qurator-spk/modstool
1
0
Fork 0
mirror of https://github.com/qurator-spk/modstool.git synced 2025-06-25 11:39:54 +02:00
Code Issues Releases Wiki Activity

✔ Test if dtypes are as expected in produced Parquet files

Browse source
This commit is contained in:
Mike Gerber 2025-06-12 09:42:29 +02:00
parent 215bfbb11f
commit ac8740c33f
5 changed files with 130 additions and 92 deletions
Download patch file Download diff file Expand all files Collapse all files

87
check_dtypes.py
View file

@ -1,87 +0,0 @@
import re
import warnings
import os
with warnings.catch_warnings():
# Filter warnings on WSL
if "Microsoft" in os.uname().release:
warnings.simplefilter("ignore")
import pandas as pd
mods_info = pd.read_parquet("mods_info_df.parquet")
page_info = pd.read_parquet("page_info_df.parquet")
alto_info = pd.read_parquet("alto_info_df.parquet")
# Check
EXPECTED_TYPES = {
# mods_info
r"mets_file": ("object", ["str"]),
r"titleInfo_title": ("object", ["str"]),
r"titleInfo_subTitle": ("object", ["str", "NoneType"]),
r"titleInfo_partName": ("object", ["str", "NoneType"]),
r"identifier-.*": ("object", ["str", "NoneType"]),
r"location_.*": ("object", ["str", "NoneType"]),
r"name\d+_.*roleTerm": ("object", ["ndarray", "NoneType"]),
r"name\d+_.*": ("object", ["str", "NoneType"]),
r"relatedItem-.*_recordInfo_recordIdentifier": ("object", ["str", "NoneType"]),
r"typeOfResource": ("object", ["str", "NoneType"]),
r"accessCondition-.*": ("object", ["str", "NoneType"]),
r"originInfo-.*": ("object", ["str", "NoneType"]),
r".*-count": ("Int64", None),
r"genre-.*": ("object", ["ndarray", "NoneType"]),
r"subject-.*": ("object", ["ndarray", "NoneType"]),
r"language_.*Term": ("object", ["ndarray", "NoneType"]),
r"classification-.*": ("object", ["ndarray", "NoneType"]),
# page_info
r"fileGrp_.*_file_FLocat_href": ("object", ["str", "NoneType"]),
r"structMap-LOGICAL_TYPE_.*": ("boolean", None),
# alto_info
r"Description_.*": ("object", ["str", "NoneType"]),
r"Layout_Page_ID": ("object", ["str", "NoneType"]),
r"Layout_Page_PHYSICAL_(IMG|IMAGE)_NR": ("object", ["str", "NoneType"]),
r"Layout_Page_PROCESSING": ("object", ["str", "NoneType"]),
r"Layout_Page_QUALITY": ("object", ["str", "NoneType"]),
r"Layout_Page_//alto:String/@WC-.*": ("Float64", None),
r"alto_xmlns": ("object", ["str", "NoneType"]),
r"Layout_Page_(WIDTH|HEIGHT)": ("Int64", None),
}
def expected_types(c):
for r, types in EXPECTED_TYPES.items():
if re.fullmatch(r, c):
edt = types[0]
einner_types = types[1]
if einner_types:
einner_types = set(einner_types)
return edt, einner_types
return None, None
def check_types(df):
for c in df.columns:
dt = df.dtypes[c]
edt, einner_types = expected_types(c)
if edt is None:
print(f"No expected dtype known for column {c} (got {dt})")
elif dt != edt:
print(f"Unexpected dtype {dt} for column {c} (expected {edt})")
if edt == "object":
inner_types = set(type(v).__name__ for v in df[c])
if any(it not in einner_types for it in inner_types):
print(f"Unexpected inner types {inner_types} for column {c} (expected {einner_types})")
check_types(mods_info)
check_types(page_info)
check_types(alto_info)

5
src/mods4pandas/alto4pandas.py
View file

@ -138,7 +138,7 @@ def walk(m):
@click.argument('alto_files', type=click.Path(exists=True), required=True, nargs=-1)
@click.option('--output', '-o', 'output_file', type=click.Path(), help='Output Parquet file',
default='alto_info_df.parquet', show_default=True)
def process(alto_files: List[str], output_file: str):
def process_command(alto_files: List[str], output_file: str):
"""
A tool to convert the ALTO metadata in INPUT to a pandas DataFrame.
@ -151,6 +151,9 @@ def process(alto_files: List[str], output_file: str):
- and a CSV file with all conversion warnings.
"""
process(alto_files, output_file)
def process(alto_files: List[str], output_file: str):
# Extend file list if directories are given
alto_files_real = []
for m in alto_files:

6
src/mods4pandas/mods4pandas.py
View file

@ -382,7 +382,7 @@ def pages_to_dict(mets, raise_errors=True) -> List[Dict]:
@click.option('--output', '-o', 'output_file', type=click.Path(), help='Output Parquet file',
default='mods_info_df.parquet', show_default=True)
@click.option('--output-page-info', type=click.Path(), help='Output page info Parquet file')
def process(mets_files: list[str], output_file: str, output_page_info: str):
def process_command(mets_files: list[str], output_file: str, output_page_info: str):
"""
A tool to convert the MODS metadata in INPUT to a pandas DataFrame.
@ -393,7 +393,9 @@ def process(mets_files: list[str], output_file: str, output_page_info: str):
Per-page information (e.g. structure information) can be output to a separate Parquet file.
"""
process(mets_files, output_file, output_page_info)
def process(mets_files: list[str], output_file: str, output_page_info: str):
# Extend file list if directories are given
mets_files_real: list[str] = []
for m in mets_files:
@ -476,7 +478,7 @@ def main():
for prefix, uri in ns.items():
ET.register_namespace(prefix, uri)
process()
process_command()
if __name__ == '__main__':

53
src/mods4pandas/tests/test_alto.py
View file

@ -1,9 +1,13 @@
from pathlib import Path
import re
from lxml import etree as ET
import pandas as pd
from mods4pandas.alto4pandas import alto_to_dict
from mods4pandas.alto4pandas import alto_to_dict, process
from mods4pandas.lib import flatten
TESTS_DATA_DIR = Path(__file__).parent / "data"
def dict_fromstring(x):
return flatten(alto_to_dict(ET.fromstring(x)))
@ -79,3 +83,50 @@ def test_String_TAGREF_counts():
""")
assert d['Layout_Page_//alto:String[@TAGREFS]-count'] == 3
assert d['Layout_Page_String-count'] == 4
def test_dtypes(tmp_path):
alto_dir = (TESTS_DATA_DIR / "alto").absolute().as_posix()
alto_info_df_parquet = (tmp_path / "test_dtypes_alto_info.parquet").as_posix()
process([alto_dir], alto_info_df_parquet)
alto_info_df = pd.read_parquet(alto_info_df_parquet)
EXPECTED_TYPES = {
r"Description_.*": ("object", ["str", "NoneType"]),
r"Layout_Page_ID": ("object", ["str", "NoneType"]),
r"Layout_Page_PHYSICAL_(IMG|IMAGE)_NR": ("object", ["str", "NoneType"]),
r"Layout_Page_PROCESSING": ("object", ["str", "NoneType"]),
r"Layout_Page_QUALITY": ("object", ["str", "NoneType"]),
r"Layout_Page_//alto:String/@WC-.*": ("Float64", None),
r".*-count": ("Int64", None),
r"alto_xmlns": ("object", ["str", "NoneType"]),
r"Layout_Page_(WIDTH|HEIGHT)": ("Int64", None),
}
def expected_types(c):
"""Return the expected types for column c."""
for r, types in EXPECTED_TYPES.items():
if re.fullmatch(r, c):
edt = types[0]
einner_types = types[1]
if einner_types:
einner_types = set(einner_types)
return edt, einner_types
return None, None
def check_types(df):
"""Check the types of the DataFrame df."""
for c in df.columns:
dt = df.dtypes[c]
edt, einner_types = expected_types(c)
print(c, dt, edt)
assert edt is not None, f"No expected dtype known for column {c} (got {dt})"
assert dt == edt, f"Unexpected dtype {dt} for column {c} (expected {edt})"
if edt == "object":
inner_types = set(type(v).__name__ for v in df[c])
assert all(it in einner_types for it in inner_types), \
f"Unexpected inner types {inner_types} for column {c} (expected {einner_types})"
check_types(alto_info_df)

71
src/mods4pandas/tests/test_mods4pandas.py
View file

@ -1,10 +1,14 @@
from pathlib import Path
import re
from lxml import etree as ET
import pandas as pd
import pytest
from mods4pandas.mods4pandas import mods_to_dict
from mods4pandas.mods4pandas import mods_to_dict, process
from mods4pandas.lib import flatten
TESTS_DATA_DIR = Path(__file__).parent / "data"
def dict_fromstring(x):
"""Helper function to parse a MODS XML string to a flattened dict"""
@ -151,3 +155,68 @@ def test_relatedItem():
""")
assert d['relatedItem-original_recordInfo_recordIdentifier-dnb-ppn'] == '1236513355'
def test_dtypes(tmp_path):
mets_files = [p.absolute().as_posix() for p in (TESTS_DATA_DIR / "mets-mods").glob("*.xml")]
mods_info_df_parquet = (tmp_path / "test_dtypes_mods_info.parquet").as_posix()
page_info_df_parquet = (tmp_path / "test_dtypes_page_info.parquet").as_posix()
process(mets_files, mods_info_df_parquet, page_info_df_parquet)
mods_info_df = pd.read_parquet(mods_info_df_parquet)
page_info_df = pd.read_parquet(page_info_df_parquet)
EXPECTED_TYPES = {
# mods_info
r"mets_file": ("object", ["str"]),
r"titleInfo_title": ("object", ["str"]),
r"titleInfo_subTitle": ("object", ["str", "NoneType"]),
r"titleInfo_partName": ("object", ["str", "NoneType"]),
r"identifier-.*": ("object", ["str", "NoneType"]),
r"location_.*": ("object", ["str", "NoneType"]),
r"name\d+_.*roleTerm": ("object", ["ndarray", "NoneType"]),
r"name\d+_.*": ("object", ["str", "NoneType"]),
r"relatedItem-.*_recordInfo_recordIdentifier": ("object", ["str", "NoneType"]),
r"typeOfResource": ("object", ["str", "NoneType"]),
r"accessCondition-.*": ("object", ["str", "NoneType"]),
r"originInfo-.*": ("object", ["str", "NoneType"]),
r".*-count": ("Int64", None),
r"genre-.*": ("object", ["ndarray", "NoneType"]),
r"subject-.*": ("object", ["ndarray", "NoneType"]),
r"language_.*Term": ("object", ["ndarray", "NoneType"]),
r"classification-.*": ("object", ["ndarray", "NoneType"]),
# page_info
r"fileGrp_.*_file_FLocat_href": ("object", ["str", "NoneType"]),
r"structMap-LOGICAL_TYPE_.*": ("boolean", None),
}
def expected_types(c):
"""Return the expected types for column c."""
for r, types in EXPECTED_TYPES.items():
if re.fullmatch(r, c):
edt = types[0]
einner_types = types[1]
if einner_types:
einner_types = set(einner_types)
return edt, einner_types
return None, None
def check_types(df):
"""Check the types of the DataFrame df."""
for c in df.columns:
dt = df.dtypes[c]
edt, einner_types = expected_types(c)
print(c, dt, edt)
assert edt is not None, f"No expected dtype known for column {c} (got {dt})"
assert dt == edt, f"Unexpected dtype {dt} for column {c} (expected {edt})"
if edt == "object":
inner_types = set(type(v).__name__ for v in df[c])
assert all(it in einner_types for it in inner_types), \
f"Unexpected inner types {inner_types} for column {c} (expected {einner_types})"
check_types(mods_info_df)
check_types(page_info_df)