def create_page(self, document, text, number=1):
record = DocumentRecord()
record.document_id = document.id
record.text = text
record.index = number
db.session.add(record)
return record
def emit_page(self, index, text):
"""Emit a plain text page."""
record = DocumentRecord()
record.document_id = self.document.id
record.text = safe_string(text)
record.index = index
db.session.add(record)
def emit_page(self, index, text):
"""Emit a plain text page."""
record = DocumentRecord()
record.document_id = self.document.id
record.text = safe_string(text)
record.index = index
db.session.add(record)
def emit_page(self, index, text):
"""Emit a plain text page."""
self.document.type = Document.TYPE_TEXT
record = DocumentRecord()
record.document_id = self.document.id
record.text = text
record.index = index
db.session.add(record)
def emit_page(self, index, text):
"""Emit a plain text page."""
text = stringify(text)
record = DocumentRecord()
record.document_id = self.document.id
record.text = text
record.index = index
db.session.add(record)
if text is not None:
self.pages.append(text)
def record(document_id, index):
enable_cache()
document = get_document(document_id)
record = DocumentRecord.by_index(document.id, index)
if record is None:
raise NotFound("No such record: %s" % index)
return jsonify(record, schema=RecordSchema)
def __init__(self, request, parser, result, schema=None):
super(RecordsQueryResult, self).__init__(request, parser, result,
schema=schema)
ids = [res.get('id') for res in self.results]
for record in DocumentRecord.find_records(ids):
for result in self.results:
if result['id'] == record.id:
result['data'] = record.data
result['text'] = record.text
def record(document_id, index):
enable_cache()
document = get_document(document_id)
if not document.has_records():
raise BadRequest("This document does not have records.")
record = DocumentRecord.by_index(document.id, index)
if record is None:
raise NotFound("No such record: %s" % index)
return jsonify(record, schema=RecordSchema)
def record(document_id, index):
enable_cache()
document = get_db_document(document_id)
if not document.supports_records:
raise BadRequest("This document does not have records.")
record = DocumentRecord.by_index(document.id, index)
if record is None:
raise NotFound("No such record: %s" % index)
return serialize_data(record, RecordSchema)
def execute_records_query(document_id, state, query):
"""Execute a query against records and return a set of results."""
result, hits, output = execute_basic(TYPE_RECORD, query)
ids = []
for rec in hits.get('hits', []):
record = rec.get('_source')
record['score'] = rec.get('_score')
record['id'] = int(rec.get('_id'))
ids.append(rec.get('_id'))
for text in rec.get('highlight', {}).get('text', []):
record['text'] = text
output['results'].append(record)
for record in DocumentRecord.find_records(document_id, ids):
for result in output['results']:
if result['id'] == record.id:
result['data'] = record.data
result['text'] = record.text
return output
def execute_records_query(document_id, state, query):
"""Execute a query against records and return a set of results."""
result, hits, output = execute_basic(TYPE_RECORD, query)
rows = []
for rec in hits.get('hits', []):
record = rec.get('_source')
record['score'] = rec.get('_score')
if record.get('row_id'):
rows.append((record.get('sheet'), record.get('row_id')))
for text in rec.get('highlight', {}).get('text', []):
record['text'] = text
output['results'].append(record)
for record in DocumentRecord.find_rows(document_id, rows):
for res in output['results']:
if res['sheet'] == record.sheet and res['row_id'] == record.row_id:
res['data'] = record.data
return output
def emit_rows(self, iterator):
"""Emit rows of a tabular iterator."""
iterator = self._emit_iterator_rows(iterator)
DocumentRecord.insert_records(self.document.id, iterator)
def emit_rows(self, iterator):
"""Emit rows of a tabular iterator."""
iterator = self._emit_iterator_rows(iterator)
DocumentRecord.insert_records(self.document.id, iterator)