From b92d1b60625592baf66b08451274e4c865028d7d Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@archive.org>
Date: Thu, 21 Jul 2022 14:37:04 -0700
Subject: more publisher_type pattern matching

---
 chocula/database.py |  4 ++--
 chocula/util.py     | 15 ++++++++++-----
 2 files changed, 12 insertions(+), 7 deletions(-)

(limited to 'chocula')

diff --git a/chocula/database.py b/chocula/database.py
index 70aa974..3840ae3 100644
--- a/chocula/database.py
+++ b/chocula/database.py
@@ -600,7 +600,7 @@ class ChoculaDatabase:
                 out["publisher_type"] = "commercial"
             elif publisher in ARCHIVE_PUBLISHERS:
                 out["publisher_type"] = "archive"
-            elif publisher in REPOSITORY_PUBLISHERS:
+            elif publisher in REPOSITORY_PUBLISHERS or "repository" in pl:
                 out["publisher_type"] = "repository"
             elif publisher in OTHER_PUBLISHERS:
                 out["publisher_type"] = "other"
@@ -614,7 +614,7 @@ class ChoculaDatabase:
                 or "ieee" in out.get("name", "")
             ):
                 out["publisher_type"] = "society"
-            elif publisher in UNI_PRESS_PUBLISHERS or "university " in pl:
+            elif publisher in UNI_PRESS_PUBLISHERS or "university " in pl or "universität" in pl:
                 out["publisher_type"] = "unipress"
             elif "scielo" in pl:
                 out["publisher_type"] = "scielo"
diff --git a/chocula/util.py b/chocula/util.py
index ed7398d..4e40889 100644
--- a/chocula/util.py
+++ b/chocula/util.py
@@ -32,14 +32,19 @@ MIMETYPE_MAP = {
 
 BIG5_PUBLISHERS = [
     "Elsevier",
-    "Informa UK (Taylor & Francis)",
+    "Springer",
     "Springer-Verlag",
-    "SAGE Publications",
-    "Wiley (Blackwell Publishing)",
-    "Wiley (John Wiley & Sons)",
     "Springer (Biomed Central Ltd.)",
     "Springer Nature",
+    "Informa UK (Taylor & Francis)",
+    "Taylor & Francis",
+    "Wiley (Blackwell Publishing)",
+    "Wiley (John Wiley & Sons)",
+    "Wiley-Blackwell",
+    "SAGE Publications",
+    # Walter de Gruyter would be a potential "6th" big publisher
 ]
+
 COMMERCIAL_PUBLISHERS = [
     "Peter Lang International Academic Publishers",
     "Walter de Gruyter GmbH",
@@ -51,7 +56,6 @@ COMMERCIAL_PUBLISHERS = [
     "Bentham Science",
     "Ovid Technologies (Wolters Kluwer)  - Lippincott Williams & Wilkins",
     "Scientific Research Publishing, Inc",
-    "MDPI AG",
     "S. Karger AG",
     "Pleiades Publishing",
     "Science Publishing Group",
@@ -94,6 +98,7 @@ SOCIETY_PUBLISHERS = [
     "International Union of Crystallography",
     "Portland Press",
     "ASME International",
+    "The Optical Society",
 ]
 UNI_PRESS_PUBLISHERS = [
     "Cambridge University Press",
-- 
cgit v1.2.3