Separate MP, MPP information into separate table

export_db.sh

@@ -1,4 +1,4 @@
 #!/bin/bash
 
 rm html/database.db
-sqlite3 database.db '.dump pbs_item pbs_item_restriction pbs_restriction pbs_restriction_criteria pbs_criteria pbs_criteria_parameter' | sqlite3 html/database.db
+sqlite3 database.db '.dump pbs_item pbs_mp pbs_mpp pbs_item_restriction pbs_restriction pbs_restriction_criteria pbs_criteria pbs_criteria_parameter' | sqlite3 html/database.db

html/index.html

@@ -76,7 +76,7 @@
 				db = new SQL.Database(new Uint8Array(buf));
 				
 				// Initialise search bar
-				const labels = execAsScalars(db.prepare('SELECT DISTINCT mp_preferred_term FROM pbs_item ORDER BY LOWER(mp_preferred_term)'));
+				const labels = execAsScalars(db.prepare('SELECT DISTINCT preferred_term FROM pbs_mp ORDER BY LOWER(preferred_term)'));
 				const data = labels.map(label => ({'label': label}));
 				const autocomplete = new Autocomplete(document.getElementById('search-input'), {
 					data: data,
@@ -88,7 +88,7 @@
 			
 			function onClickSearchItem(item) {
 				// Find matching PBS items
-				let stmt = db.prepare('SELECT * FROM pbs_item WHERE LOWER(mp_preferred_term) = ?');
+				let stmt = db.prepare('SELECT * FROM pbs_item LEFT JOIN (SELECT code AS mpp_code, preferred_term AS mpp_preferred_term, mp_code FROM pbs_mpp) AS pbs_mpp ON pbs_item.mpp_code = pbs_mpp.mpp_code LEFT JOIN (SELECT code AS mp_code, preferred_term AS mp_preferred_term FROM pbs_mp) AS pbs_mp ON pbs_mpp.mp_code = pbs_mp.mp_code WHERE LOWER(mp_preferred_term) = ?');
 				stmt.bind([item.label.toLowerCase()]);
 				const items = execAsObjects(stmt);
 				

import_pbs_xml.py

@@ -23,7 +23,13 @@ cur = con.cursor()
 
 # Init schema
 cur.execute('DROP TABLE IF EXISTS pbs_item')
-cur.execute('CREATE TABLE pbs_item (code TEXT PRIMARY KEY, mpp_preferred_term TEXT, mp_preferred_term TEXT, maximum_prescribable_units INTEGER, number_repeats INTEGER, benefit_type TEXT)')
+cur.execute('CREATE TABLE pbs_item (code TEXT PRIMARY KEY, mpp_code TEXT, maximum_prescribable_units INTEGER, number_repeats INTEGER, benefit_type TEXT)')
+
+cur.execute('DROP TABLE IF EXISTS pbs_mpp')
+cur.execute('CREATE TABLE pbs_mpp (code TEXT PRIMARY KEY, mp_code TEXT, preferred_term TEXT)')
+
+cur.execute('DROP TABLE IF EXISTS pbs_mp')
+cur.execute('CREATE TABLE pbs_mp (code TEXT PRIMARY KEY, preferred_term TEXT)')
 
 cur.execute('DROP TABLE IF EXISTS pbs_item_restriction')
 cur.execute('CREATE TABLE pbs_item_restriction (item_code TEXT, restriction_code INTEGER)')
@@ -53,6 +59,8 @@ ns = {'pbs': 'http://schema.pbs.gov.au/', 'xlink': 'http://www.w3.org/1999/xlink
 # Get General Schedule
 program = next(p for p in root.find('pbs:schedule', ns).findall('pbs:program', ns) if p.find('pbs:info', ns).find('pbs:code', ns).text == 'GE')
 
+mpps_to_parse = set()
+mps_to_parse = set()
 restrictions_to_parse = set()
 criteria_to_parse = set()
 
@@ -70,12 +78,7 @@ for item in program.findall('pbs:prescribing-rule', ns):
 	benefit = benefits[0]
 	
 	mpp_id = item.find('pbs:ready-prepared', ns).find('pbs:mpp-reference', ns).get('{http://www.w3.org/1999/xlink}href').lstrip('#')
-	mpp = root.find('pbs:drugs-list', ns).find('pbs:mpp[@xml:id="' + mpp_id + '"]', ns)
+	mpp_code = item.find('pbs:ready-prepared', ns).find('pbs:mpp-reference', ns).find('pbs:code', ns).text
-	mpp_preferred_term = mpp.find('pbs:preferred-term', ns).text
-	
-	mp_id = mpp.find('pbs:drug-references-list', ns).find('pbs:mp-reference', ns).get('{http://www.w3.org/1999/xlink}href').lstrip('#')
-	mp = root.find('pbs:drugs-list', ns).find('pbs:mp[@xml:id="' + mp_id + '"]', ns)
-	mp_preferred_term = mp.find('pbs:preferred-term[@rdf:resource="http://pbs.gov.au/clinical"]', ns).text
 	
 	max_units = item.find('pbs:ready-prepared', ns).find('pbs:maximum-prescribable[@rdf:resource="http://pbs.gov.au/reference/unit-of-use"]', ns).find('pbs:value', ns).text
 	max_repeats = item.find('pbs:ready-prepared', ns).find('pbs:number-repeats', ns).find('pbs:value', ns).text
@@ -87,7 +90,7 @@ for item in program.findall('pbs:prescribing-rule', ns):
 		'http://pbs.gov.au/benefit-type/authority-required': 'authority',
 	}[benefit.get('{http://www.w3.org/1999/02/22-rdf-syntax-ns#}resource')]
 	
-	cur.execute('INSERT INTO pbs_item (code, mpp_preferred_term, mp_preferred_term, maximum_prescribable_units, number_repeats, benefit_type) VALUES (?, ?, ?, ?, ?, ?)', (code, mpp_preferred_term, mp_preferred_term, max_units, max_repeats, benefit_type))
+	cur.execute('INSERT INTO pbs_item (code, mpp_code, maximum_prescribable_units, number_repeats, benefit_type) VALUES (?, ?, ?, ?, ?)', (code, mpp_code, max_units, max_repeats, benefit_type))
 	
 	# Get restrictions
 	if restrictions := benefit.find('pbs:restriction-references-list', ns):
@@ -98,6 +101,39 @@ for item in program.findall('pbs:prescribing-rule', ns):
 			
 			# Queue this restriction for parsing
 			restrictions_to_parse.add(restriction_id)
+	
+	# Queue the MPP for parsing
+	mpps_to_parse.add(mpp_id)
+
+# Parse MPPs
+for mpp_id in sorted(list(mpps_to_parse)):
+	mpp = root.find('pbs:drugs-list', ns).find('pbs:mpp[@xml:id="' + mpp_id + '"]', ns)
+	mpp_code = mpp.find('pbs:code', ns).text
+	mpp_preferred_term = mpp.find('pbs:preferred-term', ns).text
+	
+	mp_id = mpp.find('pbs:drug-references-list', ns).find('pbs:mp-reference', ns).get('{http://www.w3.org/1999/xlink}href').lstrip('#')
+	mp = root.find('pbs:drugs-list', ns).find('pbs:mp[@xml:id="' + mp_id + '"]', ns)
+	mp_code = mp.find('pbs:code[@rdf:resource="http://pbs.gov.au/Drug/MP"]', ns).text  # Must look this up because the <code> in <mp-reference> is only SNOMED
+	
+	cur.execute('INSERT INTO pbs_mpp (code, mp_code, preferred_term) VALUES (?, ?, ?)', (mpp_code, mp_code, mpp_preferred_term))
+	
+	# Queue the MP for parsing
+	mps_to_parse.add(mp_id)
+	
+	# Get TPPs
+	for tpp_reference in mpp.find('pbs:drug-references-list', ns).findall('pbs:tpp-reference', ns):
+		tpp_id = tpp_reference.get('{http://www.w3.org/1999/xlink}href').lstrip('#')
+		
+		# Queue the TPP for parsing
+		tpps_to_parse.add((tpp_id, mpp_code))
+
+# Parse MPs
+for mp_id in sorted(list(mps_to_parse)):
+	mp = root.find('pbs:drugs-list', ns).find('pbs:mp[@xml:id="' + mp_id + '"]', ns)
+	mp_code = mp.find('pbs:code[@rdf:resource="http://pbs.gov.au/Drug/MP"]', ns).text  # Also there are SNOMED codes but they are inconsistent
+	mp_preferred_term = mp.find('pbs:preferred-term[@rdf:resource="http://pbs.gov.au/clinical"]', ns).text
+	
+	cur.execute('INSERT INTO pbs_mp (code, preferred_term) VALUES (?, ?)', (mp_code, mp_preferred_term))
 
 # Parse restrictions
 for restriction_id in sorted(list(restrictions_to_parse)):