separate cleanups

2 files changed, 47 insertions, 0 deletions

diff --git a/fuzzycat/cleanups.py b/fuzzycat/cleanups.py
new file mode 100644
index 0000000..d806e51
--- /dev/null
+++ b/fuzzycat/cleanups.py
@@ -0,0 +1,17 @@
+
+"""
+Various shared cleanup approaches.
+"""
+
+from fuzzycat.utils import StringPipeline, normalize_whitespace, normalize_ampersand
+
+
+# These transformations should not affect the name or a journal.
+basic = StringPipeline([
+    str.lower,
+    normalize_whitespace,
+    normalize_ampersand,
+    lambda v: v.rstrip("."),
+])
+
+
diff --git a/fuzzycat/journals.py b/fuzzycat/journals.py
new file mode 100644
index 0000000..8c61f3a
--- /dev/null
+++ b/fuzzycat/journals.py
@@ -0,0 +1,30 @@
+# coding: utf-8
+
+"""
+Journal name matching. Includes names from issn database and abbreviations.
+"""
+
+import shelve
+
+class JournalLookup:
+    """
+    Lookup allows to lookup journals, using a database of real journal names.
+
+        >>> lookup = JournalLookup()
+        >>> lookup["Philosophica"]
+        {'1857-9272', '2232-299X', '2232-3007', '2232-3015'}
+
+    """
+    def __init__(self, namedb='namedb'):
+        self.db = shelve.open(namedb)
+
+    def __getitem__(self, v):
+        return self.db[v]
+
+    def get(self, v, cleanup_pipeline=None):
+        if not cleanup_pipeline:
+            return self.db[v]
+        return self.db[cleanup_pipeline(v)]
+
+    def close(self):
+        self.db.close()