Well there is an idea that might be easy to be implemented.
So the idea is to *auto-create* an edge. Let's take an example:
MySQL classes are:
- Person{id,name}
- Post{,person_id,title}
Now we also create such classes in OrientDB. Now we tell OrientDB if there
is INSERT INTO Person or Post -> create edge on "id" to "person_id". I
don't but isn't that pretty simple for a database to do?
Well I have to say all of that is possible easily done with Java API but I
really want it with ETL because there is no real need for me to use the
Java API.
Am Montag, 25. August 2014 14:15:00 UTC+2 schrieb Curtis Mosters:
>
> Could you maybe take some time to create such an example?
>
> The current "edge" is bit hard to understand:
>
> edge: {
> class: "hasPost",
> joinFieldName: "person_id",
> lookup: "person.id",
> unresolvedLinkAction: "CREATE",
> if: "person_id is not null"
> }
>
> How can I say here to Link from person on post? I just need a small
> example and also the MySQL strcuture if possible. That would make
> everything clear. Thanks Luca.
>
> Am Montag, 25. August 2014 13:23:55 UTC+2 schrieb Lvc@:
>>
>> Curtis,
>> You could create 2 or more etl scripts, some that extract vertices and
>> other that extract edges.
>>
>> Lvc@
>>
>> ᐧ
>>
>>
>> On 25 August 2014 12:25, 'Curtis Mosters' via OrientDB <
>> [email protected]> wrote:
>>
>>> Hi Luca, well your example is just reading one table from MySQL. One
>>> table with creating edges on it's entries. But in the usual cases that's
>>> not the way how to import from a MySQL. Or maybe I just didn't understood
>>> that example completely. But so far as I have seen it, you used one table
>>> from MySQL that contained {id,...,friend_id}. You created profiles on that
>>> id and then created edge on that id's. Is that right?
>>>
>>> But just once again I have those data as usually stored in another
>>> table. So that cannot work for me.
>>>
>>> Am Donnerstag, 21. August 2014 15:56:05 UTC+2 schrieb Lvc@:
>>>>
>>>> ᐧ
>>>> On 21 August 2014 14:45, 'Curtis Mosters' via OrientDB <
>>>> [email protected]> wrote:
>>>>
>>>>> This is a great example Luca, thanks first of all.
>>>>>
>>>>> But could you additionally add information about the MySQL schema and
>>>>> do I understand it correctly that you get the classes Person,Friend and
>>>>> an
>>>>> edge. But where comes the friend data?
>>>>>
>>>>
>>>>
>>>> Friend in my case is an Edge class. (create class Friend extends E).
>>>> It's useful to create ad-hoc edge classes instead that the generic "E":
>>>>
>>>> http://www.orientechnologies.com/docs/1.7.8/orientdb.wiki/
>>>> Graph-Schema.html
>>>>
>>>> Lvc@
>>>>
>>>>
>>>>
>>>>>
>>>>> Am Donnerstag, 21. August 2014 13:25:47 UTC+2 schrieb Lvc@:
>>>>>>
>>>>>> Hi Curtis,
>>>>>> ETL is very powerful and we're improving it everyday with users'
>>>>>> feedback. We want to let it to be final for 2.0, so we can bundle with
>>>>>> it.
>>>>>> By the way Enterprise Edition
>>>>>> <http://www.orientechnologies.com/orientdb-enterprise> will have a
>>>>>> Web Interface to edit it, and in the future also to debug it.
>>>>>>
>>>>>> Unfortunately I don't understand in your example table1 & 2 with both
>>>>>> name_id field. Look at this example about Friendship. I want to connect
>>>>>> a
>>>>>> social network like app.
>>>>>>
>>>>>> This is an example of ETL configuration. My comments begins with //
>>>>>> but remove them from real config file:
>>>>>>
>>>>>> {
>>>>>> config: {
>>>>>> verbose: true
>>>>>> },
>>>>>> begin: [
>>>>>> ],
>>>>>> extractor: {
>>>>>> "jdbc": { "driver": "com.mysql.jdbc.Driver",
>>>>>> "url": "jdbc:mysql://localhost/mysocialnetwork",
>>>>>> "userName": "root",
>>>>>> "userPassword": "root",
>>>>>> "query": "select * from profile"
>>>>>> }
>>>>>> },
>>>>>> transformers : [
>>>>>> {
>>>>>> // THIS IS ONLY A LOG, USEFUL TO TRACE WHAT ARRIVES FROM MYSQL
>>>>>> log: {
>>>>>> prefix: "MySQL -> "
>>>>>> }
>>>>>> },
>>>>>> {
>>>>>> merge: {
>>>>>> // LOOKUP BY ID: IF ALREADY PRESENT MERGE THE FIELDS BY
>>>>>> OVERWRITING THE DIFFERENT ONES
>>>>>> joinFieldName: "*id*",
>>>>>> lookup: "Profile.*id*"
>>>>>> }
>>>>>> },
>>>>>> {
>>>>>> vertex: {
>>>>>> // TRANSFORM IT IN A VERTEX
>>>>>> class: "Profile"
>>>>>> }
>>>>>> },
>>>>>> {
>>>>>> edge: {
>>>>>> // CONNECT THE FRIEND IN MYSQL COLUMN 'friend_id'
>>>>>> class: "Friend", // Friend is the edges' class
>>>>>> joinFieldName: "friend_id", //
>>>>>> lookup: "Profile.id",
>>>>>> unresolvedLinkAction: "CREATE", // IF DOESN'T EXIST YET,
>>>>>> CREATE THE VERTEX EMPTY. WILL BE FURTHER UPDATED
>>>>>> if: "friend_id is not null"
>>>>>> }
>>>>>> },
>>>>>> {
>>>>>> log: {
>>>>>> prefix: "Vertex -> "
>>>>>> }
>>>>>> }
>>>>>> ],
>>>>>> loader : {
>>>>>> orientdb: {
>>>>>> dbURL: "remote:localhost/yourdb",
>>>>>> dbUser: "admin",
>>>>>> dbPassword: "admin",
>>>>>> dbAutoCreate: true,
>>>>>> tx: true,
>>>>>> batchCommit: 1000,
>>>>>> dbType: "graph",
>>>>>> classes: [
>>>>>> ],
>>>>>> indexes: [
>>>>>> ]
>>>>>> }
>>>>>> }
>>>>>> }
>>>>>>
>>>>>>
>>>>>> Lvc@
>>>>>>
>>>>>>
>>>>>>
>>>>>> On 21 August 2014 13:05, 'Curtis Mosters' via OrientDB <
>>>>>> [email protected]> wrote:
>>>>>>
>>>>>>> Hey so I have now some experiences with ETL. I think this is great
>>>>>>> for import. That's why I want to improve the importing process now.
>>>>>>>
>>>>>>> 1. First of all is it possible to run several imports
>>>>>>> parallel(not with several consoles) or if one is done start another
>>>>>>> instantly?
>>>>>>> 2. How do I create an edge (https://github.com/orientechn
>>>>>>> ologies/orientdb-etl/wiki/Transformer#edge
>>>>>>>
>>>>>>> <https://github.com/orientechnologies/orientdb-etl/wiki/Transformer#edge>
>>>>>>>
>>>>>>> is not helpful for me)
>>>>>>>
>>>>>>> Let's say I have 2 tables in MySQL:
>>>>>>>
>>>>>>> table1 with {*name_id*,*name*} and table2 with {*name_id*,*text_id*,
>>>>>>> *text*}
>>>>>>>
>>>>>>> Now I want to get them connected (name_id) in OrientDB.
>>>>>>>
>>>>>>> One way would now be to import both tables as Class. That's easy and
>>>>>>> then create edges with the *name_id*. Everything is fine with that,
>>>>>>> but it would be much easier to generate edges with ETL if that is
>>>>>>> anyway
>>>>>>> possible. So I don't know what is more efficient. Someone else also
>>>>>>> said
>>>>>>> that Sails (https://github.com/vjsrinath/sails-orientdb) might be a
>>>>>>> good choice.
>>>>>>>
>>>>>>> If ETL allows that I could imagine about the following way:
>>>>>>>
>>>>>>> - read line of table1 -> safe in memory
>>>>>>> - look in table2 if name_id exists
>>>>>>> - -> if yes, create vertex Tabel1 and Table2
>>>>>>> - -> if no, create vertex Tabel1 and Table2 and create an edge
>>>>>>> - and so on
>>>>>>>
>>>>>>> Of course this is very crazy if you imagine 30 tables or so. Was
>>>>>>> just an idea, just to check if that is already possible somehow.
>>>>>>>
>>>>>>> --
>>>>>>>
>>>>>>> ---
>>>>>>> You received this message because you are subscribed to the Google
>>>>>>> Groups "OrientDB" group.
>>>>>>> To unsubscribe from this group and stop receiving emails from it,
>>>>>>> send an email to [email protected].
>>>>>>>
>>>>>>> For more options, visit https://groups.google.com/d/optout.
>>>>>>>
>>>>>>
>>>>>> --
>>>>>
>>>>> ---
>>>>> You received this message because you are subscribed to the Google
>>>>> Groups "OrientDB" group.
>>>>> To unsubscribe from this group and stop receiving emails from it, send
>>>>> an email to [email protected].
>>>>> For more options, visit https://groups.google.com/d/optout.
>>>>>
>>>>
>>>> --
>>>
>>> ---
>>> You received this message because you are subscribed to the Google
>>> Groups "OrientDB" group.
>>> To unsubscribe from this group and stop receiving emails from it, send
>>> an email to [email protected].
>>> For more options, visit https://groups.google.com/d/optout.
>>>
>>
>>
--
---
You received this message because you are subscribed to the Google Groups
"OrientDB" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
