From: Eric S. Raymond <esr@thyrsus.com>
Date: Tue, 16 Sep 2003 21:19:07 +0000 (+0000)
Subject: Polish the statistics object.
X-Git-Tag: 2.0~33
X-Git-Url: http://thyrsus.com/gitweb/?a=commitdiff_plain;h=9efef539b7a3ac7d2f36675c76bd85ccb8401e62;p=comparator.git

Polish the statistics object.
---

diff --git a/comparator.py b/comparator.py
index 1a6e15d..d2ea298 100644
--- a/comparator.py
+++ b/comparator.py
@@ -43,7 +43,7 @@ junk = {
         )),
     }
 
-def nontrivial(filetype, text):
+def nontrivial(filetype, text, debug=None):
     "Identify a shred as trivial or nontrivial."
     if filetype:
         if debug:
@@ -203,10 +203,8 @@ class CommonReport:
         filtered = []
         for clique in self.cliques:
             (type, text) = self.extract_text(clique)
-            if text:
+            if nontrivial(type, text):
                 filtered.append(clique)
-            else:
-                raise ComparatorException("couldn't extracrt file text")
         self.cliques = filtered
 
 class CommonStatistics:
@@ -214,12 +212,14 @@ class CommonStatistics:
     def __init__(self, report):
         self.chunks = report.segment_count()
         self.trees = []
+        self.normalization = report.normalization
         for (tree, totallines) in report.trees.items():
             lines = report.line_count(tree)
             self.trees.append((tree, lines, totallines))
 
     def __str__(self):
-        rep = "%d overlaps\n" % self.chunks
+        rep = "Trees: " + " ".join(map(lambda x: x[0], self.trees))
+        rep = "%d overlaps with %s normalization\n" % (self.chunks, self.normalization)
         for (tree, lines, total) in self.trees:
-            rep += "%s:%d:%d\n" % (tree, lines, total)
+            rep += "%s:\t%d of %d (%02.2f%%)\n" % (tree, lines, total, (lines * 100.0)/total)
         return rep
diff --git a/filterator b/filterator
index 608b413..5316b0d 100755
--- a/filterator
+++ b/filterator
@@ -16,12 +16,8 @@ if __name__ == '__main__':
         print "Shred-Size: %d" % state.shredsize
         if minsize:
             print "Minimum-Size: %d" % minsize
-        if stats:
-            print 75 * '-'
-            print "# %d overlaps." % state.segment_count()
-            for (tree, totallines) in state.trees.items():
-                lines = state.line_count(tree)
-                print "# %f%% of %s" % ((lines*100.0)/totallines, tree)
+        print 75 * '-'
+        sys.stdout.write(str(comparator.CommonStatistics(state)))
         print 75 * '-'
         for clique in state.cliques:
             for (file, start, end) in clique: