archive-backend

README

Backend for new archive site

Overview

Backend for new archive site including, ETLs from BB Metadata DB to Elasticsearch.

Commands

The archive-backend is meant to be executed as command line. Type archive-backend <command> -h to see how to use each command.

archive-backend server

Execute the backend api server for the new archive site.

archive-backend version

Print the version of archive-backend

Configuration

The default config file is config.toml in your current work directory.

See config.sample.toml for a sample config file.

Release and Deployment

Once development is done, all tests are green, we want to go live. All we have to do is simply execute misc/release.sh.

To add a pre-release tag, add the relevant environment variable. For example,

PRE_RELEASE=rc.1 misc/release.sh

MDB models

When MDB schema is changed we need to update the mdb package. Run this script:

misc/update_mdb_models.sh

Elasticsearch related stuff

(See the next section below for the instructions on installing Elasticsearch for Windows)

http://mrzard.github.io/blog/2015/03/25/elasticsearch-enable-mlockall-in-centos-7/

Plugins

1. Hebrew plugin: https://github.com/synhershko/elasticsearch-analysis-hebrew
2. Instead of standard analyzer for exact match (הריון to be same as היריון):

sudo bin/elasticsearch-plugin install analysis-phonetic

https://www.elastic.co/guide/en/elasticsearch/plugins/current/analysis-phonetic.html

WIP - Does not works yet.

1. ICU plugin to transliterate Russian (and others) to enable phonetic on them:

sudo bin/elasticsearch-plugin install analysis-icu

1. Ukrainial analyzer (fails for standard - Not started)

Build index

There are two more dependencies required to build index:

1. Open Office (soffice binary) - to convert all doc to docx.
2. python-docx pyton library - to get text from docx

• pip install python-docx

Elasticsearch installation for Windows

1. Download and install the Java Virtual Machine for Windows from http://www.oracle.com/technetwork/java/javase/downloads/jre8-downloads-2133155.html

1. Download and install the Elasticsearch 5.6.0 MSI from https://artifacts.elastic.co/downloads/elasticsearch/elasticsearch-5.6.0.msi

2. Open CMD as administrator

   1. Go to Elasticsearch bin directory

      cd C:\Program Files\Elastic\Elasticsearch\bin
      

   2. To install analysis-phonetic type

      elasticsearch-plugin install analysis-phonetic
      

   3. To install the hebrew plugin type

      elasticsearch-plugin install https://bintray.com/synhershko/elasticsearch-analysis-hebrew/download_file?file_path=elasticsearch-analysis-hebrew-5.6.0.zip
      

   4. Answer 'y' to the security question

      Continue with installation? [y/N]y
      

   5. To install ICU plugin type

      elasticsearch-plugin install analysis-icu
      

3. Download and install Python - version 2.7.x https://www.python.org/downloads/

4. Install python-docx (to get text from docx):

   • in CMD go to python directory

   cd C:\Python27
   

   • and type

   python -m pip install python-docx
   

5. Download and install LibreOffice (not OpenOffice!)

   https://www.libreoffice.org/donate/dl/win-x86_64/5.4.5/en-US/LibreOffice_5.4.5_Win_x64.msi

   Update 'soffice-bin' value with soffice.exe full path in config.toml, [elasticsearch] section: "C://Program Files//LibreOffice 5//program//soffice.exe"

6. Copy to config.toml the required commented-out lines from config.sample.toml that are related to Windows.

7. Updating assets:

   In order to make correct data indexing you should update the ES mapping configuration files (JSON files in /data/es/mappings):

   1. Exec. \es\mappings\make.py with python from the root path of the project. For example:

      C:\Users\[USER]\go\src\github.com\Bnei-Baruch\archive-backend>python C:\Users\[USER]\go\src\github.com\Bnei-Baruch\archive-backend\es\mappings\make.py
      

   2. From the root path of the project, type:

      go-bindata -debug data/...
      

   3. Edit bindata.go file (located in the root folder) and replace "package main" with "package bindata".
   4. Move the modified bindata.go file to /bindata folder (delete old bindata.go from /bindata if exist and make sure the bindata.go is not exist any more in the root folder).
   5. Repeat this steps any time make.py is changed and executed.

Install go-bindata

go get -u github.com/jteeuwen/go-bindata/...

License

MIT