sporada/nopaque
1
0
Fork 0
mirror of https://gitlab.ub.uni-bielefeld.de/sfb1288inf/nopaque.git synced 2024-12-25 02:44:18 +00:00
Code Issues Projects Releases Wiki Activity

Update cqi utils

Browse Source
This commit is contained in:
Patrick Jentsch 2023-08-11 13:50:56 +02:00
parent 5c00c5740e
commit 1387d80a26
1 changed files with 6 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
app/corpora/cqi_over_sio/utils.py
View File

@ -9,8 +9,7 @@ def lookups_by_cpos(corpus: CQiCorpus, cpos_list: List[int]) -> Dict:
for attr in corpus.positional_attributes.list(): for attr in corpus.positional_attributes.list():
cpos_attr_values: List[str] = attr.values_by_cpos(cpos_list) cpos_attr_values: List[str] = attr.values_by_cpos(cpos_list)
for i, cpos in enumerate(cpos_list): for i, cpos in enumerate(cpos_list):
lookups['cpos_lookup'][cpos][attr.attrs['name']] = \ lookups['cpos_lookup'][cpos][attr.name] = cpos_attr_values[i]
cpos_attr_values[i]
for attr in corpus.structural_attributes.list(): for attr in corpus.structural_attributes.list():
# We only want to iterate over non subattributes, identifiable by # We only want to iterate over non subattributes, identifiable by
# attr.has_values == False # attr.has_values == False
@ -20,19 +19,19 @@ def lookups_by_cpos(corpus: CQiCorpus, cpos_list: List[int]) -> Dict:
for i, cpos in enumerate(cpos_list): for i, cpos in enumerate(cpos_list):
if cpos_attr_ids[i] == -1: if cpos_attr_ids[i] == -1:
continue continue
lookups['cpos_lookup'][cpos][attr.attrs['name']] = cpos_attr_ids[i] lookups['cpos_lookup'][cpos][attr.name] = cpos_attr_ids[i]
occured_attr_ids = [x for x in set(cpos_attr_ids) if x != -1] occured_attr_ids = [x for x in set(cpos_attr_ids) if x != -1]
if not occured_attr_ids: if len(occured_attr_ids) == 0:
continue continue
subattrs = corpus.structural_attributes.list(filters={'part_of': attr}) subattrs = corpus.structural_attributes.list(filters={'part_of': attr})
if not subattrs: if len(subattrs) == 0:
continue continue
lookup_name: str = f'{attr.attrs["name"]}_lookup' lookup_name: str = f'{attr.name}_lookup'
lookups[lookup_name] = {} lookups[lookup_name] = {}
for attr_id in occured_attr_ids: for attr_id in occured_attr_ids:
lookups[lookup_name][attr_id] = {} lookups[lookup_name][attr_id] = {}
for subattr in subattrs: for subattr in subattrs:
subattr_name = subattr.attrs['name'][(len(attr.attrs['name']) + 1):] # noqa subattr_name = subattr.name[(len(attr.name) + 1):] # noqa
for i, subattr_value in enumerate(subattr.values_by_ids(occured_attr_ids)): # noqa for i, subattr_value in enumerate(subattr.values_by_ids(occured_attr_ids)): # noqa
lookups[lookup_name][occured_attr_ids[i]][subattr_name] = subattr_value # noqa lookups[lookup_name][occured_attr_ids[i]][subattr_name] = subattr_value # noqa
return lookups return lookups