Pubmed scraper: use JournalIssue to get publication date

gargantext/scrapers/pubmed.py

@@ -10,7 +10,10 @@ __all__ = ['PubmedScraper']
 
 class PubmedDocumentLoader(DocumentLoader):
     def parse(self, article):
+        # https://dtd.nlm.nih.gov/ncbi/pubmed/out/doc/2018/el-PubMedPubDate.html
         LD = '*[self::Year or self::Month or self::Day or self::Hour or self::Minute or self::Second]'
+        # https://dtd.nlm.nih.gov/ncbi/pubmed/out/doc/2018/el-PubDate.html
+        PD = '*[self::MedlineDate or self::Year or self::Month or self::Day or self::Season]'
 
         pubmed_id = article.xpath('PubmedData/ArticleIdList/ArticleId[@IdType="pubmed"]/text()').extract_first()
 
@@ -32,8 +35,8 @@ class PubmedDocumentLoader(DocumentLoader):
             'abstract': 'MedlineCitation/Article/Abstract/AbstractText',
             'source': 'MedlineCitation/Article/Journal/Title',
             'lang': 'MedlineCitation/Article/Language',
-            # https://www.nlm.nih.gov/bsd/licensee/elements_descriptions.html#pubdate
-            'publication': 'PubmedData/History/PubMedPubDate[@PubStatus="accepted"]/' + LD,
+            # https://www.nlm.nih.gov/bsd/licensee/elements_descriptions.html#journalissue
+            'publication': 'MedlineCitation/Article/Journal/JournalIssue/PubDate/' + PD,
             # https://www.nlm.nih.gov/bsd/licensee/elements_descriptions.html#datecompleted
             'creation': 'PubmedData/History/PubMedPubDate[@PubStatus="pubmed"]/' + LD,
         })