From 5c811c14ca378c2071de89be813881d59d0010f4 Mon Sep 17 00:00:00 2001
From: Harris Tzovanakis <me@drjova.com>
Date: Tue, 27 Feb 2024 14:30:15 +0100
Subject: [PATCH] common: add journal mapping

* Adds `journal_title` mappings.

* ref: cern-sis/issues-scoap3#308
---
 dags/common/constants.py                      |  2 +
 dags/common/parsing/generic_parsing.py        |  8 ++-
 .../common/parsing/test_generic_parsing.py    | 56 +++++++++++++++++++
 3 files changed, 65 insertions(+), 1 deletion(-)

diff --git a/dags/common/constants.py b/dags/common/constants.py
index 30262864..c4a8dc4e 100644
--- a/dags/common/constants.py
+++ b/dags/common/constants.py
@@ -12,3 +12,5 @@
 WHITE_SPACES = re.compile(r"[\n\t]{1,}" + r"\s{2,}")
 CDATA_PATTERN = re.compile(r"<\?CDATA(.*)\?>")
 FN_REGEX = re.compile(r"<fn.*<\/fn>")
+
+JOURNAL_MAPPING = {"PLB": "Physics Letters B", "NUPHB": "Nuclear Physics B"}
diff --git a/dags/common/parsing/generic_parsing.py b/dags/common/parsing/generic_parsing.py
index 377f71a8..25bba2dc 100644
--- a/dags/common/parsing/generic_parsing.py
+++ b/dags/common/parsing/generic_parsing.py
@@ -1,6 +1,8 @@
 import re
 from datetime import date
 
+from dags.common.constants import JOURNAL_MAPPING
+
 
 def take_first(arr):
     try:
@@ -82,9 +84,13 @@ def parse_thesis_supervisors(value):
 def publication_info(article):
     if "publication_info" in article:
         return article["publication_info"]
+
+    journal_title = article.get("journal_title", "")
+    journal_title = JOURNAL_MAPPING.get(journal_title, journal_title)
+
     return [
         {
-            "journal_title": article.get("journal_title", ""),
+            "journal_title": journal_title,
             "journal_volume": article.get("journal_volume", ""),
             "year": int(article.get("journal_year", 0)) or "",
             "journal_issue": article.get("journal_issue", ""),
diff --git a/tests/units/common/parsing/test_generic_parsing.py b/tests/units/common/parsing/test_generic_parsing.py
index 35759fce..6ab268a3 100644
--- a/tests/units/common/parsing/test_generic_parsing.py
+++ b/tests/units/common/parsing/test_generic_parsing.py
@@ -226,6 +226,62 @@ def test_parse_thesis_supervisors(test_input, expected):
             ],
             id="Some values populated",
         ),
+        param(
+            {
+                "journal_title": "NUPHB",
+                "journal_volume": "Test Value",
+                "journal_year": "2022",
+                "journal_issue": "Test Value",
+                "journal_artid": "",
+                "journal_fpage": "",
+                "journal_lpage": "",
+                "journal_doctype": "",
+                "pubinfo_freetext": "",
+                "another_field": "Test Another Field",
+            },
+            [
+                {
+                    "journal_title": "Nuclear Physics B",
+                    "journal_volume": "Test Value",
+                    "year": 2022,
+                    "journal_issue": "Test Value",
+                    "artid": "",
+                    "page_start": "",
+                    "page_end": "",
+                    "material": "",
+                    "pubinfo_freetext": "",
+                }
+            ],
+            id="Test journal title NUPHB to Nuclear Physics B",
+        ),
+        param(
+            {
+                "journal_title": "PLB",
+                "journal_volume": "Test Value",
+                "journal_year": "2022",
+                "journal_issue": "Test Value",
+                "journal_artid": "",
+                "journal_fpage": "",
+                "journal_lpage": "",
+                "journal_doctype": "",
+                "pubinfo_freetext": "",
+                "another_field": "Test Another Field",
+            },
+            [
+                {
+                    "journal_title": "Physics Letters B",
+                    "journal_volume": "Test Value",
+                    "year": 2022,
+                    "journal_issue": "Test Value",
+                    "artid": "",
+                    "page_start": "",
+                    "page_end": "",
+                    "material": "",
+                    "pubinfo_freetext": "",
+                }
+            ],
+            id="Test journal title PLB to Physics Letters B",
+        ),
         param(
             {
                 "journal_title": "Test Value",