qurator-spk/modstool
1
0
Fork 0
mirror of https://github.com/qurator-spk/modstool.git synced 2025-11-04 03:14:14 +01:00
Code Issues Releases Wiki Activity

🚧 ALTO: Move xpath_statistics to TagGroup class

Browse source
This commit is contained in:
Gerber, Mike 2022-05-23 19:39:21 +02:00
parent 9246519162
commit aa4e8e290d
2 changed files with 23 additions and 18 deletions
Download patch file Download diff file Expand all files Collapse all files

22
qurator/modstool/altotool.py
View file

@ -76,27 +76,13 @@ def alto_to_dict(alto, raise_errors=True):
elif localname == 'Layout': elif localname == 'Layout':
value[localname] = TagGroup(tag, group).is_singleton().has_no_attributes().descend(raise_errors) value[localname] = TagGroup(tag, group).is_singleton().has_no_attributes().descend(raise_errors)
elif localname == 'Page': elif localname == 'Page':
value['Page'] = {}
value['Page'].update(TagGroup(tag, group).is_singleton().attributes())
value['Page'].update(TagGroup(tag, group).subelement_counts())
xpath_expr = "//alto:String/@WC"
alto_namespace = ET.QName(group[0]).namespace alto_namespace = ET.QName(group[0]).namespace
namespaces={"alto": alto_namespace} namespaces={"alto": alto_namespace}
def xpath_statistics(xpath_expr, namespaces): value[localname] = {}
values = [] value[localname].update(TagGroup(tag, group).is_singleton().attributes())
for e in group: value[localname].update(TagGroup(tag, group).subelement_counts())
r = e.xpath(xpath_expr, namespaces=namespaces) value[localname].update(TagGroup(tag, group).xpath_statistics("//alto:String/@WC", namespaces))
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': elif localname == 'Styles':
pass pass

19
qurator/modstool/lib.py
View file

@ -3,6 +3,7 @@ import re
import warnings import warnings
from typing import List, Sequence, MutableMapping from typing import List, Sequence, MutableMapping
import numpy as np
from lxml import etree as ET from lxml import etree as ET
@ -206,6 +207,24 @@ class TagGroup:
counts[key] = counts.get(key, 0) + 1 counts[key] = counts.get(key, 0) + 1
return counts return counts
def xpath_statistics(self, xpath_expr, namespaces):
"""
Extract values and calculate statistics
Extract values using the given XPath expression, convert them to float and return descriptive
statistics on the values.
"""
values = []
for e in self.group:
r = e.xpath(xpath_expr, namespaces=namespaces)
values += r
values = np.array([float(v) for v in values])
statistics = {}
statistics[f'{xpath_expr}-mean'] = np.mean(values)
return statistics
def sorted_groupby(iterable, key=None): def sorted_groupby(iterable, key=None):