"""Drive 2 triage — find net-new high-value folders not present on drive 1."""
import csv

WIZ = 'clients/valleywide/app-modernization/WizTree_20260516173603.csv'

# Folders whose leaf name (last path component) we want to size
LEAVES = {
    'darv-win7-pc',     # likely Darv's actual Win7 backup
    '98_server',
    '000_asource',
    'darv',
    'kingston',
    'full',
    'virtualbox',
    'vm_vdi',
}

PARENT_HINTS = {
    'archive',     # D:\Archive\ — backup root
    'office-archive',
    'office-estimates',
}

# Also look for ALL source-relevant file types unique to this drive
hits = []
all_vbp_by_path = []

with open(WIZ, encoding='utf-8-sig', errors='replace') as f:
    r = csv.reader(f); next(r); next(r)
    for row in r:
        if not row or len(row) < 7:
            continue
        p = row[0]
        if p.endswith('\\'):
            leaf = p.rstrip('\\').rsplit('\\', 1)[-1].lower()
            if leaf in LEAVES:
                try:
                    sz = int(row[1]); files = int(row[5]); folders = int(row[6])
                except (ValueError, IndexError):
                    continue
                hits.append((sz, files, folders, row[3], p))
        else:
            if p.lower().endswith('.vbp'):
                # Capture only VWP/Orders-relevant projects (not SDK samples)
                pl = p.lower()
                if 'seagat' in pl or 'samples\\code' in pl:
                    continue
                try:
                    sz = int(row[1])
                except (ValueError, IndexError):
                    continue
                all_vbp_by_path.append((row[3], sz, p))

hits.sort(reverse=True)
print('=== Sizing key folder leaves ===')
print(f'{"GB":>8} {"Files":>8} {"Folders":>7}  Modified            Path')
for sz, files, folders, mod, p in hits[:40]:
    print(f'{sz/1024/1024/1024:>8.2f} {files:>8} {folders:>7}  {mod:<19} {p}')

print()
print('=== ALL VWP/Orders .vbp files in D:\\Archive\\Darv-Win7-PC\\ ===')
for mod, sz, p in sorted(all_vbp_by_path, reverse=True):
    if '\\archive\\darv-win7-pc\\' in p.lower():
        print(f'{mod}  {sz:>6}b  {p}')

print()
print('=== ALL .vbp files in Darv\\000_ASource\\ (any drive location) ===')
for mod, sz, p in sorted(all_vbp_by_path, reverse=True):
    if '\\darv\\000_asource\\' in p.lower():
        print(f'{mod}  {sz:>6}b  {p}')