def _outputFullTable(self,pages):
csvData = []
for page in self.document.pages:
for table in page.tables:
csvRow = []
#csvRow.append("Table")
csvData.append(csvRow)
for row in table.rows:
csvRow = []
for cell in row.cells:
csvRow.append(cell.text)
csvData.append(csvRow)
csvData.append([])
csvData.append([])
opath = "{}page-{}-tables.csv".format(self.outputPath, 'Full')
S3Helper.writeCSVRaw(csvData, self.bucketName, opath)
def _outputTable(self, page, p):
csvData = []
for table in page.tables:
csvRow = []
csvRow.append("Table")
csvData.append(csvRow)
for row in table.rows:
csvRow = []
for cell in row.cells:
csvRow.append(cell.text)
csvData.append(csvRow)
csvData.append([])
csvData.append([])
opath = "{}page-{}-tables.csv".format(self.outputPath, p)
S3Helper.writeCSVRaw(csvData, self.bucketName, opath)
self.saveItem(self.documentId, "page-{}-Tables".format(p), opath)
def _outputTable(self, page, p, no_write=False):
csvData = []
for table in page.tables:
csvRow = []
csvRow.append("Table")
csvData.append(csvRow)
for row in table.rows:
csvRow = []
for cell in row.cells:
csvRow.append(cell.text)
csvData.append(csvRow)
csvData.append([])
csvData.append([])
if no_write:
return csvData
else:
opath = "{}/page-{}/tables.csv".format(self.outputPath, p)
S3Helper.writeCSVRaw(csvData, self.bucketName, opath)
def _outputTable(self, page, p):
page_number = self.metadata['page_number']
csvData = []
for table in page.tables:
csvRow = []
csvRow.append("Table")
csvData.append(csvRow)
for row in table.rows:
csvRow = []
for cell in row.cells:
csvRow.append(cell.text)
csvData.append(csvRow)
csvData.append([])
csvData.append([])
opath = "{}page-{}-tables.csv".format(self.outputPath, page_number)
S3Helper.writeCSVRaw(csvData, self.bucketName, opath)
return opath
