Add Oxford support. Still have issues with bold font and lit cited looking bad

4 files changed, 53 insertions, 5 deletions

diff --git a/article_epub/publisher.py b/article_epub/publisher.py
index 7af9a47..7465f7f 100644
--- a/article_epub/publisher.py
+++ b/article_epub/publisher.py
@@ -72,13 +72,17 @@ class Publisher(object):
         self.journal = self.meta['container-title']
 
         if 'published-print' in self.meta.keys():
-            self.volume = str(self.meta['volume'])
-            self.pages = str(self.meta['page'])
             self.year = str(self.meta['published-print']['date-parts'][0][0])
         else:
+            self.year = str(self.meta['published-online']['date-parts'][0][0])
+        try:
+            self.volume = str(self.meta['volume'])
+        except:
             self.volume = ''
+        try:
+            self.pages = str(self.meta['page'])
+        except:
             self.pages = ''
-            self.year = str(self.meta['published-online']['date-parts'][0][0])
 
     def get_citation(self):
         
diff --git a/article_epub/publishers/__init__.py b/article_epub/publishers/__init__.py
index 3e975c4..6aa2f43 100644
--- a/article_epub/publishers/__init__.py
+++ b/article_epub/publishers/__init__.py
@@ -3,3 +3,4 @@ from article_epub.publishers.springer import Springer
 from article_epub.publishers.wiley import Wiley
 from article_epub.publishers.bioone import BioOne
 from article_epub.publishers.plosone import PLoSONE
+from article_epub.publishers.oxford import Oxford
diff --git a/article_epub/publishers/oxford.py b/article_epub/publishers/oxford.py
new file mode 100644
index 0000000..f8a95a1
--- /dev/null
+++ b/article_epub/publishers/oxford.py
@@ -0,0 +1,45 @@
+from article_epub.publisher import Publisher, register_publisher
+import copy
+
+class Oxford(Publisher):
+    """Class for Oxford articles"""
+
+    domains = ["oup.com"]
+
+    def get_doi(self):
+        if self.doi == None:
+            doi_raw = self.soup.find('div',class_='ww-citation-primary') \
+                .find('a')['href'].split('/')
+            self.doi = str(doi_raw[3]+'/'+doi_raw[4])
+
+    def get_abstract(self):
+        """Get article abstract"""
+        abstract_raw = self.soup.find('section',class_='abstract')
+        self.abstract = '<h2>Abstract<h2>\n'+str(abstract_raw)
+
+    def get_keywords(self):
+        """Get article keywords"""
+        keywords_raw = self.soup.find('div',class_='kwd-group').find_all('a')
+        self.keywords = []
+        for i in keywords_raw:
+            self.keywords.append(i.text)
+
+    def get_body(self):
+        """Get body of article"""
+        body_raw = copy.copy(self.soup.find(
+            'div',{'data-widgetname':'ArticleFulltext'}))
+        body_raw.find('h2',class_='abstract-title').decompose()
+        body_raw.find('div',class_='article-metadata-panel').decompose()
+        body_raw.find('div',class_='ref-list').decompose()
+        body_raw.find('span',{'id':'UserHasAccess'}).decompose()
+        body_raw.find('div',class_='copyright').decompose()
+        body_raw.find('h2',class_='backreferences-title').decompose()
+        self.body = body_raw
+    
+    def get_references(self):
+        """Get references list"""
+        references_title = self.soup.find('h2',class_='backreferences-title')
+        references_raw = self.soup.find('div',class_='ref-list')
+        self.references = str(references_title)+str(references_raw)
+
+register_publisher(Oxford)
diff --git a/article_epub/publishers/plosone.py b/article_epub/publishers/plosone.py
index 578826e..f096641 100644
--- a/article_epub/publishers/plosone.py
+++ b/article_epub/publishers/plosone.py
@@ -1,6 +1,4 @@
 from article_epub.publisher import Publisher, register_publisher
-import requests
-from bs4 import BeautifulSoup
 
 class PLoSONE(Publisher):
     """Class for PLoS ONE articles"""