From a52e5f39346e28bdb2eb58f17a298177659dbbfe Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@archive.org>
Date: Thu, 8 Oct 2020 17:36:07 -0700
Subject: database support for scholarsportal and cariniana preservation
 holdings

---
 chocula/common.py   | 51 +++++++++++++++++++++++++++++++++++++++++++++++++++
 chocula/database.py |  2 ++
 chocula/kbart.py    | 20 +++++++++++++++++++-
 3 files changed, 72 insertions(+), 1 deletion(-)

(limited to 'chocula')

diff --git a/chocula/common.py b/chocula/common.py
index edd48a3..94c4c7f 100644
--- a/chocula/common.py
+++ b/chocula/common.py
@@ -148,6 +148,9 @@ class KbartLoader:
                 else:
                     new_spans = [[record.start_year, record.end_year]]
                 record.year_spans = merge_spans(old_spans, new_spans)
+            elif record.year_spans:
+                old_spans = existing.year_spans or []
+                record.year_spans = merge_spans(old_spans, record.year_spans)
             kbart_dict[record.issnl] = record
 
         counts["unique-issnl"] = len(kbart_dict)
@@ -218,6 +221,54 @@ class OnixCsvLoader(KbartLoader):
         return record
 
 
+class CarinianaCsvLoader(KbartLoader):
+    """
+    Similar to the KBART loader class, but for custom CSV files instead of
+    KBART formated TSV.
+
+    CSV columns:
+      - Region
+      - Knowledge Area
+      - Publisher
+      - Title
+      - ISSN
+      - eISSN
+      - Preserved Volumes
+      - Preserved Years
+      - In Progress Volumes
+      - In Progress Years
+
+    TODO: volumes
+    """
+
+    def open_file(self) -> Iterable:
+        return csv.DictReader(open(self.file_path(), "r"))
+
+    def parse_record(self, row: dict, issn_db: IssnDatabase) -> Optional[KbartRecord]:
+
+        raw_issn = clean_issn(row["ISSN"])
+        issne = clean_issn(row["ISSN"])
+        issnl = issn_db.issn2issnl(raw_issn or issne or "")
+        # convert list of years to a set of year spans
+        years = [int(y.strip()) for y in row["Preserved Years"].split(";") if y]
+        year_spans = merge_spans([], [[y, y] for y in years])
+        record = KbartRecord(
+            issnl=issnl,
+            issne=issne,
+            issnp=None,
+            embargo=None,
+            title=clean_str(row["Title"]),
+            publisher=clean_str(row["Publisher"]),
+            url=None,
+            start_year=None,
+            end_year=None,
+            start_volume=None,
+            end_volume=None,
+            year_spans=year_spans,
+        )
+        return record
+
+
 class HathifilesLoader(KbartLoader):
     """
     Similar to the KBART loader class, but for Hathifiles bulk format.
diff --git a/chocula/database.py b/chocula/database.py
index 3e8db57..d4fbbed 100644
--- a/chocula/database.py
+++ b/chocula/database.py
@@ -837,6 +837,8 @@ class ChoculaDatabase:
                     "jstor",
                     "pkp_pln",
                     "hathitrust",
+                    "scholarsportal",
+                    "cariniana",
                 ):
                     extra["kbart"] = extra.get("kbart", {})
                     extra["kbart"][drow["slug"]] = dict(year_spans=dextra["year_spans"])
diff --git a/chocula/kbart.py b/chocula/kbart.py
index 3944430..23c5533 100644
--- a/chocula/kbart.py
+++ b/chocula/kbart.py
@@ -1,4 +1,4 @@
-from chocula.common import KbartLoader, OnixCsvLoader, HathifilesLoader
+from chocula.common import KbartLoader, OnixCsvLoader, CarinianaCsvLoader, HathifilesLoader
 
 
 class ClockssKbartLoader(KbartLoader):
@@ -41,6 +41,22 @@ class PkpPlnOnixLoader(OnixCsvLoader):
         return self.config.pkp_pln.filepath
 
 
+class CarinianaLoader(CarinianaCsvLoader):
+
+    source_slug = "cariniana"
+
+    def file_path(self) -> str:
+        return self.config.cariniana.filepath
+
+
+class ScholarsportalOnixLoader(OnixCsvLoader):
+
+    source_slug = "scholarsportal"
+
+    def file_path(self) -> str:
+        return self.config.scholarsportal.filepath
+
+
 class HathitrustLoader(HathifilesLoader):
 
     source_slug = "hathitrust"
@@ -55,5 +71,7 @@ ALL_CHOCULA_KBART_CLASSES = [
     PorticoKbartLoader,
     JstorKbartLoader,
     PkpPlnOnixLoader,
+    CarinianaLoader,
+    ScholarsportalOnixLoader,
     HathitrustLoader,
 ]
-- 
cgit v1.2.3