WIP: Starting with BCF header records (#76)

BCF header records

Author: holtgrewe

Cargo.toml

@@ -23,6 +23,7 @@ lazy_static = "0.2"
 bitflags = "0.9"
 serde = { version = "^1", optional = true }
 regex = "0.2"
+linear-map = "1.2.0"
 
 [features]
 default = []

src/bam/record_serde.rs

@@ -289,9 +289,9 @@ impl<'de> Deserialize<'de> for Record {
 
 #[cfg(test)]
 mod tests {
-    use bam::record::Record;
     use bam::Read;
     use bam::Reader;
+    use bam::record::Record;
 
     use std::path::Path;
 

src/bcf/header.rs

@@ -9,6 +9,8 @@ use std::str;
 
 use htslib;
 
+use linear_map::LinearMap;
+
 pub type SampleSubset = Vec<i32>;
 
 custom_derive! {
@@ -191,6 +193,38 @@ impl Drop for Header {
     }
 }
 
+/// A header record.
+#[derive(Debug)]
+pub enum HeaderRecord {
+    /// A `FILTER` header record.
+    Filter {
+        key: String,
+        values: LinearMap<String, String>,
+    },
+    /// An `INFO` header record.
+    Info {
+        key: String,
+        values: LinearMap<String, String>,
+    },
+    /// A `FORMAT` header record.
+    Format {
+        key: String,
+        values: LinearMap<String, String>,
+    },
+    /// A `contig` header record.
+    Contig {
+        key: String,
+        values: LinearMap<String, String>,
+    },
+    /// A structured header record.
+    Structured {
+        key: String,
+        values: LinearMap<String, String>,
+    },
+    /// A generic, unstructured header record.
+    Generic { key: String, value: String },
+}
+
 #[derive(Debug)]
 pub struct HeaderView {
     pub inner: *mut htslib::bcf_hdr_t,
@@ -340,6 +374,64 @@ impl HeaderView {
         };
         key.to_bytes().to_vec()
     }
+
+    /// Return structured `HeaderRecord`s.
+    pub fn header_records(&self) -> Vec<HeaderRecord> {
+        fn parse_kv(rec: &htslib::bcf_hrec_t) -> LinearMap<String, String> {
+            let mut result: LinearMap<String, String> = LinearMap::new();
+            for i in 0_i32..(rec.nkeys) {
+                let key = unsafe {
+                    ffi::CStr::from_ptr(*rec.keys.offset(i as isize))
+                        .to_str()
+                        .unwrap()
+                        .to_string()
+                };
+                let value = unsafe {
+                    ffi::CStr::from_ptr(*rec.vals.offset(i as isize))
+                        .to_str()
+                        .unwrap()
+                        .to_string()
+                };
+                result.insert(key, value);
+            }
+            result
+        }
+
+        let mut result: Vec<HeaderRecord> = Vec::new();
+        for i in 1_i32..unsafe { (*self.inner).nhrec } {
+            let rec = unsafe { &(**(*self.inner).hrec.offset(i as isize)) };
+            let key = unsafe { ffi::CStr::from_ptr(rec.key).to_str().unwrap().to_string() };
+            let record = match rec.type_ {
+                0 => HeaderRecord::Filter {
+                    key,
+                    values: parse_kv(rec),
+                },
+                1 => HeaderRecord::Info {
+                    key,
+                    values: parse_kv(rec),
+                },
+                2 => HeaderRecord::Format {
+                    key,
+                    values: parse_kv(rec),
+                },
+                3 => HeaderRecord::Contig {
+                    key,
+                    values: parse_kv(rec),
+                },
+                4 => HeaderRecord::Structured {
+                    key,
+                    values: parse_kv(rec),
+                },
+                5 => HeaderRecord::Generic {
+                    key,
+                    value: unsafe { ffi::CStr::from_ptr(rec.value).to_str().unwrap().to_string() },
+                },
+                _ => panic!("Unknown type: {}", rec.type_),
+            };
+            result.push(record);
+        }
+        result
+    }
 }
 
 impl Clone for HeaderView {

src/bcf/mod.rs

@@ -24,8 +24,7 @@ pub mod record;
 use bcf::header::{HeaderView, SampleSubset};
 use htslib;
 
-pub use bcf::buffer::RecordBuffer;
-pub use bcf::header::Header;
+pub use bcf::header::{Header, HeaderRecord};
 pub use bcf::record::Record;
 
 /// Redefinition of corresponding `#define` in `vcf.h.`.
@@ -839,6 +838,28 @@ mod tests {
         assert!(header.sample_to_id(b"three").is_err());
     }
 
+    #[test]
+    fn test_header_records() {
+        let vcf = Reader::from_path(&"test/test_string.vcf")
+            .ok()
+            .expect("Error opening file.");
+        let records = vcf.header().header_records();
+        assert_eq!(records.len(), 9);
+
+        match &records[0] {
+            &HeaderRecord::Filter {
+                ref key,
+                ref values,
+            } => {
+                assert_eq!(key, "FILTER");
+                assert_eq!(values["ID"], "PASS");
+            }
+            _ => {
+                assert!(false);
+            }
+        }
+    }
+
     // Helper function reading full file into string.
     fn read_all<P: AsRef<Path>>(path: P) -> String {
         let mut file = File::open(path.as_ref())

src/lib.rs

@@ -83,6 +83,8 @@ extern crate quick_error;
 extern crate regex;
 extern crate url;
 
+extern crate linear_map;
+
 #[cfg(feature = "serde")]
 extern crate serde;
 

src/sam/mod.rs

@@ -10,9 +10,9 @@ use std::path::Path;
 
 use htslib;
 
+use bam::HeaderView;
 use bam::header;
 use bam::record;
-use bam::HeaderView;
 
 /// SAM writer.
 #[derive(Debug)]
@@ -111,10 +111,10 @@ quick_error! {
 
 #[cfg(test)]
 mod tests {
-    use bam::header;
-    use bam::record;
     use bam::Read;
     use bam::Reader;
+    use bam::header;
+    use bam::record;
     use sam::Writer;
 
     #[test]