Download the PHP package itstably/clickhouse-php-client without Composer

Clickhouse Client

Package was written as client for Clickhouse.

Client uses Guzzle for sending Http requests to Clickhouse servers.

Requirements

php7.1

Install

Composer

Usage

Client works with alone server and cluster. Also, client can make async select and insert (from local files) queries.

Alone server

Cluster

Before execute any query on cluster, you should provide cluster name and client will run all queries on specified cluster.

By default client will use random server in given list of servers or in specified cluster. If you want to perform request on specified server you should use using($hostname) method on client and then run query. Client will remember hostname for next queries:

Select queries

Any SELECT query will return instance of Result. This class implements interfaces \ArrayAccess, \Countable и \Iterator, which means that it can be used as an array.

Array with result rows can be obtained via rows property

Also you can get some statistic of your query execution:

1. Number of read rows
2. Number of read bytes
3. Time of query execution
4. Rows before limit at least

Statistic can be obtained via statistic property

Sync

Using local files

You can use local files as temporary tables in Clickhouse. You should pass as third argument array of TempTable instances. instance.

In this case will be sent one file to the server from which Clickhouse will extract data to temporary table. Structure of table will be:

• number - UInt64

If you pass such an array as a structure:

Then each column from file wil be named as _1, _2, _3.

You can provide path to file or pass FileInterface instance as second argument.

There is some other types of file streams which could be used to send to server:

• File - simple file stored on disk;
• FileFromString - stream created from string. For example: new FileFromString('1'.PHP_EOL.'2'.PHP_EOL.'3'.PHP_EOL)
• MergedFiles - stream which includes many files and merges them all in one. You should pass to constructor file path, which contains list of files which should be megred in one stream.
• TempTable - wrapper to any of FileInterface instance and contains structure. Usefull to make inserts using with MergedFiles.

Async

Unlike the readOne method, which returnsResult, the read method returns an array ofResult for each executed query.

In read method, you can pass the parameter $concurrency which is responsible for the maximum simultaneous number of requests.

Using local files

As with synchronous select request you can pass files to the server:

With asynchronous requests you can pass multiple files as with synchronous request.

Insert queries

Insert queries always returns true or throws exceptions in case of error.

Data can be written row by row or from local CSV or TSV files.

In case of writeFiles queries executes asynchronously. If you have butch of files and you want to insert them in one insert query, you can use our ccat utility and MergedFiles instance instead of File. You should put list of files to insert into one file:

Building ccat

ccat sources placed into utils/ccat directory. Just run make && make install to build and install library into bin directory of package. There are already compiled binary of ccat in bin directory, but it may not work on some systems.

In writeFiles method, you can pass the parameter $concurrency which is responsible for the maximum simultaneous number of requests.

Other queries

In addition to SELECT and INSERT queries, you can execute other queries :) There is statement method for this purposes.

Testing

Roadmap

• Add ability to save query result in local file

Contributing

Please send your own pull-requests and make suggestions on how to improve anything. We will be very grateful.

Thx!