From 0e8756faae275cc7ac5a1832a4c7a5b858056e76 Mon Sep 17 00:00:00 2001
From: Lee Katz - Aspen <gzu2@cdc.gov>
Date: Wed, 21 Mar 2018 13:41:04 -0400
Subject: [PATCH] checks for /1 and /2 format

---
 src/bin/friends_ung.rs | 73 ++++++++++++++++++++++++++++++++++--------
 1 file changed, 60 insertions(+), 13 deletions(-)

diff --git a/src/bin/friends_ung.rs b/src/bin/friends_ung.rs
index 24830d69..14210c22 100644
--- a/src/bin/friends_ung.rs
+++ b/src/bin/friends_ung.rs
@@ -2,6 +2,7 @@ extern crate ross;
 extern crate regex;
 
 use ross::ross_base_options;
+use ross::logmsg;
 use regex::Regex;
 
 use std::fs::File;
@@ -12,35 +13,81 @@ use std::env;
 fn main() {
     let args: Vec<String> = env::args().collect();
     let mut opts = ross_base_options();
-    opts.optflag("x","blah","blah!");
+    opts.optopt("c","check-first","How many deflines to check to make sure the input is paired-end","INT");
     let matches = opts.parse(&args[1..]).expect("ERROR: could not parse parameters");
     
-    // If there is a match on these, then mark invalid.
-    // In other words, we are looking for a pattern that
-    // is NOT the target seq or qual
-    let seq_regex = Regex::new(r"[^a-zA-Z]").expect("malformed seq regex");
-    let qual_regex= Regex::new(r"[^!-Z]").expect("malformed qual regex");
+    let slash_r1r2_regex = Regex::new(r"(.+)/([12])$").expect("malformed qual regex");
 
     if matches.opt_present("help") {
-        println!("Convert a fastq file to a standard 4-lines-per-entry format\n{}", 
+        println!("Determine paired-end-ness in an interleaved file. Currently only checks deflines for the /1 and /2 format\n{}", 
                  opts.usage(&opts.short_usage(&args[0]))
                 );
         std::process::exit(0);
     }
 
-    let lines_per_read={
-        if matches.opt_present("paired-end") {
-            8
-        }else{
-            4
+    let check_first = { 
+        if matches.opt_present("check-first") {
+            matches.opt_str("check-first")
+                .expect("Error reading the check-first option")
+                .parse()
+                .expect("ERROR converting the check-first parameter to an integer")
+        } else {
+            200
         }
     };
 
+    let mut id1=String::new();
+    let mut id2=String::new();
+    let mut pairs_counter=0;
+
     let my_file = File::open("/dev/stdin").expect("Could not open file");
     let my_buffer=BufReader::new(my_file);
     for (i,line) in my_buffer.lines().enumerate() {
         let line = line.expect("ERROR: could not read the next line in the input");
-        //match i%lines_per_read {
+        match i%8 {
+            0=>{
+                id1=line;
+            }
+            4=>{
+                id2=line;
+                pairs_counter+=1;
+
+                let caps1 = slash_r1r2_regex.captures(&id1).expect("ERROR: could not regex against id1");
+                let caps2 = slash_r1r2_regex.captures(&id2).expect("ERROR: could not regex against id2");
+
+                // Make sure the base name matches
+                if caps1[1] != caps2[1] {
+                    let mut msg = "ID1 does not match ID2 on line ".to_string();
+                    msg.push_str(&i.to_string());
+                    msg.push_str("\n");
+                    msg.push_str(&id1);
+                    msg.push_str(" vs ");
+                    msg.push_str(&id2);
+                    logmsg(&msg);
+                    std::process::exit(1);
+                }
+                // Make sure there is a 1/2 combo
+                if &caps1[2] != "1" || &caps2[2] != "2" {
+                    let mut msg = "/1 is not followed by /2 on line ".to_string();
+                    msg.push_str(&i.to_string());
+                    msg.push_str("\n");
+                    msg.push_str(&id1);
+                    msg.push_str(" vs ");
+                    msg.push_str(&id2);
+                    logmsg(&msg);
+                    std::process::exit(1);
+                }
+
+                if pairs_counter >= check_first {
+                    break;
+                }
+
+            }
+            // We can safely ignore the seq, plus, and qual lines
+            _=>{ }
+        }
     }
+
+    logmsg("The fastq input seems to be interleaved paired-end");
 }