From: Shawn Boyette <sboyette@esilibrary.com>
Date: Mon, 8 Dec 2008 22:52:19 +0000 (+0000)
Subject: compress: now detects input sorted improperly
X-Git-Url: http://git.equinoxoli.org/?p=migration-tools.git;a=commitdiff_plain;h=c25291487fffdb20f740010b48e93fa2c58dcac0

compress: now detects input sorted improperly
---

diff --git a/compress_fingerprints b/compress_fingerprints
index 4c815fc..a436b75 100755
--- a/compress_fingerprints
+++ b/compress_fingerprints
@@ -11,6 +11,7 @@ initialize($conf);
 
 my %fps  = (); # records matching each fingerprint (and the lead)
 my %recs = (); # fingerprints belonging to each record
+my $lastscore = 0; # previous fingerprint's score
 
 open FP, '<', $ARGV[0] or die "Can't open input file: $!\n";
 
@@ -40,6 +41,13 @@ sub populate_fingerprint {
     $stripped   =~ s/[^A-Za-z0-9]//g;
     $fp{sha1}    = sha1_base64($stripped);
 
+    # make sure file is sorted properly
+    if ($lastscore and $fp{compact} > $lastscore) {
+        print "Input file is sorted improperly or unsorted.\n";
+        die "Sort descending (sort -ru) and rerun this script.\n";
+    }
+    $lastscore = $fp{compact};
+
     # populate records hash
     $recs{ $fp{id} }{ $fp{sha1} } = {};
 
@@ -80,7 +88,9 @@ sub rank_fingerprint {
 
 =head2 dump_records
 
-Writes out a 2-column file of lead and subordinate records.
+Writes out a 2-column file of lead and subordinate records. If
+posttest is enabled, a scan is also done to ensure that no recordid
+appears as both a subordinate and lead.
 
 =cut