expose abstract synthesis to let users decide when they want it done

src/rcldb/rcldb.cpp

@@ -1,5 +1,5 @@
 #ifndef lint
-static char rcsid[] = "@(#$Id: rcldb.cpp,v 1.97 2006-11-20 15:28:57 dockes Exp $ (C) 2004 J.F.Dockes";
+static char rcsid[] = "@(#$Id: rcldb.cpp,v 1.98 2006-12-05 15:17:59 dockes Exp $ (C) 2004 J.F.Dockes";
 #endif
 /*
  *   This program is free software; you can redistribute it and/or modify
@@ -107,10 +107,7 @@ class Native {
 
     string makeAbstract(Xapian::docid id, const list<string>& terms);
 
-    bool dbDataToRclDoc(std::string &data, Doc &doc, 
+    bool dbDataToRclDoc(Xapian::docid docid, std::string &data, Doc &doc);
-			int qopts,
-			Xapian::docid docid,
-			const list<string>& terms);
 
     /** Compute list of subdocuments for a given path (given by hash) 
      *  We look for all Q terms beginning with the path/hash
@@ -177,11 +174,10 @@ bool Native::subDocs(const string &hash, vector<Xapian::docid>& docids)
     return false;
 }
 
-bool Native::dbDataToRclDoc(std::string &data, Doc &doc, 
+// Turn data record from db into document fields
-			    int qopts,
+bool Native::dbDataToRclDoc(Xapian::docid docid, std::string &data, Doc &doc)
-			    Xapian::docid docid, const list<string>& terms)
 {
-    LOGDEB1(("Db::dbDataToRclDoc: opts %x data: %s\n", qopts, data.c_str()));
+    LOGDEB1(("Db::dbDataToRclDoc: data: %s\n", data.c_str()));
     ConfSimple parms(&data);
     if (!parms.ok())
 	return false;
@@ -195,17 +191,10 @@ bool Native::dbDataToRclDoc(std::string &data, Doc &doc,
     parms.get(string("abstract"), doc.abstract);
     // Possibly remove synthetic abstract indicator (if it's there, we
     // used to index the beginning of the text as abstract).
-    bool syntabs = false;
+    doc.syntabs = false;
     if (doc.abstract.find(rclSyntAbs) == 0) {
 	doc.abstract = doc.abstract.substr(rclSyntAbs.length());
-	syntabs = true;
+	doc.syntabs = true;
-    }
-    // If the option is set and the abstract is synthetic or empty , build 
-    // abstract from position data. 
-    if ((qopts & Db::QO_BUILD_ABSTRACT) && !terms.empty()) {
-	if (doc.abstract.empty() || syntabs || 
-	    (qopts & Db::QO_REPLACE_ABSTRACT))
-	    doc.abstract = makeAbstract(docid, terms);
     }
     parms.get(string("ipath"), doc.ipath);
     parms.get(string("fbytes"), doc.fbytes);
@@ -1611,11 +1600,21 @@ bool Db::getDoc(int exti, Doc &doc, int *percent)
 
     // Parse xapian document's data and populate doc fields
     string data = xdoc.get_data();
-    list<string> terms;
+    return m_ndb->dbDataToRclDoc(docid, data, doc);
-    getQueryTerms(terms);
-    return m_ndb->dbDataToRclDoc(data, doc, m_qOpts, docid, terms);
 }
 
+bool Db::makeDocAbstract(Doc &doc, string& abstract)
+{
+    LOGDEB1(("Db::makeDocAbstract: exti %d\n", exti));
+    if (!m_ndb || !m_ndb->enquire) {
+	LOGERR(("Db::makeDocAbstract: no query opened\n"));
+	return false;
+    }
+    list<string> terms;
+    getQueryTerms(terms);
+    abstract = m_ndb->makeAbstract(doc.xdocid, terms);
+    return true;
+}
 
 // Retrieve document defined by file name and internal path. 
 bool Db::getDoc(const string &fn, const string &ipath, Doc &doc, int *pc)
@@ -1651,7 +1650,7 @@ bool Db::getDoc(const string &fn, const string &ipath, Doc &doc, int *pc)
 	Xapian::Document xdoc = m_ndb->db.get_document(*docid);
 	string data = xdoc.get_data();
 	list<string> terms;
-	return m_ndb->dbDataToRclDoc(data, doc, QO_NONE, *docid, terms);
+	return m_ndb->dbDataToRclDoc(*docid, data, doc);
     } catch (const Xapian::Error &e) {
 	ermsg = e.get_msg().c_str();
     } catch (const string &s) {

src/rcldb/rcldb.h

@@ -16,7 +16,7 @@
  */
 #ifndef _DB_H_INCLUDED_
 #define _DB_H_INCLUDED_
-/* @(#$Id: rcldb.h,v 1.42 2006-11-14 13:55:43 dockes Exp $  (C) 2004 J.F.Dockes */
+/* @(#$Id: rcldb.h,v 1.43 2006-12-05 15:17:59 dockes Exp $  (C) 2004 J.F.Dockes */
 
 #include <string>
 #include <list>
@@ -74,6 +74,8 @@ class Doc {
     string title;        // Possibly set by handler
     string keywords;     // Possibly set by handler
     string abstract;     // Possibly set by handler
+    bool   syntabs;      // true if abstract is just the top of doc, not an 
+                         // explicit document attribute
     string fbytes;       // File size. Set by Db::Add
     string dbytes;       // Doc size. Set by Db::Add from text length
 
@@ -96,6 +98,7 @@ class Doc {
 	title.erase();
 	keywords.erase();
 	abstract.erase();
+	syntabs = false;
 	fbytes.erase();
 	dbytes.erase();
 
@@ -119,8 +122,7 @@ class Db {
 
     enum OpenMode {DbRO, DbUpd, DbTrunc};
     // KEEP_UPDATED is internal use by reOpen() only
-    enum QueryOpts {QO_NONE=0, QO_STEM = 1, QO_BUILD_ABSTRACT = 2,
+    enum QueryOpts {QO_NONE=0, QO_STEM = 1, QO_KEEP_UPDATED = 8};
-		    QO_REPLACE_ABSTRACT = 4, QO_KEEP_UPDATED = 8};
 
     bool open(const string &dbdir, OpenMode mode, int qops = QO_NONE);
     bool close();
@@ -184,6 +186,9 @@ class Db {
     */
     bool getDoc(int i, Doc &doc, int *percent = 0);
 
+    /* Build synthetic abstract out of query terms and term position data */
+    bool makeDocAbstract(Doc &doc, string& abstract);
+
     /** Get document for given filename and ipath */
     bool getDoc(const string &fn, const string &ipath, Doc &doc, int *percent);