[parser] add pubmed_id

src/PUBMED/Parser.hs

@@ -36,9 +36,10 @@ parseDocCount :: TL.Text -> Maybe Integer
 parseDocCount txt = TR.readMaybe $ T.unpack $ txt ^. TTL.html . TTL.allNamed (only "eSearchResult") . namedEl "Count" . TTL.contents
 
 data PubMed =
-     PubMed { pubmed_article :: PubMedArticle
-            , pubmed_date    :: PubMedDate
-     } deriving Show
+  PubMed { pubmed_id      :: Int
+         , pubmed_article :: PubMedArticle
+         , pubmed_date    :: PubMedDate
+         } deriving Show
 
 data PubMedArticle =
      PubMedArticle { pubmed_title   :: Maybe T.Text
@@ -71,11 +72,17 @@ parsePubMed :: TL.Text -> [PubMed]
 parsePubMed txt = catMaybes $ txt ^.. pubmedArticle . to pubMed
   where
     pubmedArticle = TTL.html . TTL.allNamed (only "PubmedArticleSet") . TTL.allNamed (only "PubmedArticle")
-    pubMed el = PubMed <$> el ^? article . to pubMedArticle
+    pubMed el = PubMed <$> el ^? articleId . to pubMedId
+                       <*> el ^? article . to pubMedArticle
                        <*> el ^? pubDate . to pubMedDate
       where
-        article = namedEl "MedlineCitation" . namedEl "Article"
+        medline = namedEl "MedlineCitation"
+        article = medline . namedEl "Article"
+        articleId = medline . namedEl "PMID"
         pubDate = namedEl "PubmedData" . namedEl "History" . namedEl "PubMedPubDate" . TTL.attributed (ix "PubStatus" . only "pubmed")
+    pubMedId el = case (TR.readMaybe $ T.unpack $ el ^. TTL.contents) of
+      Nothing -> panic $ "Cannot parse id: " <> (T.unpack $ el ^. TTL.contents)
+      Just id -> id
     pubMedDate el = PubMedDate { pubmedDate_date = jour y m d
                                , pubmedDate_year = y
                                , pubmedDate_month = m