Modify search to work with article contents

Co-authored-by: aider (gemini/gemini-2.5-pro) <aider@aider.chat>

apiserver/reindex.py

@@ -0,0 +1,67 @@
+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.')
+

apiserver/scripts/reindex.py

@@ -11,6 +11,7 @@ import sys
 import time
 import json
 import requests
+from bs4 import BeautifulSoup
 
 database.init()
 search.init()
@@ -45,6 +46,9 @@ if __name__ == '__main__':
             story = database.get_story(sid)
             print('Indexing {}/{} id: {} title: {}'.format(count, num_stories, sid[0], story.title))
             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)
             count += 1
 

apiserver/search.py

@@ -10,7 +10,8 @@ SEARCH_ENABLED = bool(settings.MEILI_URL)
 
 def meili_api(method, route, json=None, params=None, parse_json=True):
     try:
-        r = method(settings.MEILI_URL + route, json=json, params=params, timeout=4)
+        headers = {'Authorization': 'Bearer ' + settings.MEILI_API_KEY}
+        r = method(settings.MEILI_URL + route, json=json, params=params, headers=headers, timeout=4)
         if r.status_code > 299:
             raise Exception('Bad response code ' + str(r.status_code))
         if parse_json:
@@ -24,42 +25,40 @@ def meili_api(method, route, json=None, params=None, parse_json=True):
         logging.error('Problem with MeiliSearch api route: %s: %s', route, str(e))
         return False
 
-def create_index():
+def update_settings():
-    json = dict(uid='qotnews', primaryKey='id')
+    json = {
-    return meili_api(requests.post, 'indexes', json=json)
+        'rankingRules': ['words', 'typo', 'proximity', 'attribute', 'date:desc', 'exactness'],
-
+        'searchableAttributes': ['title', 'url', 'author', 'text'],
-def update_rankings():
+        'displayedAttributes': ['id', 'ref', 'source', 'author', 'author_link', 'score', 'date', 'title', 'link', 'url', 'num_comments', 'text'],
-    json = ['typo', 'words', 'proximity', 'date:desc', 'exactness']
+        'stopWords': ['a', 'an', 'the', 'and', 'or', 'but', 'if', 'in', 'on', 'at', 'by', 'for', 'with', 'to', 'from', 'of', 'is', 'it', 'that', 'this'],
-    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():
     if not SEARCH_ENABLED:
         logging.info('Search is not enabled, skipping init.')
         return
-    print(create_index())
+    update_settings()
-    update_rankings()
-    update_attributes()
 
 def put_story(story):
     if not SEARCH_ENABLED: return
     return meili_api(requests.post, 'indexes/qotnews/documents', [story])
 
-def search(q):
+def search(q, in_article=False):
     if not SEARCH_ENABLED: return []
-    params = dict(q=q, limit=settings.FEED_LENGTH)
+
-    r = meili_api(requests.get, 'indexes/qotnews/search', params=params, parse_json=False)
+    json = dict(q=q, limit=settings.FEED_LENGTH)
+
+    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
     
 if __name__ == '__main__':
     init()
 
-    print(update_rankings())
-
     print(search('facebook'))

apiserver/server.py

@@ -131,8 +131,9 @@ def apistats():
 @flask_app.route('/api/search', strict_slashes=False)
 def apisearch():
     q = request.args.get('q', '')
+    in_article = request.args.get('article', False)
     if len(q) >= 3:
-        results = search.search(q)
+        results = search.search(q, in_article)
     else:
         results = '[]'
     res = Response(results)
@@ -262,7 +263,7 @@ def static_story(sid):
         show_comments=request.path.endswith('/c'),
     )
 
-http_server = WSGIServer(('', 33842), flask_app)
+http_server = WSGIServer(('0.0.0.0', 33842), flask_app)
 
 def feed_thread():
     global news_index, ref_list, current_item

searchserver/.gitignore

@@ -4,3 +4,4 @@
 
 meilisearch-linux-amd64
 data.ms/
+data.ms.old/

webclient/src/Article.js

@@ -12,7 +12,6 @@ const DANGEROUS_TAGS = ['svg', 'math'];
 const latexDelimiters = [
     { left: '$$', right: '$$', display: true },
     { left: '\\[', right: '\\]', display: true },
-    { left: '$', right: '$', display: false },
     { left: '\\(', right: '\\)', display: false }
 ];
 
@@ -106,7 +105,7 @@ function Article({ cache }) {
 
 			if (v.nodeName === '#text') {
 				const text = v.data;
-				if (text.includes('\\[') || text.includes('\\(') || text.includes('$$') || text.includes('$')) {
+				if (text.includes('\\[') || text.includes('\\(') || text.includes('$$')) {
 					return <Latex key={key} delimiters={latexDelimiters}>{text}</Latex>;
 				}
 
@@ -138,8 +137,7 @@ function Article({ cache }) {
 			const textContent = v.textContent.trim();
 			const isMath = (textContent.startsWith('\\(') && textContent.endsWith('\\)')) ||
 				(textContent.startsWith('\\[') && textContent.endsWith('\\]')) ||
-				(textContent.startsWith('$$') && textContent.endsWith('$$')) ||
+				(textContent.startsWith('$$') && textContent.endsWith('$$'));
-				(textContent.startsWith('$') && textContent.endsWith('$') && textContent.indexOf('$') !== textContent.lastIndexOf('$'));
 
 			const props = { key: key };
 			if (v.hasAttributes()) {
@@ -159,13 +157,6 @@ function Article({ cache }) {
 						const firstParen = mathContent.indexOf('\\(');
 						const lastParen = mathContent.lastIndexOf('\\)');
 						mathContent = mathContent.substring(0, firstParen) + '\\[' + mathContent.substring(firstParen + 2, lastParen) + '\\]' + mathContent.substring(lastParen + 2);
-					} else if (trimmed.startsWith('$') && !trimmed.startsWith('$$')) {
-						// Replace $ with $$
-						const firstDollar = mathContent.indexOf('$');
-						const lastDollar = mathContent.lastIndexOf('$');
-						if (firstDollar !== lastDollar) {
-							mathContent = mathContent.substring(0, firstDollar) + '$$' + mathContent.substring(firstDollar + 1, lastDollar) + '$$' + mathContent.substring(lastDollar + 1);
-						}
 					}
 				}
 				return <Tag {...props}><Latex delimiters={latexDelimiters}>{mathContent}</Latex></Tag>;

webclient/src/Results.js

@@ -1,6 +1,7 @@
 import React, { useState, useEffect } from 'react';
-import { Link, useLocation } from 'react-router-dom';
+import { Link, useLocation, useHistory } from 'react-router-dom';
 import { Helmet } from 'react-helmet';
+import queryString from 'query-string';
 import { sourceLink, infoLine, logos } from './utils.js';
 import AbortController from 'abort-controller';
 
@@ -8,6 +9,19 @@ function Results() {
 	const [stories, setStories] = useState(false);
 	const [error, setError] = useState(false);
 	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(() => {
 		const controller = new AbortController();
@@ -32,11 +46,19 @@ function Results() {
 		};
 	}, [location.search]);
 
+	const searchInArticle = queryString.parse(location.search).article === 'true';
+
 	return (
 		<div className='container'>
 			<Helmet>
 				<title>Search Results | QotNews</title>
 			</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>}
 			{stories ?
 				<>
@@ -56,6 +78,10 @@ function Results() {
 									</div>
 
 									{infoLine(x)}
+
+									{!!x?._formatted &&
+										<p>{x._formatted.text.replace(/\n/g, ' ')}</p>
+									}
 								</div>
 							)
 						:

webclient/src/Search.js

@@ -15,7 +15,9 @@ function Search() {
 		const newSearch = event.target.value;
 		setSearch(newSearch);
 		if (newSearch.length >= 3) {
-			const searchQuery = queryString.stringify({ 'q': newSearch });
+			const currentQuery = queryString.parse(location.search);
+			currentQuery.q = newSearch;
+			const searchQuery = queryString.stringify(currentQuery);
 			history.replace('/search?' + searchQuery);
 		} else {
 			history.replace('/');
@@ -24,7 +26,9 @@ function Search() {
 
 	const searchAgain = (event) => {
 		event.preventDefault();
-		const searchString = queryString.stringify({ 'q': event.target[0].value });
+		const currentQuery = queryString.parse(location.search);
+		currentQuery.q = event.target[0].value;
+		const searchString = queryString.stringify(currentQuery);
 		history.push('/search?' + searchString);
 		inputRef.current.blur();
 	}