Block by result title or url using regex (#473)

Allows blocking search results using a regex filter for either
result title or result url

app/filter.py

@@ -45,12 +45,15 @@ class Filter:
     def __init__(self, user_key: str, mobile=False, config=None) -> None:
         if config is None:
             config = {}
-
         self.near = config['near'] if 'near' in config else ''
         self.dark = config['dark'] if 'dark' in config else False
         self.nojs = config['nojs'] if 'nojs' in config else False
         self.new_tab = config['new_tab'] if 'new_tab' in config else False
         self.alt_redirect = config['alts'] if 'alts' in config else False
+        self.block_title = (
+            config['block_title'] if 'block_title' in config else '')
+        self.block_url = (
+            config['block_url'] if 'block_url' in config else '')
         self.mobile = mobile
         self.user_key = user_key
         self.main_divs = ResultSet('')
@@ -87,6 +90,8 @@ class Filter:
     def clean(self, soup) -> BeautifulSoup:
         self.main_divs = soup.find('div', {'id': 'main'})
         self.remove_ads()
+        self.remove_block_titles()
+        self.remove_block_url()
         self.collapse_sections()
         self.update_styling(soup)
 
@@ -134,6 +139,28 @@ class Filter:
                        if has_ad_content(_.text)]
             _ = div.decompose() if len(div_ads) else None
 
+    def remove_block_titles(self) -> None:
+        if not self.main_divs:
+            return
+        if self.block_title == '':
+            return
+        block_title = re.compile(self.block_title)
+        for div in [_ for _ in self.main_divs.find_all('div', recursive=True)]:
+            block_divs = [_ for _ in div.find_all('h3', recursive=True)
+                          if block_title.search(_.text) is not None]
+            _ = div.decompose() if len(block_divs) else None
+
+    def remove_block_url(self) -> None:
+        if not self.main_divs:
+            return
+        if self.block_url == '':
+            return
+        block_url = re.compile(self.block_url)
+        for div in [_ for _ in self.main_divs.find_all('div', recursive=True)]:
+            block_divs = [_ for _ in div.find_all('a', recursive=True)
+                          if block_url.search(_.attrs['href']) is not None]
+            _ = div.decompose() if len(block_divs) else None
+
     def collapse_sections(self) -> None:
         """Collapses long result sections ("people also asked", "related
          searches", etc) into "details" elements

app/models/config.py

@@ -1,6 +1,7 @@
 from app.utils.misc import read_config_bool
 from flask import current_app
 import os
+import re
 
 
 class Config:
@@ -14,6 +15,8 @@ class Config:
             open(os.path.join(app_config['STATIC_FOLDER'],
                               'css/variables.css')).read())
         self.block = os.getenv('WHOOGLE_CONFIG_BLOCK', '')
+        self.block_title = os.getenv('WHOOGLE_CONFIG_BLOCK_TITLE', '')
+        self.block_url = os.getenv('WHOOGLE_CONFIG_BLOCK_URL', '')
         self.ctry = os.getenv('WHOOGLE_CONFIG_COUNTRY', '')
         self.theme = os.getenv('WHOOGLE_CONFIG_THEME', '')
         self.safe = read_config_bool('WHOOGLE_CONFIG_SAFE')

app/static/settings/translations.json

@@ -10,6 +10,10 @@
         "config-near-help": "City Name",
         "config-block": "Block",
         "config-block-help": "Comma-separated site list",
+        "config-block-title": "Block by Title",
+        "config-block-title-help": "Use regex",
+        "config-block-url": "Block by URL",
+        "config-block-url-help": "Use regex",
         "config-theme": "Theme",
         "config-nojs": "Show NoJS Links",
         "config-dark": "Dark Mode",
@@ -43,6 +47,10 @@
         "config-near-help": "Stad",
         "config-block": "Blok",
         "config-block-help": "Lijst met sites met kommas onderscheiden",
+        "config-block-title": "Blokkeren op titel",
+        "config-block-title-help": "Gebruik regex",
+        "config-block-url": "Blokkeren op URL",
+        "config-block-url-help": "Gebruik regex",
         "config-theme": "Thema",
         "config-nojs": "Laat NoJS links zien",
         "config-dark": "Donkere Modus",
@@ -76,6 +84,10 @@
         "config-near-help": "Stadt-Name",
         "config-block": "Block",
         "config-block-help": "Komma-getrennte Liste von Seiten",
+        "config-block-title": "Nach Titel blockieren",
+        "config-block-title-help": "Regex verwenden",
+        "config-block-url": "Nach URL blockieren",
+        "config-block-url-help": "Regex verwenden",
         "config-theme": "Thema",
         "config-nojs": "NoJS-Links anzeigen",
         "config-dark": "Dark Mode",
@@ -109,6 +121,10 @@
         "config-near-help": "Nombre de la Ciudad",
         "config-block": "Bloquear",
         "config-block-help": "Lista de sitios separados por comas",
+        "config-block-title": "Bloquear por título",
+        "config-block-title-help": "Usar expresiones regulares",
+        "config-block-url": "Bloquear por URL",
+        "config-block-url-help": "Usar expresiones regulares",
         "config-theme": "Tema",
         "config-nojs": "Mostrar Enlaces NoJS",
         "config-dark": "Modo Oscuro",
@@ -142,6 +158,10 @@
         "config-near-help": "Nome della città",
         "config-block": "Blocca",
         "config-block-help": "Lista di siti separati da virgole",
+        "config-block-title": "Blocca per titolo",
+        "config-block-title-help": "Usa regex",
+        "config-block-url": "Blocca per url",
+        "config-block-url-help": "Usa regex",
         "config-theme": "Tema",
         "config-nojs": "Mostra link NoJS",
         "config-dark": "Modalità Notte",
@@ -164,7 +184,7 @@
         "dark": "notte",
         "system": "impostazioni di sistema"
     },
-   "lang_pt": {
+    "lang_pt": {
         "search": "Pesquisar",
         "config": "Configuração",
         "config-country": "Filtrar Resultados por País",
@@ -175,6 +195,10 @@
         "config-near-help": "Nome da Cidade",
         "config-block": "Bloquear",
         "config-block-help": "Lista de sites separados por vírgulas",
+        "config-block-title": "Bloco por título",
+        "config-block-title-help": "Use regex",
+        "config-block-url": "Bloquear por url",
+        "config-block-url-help": "Use regex",
         "config-theme": "Tema",
         "config-nojs": "Mostrar Links NoJS",
         "config-dark": "Modo Escuro",
@@ -208,6 +232,10 @@
         "config-near-help": "城市名",
         "config-block": "屏蔽",
         "config-block-help": "逗号分隔的网站列表",
+        "config-block-title": "按网站标题屏蔽",
+        "config-block-title-help": "使用正则表达式",
+        "config-block-url": "按网站链接屏蔽",
+        "config-block-url-help": "使用正则表达式",
         "config-theme": "主题",
         "config-nojs": "显示 NoJS 链接",
         "config-dark": "深色模式",
@@ -241,6 +269,10 @@
         "config-near-help": "නගරයේ නම",
         "config-block": "අවහිර",
         "config-block-help": "අල්ප විරාම වලින් වෙන් වූ අඩවි ලැයිස්තුව",
+        "config-block-title": "මාතෘකාව අනුව අවහිර කරන්න",
+        "config-block-title-help": "රෙජෙක්ස් භාවිතා කරන්න",
+        "config-block-url": "යූආර්එල් මඟින් අවහිර කරන්න",
+        "config-block-url-help": "රෙජෙක්ස් භාවිතා කරන්න",
         "config-theme": "තේමාව",
         "config-nojs": "නෝජේඑස් සබැඳි පෙන්වන්න",
         "config-dark": "අඳුරු ආකාරය",
@@ -274,6 +306,10 @@
         "config-near-help": "Nom de ville",
         "config-block": "Bloquer",
         "config-block-help": "Liste de sites séparés pas des virgules",
+        "config-block-title": "Bloquer par titre",
+        "config-block-title-help": "Utiliser l'expression régulière",
+        "config-block-url": "Bloquer par URL",
+        "config-block-url-help": "Utiliser l'expression régulière",
         "config-theme": "Theme",
         "config-nojs": "Montrer les liens NoJS",
         "config-dark": "Mode Sombre",
@@ -307,6 +343,10 @@
         "config-near-help": "نام شهر",
         "config-block": "مسدود کردن",
         "config-block-help": "لیست سایت‌ها با ویرگول جدا می‌شود.",
+        "config-block-title": "مسدود کردن بر اساس عنوان",
+        "config-block-title-help": "از عبارت منظم استفاده کنید",
+        "config-block-url": "بلوک بر اساس URL",
+        "config-block-url-help": "از عبارت منظم استفاده کنید",
         "config-theme": "پوسته",
         "config-nojs": "نمایش پیوند‌های بدون جاوا اسکیریپت",
         "config-dark": "حالت تاریک",
@@ -340,6 +380,10 @@
         "config-near-help": "Název města",
         "config-block": "Blokovat",
         "config-block-help": "Čárkami oddělený seznam stránek",
+        "config-block-title": "Blokovat podle názvu",
+        "config-block-title-help": "Použijte regulární výraz",
+        "config-block-url": "Blokovat podle adresy URL",
+        "config-block-url-help": "Použijte regulární výraz",
         "config-theme": "Motiv",
         "config-nojs": "Zobrazit NoJS odkazy",
         "config-dark": "Tmavý motiv",
@@ -373,6 +417,10 @@
         "config-near-help": "城市名",
         "config-block": "排除",
         "config-block-help": "網址列表，以逗號分隔",
+        "config-block-title": "按標題屏蔽",
+        "config-block-title-help": "使用正則表達式",
+        "config-block-url": "按網址屏蔽",
+        "config-block-url-help": "使用正則表達式",
         "config-theme": "主題",
         "config-nojs": "顯示 NoJS 連結",
         "config-dark": "深色模式",
@@ -406,6 +454,10 @@
         "config-near-help": "Име на град",
         "config-block": "Блокирани сайтове",
         "config-block-help": "Списък сайтове, разделени със запетая",
+        "config-block-title": "Блокиране по заглавие",
+        "config-block-title-help": "Използвайте регулярно изражение",
+        "config-block-url": "Блокиране по url",
+        "config-block-url-help": "Използвайте регулярно изражение",
         "config-theme": "Стил",
         "config-nojs": "Показване на връзки без JS",
         "config-dark": "Тъмен режим",
@@ -439,6 +491,10 @@
         "config-near-help": "शहर का नाम",
         "config-block": "खंड",
         "config-block-help": "अल्पविराम से अलग की गई साइट सूची",
+        "config-block-title": "शीर्षक के अनुसार ब्लॉक करें",
+        "config-block-title-help": "रेगेक्स का प्रयोग करें",
+        "config-block-url": "url द्वारा अवरोधित करें",
+        "config-block-url-help": "रेगेक्स का प्रयोग करें",
         "config-theme": "विषय",
         "config-nojs": "NoJS लिंक दिखाएं",
         "config-dark": "डार्क मोड",
@@ -460,5 +516,5 @@
         "light": "रोशनी",
         "dark": "अंधेरा",
         "system": "प्रणाली"
-   }
+    }
 }

app/templates/index.html

@@ -122,6 +122,14 @@
 					<label for="config-block">{{ translation['config-block'] }}: </label>
 					<input type="text" name="block" id="config-block" placeholder="{{ translation['config-block-help'] }}" value="{{ config.block }}">
 				</div>
+                <div class="config-div config-div-block">
+					<label for="config-block-title">{{ translation['config-block-title'] }}: </label>
+					<input type="text" name="block_title" id="config-block" placeholder="{{ translation['config-block-title-help'] }}" value="{{ config.block_title }}">
+				</div>
+				<div class="config-div config-div-block">
+					<label for="config-block-url">{{ translation['config-block-url'] }}: </label>
+					<input type="text" name="block_url" id="config-block" placeholder="{{ translation['config-block-url-help'] }}" value="{{ config.block_url }}">
+				</div>
 				<div class="config-div config-div-nojs">
 					<label for="config-nojs">{{ translation['config-nojs'] }}: </label>
 					<input type="checkbox" name="nojs" id="config-nojs" {{ 'checked' if config.nojs else '' }}>