From 14b7ade4dbeb1c62e43ac80498534c703e0dbf43 Mon Sep 17 00:00:00 2001 From: Peter Wu Date: Tue, 20 May 2014 23:52:45 +0200 Subject: Moves preprocessing to preprocess.js The only difference in preprocess is the addition of a new options parameter to avoid using a global var. --- preprocess.js | 74 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 74 insertions(+) create mode 100644 preprocess.js (limited to 'preprocess.js') diff --git a/preprocess.js b/preprocess.js new file mode 100644 index 0000000..03af077 --- /dev/null +++ b/preprocess.js @@ -0,0 +1,74 @@ +/* "Optimizes" nodes and edges by dropping uninteresting ones. (for example, + * nodes with no edges). + */ +/* jshint devel:true */ + +'use strict'; + +function preprocess(data, options) { + // map userID to nodes + var users = {}; + data.nodes.forEach(function (user, i) { + users[user.group] = user; + }); + console.log('Initial nodes count:', data.nodes.length); + console.log('Initial edges count:', data.edges.length); + + var ratelimit_count = 0, ratelimit_max = 10; + function ratelimit() { + return ratelimit_count <= ratelimit_max; + } + // filter away invalid edges + data.edges = data.edges.filter(function (link, i) { + var invalid = false; + if (!(link.source in users)) { + if (ratelimit()) console.warn('Dropping invalid source user', + link.source, 'at line', (i + 1), link); + invalid = true; + } + if (!(link.target in users)) { + if (ratelimit()) console.warn('Dropping invalid target user', + link.target, 'at line', (i + 1), link); + invalid = true; + } + if (link.source === link.target) { + if (ratelimit()) console.warn('Dropping self-referencing user', + link.target, 'at line', (i + 1), link); + invalid = true; + } + return !invalid; + }); + if (ratelimit_count > ratelimit_max) { + console.log('Supressed', ratelimit_count, 'messages'); + } + console.log('Valid edges count:', data.edges.length); + + // find all related users by userID + var hasRelations = {}; + data.edges.forEach(function (link) { + hasRelations[link.target] = 1; + hasRelations[link.source] = 1; + }); + + if (options.kill_loners) { + var hasRelated = {}; + data.nodes = data.nodes.filter(function (d) { + return d.group in hasRelations; + }); + console.log('Nodes count (after dropping loners):', data.nodes.length); + } + + // prepare data for force layout: map user IDs to indices + var userIds_indices = {}; + data.nodes.forEach(function (user, i) { + users[user.group] = user; + userIds_indices[user.group] = i; + }); + console.log('UserID to index map:', userIds_indices); + + // change userID of relation edges to indices + data.edges.map(function (link) { + link.source = userIds_indices[link.source]; + link.target = userIds_indices[link.target]; + }); +} -- cgit v1.2.1