NRL-1666 allow table name to be configured and fix NHS number generation

Author: anjalitrace2-nhs

Makefile

@@ -15,6 +15,7 @@ ACCOUNT ?= dev
 APP_ALIAS ?= default
 HOST ?= $(TF_WORKSPACE_NAME).api.record-locator.$(ENV).national.nhs.uk
 ENV_TYPE ?= $(ENV)
+PERFTEST_TABLE_NAME ?= nhsd-nrlf--xaxel-deleteme-pointers-table
 
 export PATH := $(PATH):$(PWD)/.venv/bin
 export USE_SHARED_RESOURCES := $(shell poetry run python scripts/are_resources_shared_for_stack.py $(TF_WORKSPACE_NAME))
@@ -253,8 +254,17 @@ generate-perftest-permissions: ## Generate perftest permissions and add to nrlf_
 # Run producer performance tests with configurable HOST and ENV_TYPE
 perftest-producer:
 	@echo "Running producer performance tests with HOST=$(HOST) and ENV_TYPE=$(ENV_TYPE)"
-	k6 run tests/performance/producer/perftest.js -e HOST=$(HOST) -e ENV_TYPE=$(ENV_TYPE)
+	k6 run tests/performance/producer/perftest.js -e HOST=$(HOST) -e ENV_TYPE=$(ENV_TYPE) -e PERFTEST_TABLE_NAME=$(PERFTEST_TABLE_NAME)
 
 perftest-consumer:
 	@echo "Running consumer performance tests with HOST=$(HOST) and ENV_TYPE=$(ENV_TYPE)"
 	k6 run tests/performance/consumer/perftest.js -e HOST=$(HOST) -e ENV_TYPE=$(ENV_TYPE)
+
+# Generates input csv for a given table
+perftest-prepare:
+	mkdir -p $(DIST_PATH)
+	PYTHONPATH=. poetry run python tests/performance/perftest_environment.py generate_pointer_table_extract $(PERFTEST_TABLE_NAME)
+
+perftest-seed-data: perftest-prepare
+	@echo "Seeding dynamo with data for performance tests with HOST=$(HOST) and ENV_TYPE=$(ENV_TYPE)"
+	poetry run python tests/performance/consumer/perftest.js --table_name="" --px_with_pointers="" --pointers_per_px="" --type_dists="" --custodian_dists=""

tests/performance/README.md

@@ -0,0 +1,23 @@
+# Performance Testing
+
+some high level context short
+
+## Run perf tests
+
+### Prepare
+
+```sh
+assume dev
+make perftest-prepare PERFTEST_TABLE_NAME=<POINTER_TABLE_NAME>
+```
+
+choose existing table name/create table with this script > `PERFTEST_TABLE_NAME` env var
+makes these files
+
+### Run
+
+these/find profiles available
+
+### Outputs
+
+handy bits

tests/performance/perftest_environment.py

@@ -1,28 +1,31 @@
 import csv
 import json
+import os
 import pathlib
+import re
 
 import boto3
 
 # from nhs_number import generate
 
-# from scripts.are_resources_shared_for_stack import uses_shared_resources
 
 DYNAMODB = boto3.resource("dynamodb", region_name="eu-west-2")
 
+default_table_name = "default-table-name"
+# default_table_name = "nhsd-nrlf--xaxel-deleteme-pointers-table"
 
-def get_pointers_table_name(stack_name):
-    # if uses_shared_resources(stack_name):
-    #     env = stack_name.split("-")[0]
-    #     return f"nhsd-nrlf--{env}-pointers-table"
-    # else:
-    #     return f"nhsd-nrlf--{stack_name}-pointers-table"
-    # return f"nhsd-nrlf--perftest-pointers-table"
-    return f"nhsd-nrlf--xaxel-deleteme-pointers-table"
 
+def _get_pointers_table_name():
+    perftest_table_name = os.environ.get("PERFTEST_TABLE_NAME", default_table_name)
 
-def extract_consumer_data(stack_name, out="consumer_reference_data.json"):
-    table_name = get_pointers_table_name(stack_name)
+    if re.search("^nhsd-nrlf--.+-pointers-table$", perftest_table_name):
+        return perftest_table_name
+
+    return f"nhsd-nrlf--{perftest_table_name}-pointers-table"
+
+
+def extract_consumer_data(out="consumer_reference_data.json"):
+    table_name = _get_pointers_table_name()
     table = DYNAMODB.Table(table_name)
     scan_kwargs = {}
     done = False
@@ -84,7 +87,6 @@ def __next__(self):
 
 
 def generate_producer_data(
-    stack_name,
     proportion_existing=0.8,  # Proportion of output that should be existing NHS numbers
     total_count=1000,  # Total number of NHS numbers to output
     out="producer_reference_data.json",
@@ -96,7 +98,7 @@ def generate_producer_data(
     - total_count: total number of NHS numbers in output
     NHS numbers are generated in a semi-deterministic way, similar to the NFT seeding script.
     """
-    table_name = get_pointers_table_name(stack_name)
+    table_name = _get_pointers_table_name()
     table = DYNAMODB.Table(table_name)
     scan_kwargs = {}
     done = False
@@ -176,13 +178,12 @@ def __next__(self):
 
 
 def generate_pointer_table_extract(
-    stack_name,
     out="producer_reference_data.csv",
 ):
     """
     Generate a CSV file containing all pointer IDs, pointer type, custodian, and nhs_number (patient).
     """
-    table_name = get_pointers_table_name(stack_name)
+    table_name = _get_pointers_table_name()
     table = DYNAMODB.Table(table_name)
     scan_kwargs = {}
     done = False
@@ -197,6 +198,7 @@ def generate_pointer_table_extract(
             ["count", "pointer_id", "pointer_type", "custodian", "nhs_number"]
         )
         while not done:
+
             if start_key:
                 scan_kwargs["ExclusiveStartKey"] = start_key
             response = table.scan(**scan_kwargs)

tests/performance/producer/client_perftest.js

@@ -52,27 +52,18 @@ function randomNHSNumberInRange(start, end) {
 }
 
 function generateValidNHSNumber(first9) {
-  // NHS number with checksum logic
-  // const CHECKSUM_WEIGHTS = [10, 9, 8, 7, 6, 5, 4, 3, 2];
-  // while (true) {
-  //   const first9Str = String(first9).padStart(9, "0");
-  //   const digits = first9Str.split("").map(Number);
-  //   if (digits.some(isNaN)) {
-  //     throw new Error("Invalid digit in NHS number generation");
-  //   }
-  //   let sum = digits.reduce((acc, digit, idx) => acc + digit * CHECKSUM_WEIGHTS[idx], 0);
-  //   // Python: checksum = (sum * -1) % 11
-  //   let checksum = (sum * -1) % 11;
-  //   if (checksum === 10) {
-  //     first9++;
-  //     if (first9 > 999999999) {
-  //       throw new Error("No valid NHS number found in range.");
-  //     }
-  //     continue;
-  //   }
-  //   return first9Str + String(checksum);
-  // }
-  return "3226053469";
+  const first9Str = String(first9).padStart(9, "0");
+
+  if (first9Str.match(/^\d{9}$/)) {
+    throw new Error(
+      "bad NHS number generated - expected 9 digits",
+      first9Str,
+      first9
+    );
+  }
+
+  // NHS numbers not validated, checksum doesn't need to be legit
+  return `${first9Str}${"1"}`;
 }
 
 function pickNHSNumber() {