def update_article_description(article_id):
article = db.get_article(article_id)
if article == None:
print("There is no article with that ID. article NOT found.\n")
else:
print()
display_single_article(article, str(article.id))
article_choice = btc.read_int_ranged('1 to edit article description, 2 to leave as is: ' ,
min_value = 1, max_value = 2)
if article_choice == 1:
description_choice = btc.read_text('View article description? y/n: ')
if description_choice == 'y':
article_summary = na.get_article_summary(article.link)
print(article_summary)
# article_text = Article.get_text(article.link)
# #article_text = dm.get_cleaned_text(link)
# article_text = article_text.split()
# article_text = [i for i in article_text if de.isEnglish(i) == True]
# article_text = ' '.join(article_text)
# print(article_text)
new_description = btc.read_text('Enter new description or "." to cancel: ')
if new_description != '.':
db.update_article_description(article_id, new_description)
print('Article description updated.\n')
else:
print('Edit cancelled, article description unchanged')
else:
print('Edit cancelled, article description unchanged')
def scrape_article_name(article_id):
article = db.get_article(article_id)
if article == None:
print('There is no article with that ID. article NOT found.\n')
else:
print()
display_single_article(article, str(article.id))
article_choice = btc.read_int_ranged('1 to rescrape title, 2 to leave as is: ',
min_value = 1, max_value = 2)
if article_choice == 1:
try:
new_article_news_item = na.get_article_from_url(article.link)
new_title = new_article_news_item.title
#new_title = na.get_article_title(article.link)
#new_title = Article.get_title(article.link)
print('''
New title: {0}
Old title: {1}'''.format(new_title, article.name))
except:
new_title = 'Title scrape failed'
title_choice = btc.read_int_ranged('1 to replace title, 2 to keep original title: ',
min_value = 1, max_value = 2)
if title_choice == 1:
db.update_article_name(article_id, new_title)
elif title_choice == 2:
print('article update cancelled')
elif article_choice == 2:
print('article update cancelled')
def update_article_name(article_id):
article = db.get_article(article_id)
if article == None:
print("There is no article with that ID. article NOT found.\n")
else:
print()
display_single_article(article, str(article.id))
article_choice = btc.read_int_ranged('1 to edit article title, 2 to leave as is: ' ,
min_value = 1, max_value = 2)
if article_choice == 1:
try:
newsItem1 = na.get_article_from_url(article.link)
updated_title = newsItem1.title
except Exception as e:
print('Scrape failed because of {0}'.format(e))
updated_title = 'Invalid'
print('Rescraped title: {0}'.format(updated_title))
title_choice = btc.read_int_ranged('1 - existing title, 2 - scraped title, 3 - manual input: ', 1, 3)
if title_choice == 1:
print('Title update cancelled, article title unchanged.')
return
elif title_choice == 2:
db.update_article_name(article_id, updated_title)
print('Title update complete. Return to main menu.')
elif title_choice == 3:
new_title = btc.read_text('Enter new title or . to cancel: ')
if new_title != '.':
db.update_article_name(article_id, new_title)
else:
print('Edit cancelled, return to main menu')
return
else:
print('Edit cancelled, article title unchanged')
def export_roundup():
roundup_title = btc.read_text('Enter the roundup title or "." to cancel: ')
filename = btc.read_text('Enter the filename or "." to cancel: ')
if roundup_title != '.':
roundup_categories = db.get_categories()
for category in roundup_categories:
category.articles = db.get_articles_by_category_id(category.id)
roundup_docx.create_complete_roundup(filename=filename, roundup_title=roundup_title, categories=roundup_categories)
def export_roundup_by_year():
roundup_title = btc.read_text('Enter the roundup title: ')
roundup_year = btc.read_int_ranged('Enter roundup year: ', 1, 2100)
filename = btc.read_text('Enter roundup filename: ')
roundup_choice = btc.read_int_ranged('Enter 1 to export roundup, 2 to cancel: ', 1, 2)
if roundup_choice == 1:
roundup_categories = db.get_categories()
for category in roundup_categories:
category.articles = db.yearly_roundup_articles(roundup_year, category.id)
roundup_docx.create_complete_roundup(filename=filename, roundup_title=roundup_title, categories=roundup_categories)
#display_title()
elif roundup_choice == 2:
print('Roundup export cancelled. Return to main menu.\n')
def update_category(category_id=0):
if category_id == 0:
category_id = int(input("category ID: "))
category = db.get_category(category_id)
articles = db.get_articles_by_category_id(category_id)
display_articles(articles, category.name.upper())
new_category_name = btc.read_text("Enter new category name or '.' to cancel: ")
if new_category_name != '.':
update_choice = btc.read_int_ranged("1 to change article name to {0}, 2 to cancel: ".format(new_category_name),
1, 2)
if update_choice == 1:
db.update_category(category_id, new_category_name)
print('Category update complete\n')
elif update_choice == 2:
print('Update cancelled.\n')
def update_article_author(article_id):
article = db.get_article(article_id)
if article == None:
print("There is no article with that ID. article NOT found.\n")
else:
print()
display_single_article(article, str(article.id))
article_choice = btc.read_int_ranged('1 to edit article author, 2 to leave as is: ' ,
min_value = 1, max_value = 2)
if article_choice == 1:
new_author = btc.read_text('Enter new author name or . to cancel: ')
if new_author != '.':
db.update_article_author(article_id, new_author)
else:
print('Edit cancelled, article title unchanged')
def get_stats(command):
del command
stats_choice = btc.read_int_ranged('1 - monthly stats; 2 - yearly stats; 3 - main menu: ',
1, 3)
if stats_choice in range(1, 3):
year = btc.read_int_ranged('Enter article year: ', 1, 2100)
if stats_choice in range(1, 2):
month = btc.read_int_ranged('Enter article month: ', 1, 12)
get_monthly_category_stats(month, year)
else:
get_yearly_category_stats(year)
else:
time.sleep(0.25)
print('Returning to main menu.\n')
time.sleep(0.25)
def get_csv_in_directory():
file_list = glob.glob('*.csv')
print('Files available: ')
for item in file_list:
print(item, end='\n')
#try:
importing_file_name = btc.read_text('Enter filename from the list or ". " to cancel: ')
if importing_file_name != '.':
filename='{0}.csv'.format(importing_file_name)
csv_articles = create_csv_list(filename)
print(csv_articles)
#csv_articles = [csv_item_to_article(csv_article) for csv_article in csv_articles]
print('Articles to import:')
try:
for article in csv_articles:
try:
csv_article = csv_item_to_article(article)
try:
csv_article.name = Article.get_title(csv_article.link)
except Exception as e:
print(e)
csv_article.name = 'Not specified'
db.add_article_from_csv(csv_article)
print(csv_article.name + " was added to database.\n")
#print('Import complete, return to main menu \n')
except Exception as e:
print(e)
print('Article import failed.')
continue
print('Import complete, return to main menu')
except Exception as e:
print(e)
def finalize_desc_month(command):
if not command or command == '':
new_month = btc.read_int_ranged('Enter new month: ', min_value = 1, max_value = 12)
new_year = btc.read_int_ranged('Enter new year: ', min_value = 1, max_value = 2100)
articles_to_finalize = db.get_articles_by_month(month=new_month, year=new_year)
articles_remaining = len(articles_to_finalize)
for article in articles_to_finalize:
print('{0} unreviewed articles'.format(articles_remaining))
update_article_description(article.id)
description_choice = btc.read_int_ranged('{0} descriptions remaining. Press 1 to continue, 2 to cancel: '.format(articles_remaining),
1, 2)
articles_remaining -= 1
if description_choice == 2:
print('Update descriptions cancelled')
break
def display_articles_by_publication():
publication_snippet = btc.read_text("Publication: ")
publications = db.display_articles_by_publication(publication_snippet)
if publications == None:
print("There are no articles by that publication. article NOT found.\n")
else:
print()
display_articles(publications, "PUBLICATION: " + str(publications[0].publication))
def display_articles_by_author():
author_snippet = btc.read_text("Author Name: ")
articles = db.display_articles_by_author(author_snippet)
if articles == None:
print("There are no articles by that author. article NOT found.\n")
else:
print()
#display_single_article(article, str(article.id))
display_articles(articles, "AUTHOR: " + str(articles[0].author))
def update_article_category(article_id):
article = db.get_article(article_id)
if article == None:
print("There is no article with that ID. article NOT found.\n")
else:
print()
display_single_article(article, str(article.id))
article_choice = btc.read_int_ranged('1 to edit article category, 2 to leave as is: ' ,
min_value = 1, max_value = 2)
if article_choice == 1:
new_category_id = btc.read_int('Enter new category_id: ')
result = db.get_category(new_category_id)
if result == None:
print('There is no category with that ID, article category NOT updated.\n')
else:
db.update_article_category(article_id, new_category_id)
else:
print('Edit cancelled, article title unchanged')
def display_articles_by_name():
title_snippet = btc.read_text('Enter article title or "." to cancel: ')
if title_snippet != '.':
result = db.display_article_by_name(title_snippet)
if result == None:
print('There is no article with that name.\n')
else:
display_single_article(result, str(result.id))
else:
print('Search cancelled, returning to main menu.')
def finalize_article_descriptions(month, year=2019):
undescribed = db.finalize_descriptions(month, year)
undescribed_articles = len(undescribed)
print('{0} undescribed articles'.format(undescribed_articles))
for article in undescribed:
print('{0} undescribed articles'.format(undescribed_articles))
update_article_description(article.id)
description_choice = btc.read_int_ranged('{0} descriptions remaining. Press 1 to continue, 2 to cancel: '.format(undescribed_articles), 1, 2)
if description_choice == 2:
print('Update descriptions cancelled')
break
def get_articles_by_category():
category = btc.read_text("Enter category name or number here: ")
if category.isalpha() == True:
display_articles_by_category_name(category)
elif category.isnumeric() == True:
try:
category_id = int(category)
display_articles_by_category_id(category_id)
except:
print('Article search cancelled. Return to main menu.\n')
def delete_category():
category_id = int(input("category ID: "))
articles_in_category = db.get_articles_by_category_id(category_id)
if len(articles_in_category) > 0:
print('Category contains articles, cannot be deleted')
elif len(articles_in_category) == 0:
delete_choice = btc.read_float_ranged('Press 1 to delete, 2 to cancel: ', 1, 2)
if delete_choice == 1:
db.delete_category(category_id)
print('Category deleted.\n')
elif delete_choice == 2:
print('Delete cancelled, returning to category menu')
def update_article_date(article_id):
article = db.get_article(article_id)
if article == None:
print("There is no article with that ID. article NOT found.\n")
else:
print()
display_single_article(article, str(article.id))
article_choice = btc.read_int_ranged('1 to edit article date, 2 to leave as is: ' ,
min_value = 1, max_value = 2)
if article_choice == 1:
new_day = btc.read_int_ranged('Enter new day: ', min_value = 1, max_value = 31)
new_month = btc.read_int_ranged('Enter new month: ', min_value = 1, max_value = 12)
new_year = btc.read_int_ranged('Enter new year: ', min_value = 1, max_value = 2100)
date_choice = btc.read_int_ranged('1 to change date to: {0}/{1}/{2}, 2 to cancel: '.format(new_month, new_day, new_year),
min_value=1, max_value=2)
if date_choice == 1:
db.update_article_date(article_id, new_day, new_month, new_year)
print('Update complete.\n')
elif date_choice == 2:
print('Edit cancelled, article date unchanged')
else:
print('Edit cancelled, article date unchanged')
def date_search(search_choice):
#search_choice = btc.read_int_ranged('Options:\n1 - day\n2 - month\n3 - year\n4 - cancel\nEnter your choice: ', 1, 4)
if search_choice in range(1, 4): #user searches by year
year = btc.read_int('Year: ')
if search_choice in range(1, 3):
month = btc.read_int('Month: ')
if search_choice in range(1, 2):
day = btc.read_int('Day: ')
try:
assert datetime.date(day=day, month=month, year = year)
display_articles_by_date(year, month, day)
except ValueError:
print('Invalid date. Return to main menu.')
return
#if Article.validate_date(day=day, month=month, year=year) == True:
#display_articles_by_date(year, month, day)
#else:
# print('Invalid date selected')
else:
display_articles_by_month(month, year)
else:
display_articles_by_year(year)
else:
print('Search cancelled. ')
def finalize_title_updates(month, year):
articles = db.get_articles_by_month(month=month, year=year)
articles_remaining = len(articles)
for article in articles:
print('{0} articles remaining'.format(articles_remaining))
display_single_article(article, title_term = article.id)
strip_choice = btc.read_int_ranged('1 to update title, 2 to skip, 3 to return to main menu: ', 1, 3)
if strip_choice == 1:
update_article_name(article.id)
articles_remaining -= 1
if strip_choice == 2:
articles_remaining -= 1
print('Article title unchanged.')
if strip_choice == 3:
print('strip titles cancelled')
#display_title()
break
def manual_add(link=None):
if (link == None) or (not link):
print('Link', link)
print('No link supplied, manual_add must be followed by link.')
return
else:
print('Manual article creation/n')
print('Link: {0}'.format(link))
display_categories()
#print(link)
#if link == None:
# print('No link supplied, manual_add must be followed by link.')
new_article_category = btc.read_int('Enter category for article: ')
category = db.get_category(new_article_category)
assert category != None
new_article = Article.from_input(link=link, category=category)
if new_article == None:
print('Article creation cancelled. Returning to main menu.')
return
db.add_article(new_article)
print(new_article.title + " was added to database.\n")
def export_roundup_by_category():
display_categories()
roundup_categories = db.get_categories()
categories_remaining = len(roundup_categories)
categories_for_roundup = []
for category in roundup_categories:
print('Categories remaining: {0}'.format(categories_remaining))
print('Include {0}'.format(category.name))
category_choice = btc.read_int_ranged('1 to include, 2 to exclude: ', 1, 2)
if category_choice != 1:
categories_for_roundup.append(category)
roundup_title = btc.read_text('Enter the roundup title: ')
roundup_month = btc.read_int_ranged('Enter roundup month: ', 1, 12)
roundup_year = btc.read_int_ranged('Enter roundup year: ', 1, 2100)
filename = btc.read_text('Enter roundup filename: ')
roundup_choice = btc.read_int_ranged('Enter 1 to export roundup, 2 to cancel: ', 1, 2)
if roundup_choice == 1:
for category in categories_for_roundup:
# for category in roundup_categories:
category.articles = db.get_articles_for_roundup(roundup_month, roundup_year, category.id)
roundup_docx.create_complete_roundup(filename=filename, roundup_title=roundup_title, categories=categories_for_roundup)
#display_title()
elif roundup_choice == 2:
print('Roundup export cancelled. Return to main menu.\n')
def add_article_from_newspaper(link):
'''
Adds an article from the newspaper module after downloading it
'''
try:
#link = btc.read_text('Link or "." to cancel: ')
#try:
for i in tqdm.tqdm(range(1)):
newNewsItem = na.get_article_from_url(link)
print(newNewsItem)
#except Exception as e:
#print(e)
#get article title
try:
name = newNewsItem.title #get the title for the article
print('NameTest {0}'.format(name))
except Exception as e:
print(e)
name = btc.read_text('Please enter title: ')
#get article author
try:
author = ' '.join(newNewsItem.authors)
#get article publication
except Exception as e:
print(e)
author = btc.read_text('Please enter author: ')
try:
#works for most websites, but not Sudan Tribune
publication = newNewsItem.meta_data['og']['site_name']
except Exception as e:
print(e)
publication = btc.read_text('Please enter publication: ')
try:
year = newNewsItem.publish_date.year
except Exception as e:
print(e)
year = btc.read_int_ranged('Please enter year: ', 1, 2200)
try:
month = newNewsItem.publish_date.month
except Exception as e:
print(e)
month = btc.read_int_ranged('Please enter month: ', 1, 12)
try:
day = newNewsItem.publish_date.day
except Exception as e:
print(e)
day = btc.read_int_ranged('Please enter day: ', 1, 31)
try:
summary = newNewsItem.summary
except Exception as e:
print(e)
print('Summary download failed')
summary = 'Summary not found'
try:
keywords = ', '.join(newNewsItem.keywords)
except Exception as e:
print(e)
print('Keyword download failed')
keywords= 'keywords not found'
print('TITLE - {0} - AUTHOR {1}'.format(name, author))
print('DATE - {0}/{1}/{2} - PUBLICATION {3}'.format(month, day, year, publication))
#print(author)
#print(publication)
#print('{0}/{1}/{2}'.format(month, day, year))
#print(summary)
print('KEYWORDS: ', keywords)
display_categories()
category_id = btc.read_text("Category ID: ")
category = db.get_category(category_id)
if category == None:
print('There is no category with that ID. article NOT added.\n')
return
description_choice = btc.read_text('View article description? y/n: ')
if description_choice == 'y':
print('Title: {0}'.format(name))
print('Summary: {0}'.format(summary))
print('Keywords: {0}'.format(keywords))
description = btc.read_text("Description or '.' to cancel: ")
if description == ".":
return
else:
article = Article(name=name, year=year, month=month,day=day,
category=category, link=link, description=description,
author=author, publication=publication)
db.add_article(article)
print(name + " was added to database.\n")
except Exception as e:
print('Article download failed.', e)