routing--/update.sh

#!/usr/bin/env bash
# Pulls today's campaign snapshot from the source Google Sheet and writes
# data.json + history.json for the report page.
#
# The source spreadsheet creates a fresh tab per day named "M/D" (e.g. 9/9).
# We fetch via the gviz CSV endpoint with URL-encoded sheet name and walk
# back up to 7 days if today's tab hasn't been created yet.
set -euo pipefail

SHEET_ID="1t2Sz2W6845HbS1ZGJN1iuaj9cgjPFHt36U1fK938jT8"
SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
CSV_FILE="$SCRIPT_DIR/.sheet.csv"

log() { echo "[update.sh $(date -u +%FT%TZ)] $*" >&2; }

# Compute M/D for a day offset (portable via python3).
mmdd_of() {
  python3 -c "import datetime,sys;d=datetime.date.today()+datetime.timedelta(days=int(sys.argv[1]));print(f'{d.month}/{d.day}')" "$1"
}

# Fetch a candidate tab into CSV_FILE. Success = first header column is "Order ID"
# (gviz silently returns the first tab when the requested name doesn't exist).
fetch_sheet() {
  local name="$1"
  local encoded="${name//\//%2F}"
  local url="https://docs.google.com/spreadsheets/d/${SHEET_ID}/gviz/tq?tqx=out:csv&sheet=${encoded}"
  if ! curl --max-time 60 -sSL -f "$url" -o "$CSV_FILE"; then
    return 1
  fi
  local first_col
  first_col=$(head -1 "$CSV_FILE" | tr -d '\r' | cut -d',' -f1 | tr -d '"')
  [ "$first_col" = "Order ID" ]
}

SHEET_NAME=""
for offset in 0 -1 -2 -3 -4 -5 -6 -7; do
  candidate=$(mmdd_of "$offset")
  log "trying sheet '$candidate'"
  if fetch_sheet "$candidate"; then
    SHEET_NAME="$candidate"
    log "matched sheet '$candidate'"
    break
  fi
done

if [ -z "$SHEET_NAME" ]; then
  log "no matching sheet found in the last 8 days"
  rm -f "$CSV_FILE"
  exit 1
fi

SHEET_NAME="$SHEET_NAME" SCRIPT_DIR="$SCRIPT_DIR" CSV_FILE="$CSV_FILE" python3 <<'PYEOF'
import csv, json, os, sys, datetime

script_dir  = os.environ['SCRIPT_DIR']
csv_file    = os.environ['CSV_FILE']
sheet_name  = os.environ['SHEET_NAME']
data_path   = os.path.join(script_dir, 'data.json')
hist_path   = os.path.join(script_dir, 'history.json')

with open(csv_file, newline='', encoding='utf-8') as f:
    raw = list(csv.reader(f))
if not raw:
    print('empty CSV', file=sys.stderr); sys.exit(1)

headers = [h.strip() for h in raw[0]]
records = []
for r in raw[1:]:
    if not any((c or '').strip() for c in r):
        continue
    row = {}
    for i, h in enumerate(headers):
        row[h] = (r[i].strip() if i < len(r) else '')
    records.append(row)

def num(x):
    if x is None: return 0.0
    s = str(x).replace(',', '').replace('$', '').replace('%', '').strip()
    if not s: return 0.0
    try: return float(s)
    except ValueError: return 0.0

# Aggregate totals for the summary strip.
sum_keys = ['Valid Click', 'PB Installs', 'Revenue', 'Cost', 'Gross Profit', 'RT Reject Installs']
totals = {k: round(sum(num(r.get(k, '')) for r in records), 4) for k in sum_keys if k in headers}

# Derived rates (recomputed from totals, not averaged across rows).
if totals.get('Valid Click', 0) > 0 and 'PB Installs' in totals:
    totals['CVR'] = round(totals['PB Installs'] / totals['Valid Click'], 6)
if totals.get('Revenue', 0) > 0 and 'Gross Profit' in totals:
    totals['%Gross Profit'] = round(totals['Gross Profit'] / totals['Revenue'], 6)
if 'PB Installs' in totals and 'RT Reject Installs' in totals:
    denom = totals['PB Installs'] + totals['RT Reject Installs']
    if denom > 0:
        totals['%RT Reject Install'] = round(totals['RT Reject Installs'] / denom, 6)

# Distributions for the mini-charts.
def bucket(field):
    out = {}
    for r in records:
        key = (r.get(field) or '').strip() or 'Unknown'
        out[key] = out.get(key, 0) + 1
    return out

status_counts = bucket('Status') if 'Status' in headers else {}
geo_counts    = bucket('Geo')    if 'Geo'    in headers else {}

# Top-10 by revenue for the campaign bar chart.
def top_by(field, n=10):
    scored = [(num(r.get(field, '')), r) for r in records]
    scored.sort(key=lambda t: t[0], reverse=True)
    out = []
    for val, r in scored[:n]:
        if val <= 0: break
        out.append({
            'campaign': r.get('Campaign Name', '') or r.get('Campaign ID', ''),
            'pkg':      r.get('PKG Name', ''),
            'geo':      r.get('Geo', ''),
            'value':    val,
            'revenue':  num(r.get('Revenue', '')),
            'cost':     num(r.get('Cost', '')),
            'gp':       num(r.get('Gross Profit', '')),
        })
    return out

fetched_at = datetime.datetime.utcnow().strftime('%Y-%m-%d %H:%M:%S UTC')

payload = {
    'sheet_name':     sheet_name,
    'fetched_at':     fetched_at,
    'columns':        headers,
    'row_count':      len(records),
    'totals':         totals,
    'status_counts':  status_counts,
    'geo_counts':     geo_counts,
    'top_by_revenue': top_by('Revenue'),
    'top_by_gp':      top_by('Gross Profit'),
    'rows':           records,
}
with open(data_path, 'w', encoding='utf-8') as f:
    json.dump(payload, f, ensure_ascii=False, indent=2)

# Keep a rolling history of daily totals so the trend chart has time series.
try:
    with open(hist_path, 'r', encoding='utf-8') as f:
        history = json.load(f)
    if not isinstance(history, list):
        history = []
except (FileNotFoundError, json.JSONDecodeError):
    history = []

history = [h for h in history if h.get('sheet_name') != sheet_name]
history.append({
    'sheet_name':    sheet_name,
    'fetched_at':    fetched_at,
    'row_count':     len(records),
    'totals':        totals,
    'status_counts': status_counts,
})

def sort_key(h):
    try:
        m, d = h['sheet_name'].split('/')
        return (int(m), int(d))
    except Exception:
        return (0, 0)
history.sort(key=sort_key)
history = history[-120:]  # keep ~4 months

with open(hist_path, 'w', encoding='utf-8') as f:
    json.dump(history, f, ensure_ascii=False, indent=2)

print(f'wrote {len(records)} rows for sheet {sheet_name}; history now {len(history)} days', file=sys.stderr)
PYEOF

rm -f "$CSV_FILE"
log "done"