Merge pull request #43 from derRiesenOtter/master

Bug fix for missing first letters of words

src/default.rs

@@ -118,7 +118,7 @@ lazy_static! {
             (
                 // Deletes multiple occurrences of words using the extended fancy-regex crate
                 // syntax. Only the first mention of a word occurring multiple times is retained:
-                fancy_regex::Regex::new(r"(?i)\b(?P<first>\b\w+\b)(?P<spacer>.*)\b\k<first>").unwrap(),
+                fancy_regex::Regex::new(r"(?i)\b(?P<first>\b\w+\b)(?P<spacer>.*)\b\k<first>\b").unwrap(),
                 r"$first$spacer".to_string()
             )
         );

src/model_funcs.rs

@@ -248,6 +248,18 @@ mod tests {
                 Some(&(*CAPTURE_REPLACE_DESCRIPTION_PAIRS))
             )
         );
+
+        // Test 8 also checks that no additional letters are deleted
+        hit_words = "sp|Q6R2K2|SRF4_ARATH P Transferase Domain Containing Protein OS=Arabidopsis thaliana OX=3702 GN=SRF4 PE=2 SV=1".to_string();
+        expected = "p transferase domain containing protein";
+        assert_eq!(
+            expected, 
+            filter_stitle(
+                &hit_words,
+                &(*FILTER_REGEXS),
+                Some(&(*CAPTURE_REPLACE_DESCRIPTION_PAIRS))
+            )
+        );
 
 
     }