gedcom-d3/index.js

A GEDCOM parser, which translates GEDCOM (.ged) files into D3 capable JSON. This package is specifically designed to prepare data for use in https://github.com/oh-kay-blanket/blood-lines

var crawl = require('tree-crawl');

// from https://github.com/madprime/python-gedcom/blob/master/gedcom/__init__.py
// * Level must start with nonnegative int, no leading zeros.
// * Pointer optional, if it exists it must be flanked by '@'
// * Tag must be alphanumeric string
// * Value optional, consists of anything after a space to end of line
//   End of line defined by \n or \r
var lineRe = /\s*(0|[1-9]+[0-9]*) (@[^@]+@ |)([A-Za-z0-9_]+)( [^\n\r]*|)/;

function parse(input) {

    var start = { root: { tree: [] }, level: 0 };
    start.pointer = start.root;

    var data = input
        .split('\n')
        .map(mapLine)
        .filter(function(_) { return _; })
        .reduce(buildTree, start)
        .root;

    crawl(data, cleanUp, { getChildren });
    //console.log(data);
    //var cleanTree = data.tree.filter(hasTag('INDI'));
    //console.log(cleanTree);
    return data.tree;

    // the basic trick of this module is turning the suggested tree
    // structure of a GEDCOM file into a tree in JSON. This reduction
    // does that. The only real trick is the `.up` member of objects
    // that points to a level up in the structure. This we have to
    // censor before JSON.stringify since it creates circular references.
    function buildTree(memo, data) {
        if (data.level === memo.level) {
            memo.pointer.tree.push(data);
        } else if (data.level > memo.level) {
            var up = memo.pointer;
            memo.pointer = memo.pointer.tree[
                memo.pointer.tree.length - 1];
                memo.pointer.tree.push(data);
                memo.pointer.up = up;
                memo.level = data.level;
        } else if (data.level < memo.level) {
            // the jump up in the stack may be by more than one, so ascend
            // until we're at the right level.
            while (data.level <= memo.pointer.level && memo.pointer.up) {
                memo.pointer = memo.pointer.up;
            }
            memo.pointer.tree.push(data);
            memo.level = data.level;
        }

        return memo;
    }

    function mapLine(data) {
        var match = data.match(lineRe);
        if (!match) return null;
        return {
            level: parseInt(match[1], 10),
            pointer: match[2].trim(),
            tag: match[3].trim(),
            data: match[4].trimLeft(),
            tree: []
        };
    }

    function cleanUp(node) {
        delete node.up;
        delete node.level;
    }

    function getChildren(node) {
        //console.log(node.tree);
        return node.tree;
    }
}

module.exports.parse = parse;
module.exports.d3ize = require('./d3ize');