Normalize time series names

convert.py

@@ -46,6 +46,9 @@ YEAR_RE = re.compile(r'^([12]\d{3})$')
 MONTH_RE = re.compile(r'^([12]\d{3})-([01]\d)$')
 DAY_RE = re.compile(r'^([12]\d{3})-([01]\d)-([0-3]\d)$')
 
+# Helps normalize space
+NORM_SPACES_RE = re.compile(r'\s+')
+
 
 def write_series_file(file_path: Path, periods, observations, obs_status=None):
     """ write series tsv file from period and value list """
@@ -62,6 +65,15 @@ def write_series_file(file_path: Path, periods, observations, obs_status=None):
             fd.write('{}\n'.format('\t'.join(tup)))
 
 
+def norm_space(str_value):
+    """ Converts '  abc   de f ' into 'abc de f'
+
+    >>> norm_space("   ab\tc  de f  ")
+    'ab c de f'
+    """
+    return NORM_SPACES_RE.sub(' ', str_value).strip()
+
+
 def norm_period(period_str, freq):
     """Normalize period respecting frequency value
 
@@ -120,7 +132,7 @@ def load_series_xml_file(xml_file: Path):
             if elt.tag == SERIES_TAG:
                 info['code'] = elt.attrib['BBK_ID']
                 name_attr = 'BBK_TITLE_ENG' if 'BBK_TITLE_ENG' in elt.attrib else 'BBK_TITLE'
-                info['name'] = elt.attrib[name_attr]
+                info['name'] = norm_space(elt.attrib[name_attr])
                 info['freq'] = elt.attrib['FREQ']
                 info['periods'] = []
                 info['obs'] = []