use infer_from_path to determine filetype

This should add support for gzipped files, MzMLb and Thermo raw files.
compomics · Aug 13, 2024 · 503e0c9 · 503e0c9
1 parent 86da407
commit 503e0c9
Show file tree

Hide file tree

Showing 2 changed files with 25 additions and 18 deletions.
diff --git a/src/file_types.rs b/src/file_types.rs
@@ -1,25 +1,34 @@
+use mzdata::io::MassSpectrometryFormat;
+
 pub enum SpectrumFileType {
     MascotGenericFormat,
     MzML,
+    MzMLb,
     BrukerRaw,
-    // ThermoRaw,
+    ThermoRaw,
     Unknown,
 }
 
 pub fn match_file_type(spectrum_path: &str) -> SpectrumFileType {
-    let extension = spectrum_path.split('.').last().unwrap_or("").to_lowercase();
-    match extension.as_str() {
-        "mgf" => SpectrumFileType::MascotGenericFormat,
-        "mzml" => SpectrumFileType::MzML,
-        "d" | "ms2" => SpectrumFileType::BrukerRaw,
-        // "raw" => SpectrumFileType::ThermoRaw,
-        _ => match (
-            folder_contains_extension(spectrum_path, "bin"),
-            folder_contains_extension(spectrum_path, "parquet"),
-        ) {
-            (true, true) => SpectrumFileType::BrukerRaw,
-            _ => SpectrumFileType::Unknown,
-        },
+    match mzdata::io::infer_from_path(spectrum_path).0 {
+        MassSpectrometryFormat::MGF => SpectrumFileType::MascotGenericFormat,
+        MassSpectrometryFormat::MzML => SpectrumFileType::MzML,
+        MassSpectrometryFormat::MzMLb => SpectrumFileType::MzMLb,
+        MassSpectrometryFormat::ThermoRaw => SpectrumFileType::ThermoRaw,
+        MassSpectrometryFormat::Unknown => {
+            let extension = spectrum_path.split('.').last().unwrap_or("").to_lowercase();
+            match extension.as_str() {
+                "d" | "ms2" => SpectrumFileType::BrukerRaw,
+                _ => match (
+                    folder_contains_extension(spectrum_path, "bin"),
+                    folder_contains_extension(spectrum_path, "parquet"),
+                ) {
+                    (true, true) => SpectrumFileType::BrukerRaw,
+                    _ => SpectrumFileType::Unknown,
+                },
+            }
+        }
+        _ => SpectrumFileType::Unknown
     }
 }
 

diff --git a/src/lib.rs b/src/lib.rs
@@ -19,11 +19,10 @@ pub fn get_precursor_info(spectrum_path: String) -> PyResult<HashMap<String, Pre
     let file_type = match_file_type(&spectrum_path);
 
     let precursors = match file_type {
-        SpectrumFileType::MascotGenericFormat | SpectrumFileType::MzML => {
+        SpectrumFileType::MascotGenericFormat | SpectrumFileType::MzML | SpectrumFileType::MzMLb | SpectrumFileType:: ThermoRaw => {
             parse_mzdata::parse_precursor_info(&spectrum_path)
         }
         SpectrumFileType::BrukerRaw => parse_timsrust::parse_precursor_info(&spectrum_path),
-        // SpectrumFileType::ThermoRaw => parse_with_mzdata_thermo(&spectrum_path, file_type),
         SpectrumFileType::Unknown => return Err(PyOSError::new_err("Unsupported file type")),
     };
 
@@ -39,11 +38,10 @@ pub fn get_ms2_spectra(spectrum_path: String) -> PyResult<Vec<ms2_spectrum::MS2S
     let file_type = match_file_type(&spectrum_path);
 
     let spectra = match file_type {
-        SpectrumFileType::MascotGenericFormat | SpectrumFileType::MzML => {
+        SpectrumFileType::MascotGenericFormat | SpectrumFileType::MzML | SpectrumFileType::MzMLb | SpectrumFileType:: ThermoRaw => {
             parse_mzdata::read_ms2_spectra(&spectrum_path)
         }
         SpectrumFileType::BrukerRaw => parse_timsrust::read_ms2_spectra(&spectrum_path),
-        // SpectrumFileType::ThermoRaw => parse_with_mzdata_thermo(&spectrum_path, file_type),
         SpectrumFileType::Unknown => return Err(PyOSError::new_err("Unsupported file type")),
     };