added basic facility for result path translation

src/common/rclconfig.cpp

@@ -95,6 +95,7 @@ void RclConfig::zeroMe() {
     mimeconf = 0; 
     mimeview = 0; 
     m_fields = 0;
+    m_ptrans = 0;
     m_stopsuffixes = 0;
     m_maxsufflen = 0;
     m_stpsuffstate.init(this, 0, "recoll_noindex");
@@ -217,6 +218,8 @@ RclConfig::RclConfig(const string *argcnf)
     if (!readFieldsConfig(cnferrloc))
 	return;
 
+    m_ptrans = new ConfSimple(path_cat(m_confdir, "ptrans").c_str(), 1);
+
     m_ok = true;
     setKeyDir(cstr_null);
 
@@ -744,7 +747,7 @@ bool RclConfig::getFieldTraits(const string& _fld, const FieldTraits **ftpp)
 		 _fld.c_str(), pit->second.pfx.c_str()));
 	return true;
     } else {
-	LOGDEB1(("RclConfig::readFieldsConfig: no prefix for field [%s]\n",
+	LOGDEB1(("RclConfig::getFieldTraits: no prefix for field [%s]\n",
 		 fld.c_str()));
 	*ftpp = 0;
 	return false;
@@ -936,6 +939,41 @@ string RclConfig::getDbDir() const
     return path_canon(dbdir);
 }
 
+void RclConfig::urlrewrite(const string& dbdir, string& url) const
+{
+    LOGDEB2(("RclConfig::urlrewrite: dbdir [%s] url [%s]\n",
+	    dbdir.c_str(), url.c_str()));
+
+    // Do path translations exist for this index ?
+    if (m_ptrans == 0 || !m_ptrans->hasSubKey(dbdir)) {
+	LOGDEB2(("RclConfig::urlrewrite: no paths translations (m_ptrans %p)\n",
+		m_ptrans));
+	return;
+    }
+
+    string path = fileurltolocalpath(url);
+    if (path.empty()) {
+	LOGDEB2(("RclConfig::urlrewrite: not file url\n"));
+	return;
+    }
+
+    // For each translation check if the prefix matches the input path,
+    // replace and return the result if it does.
+    vector<string> opaths = m_ptrans->getNames(dbdir);
+    for (vector<string>::const_iterator it = opaths.begin(); 
+	 it != opaths.end(); it++) {
+	if (it->size() <= path.size() && !path.compare(0, it->size(), *it)) {
+	    string npath;
+	    // This call always succeeds because the key comes from getNames()
+	    if (m_ptrans->get(*it, npath, dbdir)) { 
+		path = path.replace(0, it->size(), npath);
+		url = "file://" + path;
+	    }
+	    break;
+	}
+    }
+}
+
 bool RclConfig::sourceChanged() const
 {
     if (m_conf && m_conf->sourceChanged())
@@ -948,6 +986,8 @@ bool RclConfig::sourceChanged() const
 	return true;
     if (m_fields && m_fields->sourceChanged())
 	return true;
+    if (m_ptrans && m_ptrans->sourceChanged())
+	return true;
     return false;
 }
 
@@ -1179,6 +1219,7 @@ void RclConfig::freeAll()
     delete mimeconf; 
     delete mimeview; 
     delete m_fields;
+    delete m_ptrans;
     delete STOPSUFFIXES;
     // just in case
     zeroMe();
@@ -1204,6 +1245,8 @@ void RclConfig::initFrom(const RclConfig& r)
 	mimeview = new ConfStack<ConfSimple>(*(r.mimeview));
     if (r.m_fields)
 	m_fields = new ConfStack<ConfSimple>(*(r.m_fields));
+    if (r.m_ptrans)
+	m_ptrans = new ConfSimple(*(r.m_ptrans));
     m_fldtotraits = r.m_fldtotraits;
     m_aliastocanon = r.m_aliastocanon;
     m_storedFields = r.m_storedFields;

src/common/rclconfig.h

@@ -150,6 +150,8 @@ class RclConfig {
     string getPidfile() const;
     /** Get indexing status file name */
     string getIdxStatusFile() const;
+    /** Do path translation according to the ptrans table */
+    void urlrewrite(const string& dbdir, string& url) const;
 
     /** Get Web Queue directory name */
     string getWebQueueDir() const;
@@ -279,6 +281,7 @@ class RclConfig {
     ConfStack<ConfSimple> *mimeconf; // but their content may depend on it.
     ConfStack<ConfSimple> *mimeview; // 
     ConfStack<ConfSimple> *m_fields;
+    ConfSimple            *m_ptrans; // Paths translations
     map<string, FieldTraits>  m_fldtotraits; // Field to field params
     map<string, string>  m_aliastocanon;
     set<string>          m_storedFields;

src/rcldb/rcldb.cpp

@@ -233,8 +233,21 @@ bool Db::Native::dbDataToRclDoc(Xapian::docid docid, std::string &data,
     if (!parms.ok())
 	return false;
 
-    // Special cases:
+    // Compute what index this comes from, and check for path translations
+    string dbdir = m_rcldb->m_basedir;
+    if (!m_rcldb->m_extraDbs.empty()) {
+	// As per trac.xapian.org/wiki/FAQ/MultiDatabaseDocumentID
+	unsigned int idxi = (docid-1) % (m_rcldb->m_extraDbs.size()+1);
+	// idxi is in [0, extraDbs.size()]. 0 is the base index, 1-n index
+	// into the additional dbs array
+	if (idxi) {
+	    dbdir = m_rcldb->m_extraDbs[idxi - 1];
+	}
+    }
     parms.get(Doc::keyurl, doc.url);
+    m_rcldb->m_config->urlrewrite(dbdir, doc.url);
+
+    // Special cases:
     parms.get(Doc::keytp, doc.mimetype);
     parms.get(Doc::keyfmt, doc.fmtime);
     parms.get(Doc::keydmt, doc.dmtime);
@@ -264,6 +277,7 @@ bool Db::Native::dbDataToRclDoc(Xapian::docid docid, std::string &data,
 	if (doc.meta.find(*it) == doc.meta.end())
 	    parms.get(*it, doc.meta[*it]);
     }
+    doc.meta[Doc::keyurl] = doc.url;
     doc.meta[Doc::keymt] = doc.dmtime.empty() ? doc.fmtime : doc.dmtime;
     return true;
 }
@@ -579,14 +593,16 @@ int Db::termDocCnt(const string& _term)
     return res;
 }
 
-bool Db::addQueryDb(const string &dir) 
+bool Db::addQueryDb(const string &_dir) 
 {
-    LOGDEB(("Db::addQueryDb: ndb %p iswritable %d db [%s]\n", m_ndb,
+    string dir = _dir;
+    LOGDEB0(("Db::addQueryDb: ndb %p iswritable %d db [%s]\n", m_ndb,
 	      (m_ndb)?m_ndb->m_iswritable:0, dir.c_str()));
     if (!m_ndb)
 	return false;
     if (m_ndb->m_iswritable)
 	return false;
+    dir = path_canon(dir);
     if (find(m_extraDbs.begin(), m_extraDbs.end(), dir) == m_extraDbs.end()) {
 	m_extraDbs.push_back(dir);
     }