First

ext_rcrs.py

@@ -0,0 +1,71 @@
+#!/usr/bin/env python3
+
+import re
+import util
+import autoproc
+
+def parse_xml_rcrs(text, summary):
+	summary = {row['uid']: row for row in summary}
+	for patient in re.finditer(r'<Patient>(.+?)</Patient>', text, re.DOTALL):
+		patient = patient.group(1)
+		if (pos := patient.find('<Tumor>')) >= 0:
+			data = parse_xml_items(patient[:pos])
+			data['tumors'] = [parse_xml_items(tumor.group(1)) for tumor in re.finditer(r'<Tumor>(.+?)</Tumor>', patient[pos:], re.DOTALL)]
+			for tumor in data['tumors']:
+				acc = tumor['accessionNumberHosp']
+				meta = tumor['meta'] = summary[acc[:4] + '-' + acc[4:] + '/' + tumor['sequenceNumberHospital']]
+			data['name'] = meta.pop('name')
+			data['last5'] = meta.pop('last5')
+		else:
+			data = parse_xml_items(patient)
+			data['tumors'] = []
+		yield data
+
+def parse_xml_items(text):
+	return {item.group(1): item.group(2) for item in re.finditer(r'<Item naaccrId="([^"]+)">([^<]+)</Item>', text)}
+
+async def cmd_patients(proc, alpha, omega):
+	"""Fetch patients from RCRS dump"""
+	async with proc.sendline, autoproc.expect_async(proc) as expect:
+		proc.sendline('^Create RCRS extract')
+		async for prompt, response in expect.prompts_any({
+				re.compile(r'^ Facility Identification Number \(FIN\): \d+// $'):	None,
+				' Select date field to be used for Start/End range: ':				'Date DX',
+				re.compile(r'^ Start, Date DX: :  \([^\)\r\n]+\): $'):				util.vista_strftime(alpha),
+				re.compile(r'^   End, Date DX: :  \([^\)\r\n]+\): TODAY// $'):		util.vista_strftime(omega),
+				' Are these settings correct? YES// ':								None,
+			}, throw=True):
+			proc.sendline(response)
+			if prompt.index == 4:
+				break
+		assert await expect.endswith('\r\n      --------------------------------------------------------------\r\n\r\n\r\n')
+		proc.sendline()	# skip delay
+		assert (m := await expect.endswith('\r\n\x1a'))
+		doc_rcrs = m.before
+		assert await expect.endswith('\r\nDEVICE: ', timeout_settle=31)
+		proc.sendline('HOME;80')	# default HOME
+		assert await expect.earliest('HOME(CRT)\r\n')
+		summary = []
+		async for prompt, response in expect.prompts_any({
+				'\x07':															None,
+				'Type <Enter> to continue or \'^\' to exit: ':					None,
+				'               Press \'RETURN\' to continue, \'^\' to stop: ':	None,
+				'Select *..Utility Options Option: ':							None,
+				'Select DHCP Tumor Registry Option: ':							None,
+			}, throw=True):
+			proc.sendline(response)
+			if prompt.index == 0 or prompt.index == 1:
+				summary.extend({k.strip(): v.strip() for k, v in row.groupdict().items()} for row in re.finditer(r'(?P<last5>[A-Z]\d{4}) (?P<name>[^\r\n]{30}) (?P<uid>[^ \r\n]+) (?P<primarySite>[^ \r\n]+) (?P<dateOfDiagnosis>\d{2}/\d{2}/\d{4})  (?P<dateCaseLastChanged>\d{2}/\d{2}/\d{4})', prompt.before))
+			elif prompt.index == 4:
+				break
+		assert await expect.endswith('\r\nSelect Patient Information and OE/RR Option: ', '\r\nSelect Patient Information and OE/RR  <TEST ACCOUNT> Option: ')
+		expect.clear()
+		for item in parse_xml_rcrs(doc_rcrs, summary):
+			yield item
+
+async def cmd_tumors(proc, alpha, omega):
+	async for patient in cmd_patients(proc, alpha, omega):
+		tumors = patient.pop('tumors')
+		for item in tumors:
+			item['patient'] = patient
+			yield item