tree__tokenizer_8h-source.html

这是一个用于数据挖掘的常用算法的模板库（数据挖掘的C++模板库for UNIX)

<a name="l00114"></a>00114             <span class="keywordflow">break</span>;
<a name="l00115"></a>00115             
<a name="l00116"></a>00116           <span class="keywordflow">default</span>:
<a name="l00117"></a>00117             <span class="keyword">typename</span> TREE_PATTERN::VERTEX_T v=el_prsr.parse_element(word);
<a name="l00118"></a>00118             
<a name="l00119"></a>00119             <span class="keywordflow">if</span>(V_P::notEq(v, V_P::convert(BK_TRK))) {
<a name="l00120"></a>00120               dfs_id++;
<a name="l00121"></a>00121               depth++;
<a name="l00122"></a>00122               <span class="keywordflow">for</span>(<span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> i=0; i&lt;obj_data.size(); i++)
<a name="l00123"></a>00123                 <span class="keywordflow">if</span>(obj_data[i].second.second)
<a name="l00124"></a>00124                   obj_data[i].second.first++;
<a name="l00125"></a>00125               obj_data.push_back(make_pair(v, make_pair(0, 1)));
<a name="l00126"></a>00126             }
<a name="l00127"></a>00127               <span class="keywordflow">else</span> {
<a name="l00128"></a>00128                 depth--;
<a name="l00129"></a>00129                 <span class="keywordflow">for</span>(<span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> i=0; i&lt;obj_data.size(); i++) {
<a name="l00130"></a>00130                   obj_data[i].second.first--;
<a name="l00131"></a>00131                   <span class="keywordflow">if</span>(obj_data[i].second.second &amp;&amp; (obj_data[i].second.first&lt;0)) {
<a name="l00132"></a>00132                     <span class="comment">// add this pattern &amp; scope to VAT</span>
<a name="l00133"></a>00133                     obj_data[i].second.second=0;
<a name="l00134"></a>00134                     p=<span class="keyword">new</span> TREE_PATTERN;
<a name="l00135"></a>00135                     p-&gt;add_vertex(obj_data[i].first);
<a name="l00136"></a>00136                     p-&gt;init_canonical_code(obj_data[i].first);
<a name="l00137"></a>00137                     
<a name="l00138"></a>00138                     <span class="keywordflow">if</span>((tvat=vat_hmap.get_vat(p))) {
<a name="l00139"></a>00139                       <span class="comment">// vat found, check if this tid occurs</span>
<a name="l00140"></a>00140                       <span class="keyword">typename</span> VAT::IT it=tvat-&gt;end()-1;
<a name="l00141"></a>00141                       <span class="keywordflow">if</span>(tvat-&gt;size() &amp;&amp; it-&gt;first==tid)
<a name="l00142"></a>00142                         it-&gt;second.push_back(<a class="code" href="classtree__instance.html">TREE_INSTANCE</a>(i, dfs_id, depth, 1));
<a name="l00143"></a>00143                       <span class="keywordflow">else</span> {
<a name="l00144"></a>00144                         <span class="keyword">typename</span> VAT::template ST&lt;TREE_INSTANCE, ALLOC&lt;TREE_INSTANCE&gt; &gt; vec_inst;
<a name="l00145"></a>00145                         vec_inst.push_back(<a class="code" href="classtree__instance.html">TREE_INSTANCE</a>(i, dfs_id, depth, 1));
<a name="l00146"></a>00146                         tvat-&gt;push_back(make_pair(tid, vec_inst));
<a name="l00147"></a>00147                       }
<a name="l00148"></a>00148                       <span class="keyword">delete</span> p;
<a name="l00149"></a>00149                     }
<a name="l00150"></a>00150                     <span class="keywordflow">else</span> {
<a name="l00151"></a>00151                       <span class="comment">// vat not found, create new one and insert it</span>
<a name="l00152"></a>00152                       tvat=<span class="keyword">new</span> <a class="code" href="classvat.html">VAT</a>;
<a name="l00153"></a>00153                       <span class="keyword">typename</span> VAT::template ST&lt;TREE_INSTANCE, ALLOC&lt;TREE_INSTANCE&gt; &gt; vec_inst;
<a name="l00154"></a>00154                       vec_inst.push_back(<a class="code" href="classtree__instance.html">TREE_INSTANCE</a>(i, dfs_id, depth, 1));
<a name="l00155"></a>00155                       tvat-&gt;push_back(make_pair(tid, vec_inst));
<a name="l00156"></a>00156                       <span class="keywordflow">if</span>(!vat_hmap.add_vat(p, tvat)) {
<a name="l00157"></a>00157                         cerr&lt;&lt;<span class="stringliteral">"tokenizer.get_length_one(tree): add_vat failed"</span>&lt;&lt;endl;
<a name="l00158"></a>00158                         <span class="keywordflow">return</span> -1;
<a name="l00159"></a>00159                       }
<a name="l00160"></a>00160                       freq_pats.push_back(p);
<a name="l00161"></a>00161                     }
<a name="l00162"></a>00162                     
<a name="l00163"></a>00163                   }<span class="comment">//end if obj_data..&lt;0</span>
<a name="l00164"></a>00164                 }<span class="comment">//end for i</span>
<a name="l00165"></a>00165               }<span class="comment">//end else v!=BK_TRK</span>
<a name="l00166"></a>00166               
<a name="l00167"></a>00167         }<span class="comment">//end switch</span>
<a name="l00168"></a>00168         
<a name="l00169"></a>00169       }<span class="comment">//end while(count&lt;..)</span>
<a name="l00170"></a>00170       
<a name="l00171"></a>00171       <span class="comment">// special case: root of tree has not been added yet</span>
<a name="l00172"></a>00172       <span class="keywordflow">if</span>(!obj_data[0].second.second) {
<a name="l00173"></a>00173         cerr&lt;&lt;<span class="stringliteral">"tokenizer.get_length_one(tree): root of tree has invalid scope"</span>&lt;&lt;endl;
<a name="l00174"></a>00174         <span class="keywordflow">return</span> -1;
<a name="l00175"></a>00175       }
<a name="l00176"></a>00176       
<a name="l00177"></a>00177       <span class="comment">// add its pattern &amp; scope to VAT</span>
<a name="l00178"></a>00178       obj_data[0].second.second=0;
<a name="l00179"></a>00179       p=<span class="keyword">new</span> TREE_PATTERN;
<a name="l00180"></a>00180       p-&gt;add_vertex(obj_data[0].first);
<a name="l00181"></a>00181       p-&gt;init_canonical_code(obj_data[0].first);
<a name="l00182"></a>00182       
<a name="l00183"></a>00183       <span class="keywordflow">if</span>((tvat=vat_hmap.get_vat(p))) {
<a name="l00184"></a>00184         <span class="comment">// vat found, check if this tid occurs</span>
<a name="l00185"></a>00185         <span class="keyword">typename</span> VAT::IT it=tvat-&gt;end()-1;
<a name="l00186"></a>00186         <span class="keywordflow">if</span>(tvat-&gt;size() &amp;&amp; it-&gt;first==tid)
<a name="l00187"></a>00187           it-&gt;second.push_back(<a class="code" href="classtree__instance.html">TREE_INSTANCE</a>(0, dfs_id, 0, 1));
<a name="l00188"></a>00188         <span class="keywordflow">else</span> {
<a name="l00189"></a>00189           <span class="keyword">typename</span> VAT::template ST&lt;TREE_INSTANCE, ALLOC&lt;TREE_INSTANCE&gt; &gt; vec_inst;
<a name="l00190"></a>00190           vec_inst.push_back(<a class="code" href="classtree__instance.html">TREE_INSTANCE</a>(0, dfs_id, 0, 1));
<a name="l00191"></a>00191           tvat-&gt;push_back(make_pair(tid, vec_inst));
<a name="l00192"></a>00192         }
<a name="l00193"></a>00193         <span class="keyword">delete</span> p;
<a name="l00194"></a>00194       }
<a name="l00195"></a>00195       <span class="keywordflow">else</span> {
<a name="l00196"></a>00196         <span class="comment">// vat not found, create new one and insert it</span>
<a name="l00197"></a>00197         tvat=<span class="keyword">new</span> <a class="code" href="classvat.html">VAT</a>;
<a name="l00198"></a>00198         <span class="keyword">typename</span> VAT::template ST&lt;TREE_INSTANCE, ALLOC&lt;TREE_INSTANCE&gt; &gt; vec_inst;
<a name="l00199"></a>00199         vec_inst.push_back(<a class="code" href="classtree__instance.html">TREE_INSTANCE</a>(0, dfs_id, 0, 1));
<a name="l00200"></a>00200         tvat-&gt;push_back(make_pair(tid, vec_inst));
<a name="l00201"></a>00201         <span class="keywordflow">if</span>(!vat_hmap.add_vat(p, tvat)) {
<a name="l00202"></a>00202           cerr&lt;&lt;<span class="stringliteral">"tokenizer.get_length_one(tree): add_vat failed"</span>&lt;&lt;endl;
<a name="l00203"></a>00203           <span class="keywordflow">return</span> -1;
<a name="l00204"></a>00204         }
<a name="l00205"></a>00205         freq_pats.push_back(p);
<a name="l00206"></a>00206       }
<a name="l00207"></a>00207       
<a name="l00208"></a>00208       <span class="comment">//cout&lt;&lt;"tokenizer(trees).parse_next_trans exit"&lt;&lt;endl;      </span>
<a name="l00209"></a>00209       <span class="keyword">delete</span>[] startline;
<a name="l00210"></a>00210       <span class="keywordflow">return</span> tid;
<a name="l00211"></a>00211       
<a name="l00212"></a>00212     }<span class="comment">//end parse_next_trans()</span>
<a name="l00213"></a>00213   
<a name="l00214"></a>00214 <span class="keyword">private</span>:
<a name="l00215"></a>00215     <span class="keyword">const</span> <span class="keywordtype">int</span> MAXLINE;
<a name="l00216"></a>00216   <a class="code" href="classelement__parser.html">element_parser&lt;typename TREE_PATTERN::VERTEX_T&gt;</a> el_prsr;
<a name="l00217"></a>00217   
<a name="l00218"></a>00218 };<span class="comment">//end class tokenizer for trees</span>
<a name="l00219"></a>00219 
<a name="l00220"></a>00220 <span class="preprocessor">#endif</span>
</pre></div><hr size="1"><address style="align: right;"><small>Generated on Wed Jul 26 14:01:08 2006 for DMTL by&nbsp;
<a href="http://www.doxygen.org/index.html">
<img src="doxygen.png" alt="doxygen" align="middle" border="0"></a> 1.4.7 </small></address>
</body>
</html>