def remove_cls_key(updater: DocumentUpdater):
"""Unset '_cls' key in documents if any"""
def by_path(ctx: ByPathContext):
ctx.collection.update_many(
{
ctx.filter_dotpath + '._cls': {
'$exists': True
},
**ctx.extra_filter
}, {'$unset': {
ctx.update_dotpath + '._cls': ''
}},
array_filters=ctx.build_array_filters())
updater.update_by_path(by_path)
def drop_field(updater: DocumentUpdater):
"""Drop field"""
def by_path(ctx: ByPathContext):
ctx.collection.update_many(
{
ctx.filter_dotpath: {
'$exists': True
},
**ctx.extra_filter
}, {'$unset': {
ctx.update_dotpath: ''
}},
array_filters=ctx.build_array_filters())
updater.update_by_path(by_path)
def change_inherit(self, updater: DocumentUpdater, diff: Diff):
"""Remove '_cls' key if EmbeddedDocument becomes non-inherit,
otherwise do nothing
"""
def by_path(ctx: ByPathContext):
ctx.collection.update_many(
{ctx.filter_dotpath + '._cls': {'$exists': True}, **ctx.extra_filter},
{'$unset': {ctx.update_dotpath + '._cls': ''}},
array_filters=ctx.build_array_filters()
)
self._check_diff(diff, False, bool)
if diff.new:
return
updater.update_by_path(by_path)
def to_url_string(updater: DocumentUpdater, check_only=False):
"""Cast fields to string and then verify if they contain URLs"""
def by_path(ctx: ByPathContext):
fltr = {
ctx.filter_dotpath: {
'$not': url_regex,
'$ne': None
},
**ctx.extra_filter
}
check_empty_result(ctx.collection, ctx.filter_dotpath, fltr)
to_string(updater)
url_regex = re.compile(
r"\A[A-Z]{3,}://[A-Z0-9\-._~:/?#\[\]@!$&'()*+,;%=]\Z", re.IGNORECASE)
if updater.migration_policy.name == 'strict':
updater.update_by_path(by_path)
def to_email_string(updater: DocumentUpdater):
def by_path(ctx: ByPathContext):
fltr = {
ctx.filter_dotpath: {
'$not': email_regex,
'$ne': None
},
**ctx.extra_filter
}
check_empty_result(ctx.collection, ctx.filter_dotpath, fltr)
to_string(updater)
email_regex = re.compile(
r"\A.*\Z", # TODO: insert email validation regex here
re.IGNORECASE)
if updater.migration_policy.name == 'strict':
updater.update_by_path(by_path)
def to_complex_datetime(updater: DocumentUpdater):
def by_path(ctx: ByPathContext):
fltr = {
ctx.filter_dotpath: {
'$not': regex,
'$ne': None
},
**ctx.extra_filter
}
check_empty_result(ctx.collection, ctx.filter_dotpath, fltr)
to_string(updater)
# We should not know which separator is used, so use '.+'
# Separator change is handled by appropriate field method
regex = r'\A' + str(
'.+'.join([r"\d{4}"] + [r"\d{2}"] * 5 + [r"\d{6}"])) + r'\Z'
if updater.migration_policy.name == 'strict':
updater.update_by_path(by_path)
def to_email_string(updater: DocumentUpdater):
def by_path(ctx: ByPathContext):
email_regex = r"\A[^\W][A-Z0-9._%+-]+@[\p{L}0-9.-]+\.\p{L}+\Z"
fltr = {
ctx.filter_dotpath: {
'$not': {
'$regex': email_regex,
'$options': 'i'
},
'$ne': None
},
**ctx.extra_filter
}
check_empty_result(ctx.collection, ctx.filter_dotpath, fltr)
to_string(updater)
if updater.migration_policy.name == 'strict':
updater.update_by_path(by_path)