1use crate::xml::builders::{build_xml_string, merge_xml_elements, reorder_root_keys};
4use crate::xml::multi_level::{ensure_segment_files_structure, load_multi_level_config};
5use crate::xml::parsers::parse_to_xml_object;
6use crate::xml::types::{MultiLevelRule, SidecarSpec, XmlElement};
7use crate::xml::utils::normalize_path_unix;
8use serde_json::Value;
9use std::collections::HashSet;
10use std::ffi::OsString;
11use std::future::Future;
12use std::path::{Path, PathBuf};
13use std::pin::Pin;
14use tokio::fs;
15
16async fn read_key_order(path: &Path) -> Option<Vec<String>> {
18 let bytes = fs::read(path).await.ok()?;
19 serde_json::from_slice::<Vec<String>>(&bytes).ok()
20}
21
22fn strip_xmlns_from_value(v: Value) -> Value {
24 match v {
25 Value::Object(obj) => {
26 Value::Object(obj.into_iter().filter(|(k, _)| k != "@xmlns").collect())
27 }
28 other => other,
29 }
30}
31
32fn deeper_candidate_rules(
40 all_rules: &[MultiLevelRule],
41 exclude_path_segment: &str,
42) -> Vec<MultiLevelRule> {
43 all_rules
44 .iter()
45 .filter(|r| r.path_segment != exclude_path_segment)
46 .cloned()
47 .collect()
48}
49
50fn is_at_base_path(dir_path: &str, base_segments: &[(String, String, bool)]) -> bool {
57 base_segments.iter().any(|(base, _, _)| dir_path == base)
58}
59
60type ProcessDirFuture<'a> = Pin<
61 Box<
62 dyn Future<Output = Result<Vec<XmlElement>, Box<dyn std::error::Error + Send + Sync>>>
63 + Send
64 + 'a,
65 >,
66>;
67
68type SegmentFuture<'a> =
69 Pin<Box<dyn Future<Output = Result<(), Box<dyn std::error::Error + Send + Sync>>> + Send + 'a>>;
70
71pub struct ReassembleXmlFileHandler;
72
73impl ReassembleXmlFileHandler {
74 pub fn new() -> Self {
75 Self
76 }
77
78 pub async fn reassemble(
79 &self,
80 file_path: &str,
81 file_extension: Option<&str>,
82 post_purge: bool,
83 sidecar_specs: Option<&[SidecarSpec]>,
84 ) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
85 let file_path = normalize_path_unix(file_path);
86 if !self.validate_directory(&file_path).await? {
87 return Ok(());
88 }
89
90 let path = Path::new(&file_path);
91 let config = load_multi_level_config(path).await;
92 if let Some(ref config) = config {
93 for (i, rule) in config.rules.iter().enumerate() {
98 let segment_path = path.join(&rule.path_segment);
99 if !segment_path.is_dir() {
100 continue;
101 }
102 let nested: Vec<MultiLevelRule> = config
103 .rules
104 .iter()
105 .enumerate()
106 .filter(|(j, _)| *j != i)
107 .map(|(_, r)| r.clone())
108 .collect();
109 self.reassemble_multi_level_segment(&segment_path, rule, &nested)
110 .await?;
111 }
112 }
113
114 let base_segments: Vec<(String, String, bool)> = config
117 .as_ref()
118 .map(|c| {
119 c.rules
120 .iter()
121 .map(|r| (file_path.clone(), r.path_segment.clone(), true))
122 .collect()
123 })
124 .unwrap_or_default();
125 let post_purge_final = post_purge || config.is_some();
127 self.reassemble_plain(
128 &file_path,
129 file_extension,
130 post_purge_final,
131 &base_segments,
132 sidecar_specs,
133 )
134 .await
135 }
136
137 fn reassemble_multi_level_segment<'a>(
162 &'a self,
163 segment_path: &'a Path,
164 rule: &'a MultiLevelRule,
165 nested_rules: &'a [MultiLevelRule],
166 ) -> SegmentFuture<'a> {
167 let segment_path = segment_path.to_path_buf();
168 let rule = rule.clone();
169 let nested_rules = nested_rules.to_vec();
170 Box::pin(async move {
171 self.reassemble_multi_level_segment_inner(&segment_path, &rule, &nested_rules)
172 .await
173 })
174 }
175
176 async fn reassemble_multi_level_segment_inner(
177 &self,
178 segment_path: &Path,
179 rule: &MultiLevelRule,
180 nested_rules: &[MultiLevelRule],
181 ) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
182 if !segment_path.is_dir() {
183 return Ok(());
184 }
185 let mut entries = Vec::new();
186 let mut read_dir = fs::read_dir(segment_path).await?;
187 while let Some(entry) = read_dir.next_entry().await? {
188 entries.push(entry);
189 }
190 entries.sort_by_key(|e| e.file_name());
191 for entry in entries {
192 let process_path = entry.path();
193 if !process_path.is_dir() {
194 continue;
195 }
196 let process_path_str = normalize_path_unix(&process_path.to_string_lossy());
197 let mut sub_entries = Vec::new();
198 let mut sub_read = fs::read_dir(&process_path).await?;
199 while let Some(e) = sub_read.next_entry().await? {
200 sub_entries.push(e);
201 }
202 sub_entries.sort_by_key(|e| e.file_name());
203
204 let mut handled: HashSet<OsString> = HashSet::new();
207 for sub_entry in &sub_entries {
208 let sub_path: PathBuf = sub_entry.path();
209 if !sub_path.is_dir() {
210 continue;
211 }
212 let sub_name = sub_path.file_name().and_then(|n| n.to_str()).unwrap_or("");
213 let Some(nested_rule) = nested_rules.iter().find(|r| r.path_segment == sub_name)
214 else {
215 continue;
216 };
217 let deeper = deeper_candidate_rules(nested_rules, &nested_rule.path_segment);
221 self.reassemble_multi_level_segment(&sub_path, nested_rule, &deeper)
222 .await?;
223 handled.insert(sub_entry.file_name());
224 }
225
226 for sub_entry in &sub_entries {
229 let sub_path = sub_entry.path();
230 if !sub_path.is_dir() {
231 continue;
232 }
233 if handled.contains(&sub_entry.file_name()) {
234 continue;
235 }
236 let sub_path_str = normalize_path_unix(&sub_path.to_string_lossy());
237 self.reassemble_plain(&sub_path_str, Some("xml"), true, &[], None)
238 .await?;
239 }
240
241 self.reassemble_plain(&process_path_str, Some("xml"), true, &[], None)
243 .await?;
244 }
245 ensure_segment_files_structure(
246 segment_path,
247 &rule.wrap_root_element,
248 &rule.path_segment,
249 &rule.wrap_xmlns,
250 )
251 .await?;
252 Ok(())
253 }
254
255 async fn reassemble_plain(
263 &self,
264 file_path: &str,
265 file_extension: Option<&str>,
266 post_purge: bool,
267 base_segments: &[(String, String, bool)],
268 sidecar_specs: Option<&[SidecarSpec]>,
269 ) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
270 let file_path = normalize_path_unix(file_path);
271 log::debug!("Parsing directory to reassemble: {}", file_path);
272 let parsed_objects = self
273 .process_files_in_directory(file_path.to_string(), base_segments.to_vec())
274 .await?;
275
276 if parsed_objects.is_empty() {
277 log::error!(
278 "No files under {} were parsed successfully. A reassembled XML file was not created.",
279 file_path
280 );
281 return Ok(());
282 }
283
284 let Some(mut merged) = merge_xml_elements(&parsed_objects) else {
288 log::error!(
289 "No usable root element found while merging files under {}. A reassembled XML file was not created.",
290 file_path
291 );
292 return Ok(());
293 };
294
295 let auto_specs: Vec<crate::xml::types::SidecarSpec>;
298 let effective_specs: Option<&[crate::xml::types::SidecarSpec]> =
299 if sidecar_specs.is_some_and(|s| !s.is_empty()) {
300 sidecar_specs
301 } else {
302 let meta_path = Path::new(&file_path).join(".sidecars.json");
303 if let Ok(content) = fs::read_to_string(&meta_path).await {
304 if let Ok(parsed) =
305 serde_json::from_str::<Vec<crate::xml::types::SidecarSpec>>(&content)
306 {
307 auto_specs = parsed;
308 Some(auto_specs.as_slice())
309 } else {
310 None
311 }
312 } else {
313 None
314 }
315 };
316
317 if let Some(specs) = effective_specs {
320 inject_sidecar_elements(&file_path, &mut merged, specs).await?;
321 }
322
323 let key_order_path = Path::new(&file_path).join(".key_order.json");
325 if let Some(reordered) = read_key_order(&key_order_path)
326 .await
327 .and_then(|order| reorder_root_keys(&merged, &order))
328 {
329 merged = reordered;
330 }
331
332 let final_xml = build_xml_string(&merged);
333 let output_path = self.get_output_path(&file_path, file_extension);
334
335 fs::write(&output_path, &final_xml).await?;
336
337 if post_purge {
341 if let Some(specs) = effective_specs {
342 let path = Path::new(&file_path);
343 let base = path
344 .file_name()
345 .and_then(|n| n.to_str())
346 .unwrap_or("output");
347 for spec in specs {
348 let sidecar = path.join(format!("{}.{}", base, spec.extension));
349 fs::remove_file(&sidecar).await.ok();
350 }
351 }
352 fs::remove_dir_all(file_path).await.ok();
353 }
354
355 Ok(())
356 }
357
358 fn process_files_in_directory<'a>(
359 &'a self,
360 dir_path: String,
361 base_segments: Vec<(String, String, bool)>,
362 ) -> ProcessDirFuture<'a> {
363 Box::pin(async move {
364 let mut parsed = Vec::new();
365 let mut entries = Vec::new();
366 let mut read_dir = fs::read_dir(&dir_path).await?;
367 while let Some(entry) = read_dir.next_entry().await? {
368 entries.push(entry);
369 }
370 entries.sort_by(|a, b| {
372 let a_name = a.file_name().to_string_lossy().to_string();
373 let b_name = b.file_name().to_string_lossy().to_string();
374 a_name.cmp(&b_name)
375 });
376
377 let is_base = is_at_base_path(&dir_path, &base_segments);
382
383 for entry in entries {
384 let path = entry.path();
385 let file_path = normalize_path_unix(&path.to_string_lossy()).to_string();
386
387 if path.is_file() {
388 let name = path.file_name().and_then(|n| n.to_str()).unwrap_or("");
389 if !name.starts_with('.') && self.is_parsable_file(name) {
390 if let Some(parsed_obj) = parse_to_xml_object(&file_path).await {
391 parsed.push(parsed_obj);
392 }
393 }
394 } else {
395 let dir_name = path.file_name().and_then(|n| n.to_str()).unwrap_or("");
398 let matched_segment = if is_base {
399 base_segments
400 .iter()
401 .find(|(_, seg_name, _)| seg_name == dir_name)
402 .cloned()
403 } else {
404 None
405 };
406 if let Some((_, segment_name, extract_inner)) = matched_segment {
407 let segment_element = self
408 .collect_segment_as_array(&file_path, &segment_name, extract_inner)
409 .await?;
410 if let Some(el) = segment_element {
411 parsed.push(el);
412 }
413 } else {
414 let sub_parsed = self
415 .process_files_in_directory(file_path, base_segments.clone())
416 .await?;
417 parsed.extend(sub_parsed);
418 }
419 }
420 }
421
422 Ok(parsed)
423 })
424 }
425
426 async fn collect_segment_as_array(
430 &self,
431 segment_dir: &str,
432 segment_name: &str,
433 extract_inner: bool,
434 ) -> Result<Option<XmlElement>, Box<dyn std::error::Error + Send + Sync>> {
435 let mut xml_files = Vec::new();
436 let mut read_dir = fs::read_dir(segment_dir).await?;
437 while let Some(entry) = read_dir.next_entry().await? {
438 let path = entry.path();
439 let name = path.file_name().and_then(|n| n.to_str()).unwrap_or("");
440 if path.is_file() && !name.starts_with('.') && self.is_parsable_file(name) {
441 xml_files.push(normalize_path_unix(&path.to_string_lossy()));
442 }
443 }
444 xml_files.sort();
445
446 let mut root_contents = Vec::new();
447 let mut first_xml: Option<(String, Option<Value>)> = None;
448 for file_path in &xml_files {
449 let Some(parsed) = parse_to_xml_object(file_path).await else {
452 continue;
453 };
454 let obj_owned = parsed.as_object().cloned().unwrap_or_default();
455 let obj = &obj_owned;
456 let Some(root_key) = obj.keys().find(|k| *k != "?xml").cloned() else {
457 continue;
458 };
459 let root_val = obj
460 .get(&root_key)
461 .cloned()
462 .unwrap_or(Value::Object(serde_json::Map::new()));
463 let mut content = if extract_inner {
464 root_val
465 .get(segment_name)
466 .cloned()
467 .unwrap_or_else(|| Value::Object(serde_json::Map::new()))
468 } else {
469 root_val
470 };
471 if extract_inner {
473 content = strip_xmlns_from_value(content);
474 }
475 root_contents.push(content);
476 if first_xml.is_none() {
477 first_xml = Some((root_key, obj.get("?xml").cloned()));
478 }
479 }
480 if root_contents.is_empty() {
481 return Ok(None);
482 }
483 let (root_key, decl_opt) = first_xml.unwrap();
484 let mut content = serde_json::Map::new();
485 content.insert(segment_name.to_string(), Value::Array(root_contents));
486 let mut top = serde_json::Map::new();
487 if let Some(decl) = decl_opt {
488 top.insert("?xml".to_string(), decl);
489 } else {
490 let mut d = serde_json::Map::new();
491 d.insert("@version".to_string(), Value::String("1.0".to_string()));
492 d.insert("@encoding".to_string(), Value::String("UTF-8".to_string()));
493 top.insert("?xml".to_string(), Value::Object(d));
494 }
495 top.insert(root_key, Value::Object(content));
496 Ok(Some(Value::Object(top)))
497 }
498
499 fn is_parsable_file(&self, file_name: &str) -> bool {
500 let lower = file_name.to_lowercase();
501 lower.ends_with(".xml")
502 || lower.ends_with(".json")
503 || lower.ends_with(".json5")
504 || lower.ends_with(".yaml")
505 || lower.ends_with(".yml")
506 }
507
508 async fn validate_directory(
509 &self,
510 path: &str,
511 ) -> Result<bool, Box<dyn std::error::Error + Send + Sync>> {
512 let meta = fs::metadata(path).await?;
513 if !meta.is_dir() {
514 log::error!(
515 "The provided path to reassemble is not a directory: {}",
516 path
517 );
518 return Ok(false);
519 }
520 Ok(true)
521 }
522
523 fn get_output_path(&self, dir_path: &str, extension: Option<&str>) -> String {
524 let path = Path::new(dir_path);
525 let parent = path.parent().unwrap_or(Path::new("."));
526 let base_name = path
527 .file_name()
528 .and_then(|n| n.to_str())
529 .unwrap_or("output");
530 let ext = extension.unwrap_or("xml");
531 parent
532 .join(format!("{}.{}", base_name, ext))
533 .to_string_lossy()
534 .to_string()
535 }
536}
537
538impl Default for ReassembleXmlFileHandler {
539 fn default() -> Self {
540 Self::new()
541 }
542}
543
544async fn inject_sidecar_elements(
555 dir_path: &str,
556 merged: &mut XmlElement,
557 specs: &[SidecarSpec],
558) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
559 let path = Path::new(dir_path);
560 let base = path
561 .file_name()
562 .and_then(|n| n.to_str())
563 .unwrap_or("output");
564
565 let root_key = merged
566 .as_object()
567 .and_then(|o| o.keys().find(|k| *k != "?xml").cloned());
568 let Some(root_key) = root_key else {
569 return Ok(());
570 };
571
572 if let Some(root_val) = merged.as_object_mut().and_then(|o| o.get_mut(&root_key)) {
573 if let Some(root_obj) = root_val.as_object_mut() {
574 for spec in specs {
575 let sidecar_path = path.join(format!("{}.{}", base, spec.extension));
576 let Ok(content) = fs::read_to_string(&sidecar_path).await else {
577 continue;
578 };
579 root_obj.insert(
580 spec.element.clone(),
581 serde_json::json!({ "#raw-text": content }),
582 );
583 }
584 }
585 }
586
587 Ok(())
588}
589
590#[cfg(test)]
591mod tests {
592 use super::*;
593 use serde_json::json;
594
595 #[test]
596 #[allow(clippy::default_constructed_unit_structs)]
597 fn reassemble_handler_default_equals_new() {
598 let _ = ReassembleXmlFileHandler::default();
599 }
600
601 #[test]
602 fn strip_xmlns_from_value_passes_non_object_through() {
603 let s = Value::String("hello".to_string());
604 assert_eq!(
605 strip_xmlns_from_value(s),
606 Value::String("hello".to_string())
607 );
608 let arr = json!([1, 2]);
609 assert_eq!(strip_xmlns_from_value(arr.clone()), arr);
610 }
611
612 #[test]
613 fn strip_xmlns_from_value_removes_xmlns_key() {
614 let obj = json!({ "@xmlns": "ns", "child": 1 });
615 let stripped = strip_xmlns_from_value(obj);
616 let map = stripped.as_object().unwrap();
617 assert!(map.get("@xmlns").is_none());
618 assert_eq!(map.get("child").and_then(|v| v.as_i64()), Some(1));
619 }
620
621 #[test]
622 fn is_parsable_file_recognises_supported_extensions() {
623 let h = ReassembleXmlFileHandler::new();
624 assert!(h.is_parsable_file("a.xml"));
625 assert!(h.is_parsable_file("a.json"));
626 assert!(h.is_parsable_file("a.json5"));
627 assert!(h.is_parsable_file("a.yaml"));
628 assert!(h.is_parsable_file("a.yml"));
629 assert!(h.is_parsable_file("A.XML"));
630 assert!(!h.is_parsable_file("a.txt"));
631 }
632
633 #[test]
634 fn get_output_path_appends_extension_and_uses_parent_dir() {
635 let h = ReassembleXmlFileHandler::new();
636 let out = h.get_output_path("/tmp/foo", Some("xml"));
637 assert!(out.ends_with("foo.xml"));
638 let out_default = h.get_output_path("/tmp/bar", None);
639 assert!(out_default.ends_with("bar.xml"));
640 assert_eq!(h.get_output_path("only", Some("json")), "only.json");
642 }
643
644 #[tokio::test]
645 async fn reassemble_multi_level_segment_noop_when_not_dir() {
646 let h = ReassembleXmlFileHandler::new();
647 let tmp = tempfile::tempdir().unwrap();
648 let file = tmp.path().join("not_a_dir.txt");
649 tokio::fs::write(&file, "hi").await.unwrap();
650 let rule = crate::xml::types::MultiLevelRule {
651 file_pattern: String::new(),
652 root_to_strip: String::new(),
653 unique_id_elements: String::new(),
654 path_segment: String::new(),
655 wrap_root_element: "Root".to_string(),
656 wrap_xmlns: String::new(),
657 };
658 h.reassemble_multi_level_segment(&file, &rule, &[])
659 .await
660 .unwrap();
661 }
662
663 #[tokio::test]
664 async fn reassemble_multi_level_segment_skips_files_in_segment_root() {
665 let h = ReassembleXmlFileHandler::new();
666 let tmp = tempfile::tempdir().unwrap();
667 let segment = tmp.path().join("segment");
668 tokio::fs::create_dir(&segment).await.unwrap();
669 tokio::fs::write(segment.join("stray.txt"), "x")
671 .await
672 .unwrap();
673 let rule = crate::xml::types::MultiLevelRule {
674 file_pattern: String::new(),
675 root_to_strip: String::new(),
676 unique_id_elements: String::new(),
677 path_segment: "segment".to_string(),
678 wrap_root_element: "Root".to_string(),
679 wrap_xmlns: "http://example.com".to_string(),
680 };
681 h.reassemble_multi_level_segment(&segment, &rule, &[])
682 .await
683 .unwrap();
684 }
685
686 #[tokio::test]
687 async fn collect_segment_as_array_returns_none_for_empty_dir() {
688 let h = ReassembleXmlFileHandler::new();
689 let tmp = tempfile::tempdir().unwrap();
690 let out = h
691 .collect_segment_as_array(tmp.path().to_str().unwrap(), "seg", true)
692 .await
693 .unwrap();
694 assert!(out.is_none());
695 }
696
697 #[tokio::test]
698 async fn collect_segment_as_array_skips_unparseable_and_empty_roots() {
699 let h = ReassembleXmlFileHandler::new();
700 let tmp = tempfile::tempdir().unwrap();
701 tokio::fs::write(tmp.path().join("bad.xml"), "<<")
703 .await
704 .unwrap();
705 tokio::fs::write(tmp.path().join("only-decl.xml"), "")
707 .await
708 .unwrap();
709 tokio::fs::write(tmp.path().join(".hidden.xml"), "<r/>")
711 .await
712 .unwrap();
713 let out = h
714 .collect_segment_as_array(tmp.path().to_str().unwrap(), "seg", false)
715 .await
716 .unwrap();
717 assert!(out.is_none());
718 }
719
720 #[tokio::test]
721 async fn collect_segment_as_array_without_xml_decl_inserts_default_decl() {
722 let h = ReassembleXmlFileHandler::new();
726 let tmp = tempfile::tempdir().unwrap();
727 tokio::fs::write(
728 tmp.path().join("a.xml"),
729 r#"<Root><seg><x>1</x></seg></Root>"#,
730 )
731 .await
732 .unwrap();
733 let out = h
734 .collect_segment_as_array(tmp.path().to_str().unwrap(), "seg", true)
735 .await
736 .unwrap()
737 .unwrap();
738 let obj = out.as_object().unwrap();
739 let decl = obj
740 .get("?xml")
741 .and_then(|v| v.as_object())
742 .expect("default declaration must be inserted when XML has none");
743 assert_eq!(decl.get("@version").and_then(|v| v.as_str()), Some("1.0"));
744 assert_eq!(
745 decl.get("@encoding").and_then(|v| v.as_str()),
746 Some("UTF-8")
747 );
748 }
749
750 #[tokio::test]
751 async fn collect_segment_as_array_with_xml_decl_preserves_it() {
752 let h = ReassembleXmlFileHandler::new();
755 let tmp = tempfile::tempdir().unwrap();
756 tokio::fs::write(
757 tmp.path().join("a.xml"),
758 r#"<?xml version="1.0" encoding="UTF-8"?><Root><seg><x>1</x></seg></Root>"#,
759 )
760 .await
761 .unwrap();
762 let out = h
763 .collect_segment_as_array(tmp.path().to_str().unwrap(), "seg", true)
764 .await
765 .unwrap()
766 .unwrap();
767 let obj = out.as_object().unwrap();
768 let decl = obj
769 .get("?xml")
770 .and_then(|v| v.as_object())
771 .expect("?xml declaration must be preserved from source");
772 assert_eq!(decl.get("@version").and_then(|v| v.as_str()), Some("1.0"));
773 assert_eq!(
774 decl.get("@encoding").and_then(|v| v.as_str()),
775 Some("UTF-8")
776 );
777 }
778
779 #[tokio::test]
780 async fn collect_segment_as_array_without_extract_inner_wraps_root() {
781 let h = ReassembleXmlFileHandler::new();
782 let tmp = tempfile::tempdir().unwrap();
783 tokio::fs::write(tmp.path().join("a.xml"), r#"<Root><child>1</child></Root>"#)
784 .await
785 .unwrap();
786 let out = h
787 .collect_segment_as_array(tmp.path().to_str().unwrap(), "seg", false)
788 .await
789 .unwrap()
790 .unwrap();
791 let obj = out.as_object().unwrap();
792 assert!(obj.contains_key("?xml"));
793 let root = obj.get("Root").and_then(|r| r.as_object()).unwrap();
794 assert!(root.get("seg").and_then(|v| v.as_array()).is_some());
795 }
796
797 fn rule_with_segment(segment: &str) -> MultiLevelRule {
798 MultiLevelRule {
799 file_pattern: String::new(),
800 root_to_strip: String::new(),
801 unique_id_elements: String::new(),
802 path_segment: segment.to_string(),
803 wrap_root_element: String::new(),
804 wrap_xmlns: String::new(),
805 }
806 }
807
808 #[test]
809 fn deeper_candidate_rules_excludes_the_matched_segment() {
810 let rules = vec![rule_with_segment("seg_a"), rule_with_segment("seg_b")];
814 let deeper = deeper_candidate_rules(&rules, "seg_a");
815 assert_eq!(deeper.len(), 1);
816 assert_eq!(deeper[0].path_segment, "seg_b");
817 }
818
819 #[test]
820 fn deeper_candidate_rules_keeps_all_when_no_segment_matches() {
821 let rules = vec![rule_with_segment("seg_a"), rule_with_segment("seg_b")];
825 let deeper = deeper_candidate_rules(&rules, "missing");
826 assert_eq!(deeper.len(), 2);
827 }
828
829 #[test]
830 fn deeper_candidate_rules_returns_empty_for_empty_input() {
831 let deeper: Vec<MultiLevelRule> = deeper_candidate_rules(&[], "anything");
832 assert!(deeper.is_empty());
833 }
834
835 #[test]
836 fn is_at_base_path_true_when_dir_matches_any_segment() {
837 let segs = vec![
838 ("/base/other".to_string(), "seg1".to_string(), false),
839 ("/base/here".to_string(), "seg2".to_string(), false),
840 ];
841 assert!(is_at_base_path("/base/here", &segs));
842 }
843
844 #[test]
845 fn is_at_base_path_false_when_dir_matches_nothing() {
846 let segs = vec![("/base/a".to_string(), "seg".to_string(), false)];
847 assert!(!is_at_base_path("/base/b", &segs));
848 }
849
850 #[test]
851 fn is_at_base_path_false_for_empty_segments() {
852 let segs: Vec<(String, String, bool)> = Vec::new();
853 assert!(!is_at_base_path("/anywhere", &segs));
854 }
855
856 #[tokio::test]
861 async fn reassemble_plain_some_empty_sidecar_specs_falls_through_to_auto_detect() {
862 let h = ReassembleXmlFileHandler::new();
863 let tmp = tempfile::tempdir().unwrap();
864 let dir = tmp.path().join("mydir");
865 tokio::fs::create_dir(&dir).await.unwrap();
866
867 tokio::fs::write(
868 dir.join("a.xml"),
869 r#"<?xml version="1.0" encoding="UTF-8"?><Root><Child>hello</Child></Root>"#,
870 )
871 .await
872 .unwrap();
873
874 tokio::fs::write(
875 dir.join(".sidecars.json"),
876 r#"[{"element":"Notes","extension":"yaml"}]"#,
877 )
878 .await
879 .unwrap();
880
881 tokio::fs::write(dir.join("mydir.yaml"), "key: value")
883 .await
884 .unwrap();
885
886 h.reassemble_plain(dir.to_str().unwrap(), Some("xml"), false, &[], Some(&[]))
887 .await
888 .unwrap();
889
890 let output = tokio::fs::read_to_string(tmp.path().join("mydir.xml"))
891 .await
892 .unwrap();
893 assert!(
894 output.contains("key: value"),
895 "sidecar content missing — auto-detect did not run:\n{output}"
896 );
897 }
898}