* added .cc program (and equivalent .xsl stylesheet) for collecting

  statistics about XML files

diff --git a/helm/papers/use_case/stats/Makefile b/helm/papers/use_case/stats/Makefile
new file mode 100644 (file)
index 0000000..fa5f5e2
--- /dev/null
+++ b/helm/papers/use_case/stats/Makefile
@@ -0,0 +1,4 @@
+
+stats: stats.cc
+       g++ -o $@ `pkg-config gdome2-cpp-smart --cflags --libs` $<
+

diff --git a/helm/papers/use_case/stats/stats.cc b/helm/papers/use_case/stats/stats.cc
new file mode 100644 (file)
index 0000000..9892e12
--- /dev/null
+++ b/helm/papers/use_case/stats/stats.cc
@@ -0,0 +1,89 @@
+
+#include <cassert>
+#include <GdomeSmartDOM.hh>
+
+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;
+
+bool
+is_blank(const std::string& s)
+{
+  for (int i = 0; i < s.length(); i++)
+    if (!isblank(s[i])) return false;
+  return true;
+}
+
+void
+visit(DOM::Node node, unsigned depth)
+{
+  assert(node);
+
+  max_depth = std::max(max_depth, depth);
+
+  switch (node.get_nodeType())
+    {
+    case DOM::Node::ELEMENT_NODE:
+      {
+       n_elements++;
+       const unsigned 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++;
+      }
+      break;
+    case DOM::Node::TEXT_NODE:
+      n_text_nodes++;
+      if (is_blank(node.get_nodeValue())) n_blank_text_nodes++;
+      break;
+    case DOM::Node::ATTRIBUTE_NODE:
+      break;
+    }
+
+  unsigned 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);
+}
+
+void
+print_results(const std::string& URI)
+{
+  std::cout << "<stats for=\"" << URI << "\">" << std::endl;
+  std::cout << "  <depth>" << max_depth << "</depth>" << std::endl;
+  std::cout << "  <elements>" << std::endl;
+  std::cout << "    <total>" << n_elements << "</total>" << std::endl;
+  std::cout << "    <leaf>" << n_leaf_elements << "</leaf>" << std::endl;
+  std::cout << "  </elements>" << std::endl;
+  std::cout << "  <text-nodes>" << std::endl;
+  std::cout << "    <total>" << n_text_nodes << "</total>" << std::endl;
+  std::cout << "    <blank>" << n_blank_text_nodes << "</blank>" << std::endl;
+  std::cout << "  </text-nodes>" << std::endl;
+  std::cout << "  <attributes>" << std::endl;
+  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[])
+{
+  DOM::DOMImplementation di;
+  DOM::Document doc = di.createDocumentFromURI(argv[1]);
+  visit(doc, 0);
+  print_results(argv[1]);
+}

diff --git a/helm/papers/use_case/stats/stats.xsl b/helm/papers/use_case/stats/stats.xsl
new file mode 100644 (file)
index 0000000..ea643f0
--- /dev/null
+++ b/helm/papers/use_case/stats/stats.xsl
@@ -0,0 +1,130 @@
+<?xml version="1.0"?>
+
+<xsl:stylesheet version="1.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform">
+
+<xsl:param name="URI" select="''"/>
+
+<xsl:output method="xml"/>
+
+<xsl:template match="/">
+<stats for="{$URI}">
+  <xsl:message>Computing max depth...</xsl:message>
+  <max-depth>
+    <xsl:call-template name="max-depth2"/>
+  </max-depth>
+  <xsl:message>Computing number of nodes...</xsl:message>
+  <number-of-elements>
+    <xsl:call-template name="number-of-elements"/>
+  </number-of-elements>
+  <number-of-text-nodes>
+    <xsl:call-template name="number-of-text-nodes"/>
+  </number-of-text-nodes>
+  <number-of-blank-text-nodes>
+    <xsl:call-template name="number-of-blank-text-nodes"/>
+  </number-of-blank-text-nodes>
+  <xsl:message>Computing number of leaves...</xsl:message>
+  <number-of-leaves>
+    <xsl:call-template name="number-of-leaves"/>
+  </number-of-leaves>
+</stats>
+</xsl:template>
+
+<xsl:template name="number-of-elements">
+  <xsl:value-of select="count(//*)"/>
+</xsl:template>
+
+<xsl:template name="number-of-text-nodes">
+  <xsl:value-of select="count(//text())"/>
+</xsl:template>
+
+<xsl:template name="number-of-blank-text-nodes">
+  <xsl:value-of select="count(//text()[normalize-space(.) = ''])"/>
+</xsl:template>
+
+<xsl:template name="number-of-leaves">
+  <xsl:value-of select="count(//text()|//*[not(child::node())])"/>
+</xsl:template>
+
+<xsl:template name="max-depth-aux">
+  <xsl:param name="nodes" select="/.."/>
+  <xsl:param name="depth" select="0"/>
+  <xsl:param name="DEPTH" select="0"/>
+  <xsl:choose>
+    <xsl:when test="$nodes">
+      <xsl:variable name="head-depth">
+        <xsl:call-template name="max-depth-aux">
+         <xsl:with-param name="nodes" select="$nodes[1]/*"/>
+         <xsl:with-param name="depth" select="$depth + 1"/>
+         <xsl:with-param name="DEPTH" select="$DEPTH"/>
+       </xsl:call-template>
+      </xsl:variable>
+      <xsl:variable name="NEW_DEPTH">
+        <xsl:choose>
+         <xsl:when test="$head-depth &gt; $DEPTH">
+           <xsl:value-of select="$head-depth"/>
+         </xsl:when>
+         <xsl:otherwise>
+           <xsl:value-of select="$DEPTH"/>
+         </xsl:otherwise>
+       </xsl:choose>
+      </xsl:variable>
+      <xsl:call-template name="max-depth-aux">
+        <xsl:with-param name="nodes" select="$nodes[position() &gt; 1]"/>
+        <xsl:with-param name="depth" select="$depth"/>
+       <xsl:with-param name="DEPTH" select="$NEW_DEPTH"/>
+      </xsl:call-template>
+    </xsl:when>
+    <xsl:otherwise>
+      <xsl:choose>
+        <xsl:when test="$depth &gt; $DEPTH">
+         <xsl:value-of select="$depth"/>
+       </xsl:when>
+       <xsl:otherwise>
+          <xsl:value-of select="$DEPTH"/>
+       </xsl:otherwise>
+      </xsl:choose>
+    </xsl:otherwise>
+  </xsl:choose>
+</xsl:template>
+
+<xsl:template name="max-depth">
+  <xsl:call-template name="max-depth-aux">
+    <xsl:with-param name="nodes" select="//node()"/>
+  </xsl:call-template>
+</xsl:template>
+
+<xsl:template name="max-depth2-aux">
+  <xsl:param name="nodes" select="/.."/>
+  <xsl:param name="DEPTH" select="0"/>
+  <xsl:choose>
+    <xsl:when test="$nodes">
+      <xsl:variable name="depth" select="count($nodes[1]/ancestor::*)"/>
+      <xsl:variable name="NEW_DEPTH">
+        <xsl:choose>
+          <xsl:when test="$depth &gt; $DEPTH">
+           <xsl:value-of select="$depth"/>
+         </xsl:when>
+         <xsl:otherwise>
+           <xsl:value-of select="$DEPTH"/>
+         </xsl:otherwise>
+        </xsl:choose>
+      </xsl:variable>
+      <xsl:call-template name="max-depth2-aux">
+        <xsl:with-param name="nodes" select="$nodes[position() &gt; 1]"/>
+       <xsl:with-param name="DEPTH" select="$NEW_DEPTH"/>
+      </xsl:call-template>
+    </xsl:when>
+    <xsl:otherwise>
+      <xsl:value-of select="$DEPTH"/>
+    </xsl:otherwise>
+  </xsl:choose>
+</xsl:template>
+
+<xsl:template name="max-depth2">
+  <xsl:call-template name="max-depth2-aux">
+    <xsl:with-param name="nodes" select="//node()[not(child::node())]"/>
+  </xsl:call-template>
+</xsl:template>
+
+</xsl:stylesheet>
+