Ahoy, I'll answer inline too. 22.09.2011 10:51, Andres Taylor wrote: > HI there, > > Answers inline, > > On Wed, Sep 21, 2011 at 8:16 PM, st.pa<st...@web.de> wrote: > >> Hi, >> I've got a Cypher query with some 10k possible source nodes and some 10k >> intended target nodes out of some 1M possible target nodes which looks >> something like the following: >> >> START source = (indexSourceNodes,"*"), >> target = (indexTargetNodes,"propertyTwo:*.txt") >> MATCH (source)-[:needsThisRelationshipType]->(target) >> WHERE (target.needsThisProperty) >> RETURN source,target >> >> Now, how can I convince this query to speed up using a >> relationship-index which I already constructed containing all the >> neccessary relationships? > > > This is something that we've planned to add to Cypher from day one, but > still haven't put in. And it's because no-one ever asked for it. I'll add it > in now, that we actually have users wanting it. Thanks for speaking up! I'm > sorry you'll have to wait for it...
no problem. using the relationshipIndex.query() and hits iteration and then checking existance of that extra needed property is probably more efficient anyway. > > >> How can I use regular expressions for an >> index-subquery like this "propertyTwo:http(s)://.*\.txt" and how and >> what needs to be escaped inside that subquery. >> > > The way to do regular expressions is with the ~= operator. More info here: > http://docs.neo4j.org/chunked/snapshot/query-where.html#where-regular-expressions > > Does that help? yes, that helps for the WHERE clause. <Example1> START source = (indexSourceNodes,"propertyOne:%REGEX%"), target = ... MATCH (source)-[relation]->(target) RETURN relation </Example1> --> I can put my regex-query in the WHERE clause so that it filters the results of an index lookup. <Example2> START source = (indexSourceNodes,"propertyOne:*"), target = ... MATCH (source)-[relation]->(target) WHERE source.propertyOne =~ /%REGEX%/ RETURN relation </Example2> but what about regexes in the START clause where lookup of some nodeIndex with *-wildcard already takes place. this is obviously not pure regex, and I did not find documentation on what other wildcards are allowed or how to search for literal "*" - so what needs to be escaped there? \*, **, "*", \"*\", \\*, _, ? other related question would be, how sophisticated is regex-support in the WHERE clause, like can I use backreferences as in regexMatchRepeatedFolder="(/.+?)(/.+?)*?\1" ... well I could find that bit out on my own. > > Andrés > _______________________________________________ St.P. _______________________________________________ Neo4j mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user
