#!/usr/bin/env python3 """Independent verification for Task 1 (review pass). Expected records below were hand-derived directly from the raw input CSVs and inputs/policy.md, NOT from reconcile.py's logic. Arithmetic uses integer cents. Run from the workspace root: python output/verify.py """ import csv, re, sys def load(p): with open(p, newline="", encoding="utf-8") as f: return list(csv.DictReader(f)) failures = [] passed = 0 def check(cond, msg): global passed passed += 1 if cond: print("PASS:", msg) else: failures.append(msg) print("FAIL:", msg) DATE = re.compile(r"^\d{4}-\d{2}-\d{2}$") MONEY = re.compile(r"^\d+\.\d{2}$") def cents(s): assert MONEY.match(s), s d, c = s.split(".") return int(d) * 100 + int(c) # --- Expected cleaned-sales (hand-derived from inputs/sales.csv + policy) --- exp_sales = [ ["S01","N1001","2026-10-02","Workshop","paid","240.00","2026-10-02T12:00:00"], ["S02","N1002","2026-10-03","Course","paid","120.00","2026-10-03T12:00:00"], # 03/10 = DD/MM ["S03;S04","N1003","2026-10-05","Workshop","paid","480.00","2026-10-05T12:00:00"], # dup collapsed ["S06","N1005","2026-10-10","Workshop","paid","240.00","2026-10-10T12:00:00"], ["S07","N1006","2026-10-11","Course","paid","240.00","2026-10-11T12:00:00"], # 11/10 = DD/MM ["S10","N1008","2026-10-15","Course","paid","180.00","2026-10-16T09:00:00"], # latest updated_at beats S09(120) ["S12","N1010","2026-10-20","Workshop","paid","720.00","2026-10-20T12:00:00"], ["S13","N1011","2026-10-22","Course","paid","360.00","2026-10-22T12:00:00"], ["S14","N1012","2026-10-31","Workshop","paid","240.00","2026-10-31T12:00:00"], ] exp_refunds = [ ["R01","RF01","N1003","2026-10-09","settled","240.00","Workshop"], ["R02;R03","RF02","N1006","2026-10-14","settled","60.00","Course"], # 14/10 = DD/MM; dup collapsed ] exp_ads = [ ["A01","AD01","2026-10-03","Workshop","search","150.00"], ["A02","AD02","2026-10-04","Course","social","90.00"], ["A03;A04","AD03","2026-10-10","Workshop","social","100.00"], # dup collapsed ["A05","AD04","2026-10-17","Course","search","60.00"], ["A06","AD05","2026-10-28","Workshop","search","50.00"], # $50.00 stripped ] exp_exceptions = { ("sales.csv","S04","collapsed_duplicate"), ("sales.csv","S09","superseded_conflict"), ("sales.csv","S05","excluded"), ("sales.csv","S08","excluded"), ("sales.csv","S11","excluded"), ("sales.csv","S15","excluded"), ("refunds.csv","R03","collapsed_duplicate"), ("refunds.csv","R04","excluded/flagged"), ("refunds.csv","R05","excluded/flagged"), ("refunds.csv","R06","excluded/flagged"), ("refunds.csv","R07","excluded/flagged"), ("ads.csv","A04","collapsed_duplicate"), ("ads.csv","A07","excluded"), } sales = load("output/cleaned-sales.csv") refunds = load("output/cleaned-refunds.csv") ads = load("output/cleaned-ads.csv") exc = load("output/exceptions.csv") check([list(r.values()) for r in sales] == exp_sales, "cleaned-sales.csv matches hand-derived expected records (all fields)") check([list(r.values()) for r in refunds] == exp_refunds, "cleaned-refunds.csv matches hand-derived expected records") check([list(r.values()) for r in ads] == exp_ads, "cleaned-ads.csv matches hand-derived expected records") check({(e["source_file"], e["row_id"], e["disposition"]) for e in exc} == exp_exceptions and len(exc) == len(exp_exceptions), "exceptions.csv: exactly the 13 expected entries/dispositions") # formats check(all(DATE.match(r["date"]) and MONEY.match(r["gross_cad"]) for r in sales), "sales: ISO dates and 2-decimal money") check(all(DATE.match(r["settled_date"]) and MONEY.match(r["amount_cad"]) for r in refunds), "refunds: ISO dates and 2-decimal money") check(all(DATE.match(r["date"]) and MONEY.match(r["spend_cad"]) for r in ads), "ads: ISO dates and 2-decimal money") # raw row-ID coverage (independent of reconcile.py's assertion) for fname, clean in [("sales.csv", sales), ("refunds.csv", refunds), ("ads.csv", ads)]: raw = {r["row_id"] for r in load("inputs/" + fname)} clean_ids = [i for r in clean for i in r["row_ids"].split(";")] dup = {i for e in exc if e["source_file"] == fname and e["disposition"] == "collapsed_duplicate" for i in e["row_id"].split(";")} excl = [i for e in exc if e["source_file"] == fname and e["disposition"] != "collapsed_duplicate" for i in e["row_id"].split(";")] ok = (set(clean_ids) | set(excl) == raw and not (set(clean_ids) & set(excl)) and len(clean_ids) + len(excl) == len(raw) and dup <= set(clean_ids)) check(ok, fname + ": all raw row IDs covered exactly once (duplicates also explained)") # unresolved value preserved: R06 amount blank in raw, not zero anywhere in outputs r06 = next(r for r in load("inputs/refunds.csv") if r["row_id"] == "R06") check(r06["amount_cad"] == "", "raw R06 amount is indeed missing") check(any("amount=''" in e["detail"] and "provisional" in e["reason"] for e in exc if e["row_id"] == "R06"), "R06 missing amount preserved as unknown and marked provisional") # superseded value preserved check(any("120 superseded by 180" in e["detail"] for e in exc if e["row_id"] == "S09"), "S09 superseded gross (120 vs 180) preserved in exceptions detail") # refund caps gross = {r["order_id"]: cents(r["gross_cad"]) for r in sales} check(all(cents(r["amount_cad"]) <= gross[r["order_id"]] for r in refunds), "applied refunds do not exceed paid gross") # totals in integer cents, computed from cleaned files def tot(prod): s = [r for r in sales if prod is None or r["product"] == prod] f = [r for r in refunds if prod is None or r["product"] == prod] a = [r for r in ads if prod is None or r["product"] == prod] n = len(s); g = sum(cents(r["gross_cad"]) for r in s); rf = sum(cents(r["amount_cad"]) for r in f) ad = sum(cents(r["spend_cad"]) for r in a); net = g - rf; co = net - ad per = round(net / n) if n else None assert n == 0 or net * 100 % n == 0 or True return n, g, rf, net, ad, co, per exp_tot = { # hand-calculated: W 240+480+240+720+240=1920; C 120+240+180+360=900 "Workshop": (5, 192000, 24000, 168000, 30000, 138000, 33600), "Course": (4, 90000, 6000, 84000, 15000, 69000, 21000), None: (9, 282000, 30000, 252000, 45000, 207000, 28000), } for p, label in [("Workshop","Workshop"), ("Course","Course"), (None,"Combined")]: check(tot(p) == exp_tot[p], f"{label} totals match hand calculation (cents): {exp_tot[p]}") # summary table values present s = open("output/summary.md", encoding="utf-8").read() need = ["1,920.00","900.00","2,820.00","240.00","60.00","300.00","1,680.00","840.00", "2,520.00","300.00","150.00","450.00","1,380.00","690.00","2,070.00", "336.00","210.00","280.00","PROVISIONAL"] check(all(v in s for v in need), "summary.md contains all expected table figures and provisional flag") check("66.7%" in s and "66% of combined" not in s, "summary contribution share corrected to 66.7%") check("settles in a later month" in s, "summary RF03 wording is period-correct") # derived claims in summary check(round(168000/30000, 1) == 5.6 and round(84000/15000, 1) == 5.6, "5.6x net-revenue-to-spend ratios correct") check(round(138000/207000*100, 1) == 66.7, "66.7% contribution share correct") check(252000 - 36000 - 12000 == 204000, "2,040.00 downside floor arithmetic correct") print() if failures: print("FAILURES:", len(failures)); sys.exit(1) print("ALL", passed, "INDEPENDENT CHECKS PASSED")