class DocumentPage(db.Model):
id = db.Column(db.BigInteger, primary_key=True)
number = db.Column(db.Integer(), nullable=False)
text = db.Column(db.Unicode(), nullable=False)
document_id = db.Column(db.Integer(), db.ForeignKey('document.id'))
document = db.relationship(
Document, backref=db.backref('pages',
cascade='all, delete-orphan')) # noqa
def __repr__(self):
return '<DocumentPage(%r,%r)>' % (self.document_id, self.number)
def text_parts(self):
"""Utility method to get all text snippets in a record."""
if self.text is not None and len(self.text):
yield self.text
def to_dict(self):
return {
'id': self.id,
'number': self.number,
'text': self.text,
'document_id': self.document_id
}
class DocumentPage(db.Model):
id = db.Column(db.BigInteger, primary_key=True)
number = db.Column(db.Integer(), nullable=False)
text = db.Column(db.Unicode(), nullable=False)
document_id = db.Column(db.Integer(),
db.ForeignKey('document.id'),
index=True) # noqa
document = db.relationship(
Document, backref=db.backref('pages',
cascade='all, delete-orphan')) # noqa
@property
def tid(self):
tid = sha1(str(self.document_id))
tid.update(str(self.id))
return tid.hexdigest()
def __repr__(self):
return '<DocumentPage(%r,%r)>' % (self.document_id, self.number)
def text_parts(self):
"""Utility method to get all text snippets in a record."""
text = string_value(self.text)
if text is not None:
yield self.text
def to_dict(self):
return {
'id': self.id,
'number': self.number,
'text': self.text,
'document_id': self.document_id
}
class DocumentRecord(db.Model):
id = db.Column(db.BigInteger, primary_key=True)
sheet = db.Column(db.Integer, nullable=False)
row_id = db.Column(db.Integer, nullable=False)
data = db.Column(JSONB)
document_id = db.Column(db.Integer(),
db.ForeignKey('document.id'),
index=True) # noqa
document = db.relationship(
Document, backref=db.backref('records',
cascade='all, delete-orphan')) # noqa
@property
def tid(self):
tid = sha1(str(self.document_id))
tid.update(str(self.sheet))
tid.update(str(self.row_id))
return tid.hexdigest()
def text_parts(self):
"""Utility method to get all text snippets in a record."""
for value in self.data.values():
text = string_value(value)
if text is not None:
yield value
def __repr__(self):
return '<DocumentRecord(%r,%r)>' % (self.document_id, self.row_id)
class Reference(db.Model, IdModel, DatedModel):
id = db.Column(db.Integer(), primary_key=True)
document_id = db.Column(db.BigInteger, db.ForeignKey('document.id'))
entity_id = db.Column(db.String(32), db.ForeignKey('entity.id'))
origin = db.Column(db.String(128))
weight = db.Column(db.Integer)
entity = db.relationship('Entity',
backref=db.backref('references', lazy='dynamic'))
document = db.relationship('Document',
backref=db.backref('references',
lazy='dynamic'))
def to_dict(self):
return {
'entity': {
'id': self.entity.id,
'name': self.entity.name,
'$schema': self.entity.type
},
'weight': self.weight,
'origin': self.origin
}
def __repr__(self):
return '<Reference(%r, %r)>' % (self.document_id, self.entity_id)
class DocumentRecord(db.Model):
id = db.Column(db.BigInteger, primary_key=True)
sheet = db.Column(db.Integer, nullable=False)
row_id = db.Column(db.Integer, nullable=False)
data = db.Column(JSONB)
document_id = db.Column(db.Integer(), db.ForeignKey('document.id'))
document = db.relationship(
Document, backref=db.backref('records',
cascade='all, delete-orphan')) # noqa
@property
def tid(self):
tid = sha1(str(self.document_id))
tid.update(str(self.sheet))
tid.update(str(self.row_id))
return tid.hexdigest()
@property
def text(self):
if self.data is None:
return []
text = [t for t in self.data.values() if t is not None]
return list(set(text))
def __repr__(self):
return '<DocumentRecord(%r,%r)>' % (self.document_id, self.row_id)
class IdModel(object):
id = db.Column(db.Integer(), primary_key=True)
def to_dict(self):
parent = super(IdModel, self)
data = parent.to_dict() if hasattr(parent, 'to_dict') else {}
data['id'] = self.id
return data
class DocumentTag(db.Model, IdModel):
"""A record reflects an entity or tag extracted from a document."""
TEXT_LENGTH = 1024
TYPE_PHONE = 'phone'
TYPE_EMAIL = 'email'
TYPE_PERSON = 'person'
TYPE_ORGANIZATION = 'organization'
TYPE_LOCATION = 'location'
TYPE_IP = 'ip'
TYPE_IBAN = 'iban'
TYPES = {
TYPE_PERSON: exactitude.names,
TYPE_ORGANIZATION: exactitude.names,
TYPE_EMAIL: exactitude.emails,
TYPE_PHONE: exactitude.phones,
TYPE_LOCATION: exactitude.addresses,
TYPE_IP: exactitude.ips,
TYPE_IBAN: exactitude.ibans,
}
id = db.Column(db.BigInteger, primary_key=True)
origin = db.Column(db.Unicode(255), nullable=False, index=True)
type = db.Column(db.Unicode(16), nullable=False)
weight = db.Column(db.Integer, default=1)
text = db.Column(db.Unicode(TEXT_LENGTH), nullable=True)
document_id = db.Column(db.Integer(),
db.ForeignKey('document.id'),
index=True) # noqa
document = db.relationship(
"Document", backref=db.backref('tags',
cascade='all, delete-orphan')) # noqa
@property
def field(self):
type_ = self.TYPES[self.type]
for (candidate, invert) in TYPES.values():
if candidate == type_:
return invert
@classmethod
def delete_by(cls, document_id=None, origin=None, type=None):
pq = db.session.query(cls)
assert document_id or origin or type
if document_id is not None:
pq = pq.filter(cls.document_id == document_id)
if origin is not None:
pq = pq.filter(cls.origin == origin)
if type is not None:
pq = pq.filter(cls.type == type)
pq.delete()
db.session.flush()
def __repr__(self):
return '<DocumentTag(%r,%r)>' % (self.document_id, self.text)
class DocumentTag(db.Model, IdModel):
"""A record reflects an entity or tag extracted from a document."""
TEXT_LENGTH = 1024
TYPE_PHONE = 'phone'
TYPE_EMAIL = 'email'
TYPE_PERSON = 'person'
TYPE_ORGANIZATION = 'organization'
TYPE_LOCATION = 'location'
TYPE_IP = 'ip'
TYPE_IBAN = 'iban'
TYPE_COUNTRY = 'country'
TYPE_LANGUAGE = 'language'
MAPPING = {
TYPE_PERSON: 'namesMentioned',
TYPE_ORGANIZATION: 'namesMentioned',
TYPE_EMAIL: 'emailMentioned',
TYPE_PHONE: 'phoneMentioned',
TYPE_LOCATION: 'locationMentioned',
TYPE_IP: 'ipMentioned',
TYPE_IBAN: 'ibanMentioned',
TYPE_COUNTRY: 'country',
TYPE_LANGUAGE: 'language'
}
id = db.Column(db.BigInteger, primary_key=True)
origin = db.Column(db.Unicode(255), nullable=False, index=True)
type = db.Column(db.Unicode(16), nullable=False)
weight = db.Column(db.Integer, default=1)
text = db.Column(db.Unicode(TEXT_LENGTH), nullable=True)
document_id = db.Column(db.Integer(), db.ForeignKey('document.id'), index=True) # noqa
document = db.relationship("Document", backref=db.backref('tags', cascade='all, delete-orphan')) # noqa
@property
def field(self):
type_ = registry.get(self.type)
if type_ is not None and type_.group is not None:
return type_.group
@classmethod
def delete_by(cls, document_id=None, origin=None, type=None):
pq = db.session.query(cls)
assert document_id or origin or type
if document_id is not None:
pq = pq.filter(cls.document_id == document_id)
if origin is not None:
pq = pq.filter(cls.origin == origin)
if type is not None:
pq = pq.filter(cls.type == type)
pq.delete()
db.session.flush()
def __repr__(self):
return '<DocumentTag(%r,%r)>' % (self.document_id, self.text)
class EntityIdentity(db.Model, IdModel, DatedModel):
CONFIRMED = 1
REJECTED = 2
UNDECIDED = 3
JUDGEMENTS = [1, 2, 3]
entity_id = db.Column(db.String(32),
db.ForeignKey('entity.id'),
index=True) # noqa
entity = db.relationship('Entity',
backref=db.backref('identities',
lazy='dynamic')) # noqa
match_id = db.Column(db.String(254), index=True, nullable=False)
judgement = db.Column(db.Integer(), nullable=False)
judge_id = db.Column(db.Integer, db.ForeignKey('role.id'), nullable=True)
@classmethod
def judgements_by_entity(cls, entity_id):
q = db.session.query(cls.match_id, cls.judgement)
q = q.filter(cls.entity_id == entity_id)
return {k: v for k, v in q.all()}
@classmethod
def entity_ids(cls, entity_id):
q = db.session.query(cls.match_id)
q = q.filter(cls.entity_id == entity_id)
q = q.filter(cls.judgement == cls.CONFIRMED)
ids = [entity_id]
for mapped_id, in q.all():
ids.append(mapped_id)
return ids
@classmethod
def by_entity_match(cls, entity_id, match_id):
q = db.session.query(cls)
q = q.filter(cls.entity_id == entity_id)
q = q.filter(cls.match_id == match_id)
return q.first()
@classmethod
def save(cls, entity_id, match_id, judgement, judge=None):
obj = cls.by_entity_match(entity_id, match_id)
if obj is None:
obj = cls()
obj.entity_id = entity_id
obj.match_id = match_id
obj.judgement = judgement
obj.judge = judge
db.session.add(obj)
return obj
def __repr__(self):
return 'EntityIdentity(%r, %r, %r)' % (self.entity_id, self.match_id,
self.judgement)
class DocumentPage(db.Model):
id = db.Column(db.BigInteger, primary_key=True)
number = db.Column(db.Integer(), nullable=False)
text = db.Column(db.Unicode(), nullable=False)
document_id = db.Column(db.Integer(), db.ForeignKey('document.id'))
document = db.relationship(
Document, backref=db.backref('pages',
cascade='all, delete-orphan')) # noqa
def __repr__(self):
return '<DocumentPage(%r,%r)>' % (self.document_id, self.number)
def to_dict(self):
return {
'id': self.id,
'number': self.number,
'text': self.text,
'document_id': self.document_id
}
class DocumentRecord(db.Model):
"""A record reflects a row or page of a document."""
id = db.Column(db.BigInteger, primary_key=True)
sheet = db.Column(db.Integer, nullable=True)
index = db.Column(db.Integer, nullable=True, index=True)
text = db.Column(db.Unicode, nullable=True)
data = db.Column(JSONB, nullable=True)
document_id = db.Column(db.Integer(),
db.ForeignKey('document.id'),
index=True) # noqa
document = db.relationship(
"Document",
backref=db.backref('records', cascade='all, delete-orphan')) # noqa
def text_parts(self):
"""Utility method to get all text snippets in a record."""
if self.data is not None:
for value in self.data.values():
text = string_value(value)
if text is not None:
yield text
text = string_value(self.text)
if text is not None:
yield text
@classmethod
def find_records(cls, document_id, ids):
if not len(ids):
return []
q = db.session.query(cls)
q = q.filter(cls.document_id == document_id)
q = q.filter(cls.id.in_(ids))
return q
def to_dict(self):
return {
'id': self.id,
'sheet': self.sheet,
'index': self.index,
'data': self.data,
'text': self.text,
'document_id': self.document_id
}
def __repr__(self):
return '<DocumentRecord(%r,%r)>' % (self.document_id, self.index)
class Reference(db.Model, IdModel, DatedModel):
id = db.Column(db.Integer(), primary_key=True)
origin = db.Column(db.String(128))
weight = db.Column(db.Integer)
document_id = db.Column(db.BigInteger,
db.ForeignKey('document.id'),
index=True) # noqa
document = db.relationship('Document',
backref=db.backref('references',
lazy='dynamic')) # noqa
entity_id = db.Column(db.String(32),
db.ForeignKey('entity.id'),
index=True) # noqa
entity = db.relationship('Entity',
backref=db.backref('references',
lazy='dynamic')) # noqa
@classmethod
def index_references(cls, document_id):
"""Helper function to get reference data for indexing."""
# cf. aleph.index.entities.generate_entities()
from aleph.model.entity import Entity
q = db.session.query(Reference.entity_id, Entity.collection_id)
q = q.filter(Reference.document_id == document_id)
q = q.filter(Entity.id == Reference.entity_id)
q = q.filter(Entity.state == Entity.STATE_ACTIVE)
return q.all()
def to_dict(self):
return {
'entity': {
'id': self.entity.id,
'name': self.entity.name,
'$schema': self.entity.type
},
'weight': self.weight,
'origin': self.origin
}
def __repr__(self):
return '<Reference(%r, %r)>' % (self.document_id, self.entity_id)
class Reference(db.Model, IdModel, DatedModel):
id = db.Column(db.Integer(), primary_key=True)
document_id = db.Column(db.BigInteger, db.ForeignKey('document.id'))
entity_id = db.Column(db.String(32), db.ForeignKey('entity.id'))
weight = db.Column(db.Integer)
entity = db.relationship(Entity,
backref=db.backref('references', lazy='dynamic'))
document = db.relationship(Document,
backref=db.backref('references',
lazy='dynamic'))
@classmethod
def delete_document(cls, document_id):
q = cls.all().filter_by(document_id=document_id)
q.delete(synchronize_session='fetch')
def __repr__(self):
return '<Reference(%r, %r)>' % (self.document_id, self.entity_id)
class DocumentRecord(db.Model):
"""A record reflects a row or page of a document."""
id = db.Column(db.BigInteger, primary_key=True)
sheet = db.Column(db.Integer, nullable=True)
index = db.Column(db.Integer, nullable=True, index=True)
text = db.Column(db.Unicode, nullable=True)
data = db.Column(JSONB, nullable=True)
document_id = db.Column(db.Integer(),
db.ForeignKey('document.id'),
index=True) # noqa
document = db.relationship(
"Document",
backref=db.backref('records', cascade='all, delete-orphan')) # noqa
@property
def texts(self):
"""Utility method to get all text snippets in a record."""
if self.data is not None:
for value in self.data.values():
yield value
yield self.text
@classmethod
def find_records(cls, ids):
if not len(ids):
return []
q = db.session.query(cls)
q = q.filter(cls.id.in_(ids))
return q
@classmethod
def by_index(cls, document_id, index):
q = db.session.query(cls)
q = db.session.query(DocumentRecord)
q = q.filter(cls.document_id == document_id)
q = q.filter(cls.index == index)
return q.first()
def __repr__(self):
return '<DocumentRecord(%r,%r)>' % (self.document_id, self.index)
class EntityTag(db.Model):
id = db.Column(db.Integer(), primary_key=True)
collection = db.Column(db.Unicode(100))
package_id = db.Column(db.Unicode(100))
entity_id = db.Column(db.Unicode(50), db.ForeignKey('entity.id'))
entity = db.relationship(Entity,
backref=db.backref('tags', lazy='dynamic'))
created_at = db.Column(db.DateTime, default=datetime.utcnow)
@classmethod
def delete_set(cls, collection, package_id):
q = db.session.query(cls)
q = q.filter_by(collection=collection)
q = q.filter_by(package_id=package_id)
q.delete()
@classmethod
def by_package(cls, collection, package_id):
etag = aliased(cls)
ent = aliased(Entity)
q = db.session.query(etag.entity_id, ent.label, ent.category,
ent.list_id)
q = q.join(ent, ent.id == etag.entity_id)
q = q.filter(etag.collection == collection)
q = q.filter(etag.package_id == package_id)
entities = []
for entity_id, label, category, lst in q.all():
entities.append({
'id': entity_id,
'entity': entity_id,
'label': label,
'category': category,
'list': lst
})
return entities
def __repr__(self):
return '<EntityTag(%r, %r)>' % (self.package_id, self.entity_id)
class DocumentRecord(db.Model):
id = db.Column(db.BigInteger, primary_key=True)
sheet = db.Column(db.Integer, nullable=False)
row_id = db.Column(db.Integer, nullable=False)
data = db.Column(JSONB)
document_id = db.Column(db.Integer(),
db.ForeignKey('document.id'),
index=True) # noqa
document = db.relationship(
Document, backref=db.backref('records',
cascade='all, delete-orphan')) # noqa
@property
def tid(self):
tid = sha1(str(self.document_id))
tid.update(str(self.sheet))
tid.update(str(self.row_id))
return tid.hexdigest()
def text_parts(self):
"""Utility method to get all text snippets in a record."""
for value in self.data.values():
text = string_value(value)
if text is not None:
yield value
@classmethod
def find_rows(cls, document_id, rows):
if not len(rows):
return []
q = db.session.query(cls)
q = q.filter(cls.document_id == document_id)
clauses = [and_(cls.sheet == r[0], cls.row_id == r[1]) for r in rows]
q = q.filter(or_(*clauses))
return q
def __repr__(self):
return '<DocumentRecord(%r,%r)>' % (self.document_id, self.row_id)
class DocumentTag(db.Model, IdModel):
"""A record reflects an entity or tag extracted from a document."""
TYPE_PHONE = 'phone'
TYPE_EMAIL = 'email'
TYPE_PERSON = 'person'
TYPE_ORGANIZATION = 'organization'
TYPE_LOCATION = 'location'
id = db.Column(db.BigInteger, primary_key=True)
origin = db.Column(db.Unicode(255), nullable=False, index=True)
type = db.Column(db.Unicode(16), nullable=False)
weight = db.Column(db.Integer, default=1)
key = db.Column(db.Unicode(1024), nullable=False, index=True)
text = db.Column(db.Unicode(1024), nullable=True)
document_id = db.Column(db.Integer(),
db.ForeignKey('document.id'),
index=True) # noqa
document = db.relationship(
"Document", backref=db.backref('tags',
cascade='all, delete-orphan')) # noqa
@classmethod
def delete_by(cls, document_id=None, origin=None, type=None):
pq = db.session.query(cls)
assert document_id or origin or type
if document_id is not None:
pq = pq.filter(cls.document_id == document_id)
if origin is not None:
pq = pq.filter(cls.origin == origin)
if type is not None:
pq = pq.filter(cls.type == type)
pq.delete()
db.session.flush()
def __repr__(self):
return '<DocumentTag(%r,%r)>' % (self.document_id, self.key)
class DocumentRecord(db.Model):
"""A record reflects a row or page of a document."""
SCHEMA_ROW = 'Row'
SCHEMA_PAGE = 'Page'
id = db.Column(db.BigInteger, primary_key=True)
index = db.Column(db.Integer, nullable=True, index=True)
text = db.Column(db.Unicode, nullable=True)
data = db.Column(JSONB, nullable=True)
document_id = db.Column(db.Integer(),
db.ForeignKey('document.id'),
index=True) # noqa
document = db.relationship(
"Document",
backref=db.backref('records', cascade='all, delete-orphan')) # noqa
def raw_texts(self):
"""Utility method to get all text snippets in a record."""
if self.data is not None:
for value in self.data.values():
yield value
yield self.text
@property
def texts(self):
yield from filter_texts(self.raw_texts())
@classmethod
def insert_records(cls, document_id, iterable, chunk_size=1000):
chunk = []
table = cls.__table__
for index, data in enumerate(iterable):
chunk.append({
'document_id': document_id,
'index': index,
'data': data
})
if len(chunk) >= chunk_size:
q = table.insert().values(chunk)
db.session.execute(q)
chunk = []
if len(chunk):
q = table.insert().values(chunk)
db.session.execute(q)
def to_proxy(self):
if self.text is not None:
proxy = model.make_entity(self.SCHEMA_PAGE)
proxy.make_id('record', self.id)
proxy.set('document', self.document_id)
proxy.set('index', self.index)
proxy.set('bodyText', stringify(self.text))
return proxy
else:
proxy = model.make_entity(self.SCHEMA_ROW)
proxy.make_id('record', self.id)
proxy.set('table', self.document_id)
proxy.set('index', self.index)
if self.data is not None:
values = [v for (k, v) in sorted(self.data.items())]
proxy.set('cells', registry.json.pack(values))
return proxy
def to_dict(self):
proxy = self.to_proxy()
data = proxy.to_full_dict()
data.update({
'document_id': self.document_id,
'bulk': False,
})
return data
def __repr__(self):
return '<DocumentRecord(%r,%r)>' % (self.document_id, self.index)
class Entity(db.Model):
id = db.Column(db.Unicode(50), primary_key=True, default=make_textid)
label = db.Column(db.Unicode)
category = db.Column(db.Enum(*CATEGORIES, name='entity_categories'),
nullable=False)
creator_id = db.Column(db.Integer(), db.ForeignKey('user.id'))
creator = db.relationship(User,
backref=db.backref('entities',
lazy='dynamic',
cascade='all, delete-orphan'))
list_id = db.Column(db.Integer(), db.ForeignKey('list.id'))
list = db.relationship('List',
backref=db.backref('entities',
lazy='dynamic',
cascade='all, delete-orphan'))
created_at = db.Column(db.DateTime, default=datetime.utcnow)
updated_at = db.Column(db.DateTime,
default=datetime.utcnow,
onupdate=datetime.utcnow)
def to_dict(self):
return {
'id': self.id,
'api_url': url_for('entities.view', id=self.id),
'label': self.label,
'category': self.category,
'creator_id': self.creator_id,
'selectors': [s.text for s in self.selectors],
'list': self.list_id,
'created_at': self.created_at,
'updated_at': self.updated_at
}
def has_selector(self, text):
normalized = Selector.normalize(text)
for selector in self.selectors:
if selector.normalized == normalized:
return True
return False
def delete(self):
db.session.delete(self)
@classmethod
def create(cls, data, user):
ent = cls()
ent.update(data)
ent.creator = user
db.session.add(ent)
return ent
def update(self, data):
data = EntityForm().deserialize(data)
self.label = data.get('label')
self.list = data.get('list')
self.category = data.get('category')
selectors = set(data.get('selectors'))
selectors.add(self.label)
existing = list(self.selectors)
for sel in list(existing):
if sel.text in selectors:
selectors.remove(sel.text)
existing.remove(sel)
for sel in existing:
db.session.delete(sel)
for text in selectors:
sel = Selector()
sel.entity = self
sel.text = text
db.session.add(sel)
@classmethod
def by_normalized_label(cls, label, lst):
q = db.session.query(cls)
q = q.filter_by(list=lst)
q = q.filter(db_compare(cls.label, label))
return q.first()
@classmethod
def by_id(cls, id):
q = db.session.query(cls).filter_by(id=id)
return q.first()
@classmethod
def by_lists(cls, lists, prefix=None):
q = db.session.query(cls)
q = q.filter(cls.list_id.in_(lists))
if prefix is not None and len(prefix):
q = q.join(Selector, cls.id == Selector.entity_id)
q = cls.apply_filter(q, Selector.normalized, prefix)
q = q.order_by(cls.label.asc())
return q
@classmethod
def by_id_set(cls, ids):
if not len(ids):
return {}
q = db.session.query(cls)
q = q.filter(cls.id.in_(ids))
entities = {}
for ent in q:
entities[ent.id] = ent
return entities
@classmethod
def apply_filter(cls, q, col, prefix):
prefix = Selector.normalize(prefix)
return q.filter(
or_(col.like('%s%%' % prefix), col.like('%% %s%%' % prefix)))
@classmethod
def suggest_prefix(cls, prefix, lists, limit=10):
from aleph.model import EntityTag
ent = aliased(Entity)
sel = aliased(Selector)
tag = aliased(EntityTag)
q = db.session.query(ent.id, ent.label, ent.category)
q = q.join(sel, ent.id == sel.entity_id)
q = q.join(tag, ent.id == tag.entity_id)
q = q.filter(ent.list_id.in_(lists))
if prefix is None or not len(prefix):
return []
q = cls.apply_filter(q, sel.normalized, prefix)
q = q.order_by(ent.label.asc())
q = q.limit(limit)
q = q.distinct()
suggestions = []
for entity_id, label, category in q.all():
suggestions.append({
'id': entity_id,
'label': label,
'category': category
})
return suggestions
@property
def terms(self):
return set([s.normalized for s in self.selectors])
def __repr__(self):
return '<Entity(%r, %r)>' % (self.id, self.label)
def __unicode__(self):
return self.label
class List(db.Model):
id = db.Column(db.Integer(), primary_key=True)
label = db.Column(db.Unicode)
public = db.Column(db.Boolean, default=False)
creator_id = db.Column(db.Integer(),
db.ForeignKey('user.id'),
nullable=True)
creator = db.relationship(User)
created_at = db.Column(db.DateTime, default=datetime.utcnow)
updated_at = db.Column(db.DateTime,
default=datetime.utcnow,
onupdate=datetime.utcnow)
users = db.relationship(User, secondary=list_user_table, backref='lists')
def to_dict(self):
return {
'id': self.id,
'api_url': url_for('lists.view', id=self.id),
'entities_api_url': url_for('entities.index', list=self.id),
'label': self.label,
'public': self.public,
'creator_id': self.creator_id,
'created_at': self.created_at,
'updated_at': self.updated_at
}
@classmethod
def create(cls, data, user):
lst = cls()
lst.update(data, user)
lst.creator = user
db.session.add(lst)
return lst
def update(self, data, user):
data = ListForm().deserialize(data)
self.label = data.get('label')
if data.get('public') is not None:
self.public = data.get('public')
users = set(data.get('users', []))
if user is not None:
users.add(user)
self.users = list(users)
def delete(self):
# for entity in self.entities:
# entity.delete()
db.session.delete(self)
@classmethod
def by_label(cls, label):
q = db.session.query(cls).filter_by(label=label)
return q.first()
@classmethod
def by_id(cls, id):
q = db.session.query(cls).filter_by(id=id)
return q.first()
@classmethod
def user_list_ids(cls, user=None, include_public=True):
logged_in = user is not None and user.is_authenticated()
q = db.session.query(cls.id)
conds = []
if include_public:
conds.append(cls.public == True) # noqa
if logged_in:
conds.append(cls.users.any(User.id == user.id))
if not len(conds):
return []
if not (logged_in and user.is_admin):
q = q.filter(or_(*conds))
return [c.id for c in q.all()]
@classmethod
def all_by_user(cls, user):
q = db.session.query(cls)
q = q.filter(cls.id.in_(cls.user_list_ids(user)))
q = q.order_by(cls.id.desc())
return q
@property
def terms(self):
from aleph.model.entity import Entity
from aleph.model.selector import Selector
q = db.session.query(Selector.normalized)
q = q.join(Entity, Entity.id == Selector.entity_id)
q = q.filter(Entity.list_id == self.id)
q = q.distinct()
return set([r[0] for r in q])
def __repr__(self):
return '<List(%r, %r)>' % (self.id, self.label)
def __unicode__(self):
return self.label
class Entity(db.Model, UuidModel, SoftDeleteModel, SchemaModel):
_schema = '/entity/entity.json#'
_schema_recurse = True
name = db.Column(db.Unicode)
type = db.Column('type', db.String(255), index=True)
summary = db.Column(db.Unicode, nullable=True)
description = db.Column(db.Unicode, nullable=True)
jurisdiction_code = db.Column(db.Unicode, nullable=True)
__mapper_args__ = {'polymorphic_on': type, 'polymorphic_identity': _schema}
collection_id = db.Column(db.Integer(), db.ForeignKey('collection.id'))
collection = db.relationship(Collection,
backref=db.backref(
'entities',
lazy='dynamic',
cascade='all, delete-orphan')) # noqa
def delete(self):
from aleph.model import Reference
q = db.session.query(Reference)
q = q.filter(Reference.entity_id == self.id)
q.delete(synchronize_session='fetch')
super(Entity, self).delete()
def update(self, data, merge=False):
self.schema_update(data, merge=merge)
@classmethod
def save(cls, data, collection_id=None, merge=False):
ent = cls.by_id(data.get('id'))
for identifier in data.get('identifiers', []):
if ent is None:
ent = cls.by_identifier(identifier.get('scheme'),
identifier.get('identifier'),
collection_id=collection_id)
if ent is None:
schema = data.get('$schema', cls._schema)
cls = cls.get_schema_class(schema)
ent = cls()
ent.id = make_textid()
if collection_id is not None:
ent.collection_id = collection_id
ent.update(data, merge=merge)
return ent
@property
def terms(self):
terms = set([self.name])
# for other_name in self.other_names:
# terms.update(other_name.terms)
return [t for t in terms if t is not None and len(t)]
@classmethod
def by_identifier(cls, scheme, identifier, collection_id=None):
ent = aliased(Entity)
q = db.session.query(ent)
q = q.filter(ent.deleted_at == None) # noqa
if collection_id is not None:
q = q.filter(ent.collection_id == collection_id)
ident = aliased(EntityIdentifier)
q = q.join(ident, ent.identifiers)
q = q.filter(ident.deleted_at == None) # noqa
q = q.filter(ident.scheme == scheme)
q = q.filter(ident.identifier == identifier)
return q.first()
@classmethod
def by_id_set(cls, ids, collection_id=None):
if not len(ids):
return {}
q = cls.all()
q = q.filter(cls.id.in_(ids))
if collection_id is not None:
q = q.filter(cls.collection_id == collection_id)
entities = {}
for ent in q:
entities[ent.id] = ent
return entities
@classmethod
def suggest_prefix(cls, prefix, collections, limit=10):
if prefix is None or not len(prefix):
return []
prefix = prefix.strip()
ent = aliased(Entity)
q = db.session.query(ent.id, ent.name, ent.type)
q = q.filter(ent.deleted_at == None) # noqa
q = q.filter(ent.collection_id.in_(collections))
q = q.filter(
or_(ent.name.ilike('%s%%' % prefix),
ent.name.ilike('%% %s%%' % prefix)))
q = q.limit(limit)
suggestions = []
for entity_id, name, schema in q.all():
suggestions.append({
'id': entity_id,
'name': name,
'$schema': schema
})
return suggestions
def __repr__(self):
return '<Entity(%r, %r)>' % (self.id, self.name)
def __unicode__(self):
return self.name
def to_dict(self):
data = super(Entity, self).to_dict()
data['collection_id'] = self.collection_id
return data
class DocumentRecord(db.Model):
"""A record reflects a row or page of a document."""
id = db.Column(db.BigInteger, primary_key=True)
sheet = db.Column(db.Integer, nullable=True)
index = db.Column(db.Integer, nullable=True, index=True)
text = db.Column(db.Unicode, nullable=True)
data = db.Column(JSONB, nullable=True)
document_id = db.Column(db.Integer(),
db.ForeignKey('document.id'),
index=True) # noqa
document = db.relationship(
"Document",
backref=db.backref('records', cascade='all, delete-orphan')) # noqa
def raw_texts(self):
"""Utility method to get all text snippets in a record."""
if self.data is not None:
for value in self.data.values():
yield value
yield self.text
@property
def texts(self):
yield from filter_texts(self.raw_texts())
@classmethod
def insert_records(cls, document_id, iterable, chunk_size=1000):
chunk = []
table = cls.__table__
for index, data in enumerate(iterable):
chunk.append({
'document_id': document_id,
'index': index,
'data': data
})
if len(chunk) >= chunk_size:
q = table.insert().values(chunk)
db.session.execute(q)
chunk = []
if len(chunk):
q = table.insert().values(chunk)
db.session.execute(q)
@classmethod
def find_records(cls, ids):
if not len(ids):
return []
q = db.session.query(cls)
q = q.filter(cls.id.in_(ids))
return q
@classmethod
def by_index(cls, document_id, index):
q = db.session.query(cls)
q = db.session.query(DocumentRecord)
q = q.filter(cls.document_id == document_id)
q = q.filter(cls.index == index)
return q.first()
def __repr__(self):
return '<DocumentRecord(%r,%r)>' % (self.document_id, self.index)
class IdModel(object):
id = db.Column(db.Integer(), primary_key=True)
class Role(db.Model, RoleMixin):
id = db.Column(db.Integer(), primary_key=True)
name = db.Column(db.String(80), unique=True)
description = db.Column(db.String(255))
from aleph.model.util import make_token
from aleph.model.forms import UserForm
from flask.ext.security import Security, SQLAlchemyUserDatastore, \
UserMixin, RoleMixin, login_required
from flask.ext.security.utils import encrypt_password, get_hmac
log = logging.getLogger(__name__)
@login_manager.user_loader
def load_user(id):
return User.query.get(int(id))
roles_users = db.Table(
'roles_users', db.Column('user_id', db.Integer(),
db.ForeignKey('user.id')),
db.Column('role_id', db.Integer(), db.ForeignKey('role.id')))
class Role(db.Model, RoleMixin):
id = db.Column(db.Integer(), primary_key=True)
name = db.Column(db.String(80), unique=True)
description = db.Column(db.String(255))
class User(db.Model):
id = db.Column(db.Integer, primary_key=True)
email = db.Column(
db.Unicode,
#following attributes are for flask-user
class CrawlerState(db.Model):
"""Report the state of a file being processed."""
TIMEOUT = timedelta(minutes=60)
STATUS_OK = 'ok'
STATUS_FAIL = 'fail'
id = db.Column(db.BigInteger, primary_key=True)
crawler_id = db.Column(db.Unicode(), index=True)
crawler_run = db.Column(db.Unicode(), nullable=True)
content_hash = db.Column(db.Unicode(65), nullable=True)
foreign_id = db.Column(db.Unicode, nullable=True)
status = db.Column(db.Unicode(10), nullable=False)
error_type = db.Column(db.Unicode(), nullable=True)
error_message = db.Column(db.Unicode(), nullable=True)
error_details = db.Column(db.Unicode(), nullable=True)
meta = db.Column(JSONB)
collection_id = db.Column(db.Integer(),
db.ForeignKey('collection.id'),
index=True)
collection = db.relationship(Collection,
backref=db.backref(
'crawl_states',
cascade='all, delete-orphan')) # noqa
created_at = db.Column(db.DateTime, default=datetime.utcnow)
@classmethod
def _from_meta(cls, meta, collection_id):
obj = cls()
obj.collection_id = collection_id
obj.crawler_id = meta.crawler
obj.crawler_run = meta.crawler_run
obj.foreign_id = meta.foreign_id
obj.content_hash = meta.content_hash
obj.meta = expand_json(meta.to_attr_dict(compute=True))
db.session.add(obj)
return obj
@classmethod
def store_stub(cls, collection_id, crawler_id, crawler_run):
obj = cls()
obj.collection_id = collection_id
obj.crawler_id = crawler_id
obj.crawler_run = crawler_run
obj.error_type = 'init'
obj.status = cls.STATUS_OK
db.session.add(obj)
return obj
@classmethod
def store_ok(cls, meta, collection_id):
obj = cls._from_meta(meta, collection_id)
obj.status = cls.STATUS_OK
return obj
@classmethod
def store_fail(cls,
meta,
collection_id,
error_type=None,
error_message=None,
error_details=None):
obj = cls._from_meta(meta, collection_id)
obj.status = cls.STATUS_FAIL
obj.error_type = error_type
obj.error_message = error_message
obj.error_details = error_details
return obj
@classmethod
def crawler_last_run(cls, crawler_id):
q = db.session.query(cls.crawler_run, cls.created_at)
q = q.filter(cls.crawler_id == crawler_id)
q = q.order_by(cls.created_at.desc())
q = q.limit(1)
res = q.first()
if res is None:
return None, None
return (res.crawler_run, res.created_at)
@classmethod
def crawler_stats(cls, crawler_id):
stats = {}
last_run_id, last_run_time = cls.crawler_last_run(crawler_id)
# Check if the crawler was active very recently, if so, don't
# allow the user to execute a new run right now.
timeout = (datetime.utcnow() - CrawlerState.TIMEOUT)
stats['running'] = last_run_time > timeout if last_run_time else False
q = db.session.query(func.count(cls.id))
q = q.filter(cls.crawler_id == crawler_id)
for section in ['last', 'all']:
data = {}
sq = q
if section == 'last':
sq = sq.filter(cls.crawler_run == last_run_id)
okq = sq.filter(cls.status == cls.STATUS_OK)
data['ok'] = okq.scalar() if last_run_id else 0
failq = sq.filter(cls.status == cls.STATUS_FAIL)
data['fail'] = failq.scalar() if last_run_id else 0
stats[section] = data
stats['last']['updated'] = last_run_time
stats['last']['run_id'] = last_run_id
return stats
@classmethod
def all(cls):
return db.session.query(CrawlerState)
def to_dict(self):
return {
'id': self.id,
'status': self.status,
'crawler_id': self.crawler_id,
'crawler_run': self.crawler_run,
'content_hash': self.content_hash,
'foreign_id': self.foreign_id,
'error_type': self.error_type,
'error_message': self.error_message,
'error_details': self.error_details,
'meta': self.meta,
'collection_id': self.collection_id,
'created_at': self.created_at
}
def __repr__(self):
return '<CrawlerState(%r,%r)>' % (self.id, self.status)
def __unicode__(self):
return self.id
class Alert(db.Model):
'''
Also consider adding:
- active/inactive
- label (short human-readable text)
'''
id = db.Column(db.Integer, primary_key=True)
user_id = db.Column(db.Integer(), db.ForeignKey('user.id'))
user = db.relationship(User, backref=db.backref('alerts'))
created_at = db.Column(db.DateTime, default=datetime.utcnow)
checked_at = db.Column(
db.DateTime,
default=None,
)
query = db.Column(db.Unicode)
label = db.Column(db.Unicode)
checking_interval = db.Column(db.Integer, default=None)
# number of days between checks. None == 'never check'
def due_to_check(self):
'''
Return True if it is time to run this query
NB We expect this script to run at nearly-but-not-precisely
the same time each day, and we want to run at an intuitive
'once per day', rather than skipping because today's run
has happened a few seconds earlier.
Therefore we allow 2 hours of wiggle room
[we aren't worried about sending duplicate alerts, because that
will be handled precisely by filtering result insert dates against
the checked_at field
'''
if self.checking_interval == None: # query is disabled
return False
if self.checked_at == None: # query is being run for the first time
return True
min_check_date = datetime.utcnow() - timedelta(
days=self.checking_interval) + timedelta(hours=2)
return self.checked_at <= min_check_date
def mark_as_checked(self):
self.checked_at = datetime.utcnow()
db.session.add(self)
db.session.commit()
def to_dict(self):
attrs = ('id', 'label', 'query', 'checking_interval', 'user_id',
'created_at', 'checked_at')
return {attr: getattr(self, attr) for attr in attrs}
@property
def search_url(self):
'''
where to go to reach the original search
'''
return 'http://search.openoil.net/#/search?q=' + urllib.parse.quote_plus(
self.query)
@classmethod
def by_id(cls, id, role=None):
q = db.session.query(cls).filter_by(id=id)
if role is not None: #only applies if we are using authz roles
q = q.filter(cls.role_id == role.id)
return q.first()
class Document(db.Model, DatedModel):
TYPE_TEXT = 'text'
TYPE_TABULAR = 'tabular'
TYPE_OTHER = 'other'
id = db.Column(db.BigInteger, primary_key=True)
content_hash = db.Column(db.Unicode(65), nullable=False, index=True)
foreign_id = db.Column(db.Unicode, unique=False, nullable=True)
type = db.Column(db.Unicode(10), nullable=False, index=True)
source_id = db.Column(db.Integer(), db.ForeignKey('source.id'), index=True)
source = db.relationship(Source,
backref=db.backref(
'documents',
lazy='dynamic',
cascade='all, delete-orphan')) # noqa
_meta = db.Column('meta', JSONB)
@property
def title(self):
return self.meta.title
@hybrid_property
def meta(self):
self._meta = self._meta or {}
self._meta['content_hash'] = self.content_hash
self._meta['foreign_id'] = self.foreign_id
return Metadata(data=self._meta or {})
@meta.setter
def meta(self, meta):
if isinstance(meta, Metadata):
self.content_hash = meta.content_hash
self.foreign_id = meta.foreign_id
meta = meta.data
self._meta = meta
flag_modified(self, '_meta')
def delete_pages(self):
pq = db.session.query(DocumentPage)
pq = pq.filter(DocumentPage.document_id == self.id)
pq.delete(synchronize_session='fetch')
db.session.refresh(self)
def delete_records(self):
pq = db.session.query(DocumentRecord)
pq = pq.filter(DocumentRecord.document_id == self.id)
pq.delete(synchronize_session='fetch')
db.session.refresh(self)
def insert_records(self, sheet, iterable, chunk_size=1000):
chunk = []
for i, data in enumerate(iterable):
chunk.append({
'document_id': self.id,
'row_id': i,
'sheet': sheet,
'data': data
})
if len(chunk) >= chunk_size:
db.session.bulk_insert_mappings(DocumentRecord, chunk)
chunk = []
if len(chunk):
db.session.bulk_insert_mappings(DocumentRecord, chunk)
def text_parts(self):
"""Utility method to get all text snippets in a document."""
if self.type == Document.TYPE_TEXT:
for page in self.pages:
if page.text is not None and len(page.text):
yield page.text, page
if self.type == Document.TYPE_TABULAR:
for record in self.records:
for value in record.data.values():
if isinstance(value, basestring) and len(value):
yield value, record
@classmethod
def get_max_id(cls):
q = db.session.query(func.max(cls.id))
return q.scalar()
def __repr__(self):
return '<Document(%r,%r,%r)>' % (self.id, self.type, self.meta.title)
def _add_to_dict(self, data):
data.update({
'id': self.id,
'type': self.type,
'source_id': self.source_id,
'created_at': self.created_at,
'updated_at': self.updated_at
})
return data
def to_dict(self):
data = self.meta.to_dict()
return self._add_to_dict(data)
def to_index_dict(self):
data = self.meta.to_index_dict()
return self._add_to_dict(data)
