🚧 ALTO: Extract a function to calculate statistics on xpath expressions

qurator/modstool/altotool.py

@@ -81,14 +81,22 @@ def alto_to_dict(alto, raise_errors=True):
             value['Page'].update(TagGroup(tag, group).subelement_counts())
 
             xpath_expr = "//alto:String/@WC"
-            values = []
+            alto_namespace = ET.QName(group[0]).namespace
-            for e in group:
+            namespaces={"alto": alto_namespace}
-                # TODO need a smart way to always have the correct namespaces for a document
+
-                alto_namespace = ET.QName(e).namespace
+            def xpath_statistics(xpath_expr, namespaces):
-                r = e.xpath(xpath_expr, namespaces={"alto": alto_namespace})
+                values = []
-                values += r
+                for e in group:
-            values = np.array([float(v) for v in values])
+                    r = e.xpath(xpath_expr, namespaces=namespaces)
-            value['Page'][f'{xpath_expr}-mean'] = np.mean(values)
+                    values += r
+                values = np.array([float(v) for v in values])
+
+                statistics = {}
+                statistics[f'{xpath_expr}-mean'] = np.mean(values)
+                return statistics
+
+            value['Page'].update(xpath_statistics(xpath_expr, namespaces))
+
 
         elif localname == 'Styles':
             pass