compare_dir/
dir_comparer.rs

1use crate::{
2    Classification, ColumnFormatter, FileComparer, FileComparisonResult, FileHasher, FileItem,
3    FileIterator, OutputFormat, Progress, ProgressBuilder, ProgressValue,
4};
5use globset::GlobSet;
6use indicatif::FormattedDuration;
7use std::{
8    cmp::Ordering,
9    io::{self, stdout},
10    path::{Path, PathBuf},
11    sync::{Arc, mpsc},
12    time,
13};
14
15#[derive(Debug, Clone)]
16enum CompareProgress {
17    StartOfComparison,
18    Progress(ProgressValue),
19    Total(ProgressValue),
20    Result(usize, FileComparisonResult),
21    Error,
22}
23
24/// Methods for comparing files.
25#[derive(Debug, Clone, Copy, PartialEq, Eq)]
26pub enum FileComparisonMethod {
27    /// Compare only size and modification time.
28    Size,
29    /// Compare by hash (BLAKE3).
30    Hash,
31    /// Compare by hash, without using the cached hashes.
32    Rehash,
33    /// Compare byte-by-byte.
34    Full,
35}
36
37/// A tool for comparing the contents of two directories.
38pub struct DirectoryComparer {
39    dir1: PathBuf,
40    dir2: PathBuf,
41    pub output_format: OutputFormat,
42    pub buffer_size: usize,
43    pub comparison_method: FileComparisonMethod,
44    pub exclude: Option<GlobSet>,
45    pub progress: Option<Arc<ProgressBuilder>>,
46    pub jobs: usize,
47}
48
49impl DirectoryComparer {
50    pub const DEFAULT_JOBS: usize = 8;
51
52    /// Creates a new `DirectoryComparer` for the two given directories.
53    pub fn new(dir1: PathBuf, dir2: PathBuf) -> Self {
54        Self {
55            dir1,
56            dir2,
57            output_format: OutputFormat::Default,
58            buffer_size: FileComparer::DEFAULT_BUFFER_SIZE,
59            comparison_method: FileComparisonMethod::Hash,
60            exclude: None,
61            progress: None,
62            jobs: Self::DEFAULT_JOBS,
63        }
64    }
65
66    /// Executes the directory comparison and prints results to stdout.
67    /// This is a convenience method for CLI usage.
68    pub fn run(&self) -> anyhow::Result<()> {
69        match self.output_format {
70            OutputFormat::Default | OutputFormat::Symbol => {}
71            _ => anyhow::bail!("Compare mode only supports default or symbol output format."),
72        }
73        if self.dir1.is_file() {
74            return self.run_file_comparer();
75        }
76
77        let mut progress = self
78            .progress
79            .as_ref()
80            .map(|progress| progress.add_spinner())
81            .unwrap_or_else(Progress::none);
82        progress.set_message("Scanning directories...");
83        let start_time = std::time::Instant::now();
84        let mut summary = ComparisonSummary::default();
85        let dir1_str = self.dir1.to_str().unwrap_or("dir1");
86        let dir2_str = self.dir2.to_str().unwrap_or("dir2");
87        let (tx, rx) = mpsc::channel();
88        std::thread::scope(|scope| {
89            scope.spawn(move || {
90                if let Err(e) = self.compare_streaming_ordered(tx) {
91                    log::error!("Error during comparison: {}", e);
92                }
93            });
94
95            // Receive results and update summary/UI
96            while let Ok(event) = rx.recv() {
97                match event {
98                    CompareProgress::StartOfComparison => {
99                        progress.set_message("Comparing files...");
100                    }
101                    CompareProgress::Total(total) => {
102                        progress.set_length(total);
103                        progress.set_message("");
104                    }
105                    CompareProgress::Result(_, result) => {
106                        summary.update(&result);
107                        if !(self.output_format == OutputFormat::Default && result.is_identical()) {
108                            progress.suspend_for(stdout(), || {
109                                result.print(self.output_format, dir1_str, dir2_str)
110                            });
111                        }
112                    }
113                    CompareProgress::Progress(value) => progress.inc(value),
114                    CompareProgress::Error => summary.num_errors += 1,
115                }
116            }
117        });
118        progress.finish();
119        eprintln!("\n--- Comparison Summary ---");
120        summary.print(&mut io::stderr(), &start_time, dir1_str, dir2_str)?;
121        Ok(())
122    }
123
124    /// Performs the directory comparison and streams results via a channel.
125    ///
126    /// # Arguments
127    /// * `tx` - A sender to transmit `FileComparisonResult` as they are computed.
128    fn compare_streaming_ordered(&self, tx: mpsc::Sender<CompareProgress>) -> anyhow::Result<()> {
129        crate::sort_stream(
130            tx,
131            |tx_unordered| self.compare_streaming(tx_unordered),
132            |event| match event {
133                CompareProgress::Result(i, _) => Some(*i),
134                _ => None,
135            },
136        )
137    }
138
139    fn compare_streaming(&self, tx: mpsc::Sender<CompareProgress>) -> anyhow::Result<()> {
140        let mut it1 = FileIterator::new(&self.dir1);
141        let mut it2 = FileIterator::new(&self.dir2);
142        it1.exclude = self.exclude.as_ref();
143        it2.exclude = self.exclude.as_ref();
144        let mut hashers = self.get_hashers(&self.dir1, &self.dir2)?;
145        if let Some((h1, h2)) = &mut hashers {
146            it1.cache = Some(h1.cache()?);
147            it2.cache = Some(h2.cache()?);
148            if self.comparison_method == FileComparisonMethod::Rehash {
149                h1.clear_cache()?;
150                h2.clear_cache()?;
151            }
152        }
153        let hashers_ref = hashers.as_ref();
154        std::thread::scope(|global_scope| {
155            let it1_rx = it1.spawn_in_scope(global_scope);
156            let it2_rx = it2.spawn_in_scope(global_scope);
157            let pool = crate::build_thread_pool(self.jobs)?;
158            pool.scope(move |scope| {
159                let mut cur1 = it1_rx.recv().ok();
160                let mut cur2 = it2_rx.recv().ok();
161                let mut index = 0;
162                let mut total = ProgressValue::default();
163                tx.send(CompareProgress::StartOfComparison)?;
164                loop {
165                    let cmp = match (&cur1, &cur2) {
166                        (Some(f1), Some(f2)) => {
167                            let rel1 = f1.relative_path(&self.dir1);
168                            let rel2 = f2.relative_path(&self.dir2);
169                            rel1.cmp(rel2)
170                        }
171                        (Some(_), None) => Ordering::Less,
172                        (None, Some(_)) => Ordering::Greater,
173                        (None, None) => break,
174                    };
175                    match cmp {
176                        Ordering::Less => {
177                            let file1 = cur1.take().unwrap();
178                            let rel1 = file1.relative_path(&self.dir1);
179                            let size = file1.size();
180                            total += ProgressValue::with_size(size);
181                            let result =
182                                FileComparisonResult::new(rel1.into(), Classification::OnlyInDir1);
183                            tx.send(CompareProgress::Result(index, result))?;
184                            tx.send(CompareProgress::Progress(ProgressValue::with_size(size)))?;
185                            index += 1;
186                            cur1 = it1_rx.recv().ok();
187                        }
188                        Ordering::Greater => {
189                            let file2 = cur2.take().unwrap();
190                            let rel2 = file2.relative_path(&self.dir2);
191                            let size = file2.size();
192                            total += ProgressValue::with_size(size);
193                            let result =
194                                FileComparisonResult::new(rel2.into(), Classification::OnlyInDir2);
195                            tx.send(CompareProgress::Result(index, result))?;
196                            tx.send(CompareProgress::Progress(ProgressValue::with_size(size)))?;
197                            index += 1;
198                            cur2 = it2_rx.recv().ok();
199                        }
200                        Ordering::Equal => {
201                            let file1 = cur1.take().unwrap();
202                            let file2 = cur2.take().unwrap();
203                            let buffer_size = self.buffer_size;
204                            let tx_clone = tx.clone();
205                            let i = index;
206                            let should_compare =
207                                self.comparison_method != FileComparisonMethod::Size;
208                            let size = file1.size();
209                            total += ProgressValue::with_size(size);
210                            scope.spawn(move |_| {
211                                let mut comparer = FileComparer::new(&file1, &file2);
212                                comparer.buffer_size = buffer_size;
213                                if let Some((h1, h2)) = hashers_ref {
214                                    comparer.hashers = Some((h1, h2));
215                                }
216                                let rel_path = file1.relative_path(&self.dir1);
217                                let mut result = FileComparisonResult::new(
218                                    rel_path.into(),
219                                    Classification::InBoth,
220                                );
221                                let event = match result.update(&comparer, should_compare) {
222                                    Ok(_) => CompareProgress::Result(i, result),
223                                    Err(error) => {
224                                        log::error!(
225                                            "Error comparing '{}': {}",
226                                            result.relative_path.display(),
227                                            error
228                                        );
229                                        CompareProgress::Error
230                                    }
231                                };
232                                if tx_clone.send(event).is_err()
233                                    || tx_clone
234                                        .send(CompareProgress::Progress(ProgressValue::with_size(
235                                            size,
236                                        )))
237                                        .is_err()
238                                {
239                                    log::error!("Send failed");
240                                }
241                            });
242                            index += 1;
243                            cur1 = it1_rx.recv().ok();
244                            cur2 = it2_rx.recv().ok();
245                        }
246                    }
247                }
248                tx.send(CompareProgress::Total(total))
249            })?;
250            Ok::<(), anyhow::Error>(())
251        })?;
252
253        Self::save_hashers(hashers)?;
254        Ok(())
255    }
256
257    fn get_hashers(
258        &self,
259        dir1: &Path,
260        dir2: &Path,
261    ) -> anyhow::Result<Option<(FileHasher, FileHasher)>> {
262        if self.comparison_method == FileComparisonMethod::Hash
263            || self.comparison_method == FileComparisonMethod::Rehash
264        {
265            let (h1_res, h2_res) = rayon::join(
266                || FileHasher::new_with_cache(&[dir1]),
267                || FileHasher::new_with_cache(&[dir2]),
268            );
269            let mut h1 = h1_res?;
270            let mut h2 = h2_res?;
271            h1.buffer_size = self.buffer_size;
272            h2.buffer_size = self.buffer_size;
273            if let Some(progress) = self.progress.as_ref() {
274                h1.progress = Some(Arc::clone(progress));
275                h2.progress = Some(Arc::clone(progress));
276            }
277            return Ok(Some((h1, h2)));
278        }
279        Ok(None)
280    }
281
282    fn save_hashers(hashers: Option<(FileHasher, FileHasher)>) -> anyhow::Result<()> {
283        if let Some((h1, h2)) = hashers {
284            let (r1, r2) = rayon::join(|| h1.save_cache(), || h2.save_cache());
285            r1?;
286            r2?;
287        }
288        Ok(())
289    }
290
291    fn run_file_comparer(&self) -> anyhow::Result<()> {
292        assert!(self.dir1.is_file());
293        let file1_path = &self.dir1;
294        let dir1 = file1_path.parent().unwrap();
295        let file1_name = file1_path.file_name().unwrap();
296        let (dir2, file2_path) = if self.dir2.is_file() {
297            (self.dir2.parent().unwrap(), self.dir2.clone())
298        } else {
299            (self.dir2.as_path(), self.dir2.join(file1_name))
300        };
301        let file1 = FileItem::try_from(file1_path.as_path())?;
302        let file2 = FileItem::try_from(file2_path.as_path())?;
303        let mut comparer = FileComparer::new(&file1, &file2);
304        comparer.buffer_size = self.buffer_size;
305        let mut hashers = self.get_hashers(dir1, dir2)?;
306        if let Some((h1, h2)) = &mut hashers {
307            if self.comparison_method == FileComparisonMethod::Rehash {
308                h1.remove_cache_entry(file1_path)?;
309                h2.remove_cache_entry(&file2_path)?;
310            }
311            comparer.hashers = Some((h1, h2));
312        }
313        let mut result = FileComparisonResult::new(PathBuf::new(), Classification::InBoth);
314        let should_compare_content = self.comparison_method != FileComparisonMethod::Size;
315        result.update(&comparer, should_compare_content)?;
316        let file1_str = file1_path.to_str().unwrap_or("file1");
317        let file2_str = file2_path.to_str().unwrap_or("file2");
318        result.print(self.output_format, file1_str, file2_str);
319        Self::save_hashers(hashers)?;
320        Ok(())
321    }
322}
323
324#[derive(Default)]
325struct ComparisonSummary {
326    pub in_both: usize,
327    pub only_in_dir1: usize,
328    pub only_in_dir2: usize,
329    pub dir1_newer: usize,
330    pub dir2_newer: usize,
331    pub dir1_larger: usize,
332    pub dir2_larger: usize,
333    pub diff_content: usize,
334    pub not_comparable: usize,
335    pub num_errors: usize,
336}
337
338impl ComparisonSummary {
339    pub fn update(&mut self, result: &FileComparisonResult) {
340        match result.classification {
341            Classification::OnlyInDir1 => self.only_in_dir1 += 1,
342            Classification::OnlyInDir2 => self.only_in_dir2 += 1,
343            Classification::InBoth => {
344                self.in_both += 1;
345                let mut is_not_comparable = false;
346                match result.modified_time_comparison {
347                    Some(Ordering::Greater) => self.dir1_newer += 1,
348                    Some(Ordering::Less) => self.dir2_newer += 1,
349                    Some(Ordering::Equal) => {}
350                    None => is_not_comparable = true,
351                }
352                match result.size_comparison {
353                    Some(Ordering::Greater) => self.dir1_larger += 1,
354                    Some(Ordering::Less) => self.dir2_larger += 1,
355                    Some(Ordering::Equal) => match result.is_content_same {
356                        Some(false) => self.diff_content += 1,
357                        Some(true) => {}
358                        None => is_not_comparable = true,
359                    },
360                    None => is_not_comparable = true,
361                }
362                if is_not_comparable {
363                    self.not_comparable += 1;
364                }
365            }
366        }
367    }
368
369    pub fn print(
370        &self,
371        mut writer: impl std::io::Write,
372        start_time: &time::Instant,
373        dir1_name: &str,
374        dir2_name: &str,
375    ) -> std::io::Result<()> {
376        let values = [
377            ("Elapsed:", 0),
378            ("Files in both:", self.in_both),
379            ("Only in left:", self.only_in_dir1),
380            ("Only in right:", self.only_in_dir2),
381            ("Left is newer:", self.dir1_newer),
382            ("Right is newer:", self.dir2_newer),
383            ("Left is larger:", self.dir1_larger),
384            ("Right is larger:", self.dir2_larger),
385            ("Different content:", self.diff_content),
386            ("Not comparable:", self.not_comparable),
387            ("Errors:", self.num_errors),
388        ];
389        let formatter = ColumnFormatter::new(values.iter().map(|(s, _)| *s));
390        formatter.write_value(&mut writer, "Left:", dir1_name)?;
391        formatter.write_value(&mut writer, "Right:", dir2_name)?;
392        formatter.write_value(
393            &mut writer,
394            values[0].0,
395            FormattedDuration(start_time.elapsed()),
396        )?;
397        formatter.write_values(&mut writer, &values[1..])?;
398        Ok(())
399    }
400}
401
402#[cfg(test)]
403mod tests {
404    use super::*;
405    use std::fs;
406    use std::io::Write;
407
408    #[test]
409    fn comparison_summary() {
410        let mut summary = ComparisonSummary::default();
411        let res1 = FileComparisonResult::new(PathBuf::from("a"), Classification::OnlyInDir1);
412        let res2 = FileComparisonResult::new(PathBuf::from("b"), Classification::OnlyInDir2);
413        let mut res3 = FileComparisonResult::new(PathBuf::from("c"), Classification::InBoth);
414        res3.modified_time_comparison = Some(Ordering::Greater);
415
416        summary.update(&res1);
417        summary.update(&res2);
418        summary.update(&res3);
419
420        assert_eq!(summary.only_in_dir1, 1);
421        assert_eq!(summary.only_in_dir2, 1);
422        assert_eq!(summary.in_both, 1);
423        assert_eq!(summary.dir1_newer, 1);
424    }
425
426    #[test]
427    fn directory_comparer_integration() -> anyhow::Result<()> {
428        let dir1 = tempfile::tempdir()?;
429        let dir2 = tempfile::tempdir()?;
430
431        // Create files in dir1
432        let file1_path = dir1.path().join("same.txt");
433        fs::write(file1_path, b"same content")?;
434
435        let only1_path = dir1.path().join("only1.txt");
436        fs::write(only1_path, b"only in dir1")?;
437
438        // Create files in dir2
439        let file2_path = dir2.path().join("same.txt");
440        fs::write(file2_path, b"same content")?;
441
442        let only2_path = dir2.path().join("only2.txt");
443        fs::write(only2_path, b"only in dir2")?;
444
445        // Create a different file
446        let diff1_path = dir1.path().join("diff.txt");
447        fs::write(diff1_path, b"content 1")?;
448        let diff2_path = dir2.path().join("diff.txt");
449        fs::write(diff2_path, b"content 222")?; // different length and content
450
451        // Same size but different content.
452        let diffc1_path = dir1.path().join("diffc.txt");
453        fs::write(diffc1_path, b"content 111")?;
454        let diffc2_path = dir2.path().join("diffc.txt");
455        fs::write(diffc2_path, b"content 222")?; // different length and content
456
457        let comparer = DirectoryComparer::new(dir1.path().to_path_buf(), dir2.path().to_path_buf());
458        let (tx, rx) = mpsc::channel();
459        comparer.compare_streaming_ordered(tx)?;
460        let mut results = Vec::new();
461        while let Ok(res) = rx.recv() {
462            if let CompareProgress::Result(_, r) = res {
463                results.push(r);
464            }
465        }
466        results.sort_by(|a, b| a.relative_path.cmp(&b.relative_path));
467        assert_eq!(results.len(), 5);
468
469        // diff.txt
470        let diff_result = &results[0];
471        assert_eq!(diff_result.relative_path.to_str().unwrap(), "diff.txt");
472        assert_eq!(diff_result.classification, Classification::InBoth);
473        assert_eq!(diff_result.size_comparison, Some(Ordering::Less));
474        assert_eq!(diff_result.is_content_same, None);
475
476        // diff2.txt
477        let diffc_result = &results[1];
478        assert_eq!(diffc_result.relative_path.to_str().unwrap(), "diffc.txt");
479        assert_eq!(diffc_result.classification, Classification::InBoth);
480        assert_eq!(diffc_result.size_comparison, Some(Ordering::Equal));
481        assert_eq!(diffc_result.is_content_same, Some(false));
482
483        // only1.txt
484        let only1_result = &results[2];
485        assert_eq!(only1_result.relative_path.to_str().unwrap(), "only1.txt");
486        assert_eq!(only1_result.classification, Classification::OnlyInDir1);
487
488        // only2.txt
489        let only2_result = &results[3];
490        assert_eq!(only2_result.relative_path.to_str().unwrap(), "only2.txt");
491        assert_eq!(only2_result.classification, Classification::OnlyInDir2);
492
493        // same.txt
494        let same_result = &results[4];
495        assert_eq!(same_result.relative_path.to_str().unwrap(), "same.txt");
496        assert_eq!(same_result.classification, Classification::InBoth);
497        assert_eq!(same_result.size_comparison, Some(Ordering::Equal));
498
499        Ok(())
500    }
501
502    #[test]
503    fn directory_comparer_size_mode() -> anyhow::Result<()> {
504        let dir1 = tempfile::tempdir()?;
505        let dir2 = tempfile::tempdir()?;
506
507        let file1_path = dir1.path().join("file.txt");
508        let mut file1 = fs::File::create(&file1_path)?;
509        file1.write_all(b"content 1")?;
510
511        let file2_path = dir2.path().join("file.txt");
512        let mut file2 = fs::File::create(&file2_path)?;
513        file2.write_all(b"content 2")?; // same length, different content
514
515        let mut comparer =
516            DirectoryComparer::new(dir1.path().to_path_buf(), dir2.path().to_path_buf());
517        comparer.comparison_method = FileComparisonMethod::Size;
518        let (tx, rx) = mpsc::channel();
519
520        comparer.compare_streaming_ordered(tx)?;
521
522        let mut results = Vec::new();
523        while let Ok(res) = rx.recv() {
524            if let CompareProgress::Result(_, r) = res {
525                results.push(r);
526            }
527        }
528
529        assert_eq!(results.len(), 1);
530        assert_eq!(results[0].relative_path.to_str().unwrap(), "file.txt");
531        assert_eq!(results[0].classification, Classification::InBoth);
532        assert_eq!(results[0].size_comparison, Some(Ordering::Equal));
533        assert_eq!(results[0].is_content_same, None);
534
535        Ok(())
536    }
537}
compare_dir/dir_comparer.rs

compare_dir/
dir_comparer.rs
