Detect duplicated words or space charactersHEADmaster

These are real typos more than 95% of time (based on current
statistics) and worth eliminating.

2 files changed, 20 insertions, 0 deletions

diff --git a/NEWS b/NEWS
index 9024244..3a8995d 100644
--- a/NEWS
+++ b/NEWS
@@ -2,6 +2,7 @@ Version 0.X
 
 - template: move CVEs into the "upstream" field in advisory.json
 - add sanity checks for CVE identifiers
+- look for typos involving duplicate words or spaces
 
 Version 0.34
 
diff --git a/lib/MGA/Advisories.pm b/lib/MGA/Advisories.pm
index e4ff62c..2f6827e 100644
--- a/lib/MGA/Advisories.pm
+++ b/lib/MGA/Advisories.pm
@@ -301,6 +301,25 @@ sub assign_id {
         return;
     }
 
+    # Look for errors in the subject and description
+    if ( $adv->{subject} =~ /\b(\w+)\s+\1\s/ ) {
+        print STDERR "Subject has a duplicated word\n";
+        return;
+    }
+    if ( $adv->{subject} =~ /\s\s/ ) {
+        print STDERR "Subject has an extra space character\n";
+        return;
+    }
+    # As of this writing, 0.03% of advisories have a word duplicated
+    # legitimately in the description, which this check would disallow. But
+    # given that this check succeeds in eliminating these typos 95% of the
+    # time, it's probably worth forcing the user to reword the description in
+    # these rare cases.
+    if ( $adv->{description} =~ /\b(\w+)\s+\1\s/ ) {
+        print STDERR "Description has a duplicated word\n";
+        return;
+    }
+
     # Turn on autoflush
     $|++;