def test_version_word_count():
#simple
assert model.Version().load({
"title": "Genesis",
"language": "he",
"versionTitle": "Tanach with Ta'amei Hamikra"
}).word_count() == 20813
assert model.Version().load({
"title": "Rashi on Shabbat",
"language": "he"
}).word_count() > 0
#complex
assert model.Version().load({
"title": "Pesach Haggadah",
"language": "he"
}).word_count() > 0
assert model.Version().load({
"title": "Orot",
"language": "he"
}).word_count() > 0
assert model.Version().load({
"title": "Ephod Bad on Pesach Haggadah"
}).word_count() > 0
#sets
assert model.VersionSet({
"title": {
"$regex": "Haggadah"
}
}).word_count() > 200000
def test_index_delete():
#Simple Text
ti = "Test Del"
i = model.Index({
"title": ti,
"heTitle": "כבכב",
"titleVariants": [ti],
"sectionNames": ["Chapter", "Paragraph"],
"categories": ["Musar"],
"lengths": [50, 501]
}).save()
new_version1 = model.Version({
"chapter": i.nodes.create_skeleton(),
"versionTitle": "Version 1 TEST",
"versionSource": "blabla",
"language": "he",
"title": i.title
})
new_version1.chapter = [[''], [''], ["לה לה לה לא חשוב על מה"]]
new_version1.save()
new_version2 = model.Version({
"chapter": i.nodes.create_skeleton(),
"versionTitle": "Version 2 TEST",
"versionSource": "blabla",
"language": "en",
"title": i.title
})
new_version2.chapter = [[], ["Hello goodbye bla bla blah"], []]
new_version2.save()
i.delete()
assert model.Index().load({'title': ti}) is None
assert model.VersionSet({'title': ti}).count() == 0
def test_index_delete():
#Simple Text
ti = "Test Del"
model.IndexSet({"title": ti}).delete()
model.VersionSet({"title": ti}).delete()
i = model.Index({
"title": ti,
"heTitle": u"כבכב",
"titleVariants": [ti],
"sectionNames": ["Chapter", "Paragraph"],
"categories": ["Musar"],
"lengths": [50, 501]
}).save()
new_version1 = model.Version(
{
"chapter": i.nodes.create_skeleton(),
"versionTitle": "Version 1 TEST",
"versionSource": "blabla",
"language": "he",
"title": i.title
}
)
new_version1.chapter = [[u''],[u''],[u"לה לה לה לא חשוב על מה"]]
new_version1.save()
new_version2 = model.Version(
{
"chapter": i.nodes.create_skeleton(),
"versionTitle": "Version 2 TEST",
"versionSource": "blabla",
"language": "en",
"title": i.title
}
)
new_version2.chapter = [[],["Hello goodbye bla bla blah"],[]]
new_version2.save()
i.delete()
assert model.Index().load({'title': ti}) is None
assert model.VersionSet({'title':ti}).count() == 0
#Commentator
from sefaria.helper.text import create_commentator_and_commentary_version
commentator_name = "Commentator Del"
he_commentator_name = u"פרשנדנן"
base_book = 'Genesis'
base_book2 = 'Pesach Haggadah'
model.IndexSet({"title": commentator_name}).delete()
model.VersionSet({"title": commentator_name + " on " + base_book}).delete()
model.VersionSet({"title": commentator_name + " on " + base_book2}).delete()
create_commentator_and_commentary_version(commentator_name, base_book, 'he', 'test', 'test', he_commentator_name)
create_commentator_and_commentary_version(commentator_name, base_book2, 'he', 'test', 'test', he_commentator_name)
ci = model.Index().load({'title': commentator_name}).delete()
assert model.Index().load({'title': commentator_name}) is None
assert model.VersionSet({'title':{'$regex': commentator_name}}).count() == 0
def merge_text_versions(version1, version2, text_title, language, warn=False):
"""
Merges the contents of two distinct text versions.
version2 is merged into version1 then deleted.
Preference is giving to version1 - if both versions contain content for a given segment,
only the content of version1 will be retained.
History entries are rewritten for version2.
NOTE: the history of that results will be incorrect for any case where the content of
version2 is overwritten - the history of those overwritten edits will remain.
To end with a perfectly accurate history, history items for segments which have been overwritten
would need to be identified and deleted.
"""
v1 = model.Version().load({
"title": text_title,
"versionTitle": version1,
"language": language
})
if not v1:
return {"error": "Version not found: %s" % version1}
v2 = model.Version().load({
"title": text_title,
"versionTitle": version2,
"language": language
})
if not v2:
return {"error": "Version not found: %s" % version2}
if warn and versions_overlap(v1.chapter, v2.chapter):
print "WARNING - %s & %s have overlapping content. Aborting." % (
version1, version2)
merged_text, sources = merge_translations([v1.chapter, v2.chapter],
[version1, version2])
v1.chapter = merged_text
v1.save()
update_version_title_in_history(version2, version1, text_title, language)
v2.delete()
return {"status": "ok"}
def setup_class(cls):
cls.simpleIndex = model.Index({
"title": cls.simpleIndexTitle,
"heTitle": "בלה1",
"titleVariants": [cls.simpleIndexTitle],
"sectionNames": ["Chapter", "Paragraph"],
"categories": ["Musar"],
"lengths": [50, 501]
}).save()
cls.simpleVersion = model.Version({
"chapter":
cls.simpleIndex.nodes.create_skeleton(),
"versionTitle":
"Version 1 TEST",
"versionSource":
"blabla",
"language":
"he",
"title":
cls.simpleIndexTitle
})
cls.simpleVersion.chapter = [['1'], ['2'], ["original text", "2nd"]]
cls.simpleVersion.save()
cls.complexIndex = model.Index({
"title": cls.complexIndexTitle,
"heTitle": "2בלה",
"titleVariants": [cls.complexIndexTitle],
"schema": {
"nodes": [{
"nodes": [{
"nodeType":
"JaggedArrayNode",
"depth":
2,
"sectionNames": ["Chapter", "Paragraph"],
"addressTypes": ["Integer", "Integer"],
"titles": [{
"text": "Node 2",
"lang": "en",
"primary": True
}, {
"text": "Node 2 he",
"lang": "he",
"primary": True
}],
"key":
"Node 2"
}],
"titles": [{
"text": "Node 1",
"lang": "en",
"primary": True
}, {
"text": "Node 1 he",
"lang": "he",
"primary": True
}],
"key":
"Node 1"
}, {
"nodeType":
"JaggedArrayNode",
"depth":
1,
"sectionNames": ["Paragraph"],
"addressTypes": ["Integer"],
"titles": [{
"text": "Node 3",
"lang": "en",
"primary": True
}, {
"text": "Node 3 he",
"lang": "he",
"primary": True
}],
"key":
"Node 3"
}],
"titles": [{
"text": cls.complexIndexTitle,
"lang": "en",
"primary": True
}, {
"text": cls.complexIndexTitle + "he",
"lang": "he",
"primary": True
}],
"key":
cls.complexIndexTitle
},
"categories": ["Musar"]
}).save()
cls.complexVersion = model.Version({
"chapter":
cls.complexIndex.nodes.create_skeleton(),
"versionTitle":
"Version 2 TEST",
"versionSource":
"blabla",
"language":
"en",
"title":
cls.complexIndexTitle
})
cls.complexVersion.chapter = {
"Node 1": {
"Node 2": [['yo'], ['', 'blah'], ["original text", "2nd"]]
},
"Node 3": ['1', '2', '3', '4']
}
cls.complexVersion.save()
