1#[cfg(test)]
8mod build_config;
9
10mod ast;
11mod diagnostics;
12#[allow(unsafe_code)]
13mod ffi;
14mod parser;
15
16pub use ast::{
17 DisplayKind, Document, MacroSet, Metadata, Node, NodeFlags, NodeKind, NormalizedListKind,
18 TableAlignment, TableCell,
19};
20pub use diagnostics::{Diagnostic, DiagnosticLevel, SourceLocation};
21pub use parser::{
22 Compression, IncludePolicy, ParseError, ParseErrorKind, ParseOptions, ParseReport, Parser,
23};
24
25pub const LIBMANDOC_VERSION: &str = "1.14.6";
27
28struct RawDocument {
30 document: Document,
31 diagnostics: String,
32}
33
34#[cfg(test)]
35mod tests {
36 use std::{fs, process};
37
38 use super::{
39 Compression, DisplayKind, Document, IncludePolicy, MacroSet, Node, NodeKind,
40 NormalizedListKind, ParseError, ParseOptions, Parser, TableAlignment,
41 };
42
43 fn source_path(label: &str) -> std::path::PathBuf {
44 std::env::temp_dir().join(format!("mant-{label}-{}.1", process::id()))
45 }
46
47 fn measured_depth(node: &Node) -> usize {
48 1 + node.children.iter().map(measured_depth).max().unwrap_or(0)
49 }
50
51 fn parse_file(path: &std::path::Path, allow_includes: bool) -> Result<Document, ParseError> {
52 Parser::new(ParseOptions {
53 includes: if allow_includes {
54 IncludePolicy::SourceTree
55 } else {
56 IncludePolicy::Deny
57 },
58 compression: Compression::Auto,
59 })
60 .parse_file(path)
61 .map(|report| report.document)
62 }
63
64 fn find_macro<'a>(node: &'a Node, name: &str) -> Option<&'a Node> {
65 (node.macro_name.as_deref() == Some(name))
66 .then_some(node)
67 .or_else(|| {
68 node.children
69 .iter()
70 .find_map(|child| find_macro(child, name))
71 })
72 }
73
74 fn find_kind(node: &Node, kind: NodeKind) -> Option<&Node> {
75 (node.kind == kind).then_some(node).or_else(|| {
76 node.children
77 .iter()
78 .find_map(|child| find_kind(child, kind))
79 })
80 }
81
82 fn find_node<'a>(node: &'a Node, predicate: &impl Fn(&Node) -> bool) -> Option<&'a Node> {
83 predicate(node).then_some(node).or_else(|| {
84 node.children
85 .iter()
86 .find_map(|child| find_node(child, predicate))
87 })
88 }
89
90 #[test]
91 fn upstream_version_is_pinned() {
92 assert_eq!(super::LIBMANDOC_VERSION, "1.14.6");
93 }
94
95 #[test]
96 fn parser_session_returns_an_owned_man_tree() {
97 let path = source_path("mandoc-session");
98 fs::write(
99 &path,
100 ".TH MANT 1 \"2026-07-19\"\n.SH NAME\nmant \\- manual viewer\n",
101 )
102 .expect("write temporary manual source");
103
104 let document = parse_file(&path, false).expect("parse temporary manual");
105 fs::remove_file(path).expect("remove temporary manual source");
106
107 assert_eq!(document.macro_set, MacroSet::Man);
108 assert_eq!(document.metadata.title.as_deref(), Some("MANT"));
109 assert_eq!(document.metadata.section.as_deref(), Some("1"));
110 assert!(document.metadata.has_body);
111 assert_eq!(document.root.kind, NodeKind::Root);
112 assert!(!document.root.children.is_empty());
113 }
114
115 #[test]
116 fn parser_decompresses_zstd_sources_before_calling_libmandoc() {
117 let path = source_path("zstd-mandoc-session").with_extension("1.zst");
118 let source = b".TH ZSTD-MANT 1 \"2026-07-20\"\n.SH NAME\nzstd-mant \\- compressed manual\n";
119 let compressed = zstd::stream::encode_all(source.as_slice(), 1).expect("compress source");
120 fs::write(&path, compressed).expect("write compressed manual source");
121
122 let report = Parser::default()
123 .parse_file(&path)
124 .expect("parse zstd manual");
125 fs::remove_file(path).expect("remove compressed manual source");
126
127 assert!(report.diagnostics.is_empty());
128 let document = report.document;
129 assert_eq!(document.macro_set, MacroSet::Man);
130 assert_eq!(document.metadata.title.as_deref(), Some("ZSTD-MANT"));
131 assert_eq!(document.metadata.section.as_deref(), Some("1"));
132 assert!(document.metadata.has_body);
133 }
134
135 #[test]
136 fn invalid_zstd_sources_fail_before_reaching_libmandoc() {
137 let path = source_path("invalid-zstd-mandoc-session").with_extension("1.zst");
138 fs::write(&path, b"not a zstd frame").expect("write invalid compressed source");
139
140 let error = parse_file(&path, false).expect_err("invalid zstd source must fail");
141 fs::remove_file(path).expect("remove invalid compressed source");
142
143 assert!(
144 error
145 .message
146 .starts_with("could not decompress zstd manual source:")
147 );
148 assert_eq!(error.kind, super::ParseErrorKind::Decompression);
149 assert!(!error.message.contains("unsupported control character"));
150 }
151
152 #[test]
153 fn zstd_sources_keep_their_original_include_root() {
154 let root = std::env::temp_dir().join(format!(
155 "mant-zstd-include-mandoc-session-{}",
156 process::id()
157 ));
158 let man1 = root.join("man1");
159 fs::create_dir_all(&man1).expect("create temporary manual tree");
160 let target = man1.join("target.1");
161 fs::write(
162 &target,
163 ".TH ZSTD-INCLUDE 1\n.SH NAME\nzstd-include \\- included manual\n",
164 )
165 .expect("write included manual");
166 let alias = man1.join("alias.1.zst");
167 let compressed =
168 zstd::stream::encode_all(b".so man1/target.1\n".as_slice(), 1).expect("compress alias");
169 fs::write(&alias, compressed).expect("write compressed alias");
170
171 let document = parse_file(&alias, true).expect("resolve include from zstd source");
172 fs::remove_dir_all(root).expect("remove temporary manual tree");
173
174 assert_eq!(document.macro_set, MacroSet::Man);
175 assert_eq!(document.metadata.title.as_deref(), Some("ZSTD-INCLUDE"));
176 assert!(document.metadata.has_body);
177 }
178
179 #[test]
180 fn parser_preserves_same_line_layout_and_next_line_content_roles() {
181 let path = source_path("line-role-mandoc-session");
182 fs::write(
183 &path,
184 ".TH LINE-ROLE 1\n.SH EXAMPLES\n.TP \\w'man\\ 'u\n.BI man \\ ls\nBody.\n",
185 )
186 .expect("write tagged paragraph source");
187
188 let document = parse_file(&path, false).expect("parse tagged paragraph source");
189 fs::remove_file(path).expect("remove tagged paragraph source");
190
191 let tagged_paragraph = find_macro(&document.root, "TP").expect("TP block");
192 let head = tagged_paragraph
193 .children
194 .iter()
195 .find(|child| child.kind == NodeKind::Head)
196 .expect("TP head");
197 assert_eq!(head.children[0].text.as_deref(), Some("96u"));
198 assert!(!head.children[0].flags.line_start);
199 assert_eq!(head.children[1].macro_name.as_deref(), Some("BI"));
200 assert!(head.children[1].flags.line_start);
201 }
202
203 #[test]
204 fn parser_preserves_mdoc_delimiter_spacing_roles() {
205 let path = source_path("delimiter-role-mandoc-session");
206 fs::write(
207 &path,
208 ".Dd August 4, 2026\n.Dt DELIMITERS 1\n.Os\n.Sh EXAMPLES\n\
209 .Dl name ( ) command\n\
210 .Dl local [ variable | - ] ...\n\
211 .Dl return [ exitstatus ]\n",
212 )
213 .expect("write delimiter-role source");
214
215 let document = parse_file(&path, false).expect("parse delimiter-role source");
216 fs::remove_file(path).expect("remove delimiter-role source");
217
218 let opening_parenthesis = find_node(&document.root, &|node| {
219 node.line == 5 && node.text.as_deref() == Some("(")
220 })
221 .expect("opening parenthesis");
222 let closing_parenthesis = find_node(&document.root, &|node| {
223 node.line == 5 && node.text.as_deref() == Some(")")
224 })
225 .expect("closing parenthesis");
226 let opening_bracket = find_node(&document.root, &|node| {
227 node.line == 7 && node.text.as_deref() == Some("[")
228 })
229 .expect("opening bracket");
230 let trailing_bracket = find_node(&document.root, &|node| {
231 node.line == 7 && node.text.as_deref() == Some("]")
232 })
233 .expect("trailing bracket");
234
235 assert!(opening_parenthesis.flags.delimiter_open);
236 assert!(closing_parenthesis.flags.delimiter_close);
237 assert!(opening_bracket.flags.delimiter_open);
238 assert!(trailing_bracket.flags.delimiter_close);
239 }
240
241 #[test]
242 fn parser_session_reports_file_errors_as_values() {
243 let path = source_path("missing-mandoc-session");
244 let error = parse_file(&path, false).expect_err("missing source must fail");
245
246 assert_eq!(error.path, path);
247 assert!(!error.message.is_empty());
248 }
249
250 #[test]
251 fn concurrent_callers_are_serialized_around_libmandoc_globals() {
252 let path = source_path("concurrent-mandoc-session");
253 fs::write(&path, ".TH THREADS 1\n.SH NAME\nthreads \\- test\n")
254 .expect("write temporary manual source");
255
256 let workers: Vec<_> = (0..4)
257 .map(|_| {
258 let path = path.clone();
259 std::thread::spawn(move || parse_file(&path, false))
260 })
261 .collect();
262 for worker in workers {
263 let document = worker
264 .join()
265 .expect("parser worker must not panic")
266 .expect("concurrent parse must succeed");
267 assert_eq!(document.metadata.title.as_deref(), Some("THREADS"));
268 }
269
270 fs::remove_file(path).expect("remove temporary manual source");
271 }
272
273 #[test]
274 fn source_relative_includes_do_not_change_process_cwd() {
275 let root =
276 std::env::temp_dir().join(format!("libmandoc-rs-relative-include-{}", process::id()));
277 fs::create_dir_all(&root).expect("create temporary manual tree");
278 let target = root.join("minimal-mdoc.1");
279 fs::write(
280 &target,
281 ".Dd July 19, 2026\n.Dt INCLUDE-FIXTURE 1\n.Os\n.Sh NAME\ninclude-fixture\n",
282 )
283 .expect("write included source");
284 let alias = root.join("alias-mdoc.1");
285 fs::write(&alias, ".so minimal-mdoc.1\n").expect("write alias source");
286 let cwd = std::env::current_dir().expect("current directory before parse");
287
288 let document = parse_file(&alias, true).expect("resolve source-relative include");
289 fs::remove_dir_all(root).expect("remove temporary manual tree");
290
291 assert_eq!(document.macro_set, MacroSet::Mdoc);
292 assert_eq!(document.metadata.title.as_deref(), Some("INCLUDE-FIXTURE"));
293 assert_eq!(
294 std::env::current_dir().expect("current directory after parse"),
295 cwd
296 );
297 }
298
299 #[test]
300 fn parser_accepts_owned_bytes_and_detects_zstd_frames() {
301 let source = b".TH BYTES 1\n.SH NAME\nbytes \\- parser input\n";
302 let plain = Parser::default()
303 .parse_bytes("memory.1", source)
304 .expect("parse plain byte input");
305 assert_eq!(plain.document.metadata.title.as_deref(), Some("BYTES"));
306
307 let compressed = zstd::stream::encode_all(source.as_slice(), 1).expect("compress source");
308 let zstd = Parser::default()
309 .parse_bytes("memory.1", &compressed)
310 .expect("detect and parse zstd byte input");
311 assert_eq!(zstd.document.metadata.title.as_deref(), Some("BYTES"));
312 }
313
314 #[test]
315 fn parser_only_expands_includes_when_policy_allows_a_root() {
316 let base = std::env::temp_dir().join(format!(
317 "libmandoc-rs-explicit-include-root-{}",
318 process::id()
319 ));
320 let includes = base.join("includes");
321 fs::create_dir_all(&includes).expect("create explicit include root");
322 fs::write(
323 includes.join("target.1"),
324 ".TH EXPLICIT-ROOT 1\n.SH NAME\nexplicit-root \\- include fixture\n",
325 )
326 .expect("write included source");
327 let alias = base.join("alias.1");
328 fs::write(&alias, ".so target.1\n").expect("write alias source");
329
330 let denied = Parser::default()
331 .parse_file(&alias)
332 .expect("parse alias without include expansion");
333 let expanded = Parser::new(ParseOptions {
334 includes: IncludePolicy::Root(includes),
335 compression: Compression::Auto,
336 })
337 .parse_file(&alias)
338 .expect("resolve alias against explicit root");
339 fs::remove_dir_all(base).expect("remove temporary manual tree");
340
341 assert_ne!(
342 denied.document.metadata.title.as_deref(),
343 Some("EXPLICIT-ROOT")
344 );
345 assert_eq!(
346 expanded.document.metadata.title.as_deref(),
347 Some("EXPLICIT-ROOT")
348 );
349 }
350
351 #[test]
352 fn explicit_include_root_does_not_fall_back_to_process_cwd() {
353 let identifier = format!("libmandoc-rs-ambient-{}", process::id());
354 let cwd_target = std::env::current_dir()
355 .expect("read test cwd")
356 .join(format!("{identifier}.1"));
357 fs::write(
358 &cwd_target,
359 ".TH AMBIENT 1\n.SH NAME\nambient \\- must not be included\n",
360 )
361 .expect("write ambient source");
362
363 let base = std::env::temp_dir().join(format!("{identifier}-root"));
364 fs::create_dir_all(&base).expect("create empty include root");
365 let alias = base.join("alias.1");
366 fs::write(&alias, format!(".so {identifier}.1\n")).expect("write alias source");
367
368 let result = Parser::new(ParseOptions {
369 includes: IncludePolicy::Root(base.clone()),
370 compression: Compression::Auto,
371 })
372 .parse_file(&alias);
373 fs::remove_file(cwd_target).expect("remove ambient source");
374 fs::remove_dir_all(base).expect("remove temporary manual tree");
375
376 match result {
377 Ok(report) => assert_ne!(report.document.metadata.title.as_deref(), Some("AMBIENT")),
378 Err(error) => assert_eq!(error.kind, super::ParseErrorKind::Parse),
379 }
380 }
381
382 #[test]
383 fn parser_returns_structured_nonfatal_diagnostics() {
384 let report = Parser::default()
385 .parse_bytes(
386 "diagnostics.1",
387 b".Dd July 19, 2026\n.Dt BAD 1\n.Os\n.Sh NAME\n.Nm bad\n.ab\n",
388 )
389 .expect("return best-effort document");
390
391 assert!(
392 report
393 .diagnostics
394 .iter()
395 .any(|diagnostic| diagnostic.level == super::DiagnosticLevel::Unsupported)
396 );
397 }
398
399 #[test]
400 fn deeply_nested_input_is_bounded_instead_of_overflowing_the_stack() {
401 let depth = 5_000;
404 let mut source = String::from(".TH DEEP 1\n.SH BODY\n");
405 for _ in 0..depth {
406 source.push_str(".RS\n");
407 }
408 source.push_str("deep\n");
409
410 let document = Parser::default()
411 .parse_bytes("deep.1", source.as_bytes())
412 .expect("deeply nested source parses")
413 .document;
414
415 assert!(
418 measured_depth(&document.root) <= 300,
419 "tree depth must be bounded by the copy cap"
420 );
421 }
422
423 #[cfg(feature = "serde")]
424 #[test]
425 fn serde_feature_round_trips_the_public_parse_report() {
426 let report = Parser::default()
427 .parse_bytes("serde.1", b".TH SERDE 1\n.SH NAME\nserde \\- fixture\n")
428 .expect("parse source for serialization");
429 let encoded = serde_json::to_string(&report).expect("serialize parse report");
430 let decoded: super::ParseReport =
431 serde_json::from_str(&encoded).expect("deserialize parse report");
432
433 assert_eq!(decoded, report);
434 }
435
436 #[test]
437 fn parser_copies_normalized_list_and_display_attributes() {
438 let path = source_path("normalized-mandoc-session");
439 fs::write(
440 &path,
441 ".Dd July 19, 2026\n.Dt NORMALIZED 1\n.Os\n.Sh ITEMS\n\
442 .Bl -tag -compact -offset indent -width 12n\n.It item\nfirst\n.El\n\
443 .Bd -literal -offset indent\ncode line\n.Ed\n",
444 )
445 .expect("write normalized mdoc source");
446
447 let document = parse_file(&path, false).expect("parse normalized mdoc source");
448 fs::remove_file(path).expect("remove normalized mdoc source");
449
450 let list = find_macro(&document.root, "Bl").expect("normalized list node");
451 assert_eq!(list.list_kind, Some(NormalizedListKind::Definition));
452 assert!(list.compact);
453 assert_eq!(list.offset.as_deref(), Some("indent"));
454 assert_eq!(list.width.as_deref(), Some("12n"));
455 let display = find_macro(&document.root, "Bd").expect("normalized display node");
456 assert_eq!(display.display_kind, Some(DisplayKind::Literal));
457 assert_eq!(display.offset.as_deref(), Some("indent"));
458 }
459
460 #[test]
461 fn parser_copies_table_cells_and_equation_text() {
462 let path = source_path("structured-payload-mandoc-session");
463 fs::write(
464 &path,
465 ".TH PAYLOAD 1\n.SH TABLE\n.TS\ntab(|);\nl r.\nleft|right\n.TE\n\
466 .SH EQUATION\n.EQ\nx sup 2\n.EN\n",
467 )
468 .expect("write table and equation source");
469
470 let document = parse_file(&path, false).expect("parse table and equation source");
471 fs::remove_file(path).expect("remove table and equation source");
472
473 let table = find_kind(&document.root, NodeKind::Table).expect("table row node");
474 assert_eq!(table.table_cells.len(), 2);
475 assert_eq!(table.table_cells[0].text.as_deref(), Some("left"));
476 assert_eq!(table.table_cells[1].alignment, TableAlignment::Right);
477 let equation = find_kind(&document.root, NodeKind::Equation).expect("equation node");
478 assert!(
479 equation
480 .equation
481 .as_deref()
482 .is_some_and(|value| value.contains('x'))
483 );
484 }
485
486 #[test]
487 fn parser_copies_validated_same_document_navigation() {
488 let path = source_path("navigation-mandoc-session");
489 fs::write(
490 &path,
491 ".Dd July 19, 2026\n.Dt NAVIGATION 1\n.Os\n.Sh FIRST\n\
492 See\n\
493 .Sx TARGET\n\
494 for details.\n\
495 .Tg explicit-target\n\
496 .Fl x\n\
497 .Sh TARGET\nTarget text.\n",
498 )
499 .expect("write navigation mdoc source");
500
501 let document = parse_file(&path, false).expect("parse navigation mdoc source");
502 fs::remove_file(path).expect("remove navigation mdoc source");
503
504 assert!(find_macro(&document.root, "Sx").is_some());
505 let explicit_target = find_node(&document.root, &|node| {
506 node.flags.deep_link_target && node.tag.as_deref() == Some("explicit-target")
507 });
508 let explicit_target = explicit_target.expect("Tg must annotate its resolved destination");
509 assert!(explicit_target.flags.permalink);
510 }
511}