X-Git-Url: http://matita.cs.unibo.it/gitweb/?a=blobdiff_plain;f=helm%2Fpapers%2Fuse_case%2Fstats%2Fstats.cc;h=f2b128faf89540d53d93ac115f6161923c53a6e6;hb=4167cea65ca58897d1a3dbb81ff95de5074700cc;hp=9892e12b67948113de7f7d61c833a7accf89e194;hpb=8fce28cde4e422993148e1fca011cae7ea230c16;p=helm.git

diff --git a/helm/papers/use_case/stats/stats.cc b/helm/papers/use_case/stats/stats.cc
index 9892e12b6..f2b128faf 100644
--- a/helm/papers/use_case/stats/stats.cc
+++ b/helm/papers/use_case/stats/stats.cc
@@ -1,39 +1,54 @@
 
+#include <vector>
 #include <cassert>
 #include <GdomeSmartDOM.hh>
 
+#include <ext/hash_map>
+namespace stdx = __gnu_cxx;
+
 namespace DOM = GdomeSmartDOM;
 
-unsigned n_elements;
-unsigned n_leaf_elements;
-unsigned n_text_nodes;
-unsigned n_blank_text_nodes;
-unsigned n_attributes;
-unsigned max_attributes;
-unsigned max_depth;
-unsigned max_children;
+int n_elements;
+int n_leaf_elements;
+int n_text_nodes;
+int n_blank_text_nodes;
+int n_attributes;
+int max_attributes;
+int max_children;
+stdx::hash_map<int,int> depths;
+std::vector<int> widths;
 
 bool
 is_blank(const std::string& s)
 {
   for (int i = 0; i < s.length(); i++)
-    if (!isblank(s[i])) return false;
+    if (!isspace(s[i])) return false;
   return true;
 }
 
 void
-visit(DOM::Node node, unsigned depth)
+add_depth(int depth)
+{
+  stdx::hash_map<int,int>::iterator p = depths.find(depth);
+  if (p != depths.end())
+    p->second++;
+  else
+    depths[depth] = 1;
+}
+
+void
+visit(DOM::Node node, int depth)
 {
   assert(node);
 
-  max_depth = std::max(max_depth, depth);
+  add_depth(depth);
 
   switch (node.get_nodeType())
     {
     case DOM::Node::ELEMENT_NODE:
       {
 	n_elements++;
-	const unsigned n_attrs = node.get_attributes().get_length();
+	const int n_attrs = node.get_attributes().get_length();
 	n_attributes += n_attrs;
 	max_attributes = std::max(max_attributes, n_attrs);
 	if (!node.get_firstChild()) n_leaf_elements++;
@@ -47,20 +62,45 @@ visit(DOM::Node node, unsigned depth)
       break;
     }
 
-  unsigned n_children = 0;
+  int n_children = 0;
   for (DOM::Node p = node.get_firstChild(); p; p = p.get_nextSibling())
     {
       visit(p, depth + 1);
       n_children++;
     }
   max_children = std::max(max_children, n_children);
+
+  if (node.get_firstChild())
+    widths.push_back(n_children);
 }
 
 void
-print_results(const std::string& URI)
+print_results(const std::string& URI, long size)
 {
+  int n_depths = 0;
+  int tot_depth = 0;
+  int max_depth = 0;
+  for (stdx::hash_map<int,int>::const_iterator p = depths.begin(); p != depths.end(); p++)
+    {
+      n_depths += p->second;
+      tot_depth += p->first * p->second;
+      max_depth = std::max(max_depth, p->first);
+    }
+
+  int tot_width = 0;
+  for (std::vector<int>::const_iterator p = widths.begin(); p != widths.end(); p++)
+    tot_width += *p;
+
   std::cout << "<stats for=\"" << URI << "\">" << std::endl;
-  std::cout << "  <depth>" << max_depth << "</depth>" << std::endl;
+  std::cout << "  <size>" << size << "</size>" << std::endl;
+  std::cout << "  <depth>" << std::endl;
+  std::cout << "    <max>" << max_depth << "</max>" << std::endl;
+  std::cout << "    <leaf-avg>" << tot_depth / ((double) n_depths) << "</leaf-avg>" << std::endl;
+  std::cout << "  </depth>" << std::endl;
+  std::cout << "  <width>" << std::endl;
+  std::cout << "    <max>" << max_children << "</max>" << std::endl;
+  std::cout << "    <inner-avg>" << tot_width / ((double) widths.size()) << "</inner-avg>" << std::endl;
+  std::cout << "  </width>" << std::endl;  
   std::cout << "  <elements>" << std::endl;
   std::cout << "    <total>" << n_elements << "</total>" << std::endl;
   std::cout << "    <leaf>" << n_leaf_elements << "</leaf>" << std::endl;
@@ -73,17 +113,19 @@ print_results(const std::string& URI)
   std::cout << "    <total>" << n_attributes << "</total>" << std::endl;
   std::cout << "    <max>" << max_attributes << "</max>" << std::endl;
   std::cout << "  </attributes>" << std::endl;
-  std::cout << "  <children>" << std::endl;
-  std::cout << "    <max>" << max_children << "</max>" << std::endl;
-  std::cout << "  </children>" << std::endl;  
   std::cout << "</stats>" << std::endl;
 }
 
 int
 main(int argc, char* argv[])
 {
+  if (argc != 3) {
+    std::cerr << "Usage: stats <URI> <size>" << std::endl;
+    return -1;
+  }
+
   DOM::DOMImplementation di;
   DOM::Document doc = di.createDocumentFromURI(argv[1]);
   visit(doc, 0);
-  print_results(argv[1]);
+  print_results(argv[1], atoi(argv[2]));
 }