Begginnings of a data generation suite for OMERO/HIC datasets & schemas

generator.py

from xml.sax.saxutils import XMLGenerator
from xml.sax.xmlreader import AttributesNSImpl
import hashlib

class SchemaGen:
	def __init__(self, output, encoding, projectID, initial_prochi):
		"""
		
		"""
		anonschema = XMLGenerator(output, encoding)
		anonschema.startDocument()
		
		self._anonschema = anonschema
		self._output = output
		self._encoding = encoding
		self._projectID = projectID
		self._initial_prochi = initial_prochi
		return

	def writeFooter(self):
		"""
		
		"""
		self._anonschema.endElementNS((None, u'project_anonymisation'), u'project_anonymisation')
		self._anonschema.endDocument()
		return

	def writeHeader(self):
		"""
		
		"""
		attr_vals = {
		    (None, u'id'): self._projectID,
		    }
		attr_qnames = {
		    (None, u'id'): self._projectID,
		    }
		attrs = AttributesNSImpl(attr_vals, attr_qnames)
		
		self._anonschema.startElementNS((None, u'project_anonymisation'), u'project_anonymisation', attrs)
		return

	def writeRow(self):
		"""
		
		"""
		prochi = self.generate_prochi()
		anochihash =  self.generate_anochihash(prochi)
		
		attr_vals = {
			(None, u'anochihash'): anochihash,
			(None, u'prochi'): prochi,
		    }
		attr_qnames = {
		    (None, u'anochihash'): anochihash,
		    (None, u'prochi'): prochi,
		    }
		attrs = AttributesNSImpl(attr_vals, attr_qnames)
		
		self._anonschema.startElementNS((None, u'map'), u'map', attrs)
		self._anonschema.endElementNS((None, u'map'), u'map')
		return

	def generate(self, numRows):
		"""
		
		"""
		self.writeHeader()
		for i in range(numRows):
			self.writeRow()
		self.writeFooter()
		return

	def generate_prochi(self):
		"""
		
		"""
		prochi = self._projectID
		self._initial_prochi += 1
		prochi += str(self._initial_prochi)
		return prochi

	def generate_anochihash(self, anochi):
		"""
		
		"""
		h = hashlib.md5()
		h.update(anochi)
		return h.hexdigest()

class DataSuite:
	def __init__(self):
		"""
		
		"""
		self._range = [10,100,1000,10000]
		return
	
	def build(self, schematype):
		"""
		
		"""
		for size in self._range:
			outname = schematype + '_' + str(size) + '.xml'
			outfile = open(outname, 'w')
			gen = SchemaGen(outfile, 'utf-8', 'AAB', 1000000)
			gen.generate(size)
			outfile.close()
		return

if __name__ == "__main__":
	ds = DataSuite()
	ds.build('anon')