From 2d9732e18bb05f76cc65430f71035af321a756b0 Mon Sep 17 00:00:00 2001 From: Ken Kellner Date: Tue, 24 Apr 2018 16:33:40 -0400 Subject: Some bugfixes for formulas --- article_epub/publishers/uchicago.py | 5 +++++ article_epub/publishers/wiley.py | 17 ++++++++++------- 2 files changed, 15 insertions(+), 7 deletions(-) diff --git a/article_epub/publishers/uchicago.py b/article_epub/publishers/uchicago.py index 9492e75..b097787 100644 --- a/article_epub/publishers/uchicago.py +++ b/article_epub/publishers/uchicago.py @@ -67,6 +67,11 @@ class UChicago(Publisher): link = 'https://www.journals.uchicago.edu'+img['src'] img['src'] = link + for i in body_raw.find_all('span',class_='formula-display'): + img = i.find('img') + link = 'https://www.journals.uchicago.edu'+img['src'] + img['src'] = link + self.body = str(body_raw) def get_references(self): diff --git a/article_epub/publishers/wiley.py b/article_epub/publishers/wiley.py index 9f08c59..ca4a64e 100644 --- a/article_epub/publishers/wiley.py +++ b/article_epub/publishers/wiley.py @@ -13,15 +13,18 @@ class Wiley(Publisher): def check_fulltext(self): full = self.soup.find('section',class_='article-section__full') - if full != None: - if full.find('div',class_='article-section__content') \ - .text == '\n\xa0\n': - sys.exit('Error: Can\'t access fulltext of article') + try: + if full != None: + if full.find('div',class_='article-section__content') \ + .text == '\n\xa0\n': + sys.exit('Error: Can\'t access fulltext of article') + else: + return(True) else: - return(True) - else: + sys.exit('Error: Can\'t access fulltext of article') + except: sys.exit('Error: Can\'t access fulltext of article') - + def get_doi(self): if self.doi == None: doi_raw = self.soup.find('a',class_='epub-doi').text.split('/') -- cgit v1.2.3