From 57db2db336c08031324e44b2d2880fbd4b6893c9 Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@archive.org>
Date: Mon, 1 Jun 2020 17:01:20 -0700
Subject: 'everything' at least partially working

---
 chocula/directories/__init__.py | 19 +++++++++++
 chocula/directories/sim.py      | 71 +++++++++++++++++++++++++++++++++++++++++
 2 files changed, 90 insertions(+)
 create mode 100644 chocula/directories/__init__.py
 create mode 100644 chocula/directories/sim.py

(limited to 'chocula/directories')

diff --git a/chocula/directories/__init__.py b/chocula/directories/__init__.py
new file mode 100644
index 0000000..4bed696
--- /dev/null
+++ b/chocula/directories/__init__.py
@@ -0,0 +1,19 @@
+
+from chocula.directories.crossref import CrossrefLoader
+from chocula.directories.doaj import DoajLoader
+from chocula.directories.entrez import EntrezLoader
+from chocula.directories.ezb import EzbLoader
+from chocula.directories.gold_oa import GoldOALoader
+from chocula.directories.norwegian import NorwegianLoader
+from chocula.directories.openapc import OpenAPCLoader
+from chocula.directories.road import RoadLoader
+from chocula.directories.sherpa_romeo import SherpaRomeoLoader
+from chocula.directories.sim import SimLoader
+from chocula.directories.szczepanski import SzczepanskiLoader
+from chocula.directories.wikidata import WikidataLoader
+
+ALL_CHOCULA_DIR_CLASSES = [
+    CrossrefLoader, DoajLoader, EntrezLoader,EzbLoader, GoldOALoader,
+    NorwegianLoader, OpenAPCLoader, RoadLoader, SherpaRomeoLoader,
+    SzczepanskiLoader, WikidataLoader, SimLoader,
+]
diff --git a/chocula/directories/sim.py b/chocula/directories/sim.py
new file mode 100644
index 0000000..c0c02df
--- /dev/null
+++ b/chocula/directories/sim.py
@@ -0,0 +1,71 @@
+
+from typing import Iterable, Optional, Dict, Any
+import csv
+
+from chocula.util import clean_str, parse_mimetypes, parse_country, parse_lang, PLATFORM_MAP, gaps_to_spans
+from chocula.common import DirectoryLoader
+from chocula.database import DirectoryInfo, HomepageUrl
+
+
+class SimLoader(DirectoryLoader):
+
+    source_slug = "sim"
+
+    def open_file(self) -> Iterable:
+        return csv.DictReader(open(self.config.sim.filepath))
+
+    def parse_record(self, row) -> Optional[DirectoryInfo]:
+
+        """
+        NA Pub Cat ID
+        Title
+        Publisher
+        ISSN
+        Impact Rank
+        Total Cities
+        Journal Impact Factor
+        Eigenfact or Score
+        First Volume
+        Last Volume
+        NA Gaps
+        "Scholarly / Peer-\n Reviewed"
+        "Peer-\n Reviewed"
+        Pub Type
+        Pub Language
+        Subjects
+        """
+        # TODO: 'Pub Type'
+
+        extra = {}
+        first_year = row['First Volume']
+        if first_year:
+            first_year = int(first_year)
+            extra['first_year'] = int(row['First Volume'])
+        else:
+            first_year = None
+        last_year = row['Last Volume']
+        if last_year:
+            last_year = int(last_year)
+            extra['last_year'] = last_year
+        else:
+            last_year = None
+        gaps = [int(g) for g in row['NA Gaps'].split(';') if g.strip()]
+        if gaps:
+            extra['gaps'] = gaps
+        if first_year and last_year:
+            extra['year_spans'] = gaps_to_spans(first_year, last_year, gaps)
+        extra['scholarly_peer_reviewed'] = row["Scholarly / Peer-\nReviewed"]
+        extra['peer_reviewed'] = row["Peer-\nReviewed"]
+        extra['pub_type'] = clean_str(row["Pub Type"])
+
+        info = DirectoryInfo(
+            directory_slug=self.source_slug,
+            name=clean_str(row['Title']),
+            publisher=clean_str(row['Publisher']),
+            raw_issn=row['ISSN'][:9],
+            custom_id=row.get('NA Pub Cat ID').strip() or None,
+            langs=[parse_lang(row['Pub Language'])],
+            extra=extra,
+        )
+        return info
+
-- 
cgit v1.2.3