GUI snippets window: add options for the max list length and for sorting the snippets by page number

src/rcldb/rclabsfromtext.cpp

@@ -138,12 +138,14 @@ public:
         if (maxtermcount && termcount++ > maxtermcount) {
             LOGINF("Rclabsfromtext: stopping because maxtermcount reached: "<<
                    maxtermcount << endl);
+            retflags |= ABSRES_TRUNC;
             return false;
         }
         // Also limit the number of fragments (just in case safety)
         if (m_fragments.size() > maxtermcount / 100) {
             LOGINF("Rclabsfromtext: stopping because maxfragments reached: "<<
                    maxtermcount/100 << endl);
+            retflags |= ABSRES_TRUNC;
             return false;
         }
         // Remember recent past
@@ -326,6 +328,10 @@ public:
 
         return;
     }
+
+    int getretflags() {
+        return retflags;
+    }
     
 private:
     // Past terms because we need to go back for context before a hit
@@ -364,6 +370,7 @@ private:
 
     unsigned int termcount{0};
     unsigned int maxtermcount{0};
+    int retflags{0};
 };
 
 int Query::Native::abstractFromText(
@@ -375,7 +382,8 @@ int Query::Native::abstractFromText(
     int ctxwords,
     unsigned int maxtotaloccs,
     vector<Snippet>& vabs,
-    Chrono& chron
+    Chrono& chron,
+    bool sortbypage
     )
 {
     (void)chron;
@@ -423,13 +431,21 @@ int Query::Native::abstractFromText(
     // Sort the fragments by decreasing weight
     const vector<MatchFragment>& res1 = splitter.getFragments();
     vector<MatchFragment> result(res1.begin(), res1.end());
-    std::sort(result.begin(), result.end(),
-              [](const MatchFragment& a,
-                 const MatchFragment& b) -> bool { 
-                  return a.coef > b.coef; 
-              }
-        );
-
+    if (sortbypage) {
+        std::sort(result.begin(), result.end(),
+                  [](const MatchFragment& a,
+                     const MatchFragment& b) -> bool { 
+                      return a.hitpos < b.hitpos; 
+                  }
+            );
+    } else {
+        std::sort(result.begin(), result.end(),
+                  [](const MatchFragment& a,
+                     const MatchFragment& b) -> bool { 
+                      return a.coef > b.coef; 
+                  }
+            );
+    }
     vector<int> vpbreaks;
     ndb->getPagePositions(docid, vpbreaks);
 
@@ -464,7 +480,7 @@ int Query::Native::abstractFromText(
         if (count++ >= maxtotaloccs)
             break;
     }
-    return ABSRES_OK;
+    return ABSRES_OK | splitter.getretflags();
 }
 
 }

src/rcldb/rclabstract.cpp

@@ -623,11 +623,12 @@ int Query::Native::abstractFromIndex(
 // @param[out] vabs the abstract is returned as a vector of snippets.
 int Query::Native::makeAbstract(Xapian::docid docid,
                                 vector<Snippet>& vabs, 
-                                int imaxoccs, int ictxwords)
+                                int imaxoccs, int ictxwords, bool sortbypage)
 {
     chron.restart();
-    LOGABS("makeAbstract: docid " << docid << " imaxoccs " <<
-           imaxoccs << " ictxwords " << ictxwords << "\n");
+    LOGDEB("makeAbstract: docid " << docid << " imaxoccs " <<
+           imaxoccs << " ictxwords " << ictxwords << " sort by page " <<
+           sortbypage << "\n");
 
     // The (unprefixed) terms matched by this document
     vector<string> matchedTerms;
@@ -675,7 +676,7 @@ int Query::Native::makeAbstract(Xapian::docid docid,
     if (ndb->m_storetext) {
         return abstractFromText(ndb, docid, matchedTerms, byQ,
                                 totalweight, ctxwords, maxtotaloccs, vabs,
-                                chron);
+                                chron, sortbypage);
     } else {
         return abstractFromIndex(ndb, docid, matchedTerms, byQ,
                                  totalweight, ctxwords, maxtotaloccs, vabs,

src/rcldb/rclquery.cpp

@@ -267,7 +267,7 @@ bool Query::getQueryTerms(vector<string>& terms)
 }
 
 int Query::makeDocAbstract(const Doc &doc, vector<Snippet>& abstract, 
-                           int maxoccs, int ctxwords)
+                           int maxoccs, int ctxwords, bool sortbypage)
 {
     LOGDEB("makeDocAbstract: maxoccs " << maxoccs << " ctxwords "  <<
            ctxwords << "\n");
@@ -276,7 +276,7 @@ int Query::makeDocAbstract(const Doc &doc, vector<Snippet>& abstract,
         return ABSRES_ERROR;
     }
     int ret = ABSRES_ERROR;
-    XAPTRY(ret = m_nq->makeAbstract(doc.xdocid, abstract, maxoccs, ctxwords),
+    XAPTRY(ret = m_nq->makeAbstract(doc.xdocid, abstract, maxoccs, ctxwords, sortbypage),
            m_db->m_ndb->xrdb, m_reason);
     if (!m_reason.empty()) {
         LOGDEB("makeDocAbstract: makeAbstract: reason: " << m_reason << "\n");

src/rcldb/rclquery.h

@@ -107,7 +107,7 @@ public:
     bool makeDocAbstract(const Doc &doc, std::vector<std::string>& abstract);
     // Returned as a vector of pair<page,snippet> page is 0 if unknown
     int makeDocAbstract(const Doc &doc, std::vector<Snippet>& abst, 
-                        int maxoccs= -1, int ctxwords = -1);
+                        int maxoccs= -1, int ctxwords = -1, bool sortbypage=false);
     /** Retrieve page number for first match for "significant" query term 
      *  @param term returns the chosen term */
     int getFirstMatchPage(const Doc &doc, std::string& term);

src/rcldb/rclquery_p.h

@@ -55,7 +55,7 @@ public:
     /** Return a list of terms which matched for a specific result document */
     bool getMatchTerms(unsigned long xdocid, std::vector<std::string>& terms);
     int makeAbstract(Xapian::docid id, std::vector<Snippet>&,
-                     int maxoccs = -1, int ctxwords = -1);
+                     int maxoccs, int ctxwords, bool sortbypage);
     int getFirstMatchPage(Xapian::docid docid, std::string& term);
     void setDbWideQTermsFreqs();
     double qualityTerms(Xapian::docid docid, 
@@ -109,7 +109,8 @@ public:
         int ctxwords,
         unsigned int maxtotaloccs,
         vector<Snippet>& vabs,
-        Chrono& chron
+        Chrono& chron,
+        bool sortbypage
         );
 };