Add new kind of aliases for field names, to be used only at query time

src/common/rclconfig.cpp

@@ -879,6 +879,21 @@ bool RclConfig::readFieldsConfig(const string& cnferrloc)
 	}
     }
 
+    // Query aliases map
+    tps = m_fields->getNames("queryaliases");
+    for (vector<string>::const_iterator it = tps.begin(); 
+         it != tps.end(); it++){
+	string canonic = stringtolower(*it); // canonic name
+	string aliases;
+	m_fields->get(canonic, aliases, "queryaliases");
+	vector<string> l;
+	stringToStrings(aliases, l);
+	for (vector<string>::const_iterator ait = l.begin();
+	     ait != l.end(); ait++) {
+	    m_aliastoqcanon[stringtolower(*ait)] = canonic;
+	}
+    }
+
 #if 0
     for (map<string, FieldTraits>::const_iterator it = m_fldtotraits.begin();
 	 it != m_fldtotraits.end(); it++) {
@@ -910,10 +925,10 @@ bool RclConfig::readFieldsConfig(const string& cnferrloc)
 }
 
 // Return specifics for field name:
-bool RclConfig::getFieldTraits(const string& _fld, const FieldTraits **ftpp)
+bool RclConfig::getFieldTraits(const string& _fld, const FieldTraits **ftpp,
-    const
+    bool isquery) const
 {
-    string fld = fieldCanon(_fld);
+    string fld = isquery ? fieldQCanon(_fld) : fieldCanon(_fld);
     map<string, FieldTraits>::const_iterator pit = m_fldtotraits.find(fld);
     if (pit != m_fldtotraits.end()) {
 	*ftpp = &pit->second;
@@ -952,6 +967,18 @@ string RclConfig::fieldCanon(const string& f) const
     return fld;
 }
 
+string RclConfig::fieldQCanon(const string& f) const
+{
+    string fld = stringtolower(f);
+    map<string, string>::const_iterator it = m_aliastoqcanon.find(fld);
+    if (it != m_aliastoqcanon.end()) {
+	LOGDEB1(("RclConfig::fieldQCanon: [%s] -> [%s]\n", 
+                f.c_str(), it->second.c_str()));
+	return it->second;
+    }
+    return fieldCanon(f);
+}
+
 vector<string> RclConfig::getFieldSectNames(const string &sk, const char* patrn)
     const
 {
@@ -1423,6 +1450,7 @@ void RclConfig::initFrom(const RclConfig& r)
 	m_ptrans = new ConfSimple(*(r.m_ptrans));
     m_fldtotraits = r.m_fldtotraits;
     m_aliastocanon = r.m_aliastocanon;
+    m_aliastoqcanon = r.m_aliastoqcanon;
     m_storedFields = r.m_storedFields;
     m_xattrtofld = r.m_xattrtofld;
     if (r.m_stopsuffixes)

src/common/rclconfig.h

@@ -241,8 +241,10 @@ class RclConfig {
     /** mimeconf: get query lang frag for named filter */
     bool getGuiFilter(const string& filtername, string& frag) const;
 
-    /** fields: get field prefix from field name */
+    /** fields: get field prefix from field name. Use additional query
-    bool getFieldTraits(const string& fldname, const FieldTraits **) const;
+       aliases if isquery is set */
+    bool getFieldTraits(const string& fldname, const FieldTraits **,
+        bool isquery = false) const;
 
     const set<string>& getStoredFields() const {return m_storedFields;}
 
@@ -251,6 +253,9 @@ class RclConfig {
     /** Get canonic name for possible alias */
     string fieldCanon(const string& fld) const;
 
+    /** Get canonic name for possible alias, including query-only aliases */
+    string fieldQCanon(const string& fld) const;
+
     /** Get xattr name to field names translations */
     const map<string, string>& getXattrToField() const {return m_xattrtofld;}
 
@@ -323,6 +328,7 @@ class RclConfig {
     ConfSimple            *m_ptrans; // Paths translations
     map<string, FieldTraits>  m_fldtotraits; // Field to field params
     map<string, string>  m_aliastocanon;
+    map<string, string>  m_aliastoqcanon;
     set<string>          m_storedFields;
     map<string, string>  m_xattrtofld;
 

src/python/recoll/pyrecoll.cpp

@@ -518,7 +518,7 @@ Doc_getattro(recoll_DocObject *self, PyObject *nameobj)
 	Py_RETURN_NONE;
     }
 
-    key = rclconfig->fieldCanon(string(name));
+    key = rclconfig->fieldQCanon(string(name));
 
     switch (key.at(0)) {
     case 'u':
@@ -640,7 +640,7 @@ Doc_setattr(recoll_DocObject *self, char *name, PyObject *value)
     }
     char* uvalue = PyBytes_AsString(putf8);
     Py_DECREF(putf8);
-    string key = rclconfig->fieldCanon(string(name));
+    string key = rclconfig->fieldQCanon(string(name));
 
     LOGDEB0(("Doc_setattr: [%s] (%s) -> [%s]\n", key.c_str(), name, uvalue));
     // We set the value in the meta array in all cases. Good idea ? or do it

src/rcldb/rcldb.cpp

@@ -1033,9 +1033,10 @@ bool Db::isopen()
 }
 
 // Try to translate field specification into field prefix. 
-bool Db::fieldToTraits(const string& fld, const FieldTraits **ftpp)
+bool Db::fieldToTraits(const string& fld, const FieldTraits **ftpp,
+                       bool isquery)
 {
-    if (m_config && m_config->getFieldTraits(fld, ftpp))
+    if (m_config && m_config->getFieldTraits(fld, ftpp, isquery))
 	return true;
 
     *ftpp = 0;

src/rcldb/rcldb.h

@@ -223,7 +223,8 @@ class Db {
     /* Return configured stop words */
     const StopList& getStopList() const {return m_stops;}
     /* Field name to prefix translation (ie: author -> 'A') */
-    bool fieldToTraits(const string& fldname, const FieldTraits **ftpp);
+    bool fieldToTraits(const string& fldname, const FieldTraits **ftpp,
+                       bool isquery = false);
 
     /* Update-related methods ******************************************/
 

src/rcldb/rclquery.cpp

@@ -160,7 +160,7 @@ void Query::setSortBy(const string& fld, bool ascending) {
     if (fld.empty()) {
 	m_sortField.erase();
     } else {
-	m_sortField = m_db->getConf()->fieldCanon(fld);
+	m_sortField = m_db->getConf()->fieldQCanon(fld);
 	m_sortAscending = ascending;
     }
     LOGDEB0(("RclQuery::setSortBy: [%s] %s\n", m_sortField.c_str(),

src/rcldb/rclterms.cpp

@@ -328,7 +328,7 @@ bool Db::idxTermMatch(int typ_sens, const string &lang, const string &root,
     string prefix;
     if (!field.empty()) {
 	const FieldTraits *ftp = 0;
-	if (!fieldToTraits(field, &ftp) || ftp->pfx.empty()) {
+	if (!fieldToTraits(field, &ftp, true) || ftp->pfx.empty()) {
             LOGDEB(("Db::termMatch: field is not indexed (no prefix): [%s]\n", 
                     field.c_str()));
         } else {

src/rcldb/searchdata.cpp

@@ -686,7 +686,7 @@ void SearchDataClauseSimple::processSimpleSpan(Rcl::Db &db, string& ermsg,
 
     string prefix;
     const FieldTraits *ftp;
-    if (!m_field.empty() && db.fieldToTraits(m_field, &ftp)) {
+    if (!m_field.empty() && db.fieldToTraits(m_field, &ftp, true)) {
 	prefix = wrap_prefix(ftp->pfx);
     }
 
@@ -743,7 +743,7 @@ void SearchDataClauseSimple::processPhraseOrNear(Rcl::Db &db, string& ermsg,
 
     string prefix;
     const FieldTraits *ftp;
-    if (!m_field.empty() && db.fieldToTraits(m_field, &ftp)) {
+    if (!m_field.empty() && db.fieldToTraits(m_field, &ftp, true)) {
 	prefix = wrap_prefix(ftp->pfx);
     }
 

src/sampleconf/fields

@@ -109,6 +109,13 @@ mtype = mime mimetype xesam:mimetype contenttype xesam:contenttype dc:format
 recipient = to xesam:recipient
 url = dc:identifier xesam:url
 
+##################
+# The queryaliases section defines aliases which are used exclusively at
+# query time: there is no risk to pick up a random field from a document
+# (e.g. an HTML meta field) and index it. 
+[queryaliases]
+#filename = fn
+
 [xattrtofields]
 ######################
 # Processing for extended file attributes.