static private QueryPart MatchToQueryPart(Match m)
{
// Looping over all Matches we have got:
// m.Groups["pm"] plus or minus sign
// m.Groups["key"] keyname
// m.Groups["quote"] quoted string
// m.Groups["midquote1"] + m.Groups["midquote2"] quoted midway string also represents unquoted string
string query = m.ToString();
// Either quote is set or midquote1 and (optionally) midquote2 is set
string text = m.Groups ["quote"].ToString() + m.Groups ["midquote1"].ToString() + m.Groups ["midquote2"].ToString();
string key = m.Groups ["key"].ToString();
bool IsProhibited = (m.Groups ["pm"].ToString() == "-");
// check for file extensions
// if match starts with *. or . and only contains letters we assume it's a file extension
if (extension_re.Match(text).Success || key.ToLower() == "ext" || key.ToLower() == "extension")
{
QueryPart_Property query_part = new QueryPart_Property();
query_part.Key = Property.FilenameExtensionPropKey;
if (text.StartsWith("*."))
{
query_part.Value = text.Substring(1).ToLower();
}
else if (text.StartsWith("."))
{
query_part.Value = text.ToLower();
}
else
{
query_part.Value = "." + text.ToLower();
}
query_part.Type = PropertyType.Keyword;
query_part.Logic = (IsProhibited ? QueryPartLogic.Prohibited : QueryPartLogic.Required);
Logger.Log.Debug("Extension query: {0}", query_part.Value);
return(query_part);
}
if (key == String.Empty)
{
Logger.Log.Debug("Parsed query '{0}' as text_query", text);
return(StringToQueryPart(text, IsProhibited));
}
// FIXME: i18n-izing "date"
if (key == "date")
{
try {
QueryPart part = DateQueryToQueryPart(text);
part.Logic = (IsProhibited ? QueryPartLogic.Prohibited : QueryPartLogic.Required);
return(part);
} catch (FormatException) {
Log.Warn("Could not parse [{0}] as date query. Assuming text.", text);
return(StringToQueryPart(text, IsProhibited));
}
}
// FIXME: i18n-izing "uri"
if (key == "uri")
{
try {
QueryPart_Uri part = new QueryPart_Uri();
part.Logic = (IsProhibited ? QueryPartLogic.Prohibited : QueryPartLogic.Required);
// uri: queries require the raw Uri as present in the index
part.Uri = new Uri(text, true);
return(part);
} catch (System.UriFormatException) {
Log.Warn("Could not parse [{0}] as uri query. Assuming text.", text);
return(StringToQueryPart(text, IsProhibited));
}
}
// Special case
if (key == "inuri")
{
QueryPart_Property inuri_part = new QueryPart_Property();
inuri_part.Logic = (IsProhibited ? QueryPartLogic.Prohibited : QueryPartLogic.Required);
inuri_part.Key = "inuri";
inuri_part.Value = text;
inuri_part.Type = PropertyType.Keyword;
Log.Debug("Handing special query 'inuri:{0}'", text);
return(inuri_part);
}
// Non-keyword queries by directly using property names
// Query of form property:namespace:name=value
// which is translated to a non-keyword query
// namespace:name=value
int pos;
if (key == "property" && ((pos = text.IndexOf('=')) != -1))
{
QueryPart_Property part = new QueryPart_Property();
part.Key = text.Substring(0, pos);
part.Value = text.Substring(pos + 1);
part.Type = PropertyType.Text;
part.Logic = (IsProhibited ? QueryPartLogic.Prohibited : QueryPartLogic.Required);
Logger.Log.Debug("Parsed query '" + query +
"' as prop query:key=" + part.Key +
", value=" + part.Value +
" and property type=" + part.Type);
return(part);
}
// keyword queries by directly using property names
// Query of form keyword:namespace:name=value
// which is translated to a keyword query
// namespace:name=value
if (key == "keyword" && ((pos = text.IndexOf('=')) != -1))
{
QueryPart_Property part = new QueryPart_Property();
part.Key = text.Substring(0, pos);
part.Value = text.Substring(pos + 1);
part.Type = PropertyType.Keyword;
part.Logic = (IsProhibited ? QueryPartLogic.Prohibited : QueryPartLogic.Required);
Logger.Log.Debug("Parsed query '" + query +
"' as prop query:key=" + part.Key +
", value=" + part.Value +
" and property type=" + part.Type);
return(part);
}
string[] prop_string = null;
bool is_present;
PropertyType[] prop_type;
int num;
is_present = PropertyKeywordFu.GetMapping(key, out num, out prop_string, out prop_type);
// if key is not present in the mapping, assume the query is a text query
// i.e. if token is foo:bar and there is no mappable property named foo,
// assume "foo:bar" as text query
// FIXME the analyzer changes the text query "foo:bar" to "foo bar"
// which might not be the right thing to do
if (!is_present)
{
Logger.Log.Warn("Could not find property, parsed query '{0}' as text_query", query);
return(StringToQueryPart(query, IsProhibited));
}
if (num == 1)
{
QueryPart_Property query_part_prop = new QueryPart_Property();
query_part_prop.Key = prop_string [0];
query_part_prop.Value = text;
query_part_prop.Type = prop_type [0];
query_part_prop.Logic = (IsProhibited ? QueryPartLogic.Prohibited : QueryPartLogic.Required);
Logger.Log.Debug("Parsed query '" + query +
"' as prop query:key=" + query_part_prop.Key +
", value=" + query_part_prop.Value +
" and property type=" + query_part_prop.Type);
return(query_part_prop);
}
// Multiple property queries are mapped to this keyword query
// Create an OR query from them
// FIXME: Would anyone want an AND query ?
QueryPart_Or query_part_or = new QueryPart_Or();
query_part_or.Logic = (IsProhibited ? QueryPartLogic.Prohibited : QueryPartLogic.Required);
Logger.Log.Debug("Parsed query '{0}' as OR of {1} queries:", query, num);
for (int i = 0; i < num; ++i)
{
QueryPart_Property query_part_prop = new QueryPart_Property();
query_part_prop.Key = prop_string [i];
query_part_prop.Value = text;
query_part_prop.Type = prop_type [i];
query_part_prop.Logic = QueryPartLogic.Required;
Log.Debug("\t:key={0}, value={1} and property type={2}", query_part_prop.Key, query_part_prop.Value, query_part_prop.Type);
query_part_or.Add(query_part_prop);
}
return(query_part_or);
}
static private QueryPart MatchToQueryPart (Match m)
{
// Looping over all Matches we have got:
// m.Groups["pm"] plus or minus sign
// m.Groups["key"] keyname
// m.Groups["quote"] quoted string
// m.Groups["midquote1"] + m.Groups["midquote2"] quoted midway string also represents unquoted string
string query = m.ToString ();
// Either quote is set or midquote1 and (optionally) midquote2 is set
string text = m.Groups ["quote"].ToString () + m.Groups ["midquote1"].ToString () + m.Groups ["midquote2"].ToString ();
string key = m.Groups ["key"].ToString ();
bool IsProhibited = (m.Groups ["pm"].ToString () == "-");
// check for file extensions
// if match starts with *. or . and only contains letters we assume it's a file extension
if (extension_re.Match (text).Success || key.ToLower () == "ext" || key.ToLower () == "extension") {
QueryPart_Property query_part = new QueryPart_Property ();
query_part.Key = Property.FilenameExtensionPropKey;
if (text.StartsWith ("*."))
query_part.Value = text.Substring (1).ToLower ();
else if (text.StartsWith ("."))
query_part.Value = text.ToLower ();
else
query_part.Value = "." + text.ToLower ();
query_part.Type = PropertyType.Keyword;
query_part.Logic = (IsProhibited ? QueryPartLogic.Prohibited : QueryPartLogic.Required);
Logger.Log.Debug ("Extension query: {0}", query_part.Value);
return query_part;
}
if (key == String.Empty) {
Logger.Log.Debug ("Parsed query '{0}' as text_query", text);
return StringToQueryPart (text, IsProhibited);
}
// FIXME: i18n-izing "date"
if (key == "date") {
try {
QueryPart part = DateQueryToQueryPart (text);
part.Logic = (IsProhibited ? QueryPartLogic.Prohibited : QueryPartLogic.Required);
return part;
} catch (FormatException) {
Log.Warn ("Could not parse [{0}] as date query. Assuming text.", text);
return StringToQueryPart (text, IsProhibited);
}
}
// FIXME: i18n-izing "uri"
if (key == "uri") {
try {
QueryPart_Uri part = new QueryPart_Uri ();
part.Logic = (IsProhibited ? QueryPartLogic.Prohibited : QueryPartLogic.Required);
part.Uri = UriFu.UserUritoEscapedUri (text);
return part;
} catch (System.UriFormatException) {
Log.Warn ("Could not parse [{0}] as uri query. Assuming text.", text);
return StringToQueryPart (text, IsProhibited);
}
}
// Special case
if (key == "inuri") {
QueryPart_Property inuri_part = new QueryPart_Property ();
inuri_part.Logic = (IsProhibited ? QueryPartLogic.Prohibited : QueryPartLogic.Required);
inuri_part.Key = "inuri";
inuri_part.Value = text;
inuri_part.Type = PropertyType.Keyword;
Log.Debug ("Handing special query 'inuri:{0}'", text);
return inuri_part;
}
// Non-keyword queries by directly using property names
// Query of form property:namespace:name=value
// which is translated to a non-keyword query
// namespace:name=value
int pos;
if (key == "property" && ((pos = text.IndexOf ('=')) != -1)) {
QueryPart_Property part = new QueryPart_Property ();
part.Key = text.Substring (0, pos);
part.Value = text.Substring (pos + 1);
part.Type = PropertyType.Text;
part.Logic = (IsProhibited ? QueryPartLogic.Prohibited : QueryPartLogic.Required);
Logger.Log.Debug ("Parsed query '" + query +
"' as prop query:key=" + part.Key +
", value=" + part.Value +
" and property type=" + part.Type);
return part;
}
// keyword queries by directly using property names
// Query of form keyword:namespace:name=value
// which is translated to a keyword query
// namespace:name=value
if (key == "keyword" && ((pos = text.IndexOf ('=')) != -1)) {
QueryPart_Property part = new QueryPart_Property ();
part.Key = text.Substring (0, pos);
part.Value = text.Substring (pos + 1);
part.Type = PropertyType.Keyword;
part.Logic = (IsProhibited ? QueryPartLogic.Prohibited : QueryPartLogic.Required);
Logger.Log.Debug ("Parsed query '" + query +
"' as prop query:key=" + part.Key +
", value=" + part.Value +
" and property type=" + part.Type);
return part;
}
if ((pos = text.IndexOf('*')) >= 0) {
QueryPart_Wildcard wild = new QueryPart_Wildcard();
wild.QueryString = text;
wild.PropertyOnly = true;
return wild;
}
string[] prop_string = null;
bool is_present;
PropertyType[] prop_type;
int num;
is_present = PropertyKeywordFu.GetMapping (key, out num, out prop_string, out prop_type);
// if key is not present in the mapping, assume the query is a text query
// i.e. if token is foo:bar and there is no mappable property named foo,
// assume "foo:bar" as text query
// FIXME the analyzer changes the text query "foo:bar" to "foo bar"
// which might not be the right thing to do
if (!is_present) {
Logger.Log.Warn ("Could not find property, parsed query '{0}' as text_query", query);
return StringToQueryPart (query, IsProhibited);
}
if (num == 1) {
QueryPart_Property query_part_prop = new QueryPart_Property ();
query_part_prop.Key = prop_string [0];
query_part_prop.Value = text;
query_part_prop.Type = prop_type [0];
query_part_prop.Logic = (IsProhibited ? QueryPartLogic.Prohibited : QueryPartLogic.Required);
Logger.Log.Debug ("Parsed query '" + query +
"' as prop query:key=" + query_part_prop.Key +
", value=" + query_part_prop.Value +
" and property type=" + query_part_prop.Type);
return query_part_prop;
}
// Multiple property queries are mapped to this keyword query
// Create an OR query from them
// FIXME: Would anyone want an AND query ?
QueryPart_Or query_part_or = new QueryPart_Or ();
query_part_or.Logic = (IsProhibited ? QueryPartLogic.Prohibited : QueryPartLogic.Required);
Logger.Log.Debug ("Parsed query '{0}' as OR of {1} queries:", query, num);
for (int i = 0; i < num; ++i) {
QueryPart_Property query_part_prop = new QueryPart_Property ();
query_part_prop.Key = prop_string [i];
query_part_prop.Value = text;
query_part_prop.Type = prop_type [i];
query_part_prop.Logic = QueryPartLogic.Required;
Log.Debug ("\t:key={0}, value={1} and property type={2}", query_part_prop.Key, query_part_prop.Value, query_part_prop.Type);
query_part_or.Add (query_part_prop);
}
return query_part_or;
}
///////// RDF fu ///////////////////////////////////////////////
// Returns a collection of Uris
// HitFilter and UriFilter are ignored for now
// They will come into play in the final FetchDocument part
// FIXME: Should RDFQuery do any query mapping using backend_query_part_hook ?
// I think it should not. QueryPart hooks are for human beings, RDF is for softwares.
public ICollection DoRDFQuery (Query _query, TextCache text_cache)
{
RDFQuery query = (RDFQuery) _query;
string subject, predicate, _object;
PropertyType pred_type;
subject = query.SubjectString;
predicate = query.Predicate;
pred_type = query.PredicateType;
_object = query.Object;
if (Debug)
Logger.Log.Debug ("###### {0}: Starting low-level queries '{1}' : '{4}:{2}' = '{3}'", IndexName, subject, predicate, _object, pred_type);
// ******** 8 cases **********
// Return all uris
if (subject == String.Empty && predicate == String.Empty && _object == String.Empty) {
ICollection hits = GetAllHitsByUri ().Values;
foreach (Hit hit in hits)
foreach (Property text_link_property in GetTextLinks (hit.Uri, text_cache))
hit.AddProperty (text_link_property);
return hits;
}
// Normal query
if (subject == String.Empty && predicate == String.Empty && _object != String.Empty) {
QueryPart_Text part = new QueryPart_Text ();
part.Text = _object;
part.SearchFullText = false; // We only search properties in RDF query
query.AddPart (part);
return DoLowLevelRDFQuery (query, pred_type, predicate, _object, text_cache);
}
// Return uris for all documents with this property
if (subject == String.Empty && predicate != String.Empty && _object == String.Empty) {
string field_name = PropertyToFieldName (pred_type, predicate);
QueryPart_Property part = new QueryPart_Property ();
part.Type = PropertyType.Internal;
part.Key = "Properties";
part.Value = field_name;
query.AddPart (part);
return DoLowLevelRDFQuery (query, pred_type, predicate, null, text_cache);
}
// Property query
if (subject == String.Empty && predicate != String.Empty && _object != String.Empty) {
QueryPart_Property part = new QueryPart_Property ();
part.Type = pred_type;
part.Key = predicate;
part.Value = _object;
query.AddPart (part);
return DoLowLevelRDFQuery (query, pred_type, predicate, _object, text_cache);
}
// Return if the URI exists
if (subject != String.Empty && predicate == String.Empty && _object == String.Empty) {
QueryPart_Uri part = new QueryPart_Uri ();
part.Uri = new Uri (subject, true); // better be URI!
query.AddPart (part);
// FIXME: Which properties to return in the hit? All or none ?
return DoLowLevelRDFQuery (query, pred_type, predicate, null, text_cache);
}
// Normal query in the document with this URI
if (subject != String.Empty && predicate == String.Empty && _object != String.Empty) {
QueryPart_Uri uri_part = new QueryPart_Uri ();
uri_part.Uri = new Uri (subject, true); // better be URI!
query.AddPart (uri_part);
QueryPart_Text part = new QueryPart_Text ();
part.Text = _object;
part.SearchFullText = false; // We only search properties in RDF query
query.AddPart (part);
return DoLowLevelRDFQuery (query, pred_type, predicate, _object, text_cache);
}
// Return URI if the document with this URI contains this property
if (subject != String.Empty && predicate != String.Empty && _object == String.Empty) {
ArrayList returned_uris = new ArrayList (1);
ArrayList uri_list = new ArrayList (1);
uri_list.Add (new Uri (subject, true));
string field_name = PropertyToFieldName (pred_type, predicate);
FieldSelector fields = new MapFieldSelector (new string[] { "Uri", "Timestamp", field_name });
ICollection hits = GetHitsForUris (uri_list, fields);
if (predicate == "TextLinks") {
foreach (Hit hit in hits)
foreach (Property text_link_property in GetTextLinks (hit.Uri, text_cache))
hit.AddProperty (text_link_property);
}
return hits;
}
// Property query in the document with this URI
if (subject != String.Empty && predicate != String.Empty && _object != String.Empty) {
QueryPart_Uri uri_part = new QueryPart_Uri ();
uri_part.Uri = new Uri (subject, true); // better be URI!
query.AddPart (uri_part);
QueryPart_Property part = new QueryPart_Property ();
part.Type = pred_type;
part.Key = predicate;
part.Value = _object;
query.AddPart (part);
return DoLowLevelRDFQuery (query, pred_type, predicate, _object, text_cache);
}
throw new Exception ("Never reaches");
}
// Remap uri in querypart_uri
private QueryPart_Uri RemapUriQueryPart (QueryPart_Uri part)
{
Uri new_uri = ExternalToInternalUri (part.Uri);
Log.Debug ("Remapping QueryPart_Uri from {0} to {1}", part.Uri, new_uri);
// Do the right thing if the uri does not exist
// Remember QueryPart_Uri can occur inside QueryPart_Or
if (new_uri == null)
new_uri = new Uri ("no-match:///"); // Will never match
QueryPart_Uri new_part = new QueryPart_Uri ();
new_part.Uri = new_uri;
new_part.Logic = part.Logic;
return new_part;
}
