1use std::collections::HashMap;
2use std::path::{Path, PathBuf};
3
4use rayon::prelude::*;
5
6use crate::entity::{Entity, Label};
7use crate::parser::ParseError;
8
9const MAX_FILENAME_LEN: usize = 200;
11
12#[derive(Debug)]
14pub struct RegistryEntry {
15 pub entity: Entity,
16 pub path: PathBuf,
17 pub tags: Vec<String>,
18}
19
20#[derive(Debug)]
24pub struct EntityRegistry {
25 entries: Vec<RegistryEntry>,
26 name_index: HashMap<String, usize>,
28 content_root: Option<PathBuf>,
30}
31
32impl EntityRegistry {
33 pub fn load(root: &Path) -> Result<Self, Vec<ParseError>> {
39 let mut entries = Vec::new();
40 let mut errors = Vec::new();
41
42 let actor_dir = root.join("people");
43 let institution_dir = root.join("organizations");
44
45 load_directory(&actor_dir, Label::Person, &mut entries, &mut errors);
46 load_directory(
47 &institution_dir,
48 Label::Organization,
49 &mut entries,
50 &mut errors,
51 );
52
53 let name_index = build_name_index(&entries, &mut errors);
55
56 if errors.iter().any(|e| e.message.starts_with("duplicate")) {
57 return Err(errors);
58 }
59
60 if errors.iter().any(|e| !e.message.starts_with("warning:")) {
64 return Err(errors);
65 }
66
67 if !errors.is_empty() {
69 for err in &errors {
70 eprintln!("{err}");
71 }
72 }
73
74 Ok(Self {
75 entries,
76 name_index,
77 content_root: Some(root.to_path_buf()),
78 })
79 }
80
81 pub fn from_entries(entries: Vec<RegistryEntry>) -> Result<Self, Vec<ParseError>> {
83 let mut errors = Vec::new();
84 let name_index = build_name_index(&entries, &mut errors);
85
86 let has_errors = errors.iter().any(|e| !e.message.starts_with("warning:"));
87 if has_errors {
88 return Err(errors);
89 }
90
91 Ok(Self {
92 entries,
93 name_index,
94 content_root: None,
95 })
96 }
97
98 pub fn get_by_name(&self, name: &str) -> Option<&RegistryEntry> {
100 self.name_index.get(name).map(|&idx| &self.entries[idx])
101 }
102
103 pub fn len(&self) -> usize {
105 self.entries.len()
106 }
107
108 pub fn is_empty(&self) -> bool {
110 self.entries.is_empty()
111 }
112
113 pub fn names(&self) -> Vec<&str> {
115 self.entries
116 .iter()
117 .map(|e| e.entity.name.as_str())
118 .collect()
119 }
120
121 pub fn entries(&self) -> &[RegistryEntry] {
123 &self.entries
124 }
125
126 pub fn slug_for(&self, entry: &RegistryEntry) -> Option<String> {
129 let root = self.content_root.as_ref()?;
130 path_to_slug(&entry.path, root)
131 }
132
133 pub fn content_root(&self) -> Option<&Path> {
135 self.content_root.as_deref()
136 }
137}
138
139pub fn path_to_slug(path: &Path, content_root: &Path) -> Option<String> {
142 let relative = path.strip_prefix(content_root).ok()?;
143 let s = relative.to_str()?;
144 Some(s.strip_suffix(".md").unwrap_or(s).to_string())
145}
146
147fn load_directory(
151 dir: &Path,
152 label: Label,
153 entries: &mut Vec<RegistryEntry>,
154 errors: &mut Vec<ParseError>,
155) {
156 let mut paths = Vec::new();
157 collect_md_files(dir, &mut paths, 0);
158
159 paths.sort();
161
162 let results: Vec<ParseResult> = paths
164 .par_iter()
165 .map(|path| parse_entity_file(path, label))
166 .collect();
167
168 for result in results {
170 if let Some(entry) = result.entry {
171 entries.push(entry);
172 }
173 errors.extend(result.errors);
174 }
175}
176
177fn collect_md_files(dir: &Path, paths: &mut Vec<PathBuf>, depth: usize) {
180 const MAX_DEPTH: usize = 2;
181 if depth > MAX_DEPTH {
182 return;
183 }
184
185 let Ok(read_dir) = std::fs::read_dir(dir) else {
186 return;
187 };
188
189 let mut dir_entries: Vec<_> = read_dir.filter_map(Result::ok).collect();
190 dir_entries.sort_by_key(std::fs::DirEntry::file_name);
191
192 for entry in dir_entries {
193 let path = entry.path();
194 if path.is_dir() {
195 collect_md_files(&path, paths, depth + 1);
196 } else if path.extension().and_then(|e| e.to_str()) == Some("md") {
197 paths.push(path);
198 }
199 }
200}
201
202struct ParseResult {
204 entry: Option<RegistryEntry>,
205 errors: Vec<ParseError>,
206}
207
208fn parse_entity_file(path: &Path, label: Label) -> ParseResult {
210 let content = match std::fs::read_to_string(path) {
211 Ok(c) => c,
212 Err(e) => {
213 return ParseResult {
214 entry: None,
215 errors: vec![ParseError {
216 line: 0,
217 message: format!("{}: error reading file: {e}", path.display()),
218 }],
219 };
220 }
221 };
222
223 let parsed = match crate::parser::parse_entity_file(&content) {
224 Ok(p) => p,
225 Err(parse_errors) => {
226 return ParseResult {
227 entry: None,
228 errors: parse_errors
229 .into_iter()
230 .map(|err| ParseError {
231 line: err.line,
232 message: format!("{}: {}", path.display(), err.message),
233 })
234 .collect(),
235 };
236 }
237 };
238
239 let mut field_errors = Vec::new();
240 let mut entity = crate::entity::parse_entity_file_body(
241 &parsed.name,
242 &parsed.body,
243 label,
244 parsed.id,
245 parsed.title_line,
246 &mut field_errors,
247 );
248 entity.tags.clone_from(&parsed.tags);
249
250 let mut errors: Vec<ParseError> = field_errors
251 .into_iter()
252 .map(|err| ParseError {
253 line: err.line,
254 message: format!("{}: {}", path.display(), err.message),
255 })
256 .collect();
257
258 validate_filename(path, &entity, &mut errors);
260
261 ParseResult {
262 entry: Some(RegistryEntry {
263 entity,
264 path: path.to_path_buf(),
265 tags: parsed.tags,
266 }),
267 errors,
268 }
269}
270
271fn build_name_index(
273 entries: &[RegistryEntry],
274 errors: &mut Vec<ParseError>,
275) -> HashMap<String, usize> {
276 let mut index = HashMap::new();
277
278 for (i, entry) in entries.iter().enumerate() {
279 let name = &entry.entity.name;
280 if let Some(&existing_idx) = index.get(name.as_str()) {
281 let existing: &RegistryEntry = &entries[existing_idx];
282 errors.push(ParseError {
283 line: entry.entity.line,
284 message: format!(
285 "duplicate entity name {name:?} in {} (first defined in {})",
286 entry.path.display(),
287 existing.path.display(),
288 ),
289 });
290 } else {
291 index.insert(name.clone(), i);
292 }
293 }
294
295 index
296}
297
298fn validate_filename(path: &Path, entity: &Entity, errors: &mut Vec<ParseError>) {
301 let Some(stem) = path.file_stem().and_then(|s| s.to_str()) else {
302 return;
303 };
304
305 if stem.len() > MAX_FILENAME_LEN {
306 errors.push(ParseError {
307 line: 0,
308 message: format!(
309 "warning: {}: filename stem exceeds {MAX_FILENAME_LEN} chars",
310 path.display()
311 ),
312 });
313 }
314
315 let expected_name = to_kebab_case(&entity.name);
316 let qualifier = entity
317 .fields
318 .iter()
319 .find(|(k, _)| k == "qualifier")
320 .and_then(|(_, v)| match v {
321 crate::entity::FieldValue::Single(s) => Some(s.as_str()),
322 crate::entity::FieldValue::List(_) => None,
323 });
324
325 let expected_stem = match qualifier {
326 Some(q) => format!("{expected_name}--{}", to_kebab_case(q)),
327 None => expected_name,
328 };
329
330 if stem != expected_stem {
331 errors.push(ParseError {
332 line: 0,
333 message: format!(
334 "warning: {}: filename {stem:?} doesn't match expected {expected_stem:?}",
335 path.display()
336 ),
337 });
338 }
339}
340
341fn to_kebab_case(s: &str) -> String {
343 s.chars()
344 .map(|c| {
345 if c.is_alphanumeric() {
346 c.to_ascii_lowercase()
347 } else {
348 '-'
349 }
350 })
351 .collect::<String>()
352 .split('-')
353 .filter(|p| !p.is_empty())
354 .collect::<Vec<_>>()
355 .join("-")
356}
357
358#[cfg(test)]
359mod tests {
360 use super::*;
361 use crate::entity::{Entity, FieldValue, Label};
362
363 fn make_entry(name: &str, label: Label, path: &str) -> RegistryEntry {
364 RegistryEntry {
365 entity: Entity {
366 name: name.to_string(),
367 label,
368 fields: Vec::new(),
369 id: None,
370 line: 1,
371 tags: Vec::new(),
372 slug: None,
373 },
374 path: PathBuf::from(path),
375 tags: Vec::new(),
376 }
377 }
378
379 #[test]
380 fn registry_from_entries_lookup() {
381 let entries = vec![
382 make_entry("Alice", Label::Person, "people/alice.md"),
383 make_entry("Corp Inc", Label::Organization, "organizations/corp-inc.md"),
384 ];
385
386 let registry = EntityRegistry::from_entries(entries).unwrap();
387 assert_eq!(registry.len(), 2);
388 assert!(registry.get_by_name("Alice").is_some());
389 assert!(registry.get_by_name("Corp Inc").is_some());
390 assert!(registry.get_by_name("Bob").is_none());
391 }
392
393 #[test]
394 fn registry_detects_duplicate_names() {
395 let entries = vec![
396 make_entry("Alice", Label::Person, "people/alice-a.md"),
397 make_entry("Alice", Label::Person, "people/alice-b.md"),
398 ];
399
400 let errors = EntityRegistry::from_entries(entries).unwrap_err();
401 assert!(errors.iter().any(|e| e.message.contains("duplicate")));
402 }
403
404 #[test]
405 fn registry_names_list() {
406 let entries = vec![
407 make_entry("Alice", Label::Person, "people/alice.md"),
408 make_entry("Bob", Label::Person, "people/bob.md"),
409 ];
410
411 let registry = EntityRegistry::from_entries(entries).unwrap();
412 let names = registry.names();
413 assert!(names.contains(&"Alice"));
414 assert!(names.contains(&"Bob"));
415 }
416
417 #[test]
418 fn to_kebab_case_conversion() {
419 assert_eq!(to_kebab_case("Mark Bonnick"), "mark-bonnick");
420 assert_eq!(to_kebab_case("Arsenal FC"), "arsenal-fc");
421 assert_eq!(
422 to_kebab_case("English Football Club"),
423 "english-football-club"
424 );
425 assert_eq!(to_kebab_case("Bob"), "bob");
426 }
427
428 #[test]
429 fn validate_filename_matching() {
430 let entity = Entity {
431 name: "Mark Bonnick".to_string(),
432 label: Label::Person,
433 fields: vec![(
434 "qualifier".to_string(),
435 FieldValue::Single("Arsenal Kit Manager".to_string()),
436 )],
437 id: None,
438 line: 1,
439 tags: Vec::new(),
440 slug: None,
441 };
442
443 let mut errors = Vec::new();
444
445 validate_filename(
447 Path::new("people/mark-bonnick--arsenal-kit-manager.md"),
448 &entity,
449 &mut errors,
450 );
451 assert!(errors.is_empty(), "errors: {errors:?}");
452
453 validate_filename(Path::new("people/wrong-name.md"), &entity, &mut errors);
455 assert!(errors.iter().any(|e| e.message.contains("warning:")));
456 }
457
458 #[test]
459 fn validate_filename_no_qualifier() {
460 let entity = Entity {
461 name: "Bob".to_string(),
462 label: Label::Person,
463 fields: Vec::new(),
464 id: None,
465 line: 1,
466 tags: Vec::new(),
467 slug: None,
468 };
469
470 let mut errors = Vec::new();
471 validate_filename(Path::new("people/bob.md"), &entity, &mut errors);
472 assert!(errors.is_empty(), "errors: {errors:?}");
473 }
474
475 #[test]
476 fn empty_registry() {
477 let registry = EntityRegistry::from_entries(Vec::new()).unwrap();
478 assert!(registry.is_empty());
479 assert_eq!(registry.len(), 0);
480 assert!(registry.get_by_name("anything").is_none());
481 }
482}