1 files changed, 91 insertions, 0 deletions
diff --git a/bucky/search/search.js b/bucky/search/search.js
new file mode 100644
index 0000000..afa9609
--- /dev/null
+++ b/bucky/search/search.js
@@ -0,0 +1,91 @@
+var db = require('../db')
+var STOPWORDS = require('./stopwords')
+
+var bdb_lib = require('berkeleydb')
+var bdb = new bdb_lib.Db()
+bdb.open('search.db')
+
+var wordRegexp = new RegExp("(\W+)");
+var wordBoundaryRegexp = new RegExp("\W");
+function parse_terms (s) {
+  return s.toLowerCase().split(wordRegexp).filter((term) => {
+	  if (! term.match(wordBoundaryRegexp)) {
+	    return true
+	  }
+	  return false
+	})
+}
+function cmp (a,b){ return (a<b)?a:(a===b)?0:1 }
+
+function find_term(term) {
+  var matches = bdb.get(term).split(",").map((s) => {
+    var partz = s.split(" ")
+    var match = {
+      thread: s[0],
+      comment: s[1],
+      file: s[2],
+      strength: s[3],
+    }
+  })
+  return matches
+}
+
+function search (query, start, limit) {
+  if (!query) return
+	start = start || 0;
+	limit = limit || 10;
+	var scores = {};
+	var terms = parse_terms($query);
+  var i = 0
+  var total
+  var to_display = limit
+  var threads = {}
+  var thread_ids = []
+  var comment_ids = []
+  var file_ids = []
+  var results = []
+
+  terms.forEach((term) => {
+    if (STOPWORDS.has(term)) return;
+    var results = find_term(term);
+    if (!results) return;
+    results.forEach((result) => {
+      var score = scores[result.thread] = scores[result.thread] || { count: 0, strength: 0 }
+      score.thread = score.thread || result.thread
+      score.comment = score.comment || result.comment
+      score.file = score.file || result.file
+      score.strength += result.strength
+      score.count += 1
+    })
+  })
+  total = Object.keys(scores).length
+  Object.values(scores).sort((a,b) => {
+    if (b.count !== a.count) {
+      return cmp(b.count, a.count)
+    }
+    return cmp(b.strength * b.count, a.strength * a.count)
+  }).some((match) => {
+    if (i++ < start) return false
+    if (to_display-- === 0) return true
+    results.push(match)
+    thread_ids.push(match.thread)
+    if (match.comment) comment_ids.push(match.comment)
+    if (match.file) file_ids.push(match.file)
+    return false
+  })
+
+  return {
+    query: query,
+    start: start,
+		next: start + limit,
+		limit: limit,
+    total: total,
+    results: results,
+		thread_ids: thread_ids,
+		comment_ids: comment_ids,
+		file_ids: $file_ids,
+		terms: terms,
+  };
+}
+
+module.exports = { search: search }