3 files changed, 83 insertions, 4 deletions
diff --git a/uploader/publications/misc.py b/uploader/publications/misc.py
index d93ecdd..fca6f71 100644
--- a/uploader/publications/misc.py
+++ b/uploader/publications/misc.py
@@ -4,7 +4,22 @@
 def publications_differences(
         filedata: tuple[dict, ...],
         dbdata: tuple[dict, ...],
-        pubmedid2pubidmap: dict[str, int]
+        pubmedid2pubidmap: tuple[dict, ...]
 ) -> tuple[dict, ...]:
     """Compute the differences between file data and db data"""
-    return tuple()
+    diff = tuple()
+    for filerow, dbrow in zip(
+            sorted(filedata, key=lambda item: (
+                item["phenotype_id"], item["xref_id"])),
+            sorted(dbdata, key=lambda item: (
+                item["PhenotypeId"], item["xref_id"]))):
+        if filerow["PubMed_ID"] == dbrow["PubMed_ID"]:
+            continue
+
+        newpubmed = filerow["PubMed_ID"]
+        diff = diff + ({
+            **dbrow,
+            "PubMed_ID": newpubmed,
+            "PublicationId": pubmedid2pubidmap.get(newpubmed)},)
+
+    return diff
diff --git a/uploader/publications/models.py b/uploader/publications/models.py
index 89da06c..3fc9542 100644
--- a/uploader/publications/models.py
+++ b/uploader/publications/models.py
@@ -1,7 +1,17 @@
 """Module to handle persistence and retrieval of publication to/from MariaDB"""
+import logging
+
+from MySQLdb.cursors import DictCursor
+
+from gn_libs.mysqldb import Connection, debug_query
+
+logger = logging.getLogger(__name__)
+
 
 def fetch_phenotype_publications(
-        conn, ids: tuple[tuple[int, int], ...]) -> tuple[dict, ...]:
+        conn: Connection,
+        ids: tuple[tuple[int, int], ...]
+) -> tuple[dict, ...]:
     """Fetch publication from database by ID."""
     paramstr = ",".join(["(%s, %s)"] * len(ids))
     query = (
@@ -13,3 +23,51 @@ def fetch_phenotype_publications(
     with conn.cursor(cursorclass=DictCursor) as cursor:
         cursor.execute(query, tuple(item for row in ids for item in row))
         return tuple(dict(row) for row in cursor.fetchall())
+
+
+def create_new_publications(
+        conn: Connection,
+        publications: tuple[dict, ...]
+) -> tuple[dict, ...]:
+    if len(publications) > 0:
+        with conn.cursor(cursorclass=DictCursor) as cursor:
+            cursor.executemany(
+                ("INSERT INTO "
+                 "Publication( "
+                 "PubMed_ID, Abstract, Authors, Title, Journal, Volume, Pages, "
+                 "Month, Year"
+                 ") "
+                 "VALUES("
+                 "%(pubmed_id)s, %(abstract)s, %(authors)s, %(title)s, "
+                 "%(journal)s, %(volume)s, %(pages)s, %(month)s, %(year)s"
+                 ") "
+                 "ON DUPLICATE KEY UPDATE "
+                 "Abstract=VALUES(Abstract), Authors=VALUES(Authors), "
+                 "Title=VALUES(Title), Journal=VALUES(Journal), "
+                 "Volume=VALUES(Volume), Pages=VALUES(pages), "
+                 "Month=VALUES(Month), Year=VALUES(Year) "
+                 "RETURNING *"),
+                publications)
+            return tuple({
+                **row, "PublicationId": row["Id"]
+            } for row in cursor.fetchall())
+        return tuple()
+
+
+def update_publications(conn: Connection , publications: tuple[dict, ...]) -> tuple[dict, ...]:
+    """Update details for multiple publications"""
+    if len(publications) > 0:
+        with conn.cursor(cursorclass=DictCursor) as cursor:
+            logger.debug("UPDATING PUBLICATIONS: %s", publications)
+            cursor.executemany(
+                ("UPDATE Publication SET "
+                 "PubMed_ID=%(pubmed_id)s, Abstract=%(abstract)s, "
+                 "Authors=%(authors)s, Title=%(title)s, Journal=%(journal)s, "
+                 "Volume=%(volume)s, Pages=%(pages)s, Month=%(month)s, "
+                 "Year=%(year)s "
+                 "WHERE Id=%(publication_id)s"),
+                publications)
+            debug_query(cursor, logger)
+            return publications
+        return tuple()
+    return tuple()
diff --git a/uploader/publications/pubmed.py b/uploader/publications/pubmed.py
index d984d99..ed9b652 100644
--- a/uploader/publications/pubmed.py
+++ b/uploader/publications/pubmed.py
@@ -1,4 +1,10 @@
 """Module to interact with NCBI's PubMed"""
+import logging
+
+import requests
+from lxml import etree
+
+logger = logging.getLogger(__name__)
 
 
 def __pub_date__(pubdate: etree.Element):
@@ -44,7 +50,7 @@ def __abstract__(article: etree.Element) -> str:
 def __article__(pubmed_article: etree.Element) -> dict:
     article = pubmed_article.find("MedlineCitation/Article")
     return {
-        "pubmed_id": pubmed_article.find("MedlineCitation/PMID").text,
+        "pubmed_id": int(pubmed_article.find("MedlineCitation/PMID").text),
         "title": article.find("ArticleTitle").text,
         **__journal__(article.find("Journal")),
         "abstract": __abstract__(article),