Merge pull request #10 from sebastian-wehner-bruker/feature/updated_m…

…iniTDF_conventions

feat: updated to miniTDF conventions. Adapted test names accordingly

README.md

@@ -28,12 +28,18 @@ Two primary data types are exposed through TimsRust:
 ### File formats
 
 Two file formats are supported:
-* Bruker .d folder containing:
+* TDF - Bruker .d folder containing:
     * analysis.tdf
     * analysis.tdf_bin
-* Bruker .ms2 folder containing:
-    * converter.ms2.bin
-    * converter.MS2Spectra.ms2.parquet
+
+
+* miniTDF - ProteoScape optimized Bruker file-format. Similar to TDF, miniTDF consists of multiple files: a binary '.bin'
+ and an index '.parquet' file. The file-names are made up to the following convention: `<producing-engine-name>.<domain-name>.<extension>`.
+ e.g. for MS2 spectrum information: `<producing-engine-name>.ms2spectrum.<extension>`. Therefor the following files are expected 
+ in the provided ms2 folder:
+    * *.ms2spectrum.bin
+    * *.ms2spectrum.parquet
+
 
 ## Python bindings
 

src/file_readers/file_formats.rs

@@ -20,15 +20,7 @@ impl FileFormat {
             .unwrap_or_default();
         let format = match extension {
             "d" => Self::DFolder(path),
-            "ms2" => Self::MS2Folder(path),
-            _ => {
-                if let Some(path) = path.parent() {
-                    // Only recurse if there is a valid parent section,
-                    // otherwise we'll get a stack overflow
-                    return Self::parse(path);
-                }
-                return Err(FileFormatError::NoParentWithBrukerExtension);
-            },
+            _ => Self::MS2Folder(path)
         };
         format.is_valid()?;
         Ok(format)

src/file_readers/spectrum_readers/mini_tdf_reader.rs

@@ -1,3 +1,4 @@
+use std::fs;
 use {
     crate::{
         file_readers::{
@@ -13,6 +14,7 @@ use {
     rayon::prelude::*,
     std::path::PathBuf,
 };
+use crate::file_readers::FileFormatError;
 
 #[derive(Debug, Default, Clone)]
 pub struct MiniTDFReader {
@@ -23,6 +25,24 @@ pub struct MiniTDFReader {
     frame_reader: BinFileReader,
 }
 
+fn find_ms2spectrum_file(ms2_dir_path: &str, extension: String) -> Result<String, FileFormatError> {
+    let files = fs::read_dir(ms2_dir_path).unwrap();
+    for file in files {
+        let filename = file.unwrap().path().file_name().unwrap().to_str().unwrap().to_owned();
+        if filename.ends_with( std::format!("ms2spectrum.{}", extension).as_str()) {
+            return Ok(filename)
+        }
+    }
+    let err =  match extension.as_str() {
+        "parquet" => FileFormatError::MetadataFilesAreMissing,
+        "bin" => FileFormatError::BinaryFilesAreMissing,
+        _ => FileFormatError::BinaryFilesAreMissing
+    };
+    println!("{}", format!("No '*.ms2spectrum.{}' file found in '{}'", extension, ms2_dir_path));
+    return Err(err);
+}
+
+
 impl MiniTDFReader {
     pub fn new(path_name: String) -> Self {
         let mut reader: MiniTDFReader = Self::default();
@@ -35,7 +55,8 @@ impl MiniTDFReader {
 
     fn read_parquet_file_name(&mut self) {
         let mut path: PathBuf = PathBuf::from(&self.path_name);
-        path.push("converter.MS2Spectra.ms2.parquet");
+        let ms2_parquet_file = find_ms2spectrum_file(&self.path_name, "parquet".to_owned()).unwrap();
+        path.push(ms2_parquet_file);
         self.parquet_file_name = path.to_string_lossy().into_owned();
     }
 
@@ -45,7 +66,8 @@ impl MiniTDFReader {
     }
     fn set_spectrum_reader(&mut self) {
         let mut path: PathBuf = PathBuf::from(&self.path_name);
-        path.push("converter.ms2.bin");
+        let ms2_bin_file = find_ms2spectrum_file(&self.path_name, "bin".to_owned()).unwrap();
+        path.push(ms2_bin_file);
         let file_name: String = path.to_string_lossy().into_owned();
         self.frame_reader =
             BinFileReader::new(String::from(&file_name), self.offsets.clone());