refactored MeSH handling code

mesh.py

+import collections
+import re
+
+
+_prop_pattern = re.compile('([1A-Z_ ]+) = (.*)')
+
+
+class Values(list):
+    def __init__(self):
+        super().__init__()
+
+    def __bool__(self):
+        return True
+
+
+class Record(collections.defaultdict):
+    def __init__(self):
+        super().__init__(Values)
+
+    def __getattr__(self, item):
+        return self[item]
+
+    @property
+    def uid(self):
+        return self.UI[0]
+
+
+def records(path):
+    current = Record()
+    with open(path) as f:
+        for line in f:
+            line = line.strip()
+            if line == '':
+                continue
+            if line == '*NEWRECORD':
+                if 'UI' in current:
+                    yield current
+                current = Record()
+                continue
+            m = _prop_pattern.fullmatch(line)
+            if m is None:
+                raise RuntimeError('could not parse: ' + line)
+            key = m.group(1)
+            value = m.group(2)
+            current[key].append(value)
+    if 'UI' in current:
+        yield current

scripts/roots2mesh-tree.py

@@ -3,8 +3,7 @@
 
 import sys
 import yaml
-import re
-import collections
+import mesh
 
 
 MESH_FILE = sys.argv[1]
@@ -15,41 +14,18 @@ with open(ROOTS_FILE) as f:
     ROOTS = dict((e['mesh-id'], e['name']) for e in yaml.load(f, Loader=yaml.FullLoader) if 'mesh-id' in e)
 
 
-LINE = re.compile('([1A-Z_ ]+) = (.*)')
-
-
 SEEN = set()
 
 
-def finish(rec):
-    if 'UI' not in rec:
-        return
-    if rec['UI'][0] not in ROOTS:
-        return
-    if 'MN' not in rec:
-        raise RuntimeError('no tree for ' + rec['UI'][0])
-    for mtree in rec['MN']:
+for rec in mesh.records(MESH_FILE):
+    mid = rec.uid
+    if mid not in ROOTS:
+        continue
+    if not rec.MN:
+        raise RuntimeError('no tree for ' + mid)
+    for mtree in rec.MN:
         print(mtree)
-    SEEN.add(rec['UI'][0])
-
-
-current = collections.defaultdict(list)
-with open(MESH_FILE) as f:
-    for line in f:
-        line = line.strip()
-        if line == '':
-            continue
-        if line == '*NEWRECORD':
-            finish(current)
-            current = collections.defaultdict(list)
-            continue
-        m = LINE.fullmatch(line)
-        if m is None:
-            raise RuntimeError('could not parse: ' + line)
-        key = m.group(1)
-        value = m.group(2)
-        current[key].append(value)
-finish(current)
+    SEEN.add(mid)
 
 
 for meshid, name in ROOTS.items():