Merge branch 'dev' of github.com:hhandika/segul into dev

README.md

@@ -93,15 +93,15 @@ Learn more about using SEGUL API in the [documentation](https://www.segul.app/do
 
 | Feature                        | Quick Link                                                                                                             |
 | ------------------------------ | ---------------------------------------------------------------------------------------------------------------------- |
-| Alignment concatenation        | [CLI](https://www.segul.app/docs/cli-usage/concat) / [GUI](https://www.segul.app/docs/gui-usage/align-concat)          |
+| Alignment concatenation        | [CLI](https://www.segul.app/docs/cli-usage/alignments) / [GUI](https://www.segul.app/docs/gui-usage/align-concat)      |
 | Alignment conversion           | [CLI](https://www.segul.app/docs/cli-usage/convert) / [GUI](https://www.segul.app/docs/gui-usage/align-convert)        |
 | Alignment filtering            | [CLI](https://www.segul.app/docs/cli-usage/filter) / [GUI](https://www.segul.app/docs/gui-usage/align-filter)          |
 | Alignment splitting            | [CLI](https://www.segul.app/docs/cli-usage/split) / [GUI](https://www.segul.app/docs/gui-usage/align-split)            |
 | Alignment partition conversion | [CLI](https://www.segul.app/docs/cli-usage/part) / [GUI](https://www.segul.app/docs/gui-usage/align-partition)         |
 | Alignment summary statistics   | [CLI](https://www.segul.app/docs/cli-usage/summary) / [GUI](https://www.segul.app/docs/gui-usage/align-summary)        |
 | Genomic summary statistics     | [CLI](https://www.segul.app/docs/cli-usage/genomic) / [GUI](https://www.segul.app/docs/gui-usage/genomic)              |
 | Sequence extraction            | [CLI](https://www.segul.app/docs/cli-usage/extract) / [GUI](https://www.segul.app/docs/gui-usage/sequence-extract)     |
-| Sequence filtering             | [CLI](https://www.segul.app/docs/cli-usage/seq_filter) / GUI feature in development       |
+| Sequence filtering             | [CLI](https://www.segul.app/docs/cli-usage/seq_filter) / GUI feature in development                                    |
 | Sequence ID extraction         | [CLI](https://www.segul.app/docs/cli-usage/id) / [GUI](https://www.segul.app/docs/gui-usage/sequence-id)               |
 | Sequence ID mapping            | [CLI](https://www.segul.app/docs/cli-usage/map) / [GUI](https://www.segul.app/docs/gui-usage/sequence-id-map)          |
 | Sequence ID renaming           | [CLI](https://www.segul.app/docs/cli-usage/rename) / [GUI](https://www.segul.app/docs/gui-usage/sequence-rename)       |

src/core/align/unalign.rs

@@ -76,7 +76,7 @@ impl<'a> UnalignAlignment<'a> {
     ///
     /// let input_fmt = InputFmt::Nexus;
     /// let datatype = DataType::Dna;
-    /// let input_dir = Path::new("tests/files/concat");
+    /// let input_dir = Path::new("tests/files/alignments");
     /// // Find matching alignment files in the input directory
     /// let files = SeqFileFinder::new(Path::new(input_dir)).find(&input_fmt);
     /// // Replace the temp directory with your own directory.
@@ -93,10 +93,6 @@ impl<'a> UnalignAlignment<'a> {
     ///     );
     /// handle.unalign();
     pub fn unalign(&self) {
-        if self.output_fmt != &OutputFmt::Fasta || self.output_fmt != &OutputFmt::FastaInt {
-            log::warn!("Unalign feature only supports fasta or fasta-int output format");
-            return;
-        }
         let spin = utils::set_spinner();
         spin.set_message("Converting un-aligned sequence files...");
         self.input_files.par_iter().for_each(|file| {

src/core/sequence/filter.rs

@@ -97,7 +97,7 @@ impl<'a> SequenceFiltering<'a> {
     ///
     /// let input_fmt = InputFmt::Nexus;
     /// let datatype = DataType::Dna;
-    /// let input_dir = Path::new("tests/files/concat");
+    /// let input_dir = Path::new("tests/files/alignments");
     /// let files = SeqFileFinder::new(Path::new(input_dir)).find(&input_fmt);
     /// // Replace the temp directory with your own directory.
     /// let output = TempDir::new("tempt").unwrap();
@@ -224,7 +224,7 @@ mod tests {
 
     #[test]
     fn test_filter_sequences_by_length() {
-        let dir = Path::new("tests/files/concat");
+        let dir = Path::new("tests/files/alignments");
         let params = SeqFilteringParameters::MinSequenceLength(7);
         setup!(dir, handle, params, output);
         handle.filter();

src/core/sequence/id.rs

@@ -59,8 +59,8 @@ impl<'a> SequenceID<'a> {
     /// use segul::helper::types::{DataType, InputFmt};
     /// use tempdir::TempDir;
     ///
-    /// let alignment_2 = PathBuf::from("tests/files/concat/gene_2.nex");
-    /// let alignment_1 = PathBuf::from("tests/files/concat/gene_1.nex");
+    /// let alignment_2 = PathBuf::from("tests/files/alignments/gene_2.nex");
+    /// let alignment_1 = PathBuf::from("tests/files/alignments/gene_1.nex");
     /// let files = vec![alignment_1, alignment_2];
     /// let output = TempDir::new("tempt").unwrap();
     /// let handle = SequenceID::new(&files, &InputFmt::Auto, &DataType::Dna, Path::new(output.path()), None);

src/helper/concat.rs

@@ -42,8 +42,8 @@ impl ConcatParams {
 /// use segul::helper::types::{DataType, InputFmt};
 ///
 /// let mut files = vec![
-///     PathBuf::from("tests/files/concat/gene_1.nex"),
-///    PathBuf::from("tests/files/concat/gene_2.nex"),
+///     PathBuf::from("tests/files/alignments/gene_1.nex"),
+///    PathBuf::from("tests/files/alignments/gene_2.nex"),
 /// ];
 /// let spinner = segul::helper::utils::set_spinner();
 /// let input_fmt = InputFmt::Nexus;
@@ -188,7 +188,7 @@ mod test {
 
     #[test]
     fn test_concat_nexus() {
-        let path = Path::new("tests/files/concat/");
+        let path = Path::new("tests/files/alignments/");
         let mut files = SeqFileFinder::new(path).find(&InputFmt::Nexus);
         let mut concat = Concat::new(&mut files, &InputFmt::Nexus, &DNA);
         let spin = utils::set_spinner();
@@ -199,15 +199,15 @@ mod test {
     #[test]
     #[should_panic]
     fn test_get_alignment_panic() {
-        let path = Path::new("tests/files/concat/");
+        let path = Path::new("tests/files/alignments/");
         let mut files = SeqFileFinder::new(path).find(&InputFmt::Nexus);
         let concat = Concat::new(&mut files, &InputFmt::Nexus, &DNA);
         concat.get_alignment(Path::new("."));
     }
 
     #[test]
     fn test_concat_check_result() {
-        let path = Path::new("tests/files/concat/");
+        let path = Path::new("tests/files/alignments/");
         let mut files = SeqFileFinder::new(path).find(&InputFmt::Nexus);
         let mut concat = Concat::new(&mut files, &InputFmt::Nexus, &DNA);
         let spin = utils::set_spinner();
@@ -219,7 +219,7 @@ mod test {
 
     #[test]
     fn test_concat_partition() {
-        let path = Path::new("tests/files/concat/");
+        let path = Path::new("tests/files/alignments/");
         let mut files = SeqFileFinder::new(path).find(&InputFmt::Nexus);
         let mut concat = Concat::new(&mut files, &InputFmt::Nexus, &DNA);
         let spin = utils::set_spinner();
@@ -246,7 +246,7 @@ mod test {
 
     #[test]
     fn test_header_datatype() {
-        let path = Path::new("tests/files/concat/");
+        let path = Path::new("tests/files/alignments/");
         let mut files = SeqFileFinder::new(path).find(&InputFmt::Nexus);
         let mut concat = Concat::new(&mut files, &InputFmt::Nexus, &DataType::Aa);
         concat.match_header_datatype();

src/helper/finder.rs

@@ -227,7 +227,7 @@ impl<'a> SeqFileFinder<'a> {
     /// use segul::helper::types::InputFmt;
     /// use segul::helper::finder::SeqFileFinder;
     ///
-    /// let dir = Path::new("tests/files/concat");
+    /// let dir = Path::new("tests/files/alignments");
     /// let input_fmt = InputFmt::Nexus;
     /// let files = SeqFileFinder::new(&dir).find(&input_fmt);
     /// assert_eq!(files.len(), 4);
@@ -250,17 +250,39 @@ impl<'a> SeqFileFinder<'a> {
     /// # Example
     /// ```
     /// use std::path::Path;
-    /// use segul::helper::types::InputFmt;
     /// use segul::helper::finder::SeqFileFinder;
     ///
-    /// let dir = Path::new("tests/files/concat");
+    /// let dir = Path::new("tests/files/alignments");
     /// let files = SeqFileFinder::new(&dir).find_recursive();
     /// assert_eq!(files.len(), 4);
     /// ```
     pub fn find_recursive(&self) -> Vec<PathBuf> {
         walk_dir!(self, re_match_sequence_lazy)
     }
 
+    /// Find input files for sequence and alignment, recursively.
+    /// Limit search to only the input format.
+    ///
+    /// # Example
+    /// ```
+    /// use std::path::Path;
+    /// use segul::helper::types::InputFmt;
+    /// use segul::helper::finder::SeqFileFinder;
+    ///
+    /// let dir = Path::new("tests/files/alignments");
+    /// let files = SeqFileFinder::new(&dir).find_recursive_only(&InputFmt::Nexus);
+    /// assert_eq!(files.len(), 4);
+    pub fn find_recursive_only(&self, input_fmt: &'a InputFmt) -> Vec<PathBuf> {
+        let files = match input_fmt {
+            InputFmt::Fasta => walk_dir!(self, re_matches_fasta_lazy),
+            InputFmt::Nexus => walk_dir!(self, re_match_nexus_lazy),
+            InputFmt::Phylip => walk_dir!(self, re_match_phylip_lazy),
+            _ => unreachable!(),
+        };
+
+        files
+    }
+
     fn check_results(&self, files: &[PathBuf]) {
         if files.is_empty() {
             panic!(
@@ -305,6 +327,22 @@ fn re_match_sequence_lazy(fname: &str) -> bool {
     RE.is_match(fname)
 }
 
+fn re_match_nexus_lazy(fname: &str) -> bool {
+    lazy_static! {
+        static ref RE: Regex = Regex::new(r"(?i)(.nex*|.nxs)(?:.*)").unwrap();
+    }
+
+    RE.is_match(fname)
+}
+
+fn re_match_phylip_lazy(fname: &str) -> bool {
+    lazy_static! {
+        static ref RE: Regex = Regex::new(r"(?i)(.phy*|.fna|.fa*)(?:.*)").unwrap();
+    }
+
+    RE.is_match(fname)
+}
+
 /// Parse IDs from input sequence files.
 /// # Example
 /// ```
@@ -314,8 +352,8 @@ fn re_match_sequence_lazy(fname: &str) -> bool {
 /// use indexmap::IndexSet;
 ///
 /// let files = vec![
-///    PathBuf::from("tests/files/concat/gene_1.nex"),
-///    PathBuf::from("tests/files/concat/gene_2.nex"),
+///    PathBuf::from("tests/files/alignments/gene_1.nex"),
+///    PathBuf::from("tests/files/alignments/gene_2.nex"),
 /// ];
 ///
 /// let input_fmt = InputFmt::Nexus;
@@ -404,7 +442,7 @@ mod test {
 
     macro_rules! input {
         ($files: ident) => {
-            let path = Path::new("tests/files/concat");
+            let path = Path::new("tests/files/alignments");
 
             let mut $files = SeqFileFinder::new(path);
         };
@@ -451,7 +489,7 @@ mod test {
         input!(files);
         let fmt = InputFmt::Nexus;
         files.pattern(&fmt);
-        assert_eq!("tests/files/concat/*.nex*", files.pattern);
+        assert_eq!("tests/files/alignments/*.nex*", files.pattern);
     }
 
     #[test]

tests/concat.rs

@@ -9,7 +9,7 @@ use segul::helper::types::InputFmt;
 
 #[test]
 fn test_concat() {
-    initiate_cmd!(cmd, "align", "concat", "tests/files/concat/", tmp_dir);
+    initiate_cmd!(cmd, "align", "concat", "tests/files/alignments/", tmp_dir);
     cmd.arg("--partition-format")
         .arg("raxml")
         .assert()
@@ -19,7 +19,7 @@ fn test_concat() {
 
 #[test]
 fn test_concat_nexus_part() {
-    initiate_cmd!(cmd, "align", "concat", "tests/files/concat/", tmp_dir);
+    initiate_cmd!(cmd, "align", "concat", "tests/files/alignments/", tmp_dir);
     cmd.arg("--partition-format")
         .arg("nexus")
         .assert()

tests/convert.rs

@@ -9,7 +9,7 @@ use segul::helper::types::InputFmt;
 
 #[test]
 fn test_convert() {
-    initiate_cmd!(cmd, "align", "convert", "tests/files/concat/", tmp_dir);
+    initiate_cmd!(cmd, "align", "convert", "tests/files/alignments/", tmp_dir);
     cmd.arg("-F").arg("phylip").assert().success();
     test_results!(4, tmp_dir, "Align-Convert", Phylip);
 }

tests/extract.rs

@@ -9,15 +9,15 @@ use segul::helper::types::InputFmt;
 
 #[test]
 fn test_extract() {
-    initiate_cmd!(cmd, "sequence", "extract", "tests/files/concat/", tmp_dir);
+    initiate_cmd!(cmd, "sequence", "extract", "tests/files/alignments/", tmp_dir);
     cmd.arg("--id=ABCD").assert().success();
     test_results!(4, tmp_dir, "Sequence-Extract", Nexus);
 }
 
 #[test]
 #[should_panic]
 fn test_conflicting_extract_cmd() {
-    initiate_cmd!(cmd, "sequence", "extract", "tests/files/concat/", tmp_dir);
+    initiate_cmd!(cmd, "sequence", "extract", "tests/files/alignments/", tmp_dir);
     cmd.arg("--id")
         .arg("ABCD")
         .arg("--re=^AB")
@@ -28,18 +28,18 @@ fn test_conflicting_extract_cmd() {
 #[test]
 #[should_panic]
 fn test_no_extract_cmd() {
-    initiate_cmd!(cmd, "sequence", "extract", "tests/files/concat/", tmp_dir);
+    initiate_cmd!(cmd, "sequence", "extract", "tests/files/alignments/", tmp_dir);
     cmd.arg("--id")
         .arg("ABCD")
         .arg("--file")
-        .arg("tests/files/concat/concat.nex")
+        .arg("tests/files/alignments/alignments.nex")
         .assert()
         .success();
 }
 
 #[test]
 fn test_extract_re() {
-    initiate_cmd!(cmd, "sequence", "extract", "tests/files/concat/", tmp_dir);
+    initiate_cmd!(cmd, "sequence", "extract", "tests/files/alignments/", tmp_dir);
     cmd.arg("--re=^AB").assert().success();
     test_results!(4, tmp_dir, "Sequence-Extract", Nexus);
 }

tests/id.rs

@@ -8,7 +8,7 @@ use segul::parser::txt;
 
 #[test]
 fn test_id_success() {
-    initiate_cmd!(cmd, "sequence", "id", "tests/files/concat/", tmp_dir);
+    initiate_cmd!(cmd, "sequence", "id", "tests/files/alignments/", tmp_dir);
     cmd.assert().success();
     let pred = predicates::path::is_file();
     let res_path = tmp_dir.path().join("SEGUL-ID").join("id.txt");

tests/remove.rs

@@ -9,14 +9,14 @@ use segul::helper::types::InputFmt;
 
 #[test]
 fn test_remove_id() {
-    initiate_cmd!(cmd, "sequence", "remove", "tests/files/concat/", tmp_dir);
+    initiate_cmd!(cmd, "sequence", "remove", "tests/files/alignments/", tmp_dir);
     cmd.arg("--id=ABCD").assert().success();
     test_results!(3, tmp_dir, "Sequence-Remove", Nexus);
 }
 
 #[test]
 fn test_remove_re() {
-    initiate_cmd!(cmd, "sequence", "remove", "tests/files/concat/", tmp_dir);
+    initiate_cmd!(cmd, "sequence", "remove", "tests/files/alignments/", tmp_dir);
     cmd.arg("--re=E$").assert().success();
     test_results!(4, tmp_dir, "Sequence-Remove", Nexus);
 }

tests/rename.rs

@@ -9,7 +9,7 @@ use segul::helper::types::InputFmt;
 
 #[test]
 fn test_rename() {
-    initiate_cmd!(cmd, "sequence", "rename", "tests/files/concat/", tmp_dir);
+    initiate_cmd!(cmd, "sequence", "rename", "tests/files/alignments/", tmp_dir);
     cmd.arg("--remove=D").assert().success();
     test_results!(4, tmp_dir, "Sequence-Rename", Nexus);
 }

tests/translate.rs

@@ -9,7 +9,13 @@ use segul::helper::types::InputFmt;
 
 #[test]
 fn test_dna_translation() {
-    initiate_cmd!(cmd, "sequence", "translate", "tests/files/concat/", tmp_dir);
+    initiate_cmd!(
+        cmd,
+        "sequence",
+        "translate",
+        "tests/files/alignments/",
+        tmp_dir
+    );
     cmd.assert().success();
     test_results!(4, tmp_dir, "Sequence-Translate", Nexus);
 }