skip_spaces(characters: &mut Scan<'_>) {
404
+ while matches!(characters.peek(), Some(' ' | '\t')) {
405
+ characters.next();
406
+ }
407
+ }
408
+
409
+ fn fold_break(result: &mut String, characters: &mut Scan<'_>) {
410
+ while result.ends_with(' ') || result.ends_with('\t') {
411
+ result.pop();
412
+ }
413
+
414
+ let mut breaks = 1;
415
+
416
+ loop {
417
+ skip_spaces(characters);
418
+
419
+ match characters.peek() {
420
+ Some('\n') => {
421
+ characters.next();
422
+ breaks += 1;
423
+ }
424
+
425
+ Some('\r') => {
426
+ characters.next();
427
+ characters.next_if_eq(&'\n');
428
+ breaks += 1;
429
+ }
430
+
431
+ _ => break,
432
+ }
433
+ }
434
+
435
+ if breaks == 1 {
436
+ result.push(' ');
437
+ } else {
438
+ result.extend(std::iter::repeat_n('\n', breaks - 1));
439
+ }
440
+ }
441
+
442
+ pub fn fold_flow_scalar(text: &str) -> String {
443
+ if !text.contains('\n') && !text.contains('\r') {
444
+ return text.to_string();
445
+ }
446
+
447
+ let mut result = String::with_capacity(text.len());
448
+ let mut characters = text.chars().peekable();
449
+
450
+ while let Some(character) = characters.next() {
451
+ match character {
452
+ '\n' => fold_break(&mut result, &mut characters),
453
+
454
+ '\r' => {
455
+ characters.next_if_eq(&'\n');
456
+ fold_break(&mut result, &mut characters);
457
+ }
458
+
459
+ _ => result.push(character),
460
+ }
461
+ }
462
+
463
+ result
464
+ }
465
+
215
466
  pub fn unescape_double_quoted(text: &str) -> String {
216
467
  let mut result = String::with_capacity(text.len());
217
- let mut chars = text.chars();
468
+ let mut chars = text.chars().peekable();
218
469
 
219
470
  while let Some(character) = chars.next() {
471
+ if character == '\n' {
472
+ fold_break(&mut result, &mut chars);
473
+ continue;
474
+ }
475
+
476
+ if character == '\r' {
477
+ chars.next_if_eq(&'\n');
478
+ fold_break(&mut result, &mut chars);
479
+ continue;
480
+ }
481
+
220
482
  if character == '\\' {
221
483
  match chars.next() {
222
484
  Some('n') => result.push('\n'),
@@ -232,7 +494,19 @@ pub fn unescape_double_quoted(text: &str) -> String {
232
494
  Some('v') => result.push('\u{0b}'),
233
495
  Some(' ') => result.push(' '),
234
496
  Some('_') => result.push('\u{a0}'),
235
- Some('\n') => {} // line continuation: skip newline and leading whitespace
497
+ Some('N') => result.push('\u{85}'),
498
+ Some('L') => result.push('\u{2028}'),
499
+ Some('P') => result.push('\u{2029}'),
500
+ Some('x') => push_hex_escape(&mut result, &mut chars, 2),
501
+ Some('u') => push_hex_escape(&mut result, &mut chars, 4),
502
+ Some('U') => push_hex_escape(&mut result, &mut chars, 8),
503
+ Some('\n') => skip_spaces(&mut chars),
504
+
505
+ Some('\r') => {
506
+ chars.next_if_eq(&'\n');
507
+ skip_spaces(&mut chars);
508
+ }
509
+
236
510
  Some(other) => {
237
511
  result.push('\\');
238
512
  result.push(other);
@@ -248,7 +522,7 @@ pub fn unescape_double_quoted(text: &str) -> String {
248
522
  }
249
523
 
250
524
  pub fn unescape_single_quoted(text: &str) -> String {
251
- text.replace("''", "'")
525
+ fold_flow_scalar(&text.replace("''", "'"))
252
526
  }
253
527
 
254
528
  pub fn line_at(source: &str, offset: usize) -> usize {
@@ -0,0 +1,30 @@
1
+ use crate::syntax::is_control_character;
2
+
3
+ #[derive(Debug, Clone)]
4
+ pub struct ControlCharacter {
5
+ pub character: char,
6
+ pub line: usize,
7
+ pub column: usize,
8
+ }
9
+
10
+ impl std::fmt::Display for ControlCharacter {
11
+ fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
12
+ write!(f, "U+{:04X} at line {}, column {}", self.character as u32, self.line, self.column)
13
+ }
14
+ }
15
+
16
+ pub fn find_control_characters(source: &str) -> Vec<ControlCharacter> {
17
+ source
18
+ .lines()
19
+ .enumerate()
20
+ .flat_map(|(index, line)| {
21
+ line.chars().enumerate().filter_map(move |(column, character)| {
22
+ is_control_character(character).then_some(ControlCharacter {
23
+ character,
24
+ line: index + 1,
25
+ column: column + 1,
26
+ })
27
+ })
28
+ })
29
+ .collect()
30
+ }
@@ -10,6 +10,8 @@ pub fn json_to_yaml_text(value: &Value, quote_style: &QuoteStyle, indent: usize)
10
10
  map
11
11
  .iter()
12
12
  .map(|(k, v)| match v {
13
+ Value::Array(arr) if arr.is_empty() => format!("{}{}: []", prefix, k),
14
+
13
15
  Value::Array(arr) => {
14
16
  let items: Vec<String> = arr
15
17
  .iter()
@@ -26,6 +28,8 @@ pub fn json_to_yaml_text(value: &Value, quote_style: &QuoteStyle, indent: usize)
26
28
  format!("{}{}:\n{}", prefix, k, items.join("\n"))
27
29
  }
28
30
 
31
+ Value::Object(inner_map) if inner_map.is_empty() => format!("{}{}: {{}}", prefix, k),
32
+
29
33
  Value::Object(_) => {
30
34
  let inner = json_to_yaml_text(v, quote_style, indent + 2);
31
35
  format!("{}{}:\n{}", prefix, k, inner)
@@ -69,8 +73,8 @@ pub fn yaml_value_to_flow_text(value: &yaml_serde::Value) -> String {
69
73
  yaml_serde::Value::Number(number) => number.to_string(),
70
74
 
71
75
  yaml_serde::Value::String(string) => {
72
- if crate::syntax::is_yaml_non_string(string) {
73
- format!("\"{}\"", string.replace('"', "\\\""))
76
+ if crate::syntax::needs_quoting_in_flow(string) {
77
+ crate::syntax::format_scalar_value(string, yaml_parser::SyntaxKind::DOUBLE_QUOTED_SCALAR)
74
78
  } else {
75
79
  string.clone()
76
80
  }
@@ -149,31 +153,15 @@ fn format_yaml_scalar(value: &Value, quote_style: &QuoteStyle) -> String {
149
153
  Value::Bool(boolean) => boolean.to_string(),
150
154
  Value::Number(number) => number.to_string(),
151
155
  Value::String(string) => match quote_style {
152
- QuoteStyle::Double => {
153
- let escaped = string.replace('\\', "\\\\").replace('"', "\\\"");
154
-
155
- format!("\"{}\"", escaped)
156
- }
157
-
158
- QuoteStyle::Single => {
156
+ QuoteStyle::Single if crate::syntax::is_single_quotable(string) => {
159
157
  let escaped = string.replace('\'', "''");
160
158
 
161
159
  format!("'{}'", escaped)
162
160
  }
163
161
 
164
- QuoteStyle::Plain => {
165
- if crate::syntax::is_yaml_non_string(string) {
166
- format!("\"{}\"", string.replace('"', "\\\""))
167
- } else {
168
- string.clone()
169
- }
170
- }
162
+ QuoteStyle::Plain if !crate::syntax::needs_quoting(string) => string.clone(),
171
163
 
172
- _ => {
173
- let escaped = string.replace('\\', "\\\\").replace('"', "\\\"");
174
-
175
- format!("\"{}\"", escaped)
176
- }
164
+ _ => crate::syntax::format_scalar_value(string, yaml_parser::SyntaxKind::DOUBLE_QUOTED_SCALAR),
177
165
  },
178
166
 
179
167
  Value::Array(array) => {