Re: [Neo4j] Indexing - Rest Api
Hi Romiko,
Out of the box auto-indexing only supports exact matches, rather than full text
searches. The reason is that auto indexes are created with the default
configuration (http://docs.neo4j.org/chunked/snapshot/indexing-create.html) the
first time you access them. But this gives us a little wiggle room to cheat the
lifecycle of the server.
Warning: if you read past this point, what I'm about to suggest might make you
rip out your own eyeballs and feed them to a passing alley cat.
Seriously, you've been warned, this is a bit of a hack :-)
So, before we do anything, let's make sure we've got auto-indexing enabled for
the server. Make sure that you've added some config like this into your
neo4j.properties file:
node_keys_indexable=name,phone
relationship_keys_indexable=since
node_auto_indexing=true
relationship_auto_indexing=true
Then bring up your server.
Next up, we want to pre-empt the creation of an auto-index, by telling the
server to create an apparently manual index which has the same name as the node
(or rel) auto-index (in this case we're making a node auto index so the index
name is node_auto_index), like so:
POST /db/data/index/node HTTP/1.1
Host: localhost:7474
Content-Length: 76
Content-Type: application/json
{name:node_auto_index, config:{type:fulltext,provider:lucene}}
This triggers the creation of an index which happens to have the same name as
the auto index that the database will create for itself. Now when we interact
with the database, the index is created so the state machine skips over that
step and just gets on with normal day-to-day auto-indexing.
You have to do this early in your server lifecycle, otherwise you run the risk
of creating a normal auto index as a side effect of doing normal work.
See, I told you it was yucky.
Jim
PS - cheers to Chris Gioran for the tip that this is possible
PPS - double cheers to Chris for wanting to make this sane in future releases
:-)
___
Neo4j mailing list
User@lists.neo4j.org
https://lists.neo4j.org/mailman/listinfo/user
Re: [Neo4j] Indexing - Rest Api
Hi Jim Thanks for the response. I have built a .Net API to manage manual indexes. So basically when I create a node I have an option to parse in a collection of key/Val pairs to index the node. I'm using Lucene and fulltext. However it sounds like auto indexing might be a better option? I avoided it as the docs mention experimental and I was not sure if you can configure exactly which fields to index, but it seems after reading your profile that this is possible via config. So is it possible via config to do this: Imagine a customer node, with firstname, lastname, preferredname. In my current implementtion I add the following into ONE fulltext index Key:Name Value permutations for same key: Fn Ln Pn Fn ln Pn ln This means people can search bob brown Bobby brown Brown Bob Bobby Bo~ browns~ How would you configure this in the config file for autoindexing? I did not know that autoindexing could be fine grained control, which will make code maintenance much easier. The last question then is when should manual be used over auto? Much appreciated. Sent from my iPhonen On 17/08/2011, at 12:19 AM, Jim Webber j...@neotechnology.com wrote: G'day Romiko, * Configuration File: Autoindexing enabled here is related to automatically indexing all nodes and relationships? * If I want to add a node to an index following the guidelines, what convention should be used for the Key/Value naming? Imagine we have nodes of type Customers, and we have 1 customers, what sort of key/value would we set for each customer, as I would guess that all properties are indexed for the node? Would it be perhaps a unique auto increment number for example that is in no way related to the data node? The basic mechanics are straightforward: Assuming you've configured your indexes (http://docs.neo4j.org/chunked/snapshot/auto-indexing.html#auto-indexing-config), then there's nothing more to do. Every time you encounter a node (or relationship) with one of the properties that you declared in your config, it'll be added to the index if it wasn't before, and the property value will be in sync with the value in the node (or relationship). Only the properties that you ask to be auto-indexed will be. So in your case, you might well have an auto index config that indexes only (say) customer_id and doesn't care about other properties. * Also, how can we control the properties/fields to index when we add a node to an index with the RestApi, I would imagine we would not want to index all fields on a node via the auto indexing feature? At the moment you can't control auto indexing through the REST API, you can only use auto indexes that have been configured on the server. I recall Jake (I think) mentioning this before, and it boils down to the fact that we don't mutate config (and so changes to auto index config isn't persisted across DB restarts). Once we figure out how to persist things like config in a database, we'll get stuck into that :-) Jim PS - thanks for prompting on this, it's triggered a good refactoring and cleanup of the index functional tests. ___ Neo4j mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user ___ Neo4j mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user
Re: [Neo4j] Indexing - Rest Api
Jim, To try answer my own question for auto bs manual. One think I like with manual is a situation where a client had multiple alsoknownas names which is a relationship IS_ALSO_KNOWN_AS Now in this case with manual index I can actualy index a reference to client node using fields from nodes related to the client Fn Ln Fn ln Aka Aka ln So the last two index entries point to the Client Node address, this is great since in rest API we can control the node to point to for manual indexes, and thus I guess is why they perhaps are preferred over automatic. But automatic is good for simple one to one non composite values?? Cheers Sent from my iPhone On 20/08/2011, at 9:10 AM, Romiko Derbynew romiko.derby...@readify.net wrote: Hi Jim Thanks for the response. I have built a .Net API to manage manual indexes. So basically when I create a node I have an option to parse in a collection of key/Val pairs to index the node. I'm using Lucene and fulltext. However it sounds like auto indexing might be a better option? I avoided it as the docs mention experimental and I was not sure if you can configure exactly which fields to index, but it seems after reading your profile that this is possible via config. So is it possible via config to do this: Imagine a customer node, with firstname, lastname, preferredname. In my current implementtion I add the following into ONE fulltext index Key:Name Value permutations for same key: Fn Ln Pn Fn ln Pn ln This means people can search bob brown Bobby brown Brown Bob Bobby Bo~ browns~ How would you configure this in the config file for autoindexing? I did not know that autoindexing could be fine grained control, which will make code maintenance much easier. The last question then is when should manual be used over auto? Much appreciated. Sent from my iPhonen On 17/08/2011, at 12:19 AM, Jim Webber j...@neotechnology.com wrote: G'day Romiko, * Configuration File: Autoindexing enabled here is related to automatically indexing all nodes and relationships? * If I want to add a node to an index following the guidelines, what convention should be used for the Key/Value naming? Imagine we have nodes of type Customers, and we have 1 customers, what sort of key/value would we set for each customer, as I would guess that all properties are indexed for the node? Would it be perhaps a unique auto increment number for example that is in no way related to the data node? The basic mechanics are straightforward: Assuming you've configured your indexes (http://docs.neo4j.org/chunked/snapshot/auto-indexing.html#auto-indexing-config), then there's nothing more to do. Every time you encounter a node (or relationship) with one of the properties that you declared in your config, it'll be added to the index if it wasn't before, and the property value will be in sync with the value in the node (or relationship). Only the properties that you ask to be auto-indexed will be. So in your case, you might well have an auto index config that indexes only (say) customer_id and doesn't care about other properties. * Also, how can we control the properties/fields to index when we add a node to an index with the RestApi, I would imagine we would not want to index all fields on a node via the auto indexing feature? At the moment you can't control auto indexing through the REST API, you can only use auto indexes that have been configured on the server. I recall Jake (I think) mentioning this before, and it boils down to the fact that we don't mutate config (and so changes to auto index config isn't persisted across DB restarts). Once we figure out how to persist things like config in a database, we'll get stuck into that :-) Jim PS - thanks for prompting on this, it's triggered a good refactoring and cleanup of the index functional tests. ___ Neo4j mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user ___ Neo4j mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user
Re: [Neo4j] Indexing - Rest Api
G'day Romiko, * Configuration File: Autoindexing enabled here is related to automatically indexing all nodes and relationships? * If I want to add a node to an index following the guidelines, what convention should be used for the Key/Value naming? Imagine we have nodes of type Customers, and we have 1 customers, what sort of key/value would we set for each customer, as I would guess that all properties are indexed for the node? Would it be perhaps a unique auto increment number for example that is in no way related to the data node? The basic mechanics are straightforward: Assuming you've configured your indexes (http://docs.neo4j.org/chunked/snapshot/auto-indexing.html#auto-indexing-config), then there's nothing more to do. Every time you encounter a node (or relationship) with one of the properties that you declared in your config, it'll be added to the index if it wasn't before, and the property value will be in sync with the value in the node (or relationship). Only the properties that you ask to be auto-indexed will be. So in your case, you might well have an auto index config that indexes only (say) customer_id and doesn't care about other properties. * Also, how can we control the properties/fields to index when we add a node to an index with the RestApi, I would imagine we would not want to index all fields on a node via the auto indexing feature? At the moment you can't control auto indexing through the REST API, you can only use auto indexes that have been configured on the server. I recall Jake (I think) mentioning this before, and it boils down to the fact that we don't mutate config (and so changes to auto index config isn't persisted across DB restarts). Once we figure out how to persist things like config in a database, we'll get stuck into that :-) Jim PS - thanks for prompting on this, it's triggered a good refactoring and cleanup of the index functional tests. ___ Neo4j mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user
[Neo4j] Indexing - Rest Api
Hi, I am planning to extend the .Net client that we use for CRUD operation and now add indexing support. I am keen to go the route of AutoIndexing, so we just query the index and it automatically gets create. Just a few questions. * Configuration File: Autoindexing enabled here is related to automatically indexing all nodes and relationships? * If I want to add a node to an index following the guidelines, what convention should be used for the Key/Value naming? Imagine we have nodes of type Customers, and we have 1 customers, what sort of key/value would we set for each customer, as I would guess that all properties are indexed for the node? Would it be perhaps a unique auto increment number for example that is in no way related to the data node? e.g. Example request * POST http://localhost:7474/db/data/index/node/favorites/key/the%20value * Accept: application/json * Content-Type: application/json http://localhost:7474/db/data/node/0 * Also, how can we control the properties/fields to index when we add a node to an index with the RestApi, I would imagine we would not want to index all fields on a node via the auto indexing feature? Our plan is to use Lucene Search Queries so we can do funzzy searches, however, we would like to leverage autoindexing but have the option to choose exactly what fields we want to index on a node when we add nodes to indexes. Thank You. ___ Neo4j mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user
