worldbrain-data-converter

Convert from a CSV containing page data to WorldBrain web extension compatible JSON data. The JSON data can be imported into the PouchDB via the web extension's user interface.

Installation

With yarn:

$ yarn global add worldbrain-data-converter

or with NPM:

$ npm i -g worldbrain-data-converter

Usage

To convert a CSV containing relevant page data fields:

$ worldbrain-data-converter -i /path/to/file.csv -o /path/to/output.txt

Or simply redirect IO files on stdout/stdin:

$ worldbrain-data-converter < /path/to/file.csv > /path/to/output.txt

Outputting to batches of 10000 docs/lines named: output_aa, output_ab, etc:

$ worldbrain-data-converter < /path/to/file.csv | split -l 10000 - output_

Full options:

$ worldbrain-data-converter --help
Usage: worldbrain-data-converter

  -o, --output=ARG          name of output file (default: stdout)
  -i, --input=ARG           name of input CSV file (default: stdin)
  -v, --maxVisits=ARG       max number of visit docs to generate per page (default: 10)
  -b, --bookmarkChance=ARG  percentage chance that a bookmark doc will be created for input row (default: 1)
  -s, --imports             schedule converted docs for imports for later filling out
  -h, --help                display this help

Details on input format

Should be a CSV containing columns url, title, and body. From each row in the CSV, the following WorldBrain data will be produced:

• 1 page doc
• 0-maxVisits visit docs
• 0-1 bookmark docs (change to generate specified via bookmarkChance)

body is the page text content.

Maybe later have the option of being able to accept other fields, or miss them out (only url really matters).

Details on output format

Should be a new-line delimited JSON file, which each line containing one JSON object representing a document in the web extension's PouchDB. Data will be derived from input file and also generated for unavailable fields.

The following data is generated:

• _id for all doc types
• content.keywords for page docs
• content.description for page docs

Maybe will have a flag to disable data generation of these, and add support in the input file format.

Split output for importing into extension

As the extension is currently running entirely within the browser, things like memory management with file IO are painful. It's recommended to pipe the output of this script to something like split to split up the output files. Split should happen per line, as the output format is new-line delimited JSON. Splitting by bytes is not handled by the extension's import process.

Example:

$ worldbrain-data-converter < /path/to/file.csv | split -l ${DOCS_PER_FILE} - ${OUTPUT_FILE_PREFIX}