Compare commits
5 Commits
| Author | SHA1 | Date | |
|---|---|---|---|
| 8700bda104 | |||
| 3160e02d41 | |||
| cfa61c2066 | |||
| 27faea1823 | |||
| df76e34c71 |
@@ -1,67 +0,0 @@
|
|||||||
import logging
|
|
||||||
logging.basicConfig(
|
|
||||||
format='%(asctime)s - %(name)s - %(levelname)s - %(message)s',
|
|
||||||
level=logging.INFO)
|
|
||||||
|
|
||||||
import database
|
|
||||||
from sqlalchemy import select
|
|
||||||
import search
|
|
||||||
import sys
|
|
||||||
|
|
||||||
import time
|
|
||||||
import json
|
|
||||||
import requests
|
|
||||||
from bs4 import BeautifulSoup
|
|
||||||
|
|
||||||
database.init()
|
|
||||||
search.init()
|
|
||||||
|
|
||||||
BATCH_SIZE = 1000
|
|
||||||
|
|
||||||
def put_stories(stories):
|
|
||||||
return search.meili_api(requests.post, 'indexes/qotnews/documents', stories)
|
|
||||||
|
|
||||||
def get_update(update_id):
|
|
||||||
return search.meili_api(requests.get, 'tasks/{}'.format(update_id))
|
|
||||||
|
|
||||||
if __name__ == '__main__':
|
|
||||||
num_stories = database.count_stories()
|
|
||||||
|
|
||||||
print('Reindex {} stories?'.format(num_stories))
|
|
||||||
print('Press ENTER to continue, ctrl-c to cancel')
|
|
||||||
input()
|
|
||||||
|
|
||||||
story_list = database.get_story_list()
|
|
||||||
|
|
||||||
count = 1
|
|
||||||
while len(story_list):
|
|
||||||
stories = []
|
|
||||||
|
|
||||||
for _ in range(BATCH_SIZE):
|
|
||||||
try:
|
|
||||||
sid = story_list.pop()
|
|
||||||
except IndexError:
|
|
||||||
break
|
|
||||||
|
|
||||||
story = database.get_story(sid)
|
|
||||||
print('Indexing {}/{} id: {} title: {}'.format(count, num_stories, sid[0], story.title))
|
|
||||||
story_obj = json.loads(story.full_json)
|
|
||||||
story_obj.pop('comments', False)
|
|
||||||
if 'text' in story_obj and story_obj['text']:
|
|
||||||
soup = BeautifulSoup(story_obj['text'], 'html.parser')
|
|
||||||
story_obj['text'] = soup.get_text()
|
|
||||||
stories.append(story_obj)
|
|
||||||
count += 1
|
|
||||||
|
|
||||||
res = put_stories(stories)
|
|
||||||
update_id = res['taskUid']
|
|
||||||
|
|
||||||
print('Waiting for processing', end='')
|
|
||||||
while get_update(update_id)['status'] != 'succeeded':
|
|
||||||
time.sleep(0.5)
|
|
||||||
print('.', end='', flush=True)
|
|
||||||
|
|
||||||
print()
|
|
||||||
|
|
||||||
print('Done.')
|
|
||||||
|
|
||||||
@@ -11,7 +11,6 @@ import sys
|
|||||||
import time
|
import time
|
||||||
import json
|
import json
|
||||||
import requests
|
import requests
|
||||||
from bs4 import BeautifulSoup
|
|
||||||
|
|
||||||
database.init()
|
database.init()
|
||||||
search.init()
|
search.init()
|
||||||
@@ -46,9 +45,6 @@ if __name__ == '__main__':
|
|||||||
story = database.get_story(sid)
|
story = database.get_story(sid)
|
||||||
print('Indexing {}/{} id: {} title: {}'.format(count, num_stories, sid[0], story.title))
|
print('Indexing {}/{} id: {} title: {}'.format(count, num_stories, sid[0], story.title))
|
||||||
story_obj = json.loads(story.meta_json)
|
story_obj = json.loads(story.meta_json)
|
||||||
if 'text' in story_obj and story_obj['text']:
|
|
||||||
soup = BeautifulSoup(story_obj['text'], 'html.parser')
|
|
||||||
story_obj['text'] = soup.get_text()
|
|
||||||
stories.append(story_obj)
|
stories.append(story_obj)
|
||||||
count += 1
|
count += 1
|
||||||
|
|
||||||
|
|||||||
+23
-22
@@ -10,8 +10,7 @@ SEARCH_ENABLED = bool(settings.MEILI_URL)
|
|||||||
|
|
||||||
def meili_api(method, route, json=None, params=None, parse_json=True):
|
def meili_api(method, route, json=None, params=None, parse_json=True):
|
||||||
try:
|
try:
|
||||||
headers = {'Authorization': 'Bearer ' + settings.MEILI_API_KEY}
|
r = method(settings.MEILI_URL + route, json=json, params=params, timeout=4)
|
||||||
r = method(settings.MEILI_URL + route, json=json, params=params, headers=headers, timeout=4)
|
|
||||||
if r.status_code > 299:
|
if r.status_code > 299:
|
||||||
raise Exception('Bad response code ' + str(r.status_code))
|
raise Exception('Bad response code ' + str(r.status_code))
|
||||||
if parse_json:
|
if parse_json:
|
||||||
@@ -25,40 +24,42 @@ def meili_api(method, route, json=None, params=None, parse_json=True):
|
|||||||
logging.error('Problem with MeiliSearch api route: %s: %s', route, str(e))
|
logging.error('Problem with MeiliSearch api route: %s: %s', route, str(e))
|
||||||
return False
|
return False
|
||||||
|
|
||||||
def update_settings():
|
def create_index():
|
||||||
json = {
|
json = dict(uid='qotnews', primaryKey='id')
|
||||||
'rankingRules': ['words', 'typo', 'proximity', 'attribute', 'date:desc', 'exactness'],
|
return meili_api(requests.post, 'indexes', json=json)
|
||||||
'searchableAttributes': ['title', 'url', 'author', 'text'],
|
|
||||||
'displayedAttributes': ['id', 'ref', 'source', 'author', 'author_link', 'score', 'date', 'title', 'link', 'url', 'num_comments', 'text'],
|
def update_rankings():
|
||||||
'stopWords': ['a', 'an', 'the', 'and', 'or', 'but', 'if', 'in', 'on', 'at', 'by', 'for', 'with', 'to', 'from', 'of', 'is', 'it', 'that', 'this'],
|
json = ['typo', 'words', 'proximity', 'date:desc', 'exactness']
|
||||||
}
|
return meili_api(requests.post, 'indexes/qotnews/settings/ranking-rules', json=json)
|
||||||
return meili_api(requests.patch, 'indexes/qotnews/settings', json=json)
|
|
||||||
|
def update_attributes():
|
||||||
|
json = ['title', 'url', 'author']
|
||||||
|
r = meili_api(requests.post, 'indexes/qotnews/settings/searchable-attributes', json=json)
|
||||||
|
json = ['id', 'ref', 'source', 'author', 'author_link', 'score', 'date', 'title', 'link', 'url', 'num_comments']
|
||||||
|
r = meili_api(requests.post, 'indexes/qotnews/settings/displayed-attributes', json=json)
|
||||||
|
return r
|
||||||
|
|
||||||
def init():
|
def init():
|
||||||
if not SEARCH_ENABLED:
|
if not SEARCH_ENABLED:
|
||||||
logging.info('Search is not enabled, skipping init.')
|
logging.info('Search is not enabled, skipping init.')
|
||||||
return
|
return
|
||||||
update_settings()
|
print(create_index())
|
||||||
|
update_rankings()
|
||||||
|
update_attributes()
|
||||||
|
|
||||||
def put_story(story):
|
def put_story(story):
|
||||||
if not SEARCH_ENABLED: return
|
if not SEARCH_ENABLED: return
|
||||||
return meili_api(requests.post, 'indexes/qotnews/documents', [story])
|
return meili_api(requests.post, 'indexes/qotnews/documents', [story])
|
||||||
|
|
||||||
def search(q, in_article=False):
|
def search(q):
|
||||||
if not SEARCH_ENABLED: return []
|
if not SEARCH_ENABLED: return []
|
||||||
|
params = dict(q=q, limit=settings.FEED_LENGTH)
|
||||||
json = dict(q=q, limit=settings.FEED_LENGTH)
|
r = meili_api(requests.get, 'indexes/qotnews/search', params=params, parse_json=False)
|
||||||
|
|
||||||
if True:
|
|
||||||
json['attributesToSearchOn'] = ['text']
|
|
||||||
json['attributesToCrop'] = ['text']
|
|
||||||
json['attributesToRetrieve'] = ['id', 'ref', 'source', 'author', 'author_link', 'score', 'date', 'title', 'link', 'url', 'num_comments']
|
|
||||||
json['cropLength'] = 80
|
|
||||||
|
|
||||||
r = meili_api(requests.post, 'indexes/qotnews/search', json=json, parse_json=False)
|
|
||||||
return r
|
return r
|
||||||
|
|
||||||
if __name__ == '__main__':
|
if __name__ == '__main__':
|
||||||
init()
|
init()
|
||||||
|
|
||||||
|
print(update_rankings())
|
||||||
|
|
||||||
print(search('facebook'))
|
print(search('facebook'))
|
||||||
|
|||||||
+2
-3
@@ -131,9 +131,8 @@ def apistats():
|
|||||||
@flask_app.route('/api/search', strict_slashes=False)
|
@flask_app.route('/api/search', strict_slashes=False)
|
||||||
def apisearch():
|
def apisearch():
|
||||||
q = request.args.get('q', '')
|
q = request.args.get('q', '')
|
||||||
in_article = request.args.get('article', False)
|
|
||||||
if len(q) >= 3:
|
if len(q) >= 3:
|
||||||
results = search.search(q, in_article)
|
results = search.search(q)
|
||||||
else:
|
else:
|
||||||
results = '[]'
|
results = '[]'
|
||||||
res = Response(results)
|
res = Response(results)
|
||||||
@@ -263,7 +262,7 @@ def static_story(sid):
|
|||||||
show_comments=request.path.endswith('/c'),
|
show_comments=request.path.endswith('/c'),
|
||||||
)
|
)
|
||||||
|
|
||||||
http_server = WSGIServer(('0.0.0.0', 33842), flask_app)
|
http_server = WSGIServer(('', 33842), flask_app)
|
||||||
|
|
||||||
def feed_thread():
|
def feed_thread():
|
||||||
global news_index, ref_list, current_item
|
global news_index, ref_list, current_item
|
||||||
|
|||||||
@@ -4,4 +4,3 @@
|
|||||||
|
|
||||||
meilisearch-linux-amd64
|
meilisearch-linux-amd64
|
||||||
data.ms/
|
data.ms/
|
||||||
data.ms.old/
|
|
||||||
|
|||||||
@@ -137,7 +137,8 @@ function Article({ cache }) {
|
|||||||
const textContent = v.textContent.trim();
|
const textContent = v.textContent.trim();
|
||||||
const isMath = (textContent.startsWith('\\(') && textContent.endsWith('\\)')) ||
|
const isMath = (textContent.startsWith('\\(') && textContent.endsWith('\\)')) ||
|
||||||
(textContent.startsWith('\\[') && textContent.endsWith('\\]')) ||
|
(textContent.startsWith('\\[') && textContent.endsWith('\\]')) ||
|
||||||
(textContent.startsWith('$$') && textContent.endsWith('$$'));
|
(textContent.startsWith('$$') && textContent.endsWith('$$')) ||
|
||||||
|
(textContent.startsWith('$') && textContent.endsWith('$') && textContent.indexOf('$') !== textContent.lastIndexOf('$') && !/\s/.test(textContent.charAt(textContent.length - 2)));
|
||||||
|
|
||||||
const props = { key: key };
|
const props = { key: key };
|
||||||
if (v.hasAttributes()) {
|
if (v.hasAttributes()) {
|
||||||
|
|||||||
@@ -1,7 +1,6 @@
|
|||||||
import React, { useState, useEffect } from 'react';
|
import React, { useState, useEffect } from 'react';
|
||||||
import { Link, useLocation, useHistory } from 'react-router-dom';
|
import { Link, useLocation } from 'react-router-dom';
|
||||||
import { Helmet } from 'react-helmet';
|
import { Helmet } from 'react-helmet';
|
||||||
import queryString from 'query-string';
|
|
||||||
import { sourceLink, infoLine, logos } from './utils.js';
|
import { sourceLink, infoLine, logos } from './utils.js';
|
||||||
import AbortController from 'abort-controller';
|
import AbortController from 'abort-controller';
|
||||||
|
|
||||||
@@ -9,19 +8,6 @@ function Results() {
|
|||||||
const [stories, setStories] = useState(false);
|
const [stories, setStories] = useState(false);
|
||||||
const [error, setError] = useState(false);
|
const [error, setError] = useState(false);
|
||||||
const location = useLocation();
|
const location = useLocation();
|
||||||
const history = useHistory();
|
|
||||||
|
|
||||||
const handleFilterChange = e => {
|
|
||||||
const isChecked = e.target.checked;
|
|
||||||
|
|
||||||
const currentQuery = queryString.parse(location.search);
|
|
||||||
if (isChecked) {
|
|
||||||
currentQuery.article = 'true';
|
|
||||||
} else {
|
|
||||||
delete currentQuery.article;
|
|
||||||
}
|
|
||||||
history.push('/search?' + queryString.stringify(currentQuery));
|
|
||||||
};
|
|
||||||
|
|
||||||
useEffect(() => {
|
useEffect(() => {
|
||||||
const controller = new AbortController();
|
const controller = new AbortController();
|
||||||
@@ -46,19 +32,11 @@ function Results() {
|
|||||||
};
|
};
|
||||||
}, [location.search]);
|
}, [location.search]);
|
||||||
|
|
||||||
const searchInArticle = queryString.parse(location.search).article === 'true';
|
|
||||||
|
|
||||||
return (
|
return (
|
||||||
<div className='container'>
|
<div className='container'>
|
||||||
<Helmet>
|
<Helmet>
|
||||||
<title>Search Results | QotNews</title>
|
<title>Search Results | QotNews</title>
|
||||||
</Helmet>
|
</Helmet>
|
||||||
|
|
||||||
<div style={{marginBottom: '1rem'}}>
|
|
||||||
<input type="checkbox" id="search-in-article" className="checkbox" checked={searchInArticle} onChange={handleFilterChange} />
|
|
||||||
<label htmlFor="search-in-article">Search in article</label>
|
|
||||||
</div>
|
|
||||||
|
|
||||||
{error && <p>Connection error?</p>}
|
{error && <p>Connection error?</p>}
|
||||||
{stories ?
|
{stories ?
|
||||||
<>
|
<>
|
||||||
@@ -78,10 +56,6 @@ function Results() {
|
|||||||
</div>
|
</div>
|
||||||
|
|
||||||
{infoLine(x)}
|
{infoLine(x)}
|
||||||
|
|
||||||
{!!x?._formatted &&
|
|
||||||
<p>{x._formatted.text.replace(/\n/g, ' ')}</p>
|
|
||||||
}
|
|
||||||
</div>
|
</div>
|
||||||
)
|
)
|
||||||
:
|
:
|
||||||
|
|||||||
@@ -15,9 +15,7 @@ function Search() {
|
|||||||
const newSearch = event.target.value;
|
const newSearch = event.target.value;
|
||||||
setSearch(newSearch);
|
setSearch(newSearch);
|
||||||
if (newSearch.length >= 3) {
|
if (newSearch.length >= 3) {
|
||||||
const currentQuery = queryString.parse(location.search);
|
const searchQuery = queryString.stringify({ 'q': newSearch });
|
||||||
currentQuery.q = newSearch;
|
|
||||||
const searchQuery = queryString.stringify(currentQuery);
|
|
||||||
history.replace('/search?' + searchQuery);
|
history.replace('/search?' + searchQuery);
|
||||||
} else {
|
} else {
|
||||||
history.replace('/');
|
history.replace('/');
|
||||||
@@ -26,9 +24,7 @@ function Search() {
|
|||||||
|
|
||||||
const searchAgain = (event) => {
|
const searchAgain = (event) => {
|
||||||
event.preventDefault();
|
event.preventDefault();
|
||||||
const currentQuery = queryString.parse(location.search);
|
const searchString = queryString.stringify({ 'q': event.target[0].value });
|
||||||
currentQuery.q = event.target[0].value;
|
|
||||||
const searchString = queryString.stringify(currentQuery);
|
|
||||||
history.push('/search?' + searchString);
|
history.push('/search?' + searchString);
|
||||||
inputRef.current.blur();
|
inputRef.current.blur();
|
||||||
}
|
}
|
||||||
|
|||||||
@@ -69,7 +69,7 @@
|
|||||||
}
|
}
|
||||||
|
|
||||||
.black .checkbox:checked + label::after {
|
.black .checkbox:checked + label::after {
|
||||||
border-color: #ddd;
|
border-color: #eee;
|
||||||
}
|
}
|
||||||
|
|
||||||
.black .copy-button {
|
.black .copy-button {
|
||||||
|
|||||||
@@ -65,7 +65,7 @@
|
|||||||
}
|
}
|
||||||
|
|
||||||
.dark .checkbox:checked + label::after {
|
.dark .checkbox:checked + label::after {
|
||||||
border-color: #ddd;
|
border-color: #eee;
|
||||||
}
|
}
|
||||||
|
|
||||||
.dark .copy-button {
|
.dark .copy-button {
|
||||||
|
|||||||
@@ -355,7 +355,8 @@ button.comment {
|
|||||||
top: 0.2em;
|
top: 0.2em;
|
||||||
width: 0.3rem;
|
width: 0.3rem;
|
||||||
height: 0.6rem;
|
height: 0.6rem;
|
||||||
border: solid #000;
|
border-style: solid;
|
||||||
|
border-color: #000;
|
||||||
border-width: 0 2px 2px 0;
|
border-width: 0 2px 2px 0;
|
||||||
transform: rotate(45deg);
|
transform: rotate(45deg);
|
||||||
}
|
}
|
||||||
|
|||||||
@@ -87,7 +87,7 @@
|
|||||||
}
|
}
|
||||||
|
|
||||||
.red .checkbox:checked + label::after {
|
.red .checkbox:checked + label::after {
|
||||||
border-color: #aa0000;
|
border-color: #dd0000;
|
||||||
}
|
}
|
||||||
|
|
||||||
.red .copy-button {
|
.red .copy-button {
|
||||||
|
|||||||
Reference in New Issue
Block a user