From d285e4e4859fb7cf8cc45bb81848635691af3630 Mon Sep 17 00:00:00 2001
From: BelfordZ <belfordz66@gmail.com>
Date: Thu, 12 Oct 2017 20:49:00 -0700
Subject: [PATCH] fix(search) Update getAutoComplete
---
package.json | 2 +-
server/controllers/lighthouse.js | 72 ++++++++++++++++++++++----------
server/utils/importer/index.js | 2 +-
3 files changed, 52 insertions(+), 24 deletions(-)
diff --git a/package.json b/package.json
index fac36c6..cf377a2 100644
--- a/package.json
+++ b/package.json
@@ -2,7 +2,7 @@
"name": "lighthouse",
"description": "Lighthouse is a lightning-fast advanced search engine API for publications on the lbrycrd with autocomplete capabilities.",
"version": "0.0.1",
- "author": "filipnyquist <filip@lbry.io>,
+ "author": "filipnyquist <filip@lbry.io>",
"keywords": [
"lbry",
"search",
diff --git a/server/controllers/lighthouse.js b/server/controllers/lighthouse.js
index d69a5e3..fbdf141 100644
--- a/server/controllers/lighthouse.js
+++ b/server/controllers/lighthouse.js
@@ -18,6 +18,7 @@ const eclient = new elasticsearch.Client({
},
});
+
function getResults (input) {
if (input.size === undefined) input.size = 10;
if (input.from === undefined) input.from = 0;
@@ -46,30 +47,57 @@ function getResults (input) {
});
}
-function getAutoComplete (input) {
- if (input.size === undefined) input.size = 10;
- if (input.from === undefined) input.from = 0;
+function getIndex() {
+ // ideally, data is inserted into elastic search with an index that helps us query it faster/better results
+ // A simple start is to default queries to be within the n months, and to make a new index each month.
+ return;
+}
+
+function getRoutingKey() {
+ // This is the most important field for performance. Being able to route the queries ahead of time can make typedowns insanely good.
+ return;
+}
+
+function getAutoCompleteQuery(query) {
+ return {
+ multi_match: {
+ query: query.s.trim(),
+ type: 'phrase_prefix',
+ slop: 5,
+ max_expansions: 50,
+ fields: [
+ 'name',
+ 'value.stream.metadata.author',
+ 'value.stream.metadata.title',
+ 'value.stream.metadata.description'
+ ]
+ }
+ };
+}
+
+function getFilter(query) {
+ // this is the best place for putting things like filtering on the type of content
+ // Perhaps we can add search param that will filter on how people have categorized / tagged their content
+ return;
+}
+
+function getAutoComplete(query) {
return eclient.search({
- index : 'claims',
- _source: ['name', 'value.stream.metadata.title', 'value.stream.metadata.author'],
- body : {
- 'query': {
- 'bool': {
- 'must': {
- 'query_string': {
- 'query' : '*' + input.s.trim() + '*',
- 'fields': [
- 'name',
- 'value.stream.metadata.title',
- 'value.stream.metadata.author',
- ],
- },
- },
- },
- },
- size: input.size,
- from: input.from,
+ index: getIndex(query) || 'claims',
+ routing: getRoutingKey(query),
+ ignore_unavailable: true, // ignore error when date index does not exist
+ body: {
+ size: query.size || 5,
+ from: query.from || 0,
+ query: {
+ bool: {
+ must: getAutoCompleteQuery(query),
+ filter: getFilter(query)
+ }
+ }
},
+ size: query.size,
+ from: query.from,
});
}
diff --git a/server/utils/importer/index.js b/server/utils/importer/index.js
index a74b585..f20dc0d 100644
--- a/server/utils/importer/index.js
+++ b/server/utils/importer/index.js
@@ -124,7 +124,7 @@ export async function sync () {
// Done adding, update our claimTrie cache to latest and wait a bit...
await saveJSON(path.join(appRoot.path, 'claimTrieCache.json'), latestClaimTrie);
status.info = 'upToDate';
- await sleep(300000);
+ await sleep(600000);
sync();
} catch (err) {
winston.log(err);
--
2.45.3