Merge remote-tracking branch 'refs/remotes/nyaadevs/master'

import_to_es.py

@@ -8,6 +8,7 @@ with a cron job or some binlog-reading thing (TODO)
 from nyaa import app
 from nyaa.models import Torrent
 from elasticsearch import Elasticsearch
+from elasticsearch.client import IndicesClient
 from elasticsearch import helpers
 import progressbar
 import sys
@@ -21,7 +22,8 @@ bar = progressbar.ProgressBar(
             ' (', progressbar.ETA(), ') ',
             ])
 
-es = Elasticsearch()
+es = Elasticsearch(timeout=30)
+ic = IndicesClient(es)
 
 # turn into thing that elasticsearch indexes. We flatten in
 # the stats (seeders/leechers) so we can order by them in es naturally.
@@ -89,4 +91,10 @@ def page_query(query, limit=sys.maxsize, batch_size=10000):
         bar.update(start)
         start = min(limit, start + batch_size)
 
+# turn off refreshes while bulk loading
+ic.put_settings(body={'index': {'refresh_interval': '-1'}}, index=app.config['ES_INDEX_NAME'])
+
 helpers.bulk(es, (mk_es(t) for t in page_query(Torrent.query)), chunk_size=10000)
+
+# restore to near-enough real time
+ic.put_settings(body={'index': {'refresh_interval': '30s'}}, index=app.config['ES_INDEX_NAME'])

nyaa/routes.py

@@ -116,15 +116,13 @@ def home(rss):
     if flask.request.args.get('page') == 'rss':
         rss = True
 
-    term = flask.request.args.get('q')
+    term = flask.request.args.get('q', flask.request.args.get('term'))
     sort = flask.request.args.get('s')
     order = flask.request.args.get('o')
-    category = flask.request.args.get('c')
-    quality_filter = flask.request.args.get('f')
-    user_name = flask.request.args.get('u')
-    page = flask.request.args.get('p')
-    if page:
-        page = int(page)
+    category = flask.request.args.get('c', flask.request.args.get('cats'))
+    quality_filter = flask.request.args.get('f', flask.request.args.get('filter'))
+    user_name = flask.request.args.get('u', flask.request.args.get('user'))
+    page = flask.request.args.get('p', flask.request.args.get('offset', 1, int), int)
 
     per_page = app.config.get('RESULTS_PER_PAGE')
     if not per_page:
@@ -143,7 +141,7 @@ def home(rss):
         'order': order or 'desc',
         'category': category or '0_0',
         'quality_filter': quality_filter or '0',
-        'page': page or 1,
+        'page': page,
         'rss': rss,
         'per_page': per_page
     }

nyaa/static/js/main.js

@@ -105,8 +105,13 @@ document.addEventListener("DOMContentLoaded", function() {
     var previewTabEl = markdownEditor.querySelector(previewTabSelector);
     var targetEl = markdownEditor.querySelector(targetSelector);
 
+    var reader = new commonmark.Parser({safe: true});
+    var writer = new commonmark.HtmlRenderer({safe: true});
+    writer.softbreak = '<br />';
+
     previewTabEl.addEventListener('click', function () {
-      targetEl.innerHTML = marked(sourceSelector.value.trim(), { sanitize: true, breaks:true });
+      var parsed = reader.parse(sourceSelector.value.trim());
+      targetEl.innerHTML = writer.render(parsed);
     });
   });
 });

nyaa/templates/layout.html

@@ -8,6 +8,7 @@
 		<meta http-equiv="X-UA-Compatible" content="IE=edge">
 		<link rel="shortcut icon" type="image/png" href="/static/favicon.png">
 		<link rel="icon" type="image/png" href="/static/favicon.png">
+		<link rel="mask-icon" href="/static/pinned-tab.svg" color="#3582F7">
 		<link rel="alternate" type="application/rss+xml" href="{% if rss_filter %}{{ url_for('home', page='rss', _external=True, **rss_filter) }}{% else %}{{ url_for('home', page='rss', _external=True) }}{% endif %}" />
 
 		<!-- Bootstrap core CSS -->
@@ -34,7 +35,7 @@
 		<!-- Modified to not apply border-radius to selectpickers and stuff so our navbar looks cool -->
 		<script src="/static/js/bootstrap-select.js"></script>
 		<script src="/static/js/main.js"></script>
-		<script src="//cdnjs.cloudflare.com/ajax/libs/marked/0.3.6/marked.min.js"></script>
+		<script src="//cdnjs.cloudflare.com/ajax/libs/commonmark/0.27.0/commonmark.min.js"></script>
 
 		<!-- HTML5 shim and Respond.js for IE8 support of HTML5 elements and media queries -->
 		<!--[if lt IE 9]>

nyaa/templates/view.html

@@ -122,8 +122,11 @@
 <script>
 	var target = document.getElementById('torrent-description');
 	var text = target.innerHTML;
-	var html = marked(text.trim(), { sanitize: true, breaks:true });
-	target.innerHTML = html;
+	var reader = new commonmark.Parser({safe: true});
+	var writer = new commonmark.HtmlRenderer({safe: true});
+	writer.softbreak = '<br />';
+	var parsed = reader.parse(text.trim());
+	target.innerHTML = writer.render(parsed);
 </script>
 
 {% endblock %}

sync_es.py

@@ -23,6 +23,7 @@ changes that happen while the import_to_es script is dumping stuff from the
 database into es, at the expense of redoing a (small) amount of indexing.
 """
 from elasticsearch import Elasticsearch
+from elasticsearch.helpers import bulk
 from pymysqlreplication import BinLogStreamReader
 from pymysqlreplication.row_event import UpdateRowsEvent, DeleteRowsEvent, WriteRowsEvent
 from datetime import datetime
@@ -50,7 +51,7 @@ NT_DB = 'nyaav2'
 with open(SAVE_LOC) as f:
     pos = json.load(f)
 
-es = Elasticsearch()
+es = Elasticsearch(timeout=30)
 
 stream = BinLogStreamReader(
         # TODO parse out from config.py or something
@@ -58,8 +59,7 @@ stream = BinLogStreamReader(
             'host': MYSQL_HOST,
             'port': MYSQL_PORT,
             'user': MYSQL_USER,
-            'passwd': MYSQL_PW,
-
+            'passwd': MYSQL_PW
         },
         server_id=10, # arbitrary
         # only care about this database currently
@@ -108,60 +108,76 @@ def reindex_torrent(t, index_name):
         "has_torrent": bool(t['has_torrent']),
     }
     # update, so we don't delete the stats if present
-    es.update(
-        index=index_name,
-        doc_type='torrent',
-        id=t['id'],
-        body={"doc": doc, "doc_as_upsert": True})
+    return {
+        '_op_type': 'update',
+        '_index': index_name,
+        '_type': 'torrent',
+        '_id': str(t['id']),
+        "doc": doc,
+        "doc_as_upsert": True
+    }
 
 def reindex_stats(s, index_name):
-    es.update(
-        index=index_name,
-        doc_type='torrent',
-        id=s['torrent_id'],
-        body={
+    # update the torrent at torrent_id, assumed to exist;
+    # this will always be the case if you're reading the binlog
+    # in order; the foreign key constraint on torrrent_id prevents
+    # the stats row rom existing if the torrent isn't around.
+    return {
+        '_op_type': 'update',
+        '_index': index_name,
+        '_type': 'torrent',
+        '_id': str(s['torrent_id']),
         "doc": {
             "stats_last_updated": s["last_updated"],
             "download_count": s["download_count"],
             "leech_count": s['leech_count'],
             "seed_count": s['seed_count'],
-            }, "doc_as_upsert": True})
+        }}
+
+def delet_this(row, index_name):
+    return {
+        "_op_type": 'delete',
+        '_index': index_name,
+        '_type': 'torrent',
+        '_id': str(row['values']['id'])}
 
 n = 0
 last_save = time.time()
 
 for event in stream:
-    for row in event.rows:
     if event.table == "nyaa_torrents" or event.table == "sukebei_torrents":
         if event.table == "nyaa_torrents":
             index_name = "nyaa"
         else:
             index_name = "sukebei"
         if type(event) is WriteRowsEvent:
-                reindex_torrent(row['values'], index_name)
+            bulk(es, (reindex_torrent(row['values'], index_name) for row in event.rows))
         elif type(event) is UpdateRowsEvent:
-                reindex_torrent(row['after_values'], index_name)
+            # UpdateRowsEvent includes the old values too, but we don't care
+            bulk(es, (reindex_torrent(row['after_values'], index_name) for row in event.rows))
         elif type(event) is DeleteRowsEvent:
-                # just delete it
-                es.delete(index=index_name, doc_type='torrent', id=row['values']['id'])
+            # ok, bye
+            bulk(es, (delet_this(row, index_name) for row in event.rows))
         else:
             raise Exception(f"unknown event {type(event)}")
     elif event.table == "nyaa_statistics" or event.table == "sukebei_statistics":
-            if event.table == "nyaa_torrents":
+        if event.table == "nyaa_statistics":
             index_name = "nyaa"
         else:
             index_name = "sukebei"
         if type(event) is WriteRowsEvent:
-                reindex_stats(row['values'], index_name)
+            bulk(es, (reindex_stats(row['values'], index_name) for row in event.rows))
         elif type(event) is UpdateRowsEvent:
-                reindex_stats(row['after_values'], index_name)
+            bulk(es, (reindex_stats(row['after_values'], index_name) for row in event.rows))
         elif type(event) is DeleteRowsEvent:
-                # uh ok. assume that the torrent row will get deleted later.
+            # uh ok. assume that the torrent row will get deleted later,
+            # which will clean up the entire es "torrent" document
             pass
         else:
             raise Exception(f"unknown event {type(event)}")
     else:
       raise Exception(f"unknown table {s.table}")
+
     n += 1
     if n % 100 == 0 or time.time() - last_save > 30:
         log.info(f"saving position {stream.log_file}/{stream.log_pos}")