graph__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="l00127"></a>00127           <span class="keywordflow">if</span>(!(line=parse_word()(line, word))) {
<a name="l00128"></a>00128             <span class="comment">// parse_word() failed</span>
<a name="l00129"></a>00129             <span class="keyword">delete</span>[] startline;
<a name="l00130"></a>00130             <span class="keywordflow">return</span> -1;
<a name="l00131"></a>00131           }
<a name="l00132"></a>00132           <span class="keywordflow">switch</span>(count) {
<a name="l00133"></a>00133             <span class="keywordflow">case</span> 0: vid=atoi(word); <span class="keywordflow">break</span>;
<a name="l00134"></a>00134             <span class="keywordflow">case</span> 1:
<a name="l00135"></a>00135               v_lbl=el_prsr.parse_element(word); 
<a name="l00142"></a>00142               vid_to_lbl.insert(make_pair(vid, v_lbl));
<a name="l00143"></a>00143           }
<a name="l00144"></a>00144           count++;
<a name="l00145"></a>00145           
<a name="l00146"></a>00146         }<span class="comment">//while(count&lt;..)</span>
<a name="l00147"></a>00147         
<a name="l00148"></a>00148       }<span class="comment">//if word[0]=='v'</span>
<a name="l00149"></a>00149       <span class="keywordflow">else</span> <span class="keywordflow">if</span>(word[0]==<span class="charliteral">'e'</span>) { <span class="comment">// undirected edge</span>
<a name="l00153"></a>00153 <span class="comment"></span>        <span class="keywordtype">int</span> vid1, vid2;
<a name="l00154"></a>00154         <span class="keyword">typename</span> GRAPH_PATTERN::EDGE_T e_lbl;
<a name="l00155"></a>00155         <span class="keyword">typename</span> GRAPH_PATTERN::VERTEX_T v_lbl1, v_lbl2;
<a name="l00156"></a>00156         num_items=3; <span class="comment">// 3 more words to be parsed</span>
<a name="l00157"></a>00157         <span class="keywordtype">bool</span> swap_vids; <span class="comment">// flag=false if v_lbl1&lt;v_lbl2</span>
<a name="l00158"></a>00158         
<a name="l00159"></a>00159         <span class="keywordflow">while</span>(count&lt;num_items) {
<a name="l00160"></a>00160           <span class="keywordflow">if</span>(!(line=parse_word()(line, word))) {
<a name="l00161"></a>00161             <span class="comment">// parse_word() failed</span>
<a name="l00162"></a>00162             <span class="keyword">delete</span>[] startline;
<a name="l00163"></a>00163             <span class="keywordflow">return</span> -1;
<a name="l00164"></a>00164           }
<a name="l00165"></a>00165           
<a name="l00166"></a>00166           <span class="keywordflow">switch</span>(count) {
<a name="l00167"></a>00167             <span class="keywordflow">case</span> 0: 
<a name="l00168"></a>00168               vid1=atoi(word); 
<a name="l00169"></a>00169               <span class="keywordflow">if</span>((tmp_it=vid_to_lbl.find(vid1))==vid_to_lbl.end()) {
<a name="l00170"></a>00170                 cerr&lt;&lt;<span class="stringliteral">"graph_tokenizer.parse_next_trans: vid "</span>&lt;&lt;vid1&lt;&lt;<span class="stringliteral">" not found in vid_to_lbl"</span>&lt;&lt;endl;
<a name="l00171"></a>00171                 <span class="keywordflow">return</span> -1;
<a name="l00172"></a>00172               }
<a name="l00173"></a>00173                 v_lbl1=tmp_it-&gt;second;
<a name="l00174"></a>00174               <span class="keywordflow">break</span>;
<a name="l00175"></a>00175               
<a name="l00176"></a>00176             <span class="keywordflow">case</span> 1: 
<a name="l00177"></a>00177               vid2=atoi(word);
<a name="l00178"></a>00178               <span class="keywordflow">if</span>((tmp_it=vid_to_lbl.find(vid2))==vid_to_lbl.end()) {
<a name="l00179"></a>00179                 cerr&lt;&lt;<span class="stringliteral">"graph_tokenizer.parse_next_trans: vid "</span>&lt;&lt;vid2&lt;&lt;<span class="stringliteral">" not found in vid_to_lbl"</span>&lt;&lt;endl;
<a name="l00180"></a>00180                 <span class="keywordflow">return</span> -1;
<a name="l00181"></a>00181               }
<a name="l00182"></a>00182                 v_lbl2=tmp_it-&gt;second;
<a name="l00183"></a>00183               <span class="keywordflow">break</span>;
<a name="l00184"></a>00184               
<a name="l00185"></a>00185             <span class="keywordflow">case</span> 2: 
<a name="l00186"></a>00186               e_lbl=edge_prsr.parse_element(word); 
<a name="l00193"></a>00193               
<a name="l00195"></a>00195               g1=<span class="keyword">new</span> GRAPH_PATTERN;
<a name="l00196"></a>00196               <span class="keywordflow">if</span>(v_lbl1&lt;=v_lbl2) {
<a name="l00197"></a>00197                 make_edge(g1, v_lbl1, v_lbl2, e_lbl);
<a name="l00198"></a>00198                 swap_vids=0;
<a name="l00199"></a>00199               }
<a name="l00200"></a>00200                 <span class="keywordflow">else</span> {
<a name="l00201"></a>00201                   make_edge(g1, v_lbl2, v_lbl1, e_lbl);
<a name="l00202"></a>00202                   swap_vids=1;
<a name="l00203"></a>00203                 }
<a name="l00204"></a>00204                 
<a name="l00209"></a>00209                 
<a name="l00210"></a>00210                 <span class="keywordflow">if</span>(!(gvat=vat_hmap.get_vat(g1))) { <span class="comment">// vat not found</span>
<a name="l00211"></a>00211                   gvat=<span class="keyword">new</span> <a class="code" href="classvat.html">VAT</a>;
<a name="l00212"></a>00212                   <span class="keywordflow">if</span>(!swap_vids)
<a name="l00213"></a>00213                     gvat-&gt;insert_occurrence_tid(tid, make_pair(vid1, vid2));
<a name="l00214"></a>00214                   <span class="keywordflow">else</span>
<a name="l00215"></a>00215                     gvat-&gt;insert_occurrence_tid(tid, make_pair(vid2, vid1));
<a name="l00216"></a>00216                   
<a name="l00217"></a>00217                   gvat-&gt;insert_vid_tid(vid1);
<a name="l00218"></a>00218                   gvat-&gt;insert_vid(vid2);
<a name="l00219"></a>00219                   vat_hmap.add_vat(g1, gvat); <span class="comment">// add pattern-vat mapping</span>
<a name="l00220"></a>00220                   freq_pats.push_back(g1); <span class="comment">// this is the first time </span>
<a name="l00221"></a>00221                                <span class="comment">// this pattern has been encountered, so add it</span>
<a name="l00222"></a>00222                 }
<a name="l00223"></a>00223                 <span class="keywordflow">else</span> <span class="keywordflow">if</span>(gvat-&gt;back().first!=tid) { <span class="comment">// or, new tid</span>
<a name="l00224"></a>00224                   <span class="keywordflow">if</span>(!swap_vids)
<a name="l00225"></a>00225                     gvat-&gt;insert_occurrence_tid(tid, make_pair(vid1, vid2));
<a name="l00226"></a>00226                   <span class="keywordflow">else</span>
<a name="l00227"></a>00227                     gvat-&gt;insert_occurrence_tid(tid, make_pair(vid2, vid1));
<a name="l00228"></a>00228                   
<a name="l00229"></a>00229                   gvat-&gt;insert_vid_tid(vid1);
<a name="l00230"></a>00230                   gvat-&gt;insert_vid(vid2);
<a name="l00231"></a>00231                   <span class="keyword">delete</span> g1;
<a name="l00232"></a>00232                 }
<a name="l00233"></a>00233                 <span class="keywordflow">else</span> { <span class="comment">// assert: gvat-&gt;back().first=tid</span>
<a name="l00234"></a>00234                   <span class="keywordflow">if</span>(!swap_vids)
<a name="l00235"></a>00235                     gvat-&gt;insert_occurrence(make_pair(vid1, vid2));
<a name="l00236"></a>00236                   <span class="keywordflow">else</span>
<a name="l00237"></a>00237                     gvat-&gt;insert_occurrence(make_pair(vid2, vid1));
<a name="l00238"></a>00238                   
<a name="l00239"></a>00239                   gvat-&gt;insert_vid_hs(vid1);
<a name="l00240"></a>00240                   gvat-&gt;insert_vid(vid2);
<a name="l00241"></a>00241                   <span class="keyword">delete</span> g1;
<a name="l00242"></a>00242                 }
<a name="l00243"></a>00243                 
<a name="l00244"></a>00244           }<span class="comment">//switch</span>
<a name="l00245"></a>00245           count++;
<a name="l00246"></a>00246         }<span class="comment">//while(count&lt;..)</span>
<a name="l00247"></a>00247         
<a name="l00248"></a>00248       }<span class="comment">//if(word[0]=='u')</span>
<a name="l00249"></a>00249       <span class="keywordflow">else</span> {
<a name="l00250"></a>00250         cerr&lt;&lt;<span class="stringliteral">"graph.tokenizer.parse_next_trans: Unidentifiable line="</span>&lt;&lt;line&lt;&lt;endl;
<a name="l00251"></a>00251         <span class="keywordflow">return</span> -1;
<a name="l00252"></a>00252       }
<a name="l00253"></a>00253     }<span class="comment">//while(1)</span>
<a name="l00254"></a>00254     
<a name="l00255"></a>00255     <span class="keywordflow">return</span> tid;
<a name="l00256"></a>00256     
<a name="l00257"></a>00257   }<span class="comment">//parse_next_trans()</span>
<a name="l00258"></a>00258   
<a name="l00259"></a>00259 <span class="keyword">private</span>:
<a name="l00260"></a>00260   <span class="keywordtype">int</span> MAXLINE; 
<a name="l00261"></a>00261   <a class="code" href="classelement__parser.html">element_parser&lt;typename GRAPH_PATTERN::VERTEX_T&gt;</a> el_prsr; 
<a name="l00262"></a>00262   <a class="code" href="classelement__parser.html">element_parser&lt;typename GRAPH_PATTERN::EDGE_T&gt;</a> edge_prsr; 
<a name="l00264"></a>00264 }; <span class="comment">//end class tokenizer</span>
<a name="l00265"></a>00265 
<a name="l00266"></a>00266 <span class="preprocessor">#endif</span>
<a name="l00267"></a>00267 <span class="preprocessor"></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>