Update app.py
Browse files
app.py
CHANGED
|
@@ -61,6 +61,7 @@ def load_file_id_mapping():
|
|
| 61 |
file_id_mapping = load_file_id_mapping()
|
| 62 |
|
| 63 |
selected_websites = {}
|
|
|
|
| 64 |
|
| 65 |
for website, details in config.items():
|
| 66 |
if st.checkbox(website, key=website):
|
|
@@ -70,8 +71,11 @@ for website, details in config.items():
|
|
| 70 |
language = st.selectbox(f'Choose language for {website}', list(languages.keys()), key=f'lang_{website}')
|
| 71 |
selected_websites[website] = f"{website}_{language}" # like: hespress_en
|
| 72 |
else:
|
|
|
|
| 73 |
selected_websites[website] = website # like: akhbarona
|
| 74 |
|
|
|
|
|
|
|
| 75 |
# Category selection
|
| 76 |
categories = languages.get(language, {})
|
| 77 |
if categories:
|
|
@@ -90,7 +94,11 @@ if st.button('Start Scraping'):
|
|
| 90 |
for website, module_name in selected_websites.items():
|
| 91 |
scraper_module = importlib.import_module(module_name)
|
| 92 |
for category in selected_categories.get(website, []):
|
| 93 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 94 |
if 'category_name' in config[website]:
|
| 95 |
category_name = config[website]['category_name'].get(category, 'default_category_name')
|
| 96 |
file_path = scraper_module.scrape_category(category_url, num_articles)
|
|
|
|
| 61 |
file_id_mapping = load_file_id_mapping()
|
| 62 |
|
| 63 |
selected_websites = {}
|
| 64 |
+
selected_languages = {}
|
| 65 |
|
| 66 |
for website, details in config.items():
|
| 67 |
if st.checkbox(website, key=website):
|
|
|
|
| 71 |
language = st.selectbox(f'Choose language for {website}', list(languages.keys()), key=f'lang_{website}')
|
| 72 |
selected_websites[website] = f"{website}_{language}" # like: hespress_en
|
| 73 |
else:
|
| 74 |
+
language = next(iter(languages.keys()))
|
| 75 |
selected_websites[website] = website # like: akhbarona
|
| 76 |
|
| 77 |
+
selected_languages[website] = language
|
| 78 |
+
|
| 79 |
# Category selection
|
| 80 |
categories = languages.get(language, {})
|
| 81 |
if categories:
|
|
|
|
| 94 |
for website, module_name in selected_websites.items():
|
| 95 |
scraper_module = importlib.import_module(module_name)
|
| 96 |
for category in selected_categories.get(website, []):
|
| 97 |
+
try:
|
| 98 |
+
language = selected_languages[website]
|
| 99 |
+
category_url = config[website]['languages'][language][category]
|
| 100 |
+
except KeyError:
|
| 101 |
+
st.error(f"KeyError: {category} not found for {language} language in {website}.")
|
| 102 |
if 'category_name' in config[website]:
|
| 103 |
category_name = config[website]['category_name'].get(category, 'default_category_name')
|
| 104 |
file_path = scraper_module.scrape_category(category_url, num_articles)
|