Changeset 22923

Timestamp:

2010-09-20T10:52:00+12:00 (13 years ago)

Author:

davidb

Message:

Main change is to use reverse iterator (see comment in code), but also added a bit more white space, and { } on if-statements etc.

File:

• main/trunk/greenstone2/runtime-src/src/recpt/summarise.cpp (modified) (6 diffs)

main/trunk/greenstone2/runtime-src/src/recpt/summarise.cpp

@@ -33 +33 @@
 #include <string.h>
 
+#include <iostream>
+using namespace std;
 
 /* **************** LOCAL PROTOTYPES **************** */
@@ -99 +101 @@
 *****************************************************/
 
-text_t summarise_keywords(text_t &htmlstr, text_t &query, int summaryLength) {
+text_t summarise_keywords(text_t &htmlstr, text_t &query, int summaryLength) 
+{
 
   if ((query.size()==0) || (htmlstr.size()==0)) {
@@ -109 +112 @@
 
   // consider only non-empty terms
-  for(text_tarray::iterator term = allterms.begin();
+  for (text_tarray::iterator term = allterms.begin();
       term < allterms.end(); ++term) {
-    if(!(*term).empty())
+    if (!(*term).empty())
       terms.push_back(*term);
   }
@@ -130 +133 @@
     //   answers[0] contains sentences with 1 keyword
     //   answers[1] contains sentences with 2 keywords, etc.
+
   vector<int> answersSize(terms.size());
     // answersSize[0] is the combined size of sentences with 1 keyword, etc.
   for(vector<int>::iterator size = answersSize.begin();
-      size<answersSize.end(); ++size)
+      size<answersSize.end(); ++size) {
     *size = 0; // initialise sentence size
+  }
 
   int totfound  = 0;
   text_t::iterator str_current = str_start;
-  while(str_current<str_end && answersSize[terms.size()-1]<summaryLength) {
+  while (str_current<str_end && answersSize[terms.size()-1]<summaryLength) {
     // if the size of best sentences is greater than summary, that's enough!
     text_t sentence = next_sentence(str_current,str_end);
@@ -144 +149 @@
     text_tarray::iterator terms_current = terms_start;
     int nFound = 0;
-    while(terms_current!=terms_end) {
+    while (terms_current!=terms_end) {
       text_t::iterator word = findword(sentence.begin(),sentence.end(),
                        *terms_current);
-      if(word!=sentence.end())
-    {        ++nFound; ++totfound; }
+      if (word!=sentence.end()) {
+        ++nFound; 
+    ++totfound; 
+      }
       ++terms_current;
     }
 
-    if(nFound>0 && answersSize[nFound-1]<summaryLength) {
+    if (nFound>0 && answersSize[nFound-1]<summaryLength) {
       answers[nFound-1].push_back(sentence);
       answersSize[nFound-1] += sentence.size();
@@ -159 +166 @@
 
   text_t answer;
-  for(vector<text_tarray>::iterator sentarray = answers.end()-1;
-      sentarray>=answers.begin(); --sentarray)
-    for(text_tarray::iterator sentence = (*sentarray).begin(); 
+
+  // Changed to using reverse iterator, as there is some concern as to
+  // whether the operations encoded with the usual iterator -- e.g. 
+  // answers.end()-1 and so forth -- are safe.  Certainly the code
+  // works out tidier using the reverse iterator and the segmentation
+  // fault that was occurring in this block went away
+
+  for (vector<text_tarray>::reverse_iterator sentarray = answers.rbegin();
+      sentarray<answers.rend(); ++sentarray) {
+    for (text_tarray::iterator sentence = (*sentarray).begin(); 
         sentence < (*sentarray).end(); ++sentence) {
       answer.append(*sentence);
-      if(answer.size()>=summaryLength)
+
+      if(answer.size()>=summaryLength) {
         return answer;
-    }
-
-  if(!answer.empty())
+      }
+    }
+  }
+
+  if (!answer.empty()) {
     return answer;
+  }
 
   return summarise_startend(htmlstr,summaryLength);