From 8b8059b414849570870fbd8faffeccd10533e1aa Mon Sep 17 00:00:00 2001
From: Miller <millerd15@email.chop.edu>
Date: Sat, 27 Jan 2024 12:55:34 -0500
Subject: [PATCH] :construction: add description scraper

:pencil: fix borrows

:bug: fix quote injection
---
 src/commands/encoder_cmd.rs | 24 +++++++++++++++++++++++-
 src/lib/echtvar.rs          |  6 +++---
 2 files changed, 26 insertions(+), 4 deletions(-)

diff --git a/src/commands/encoder_cmd.rs b/src/commands/encoder_cmd.rs
index ef1f723..31a1e27 100644
--- a/src/commands/encoder_cmd.rs
+++ b/src/commands/encoder_cmd.rs
@@ -1,6 +1,6 @@
 use bincode::Options;
 use echtvar_lib::{echtvar::bstrip_chr, fields, kmer16, var32, zigzag};
-use rust_htslib::bcf::header::{TagLength, TagType};
+use rust_htslib::bcf::header::{TagLength, TagType, HeaderRecord};
 use rust_htslib::bcf::record::{Buffer, Record};
 use rust_htslib::bcf::{Read as BCFRead, Reader};
 use stream_vbyte::{encode::encode, x86::Sse41};
@@ -160,6 +160,25 @@ fn is_sorted<T: std::cmp::PartialOrd>(data: &Vec<T>) -> bool {
     return true;
 }
 
+fn hdr_info_id2description(
+    mut hrecs: Vec<HeaderRecord>,
+    id: &String,
+    default: &std::string::String,
+) -> std::string::String {
+    hrecs.retain(|rec| match rec {
+        HeaderRecord::Info {key: _, values: v} => &v["ID"] == id,
+        _ => false}
+    );
+    if hrecs.len() != 1 {
+        panic!("Field {} is either not present in the header or present multiple times!", id);
+    };
+    let description = match hrecs.first().unwrap() {
+        HeaderRecord::Info {key: _, values: v} => if v.contains_key("Description") { &v["Description"] } else { default },
+        _ => default,
+    };
+    return description.to_string();
+}
+
 pub fn encoder_main(vpaths: Vec<&str>, opath: &str, jpath: &str) {
     let zpath = std::path::Path::new(opath);
     let jpath = std::path::Path::new(jpath);
@@ -228,6 +247,9 @@ pub fn encoder_main(vpaths: Vec<&str>, opath: &str, jpath: &str) {
                 tl, f.field
             ),
         };
+        println!("Old description for field {}: {}", f.field, f.description);
+        f.description = hdr_info_id2description(header.header_records(), &f.field, &f.description);
+        println!("New description for field {}: {}", f.field, f.description);
     }
 
     let zfile = std::fs::File::create(&zpath).unwrap();
diff --git a/src/lib/echtvar.rs b/src/lib/echtvar.rs
index 1e5bd00..1825d86 100644
--- a/src/lib/echtvar.rs
+++ b/src/lib/echtvar.rs
@@ -177,7 +177,7 @@ impl EchtVars {
         for e in &self.fields {
             header.push_record(
                 format!(
-                    "##INFO=<ID={},Number={},Type={},Description=\"{}\">",
+                    "##INFO=<ID={},Number={},Type={},Description={}>",
                     e.alias,
                     if vec!["A", "R", "G"].iter().any(|n| n == &e.number) {
                         "1"
@@ -191,8 +191,8 @@ impl EchtVars {
                     } else {
                         "Float"
                     },
-                    if &e.description.to_string() == "added by echtvar" {
-                        format!("added by echtvar from {}", path)
+                    if &e.description.to_string() == "added by echtvar"{
+                        format!("\"added by echtvar from {}\"", path)
                     } else {
                         format!("added by echtvar {}", e.description.to_string())
                     }