84 lines
3.5 KiB
Markdown
84 lines
3.5 KiB
Markdown
## batch processing with python-client
|
|
|
|
There are some client libraries for OpenRefine that communicate with the [OpenRefine API](https://github.com/OpenRefine/OpenRefine/wiki/OpenRefine-API). I have prepared a docker container on top of the [Python Library from PaulMakepeace](https://github.com/PaulMakepeace/refine-client-py/) and extended the CLI with some options to create new OpenRefine projects from files.
|
|
|
|
### basic usage
|
|
|
|
**1) start server:**
|
|
> docker run -d --name=openrefine-server felixlohmeier/openrefine
|
|
|
|
**2) run client with one of the following commands:**
|
|
|
|
list projects:
|
|
> docker run --rm --link openrefine-server felixlohmeier/openrefine-client --list
|
|
|
|
create project from file:
|
|
> docker run --rm --link openrefine-server felixlohmeier/openrefine-client --create [FILE]
|
|
|
|
apply [rules from json file](http://kb.refinepro.com/2012/06/google-refine-json-and-my-notepad-or.html):
|
|
> docker run --rm --link openrefine-server felixlohmeier/openrefine-client --apply [FILE.json] [PROJECTID]
|
|
|
|
export project to file:
|
|
> docker run --rm --link openrefine-server felixlohmeier/openrefine-client --export [PROJECTID] --output=FILE.tsv
|
|
|
|
check help screen for more options:
|
|
> docker run --rm --link openrefine-server felixlohmeier/openrefine-client --help
|
|
|
|
**3) cleanup:**
|
|
> docker stop openrefine-server && docker rm openrefine-server
|
|
|
|
### example for customized run commands in interactive mode (e.g. for usage in terminals)
|
|
|
|
**1) start server in terminal A:**
|
|
|
|
```docker run --rm --name=openrefine-server -p 80:3333 -v /home/felix/refine:/data:z felixlohmeier/openrefine -i 0.0.0.0 -m 4G -d /data```
|
|
|
|
* automatically remove docker container when it exits
|
|
* set name "openrefine" for docker container
|
|
* publish internal port 3333 to host port 80
|
|
* mount host directory /home/felix/refine as working directory
|
|
* make openrefine available in the network
|
|
* increase java heap size to 4 GB
|
|
* set refine workspace to /data
|
|
* OpenRefine should be available at http://localhost
|
|
|
|
**2) start client in terminal B (prints help screen):**
|
|
|
|
```docker run --rm --link openrefine-server -v /home/felix/refine:/data:z felixlohmeier/openrefine-client```
|
|
|
|
* automatically remove docker container when it exits
|
|
* build up network connection with docker container "openrefine"
|
|
* mount host directory /home/felix/refine as working directory
|
|
* apply history in file /home/felix/refine/history.json to project with id 1234567890123
|
|
|
|
### example for customized run commands in detached mode (e.g. for usage in shell scripts)
|
|
|
|
**1) define variables (bring your own example data)**
|
|
> workingdir=/home/felix/refine
|
|
> inputfile=example.csv
|
|
> jsonfile=test.json
|
|
|
|
**2) start server**
|
|
|
|
```docker run -d --name=openrefine-server -v ${workingdir}:/data:z felixlohmeier/openrefine -i 0.0.0.0 -m 4G -d /data```
|
|
|
|
**3) create project (import file)**
|
|
|
|
```docker run --rm --link openrefine-server -v ${workingdir}:/data:z felixlohmeier/openrefine-client --create $inputfile```
|
|
|
|
**4) get project id**
|
|
|
|
```project=($(docker run --rm --link openrefine-server -v ${workingdir}:/data felixlohmeier/openrefine-client --list | cut -c 2-14))```
|
|
|
|
**5) apply transformations from json file**
|
|
|
|
```docker run --rm --link openrefine-server -v ${workingdir}:/data felixlohmeier/openrefine-client --apply ${jsonfile} ${project}```
|
|
|
|
**6) export project to file**
|
|
|
|
```docker run --rm --link openrefine-server -v ${workingdir}:/data felixlohmeier/openrefine-client --export --output=${project}.tsv ${project}```
|
|
|
|
**7) cleanup**
|
|
|
|
```docker stop -t=500 openrefine-server && docker rm openrefine-server```
|