class TestApplicator(unittest.TestCase):
def setUp(self):
self.a = EADDir(input_dir=r'C:\Users\wboyle\PycharmProjects\bentley_code\utilities')
def _method_for_testing_write(self, ead):
texts = ead.tree.xpath("//text")
for text in texts:
text.text = "yo"
def _method_for_testing_characterize(self, ead):
results = []
texts = ead.tree.xpath("//text")
for text in texts:
results.append(text.text)
return results
def test_ead_list(self):
self.assertEquals(self.a.ead_files, ['ead_appended.xml', 'ead_messy.xml', 'ead_pretty.xml'])
def test_characterize_directory(self):
intended_results = [['text'], ['text'], ['text']]
self.assertEquals(self.a.characterize_dir(function=self._method_for_testing_characterize), intended_results)
def test_apply_to_directory(self):
output_dir = os.path.join(self.a.input_dir, "output")
self.a.apply_function_to_dir(function=self._method_for_testing_write, output_dir=output_dir)
b = EADDir(output_dir)
intended_results = [['yo'], ['yo'], ['yo']]
self.assertEquals(b.characterize_dir(self._method_for_testing_characterize), intended_results)
for ead in os.listdir(output_dir):
os.remove(os.path.join(output_dir, ead))
def test_apply_to_directory(self):
output_dir = os.path.join(self.a.input_dir, "output")
self.a.apply_function_to_dir(function=self._method_for_testing_write, output_dir=output_dir)
b = EADDir(output_dir)
intended_results = [['yo'], ['yo'], ['yo']]
self.assertEquals(b.characterize_dir(self._method_for_testing_characterize), intended_results)
for ead in os.listdir(output_dir):
os.remove(os.path.join(output_dir, ead))
def test_apply_to_directory(self):
output_dir = os.path.join(self.a.input_dir, "output")
self.a.apply_function_to_dir(function=self._method_for_testing_write,
output_dir=output_dir)
b = EADDir(output_dir)
intended_results = [['yo'], ['yo'], ['yo']]
self.assertEquals(
b.characterize_dir(self._method_for_testing_characterize),
intended_results)
for ead in os.listdir(output_dir):
os.remove(os.path.join(output_dir, ead))
class TestApplicator(unittest.TestCase):
def setUp(self):
self.a = EADDir(
input_dir=r'C:\Users\wboyle\PycharmProjects\bentley_code\utilities'
)
def _method_for_testing_write(self, ead):
texts = ead.tree.xpath("//text")
for text in texts:
text.text = "yo"
def _method_for_testing_characterize(self, ead):
results = []
texts = ead.tree.xpath("//text")
for text in texts:
results.append(text.text)
return results
def test_ead_list(self):
self.assertEquals(
self.a.ead_files,
['ead_appended.xml', 'ead_messy.xml', 'ead_pretty.xml'])
def test_characterize_directory(self):
intended_results = [['text'], ['text'], ['text']]
self.assertEquals(
self.a.characterize_dir(
function=self._method_for_testing_characterize),
intended_results)
def test_apply_to_directory(self):
output_dir = os.path.join(self.a.input_dir, "output")
self.a.apply_function_to_dir(function=self._method_for_testing_write,
output_dir=output_dir)
b = EADDir(output_dir)
intended_results = [['yo'], ['yo'], ['yo']]
self.assertEquals(
b.characterize_dir(self._method_for_testing_characterize),
intended_results)
for ead in os.listdir(output_dir):
os.remove(os.path.join(output_dir, ead))
def get_all_agents(input_dir):
"""
Directs extraction of controlaccess terms from a directory of EADs.
:param input_dir: filepath to the input director
:return: a dictionary in the form {"corpname": {"Apple Computer": [authid, naming_source], etc.},
"persname": {"Jane Doe (1900-1911)": [authid, naming_source], etc.},
"famname": {"Adams family": [authid, _naming_source], etc.}}
"""
agent_types = ["corpname", "persname", "famname"]
agents = dict(zip(agent_types, [{}, {}, {}]))
ead_dir = EADDir(input_dir=input_dir)
for ead in tqdm(ead_dir.ead_files, desc="grabbing all agents from eads"):
tree = etree.parse(os.path.join(ead_dir.input_dir, ead))
all_agents = get_agents_from_ead(tree)
for key, value in all_agents.items():
agents[key].update(value)
return agents
for text in texts:
if text[0] == tag.tag and text[1] == tag.text and tag.get("authfilenumber"):
index = dct["normalize to"]
# deleting authfilenumbers if the normalization index is "x"
if index == "x":
if tag.text == "University of Michigan.":
continue
del tag.attrib["authfilenumber"]
continue
# normalizing terms
normal_form = texts[int(index) - 1]
normal_tag = normal_form[0]
normal_text = normal_form[1]
tag.tag = normal_tag
tag.text = normal_text
if __name__ == "__main__":
directory = r'C:\Users\wboyle\PycharmProjects\without-reservations\Real_Masters_all'
ead_dir = EADDir(input_dir=directory)
data = load_data("normalization data.csv")
ead_dir.apply_function_to_dir(replace_subjects, output_dir=directory)
import csv
from lxml import etree
from utilities import EADDir
def find_multiple_physdescs(ead):
results = []
parents = ead.tree.xpath("//physdesc")
for parent in parents:
tags = parent.xpath("extent")
if len(tags) > 1:
results.append([ead.filename, ead.tree.getpath(parent), len(tags), etree.tostring(etree.fromstring(etree.tostring(parent)))])
return results
if __name__ == "__main__":
input_dir = r'C:\Users\wboyle\PycharmProjects\vandura\Real_Masters_all'
e = EADDir(input_dir)
results = e.characterize_dir(find_multiple_physdescs)
with open("eads_with_multiple_extents.csv", mode="wb") as f:
writer = csv.writer(f)
for result in results:
writer.writerows(result)
def setUp(self):
self.a = EADDir(input_dir=r'C:\Users\wboyle\PycharmProjects\bentley_code\utilities')
def setUp(self):
self.a = EADDir(
input_dir=r'C:\Users\wboyle\PycharmProjects\bentley_code\utilities'
)
