"""Drive 3 — find SourceSafe artifacts + size up the VWIN7 and VWP-FIN top-level dirs that weren't on drive 1."""
import csv

WIZ = 'clients/valleywide/app-modernization/WizTree_20260516174356.csv'

ss_hits = []
top_level_sizes = {}  # path -> (size, files, folders, mod)
TARGETS = {
    r'D:\VWIN7\\',
    r'D:\VWP-FIN\\',
    r'D:\Archive\\',
    r'D:\WIN7-Orders\Darv-2\\',
}

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) < 4: continue
        p = row[0]
        pl = p.lower()
        if pl.endswith('srcsafe.ini') or pl.endswith('.scc'):
            try: sz = int(row[1])
            except (ValueError, IndexError): sz = 0
            ss_hits.append((row[3], sz, p))
        if '\\sourcesafe' in pl or '\\vss\\' in pl:
            try: sz = int(row[1])
            except (ValueError, IndexError): sz = 0
            ss_hits.append((row[3], sz, p))
        if p.endswith('\\') and p in TARGETS:
            try:
                sz = int(row[1]); files = int(row[5]); folders = int(row[6])
                top_level_sizes[p] = (sz, files, folders, row[3])
            except (ValueError, IndexError):
                pass

print('=== Top-level dirs unique-ish to drive 3 ===')
print(f'{"GB":>8} {"Files":>8} {"Folders":>7}  Modified            Path')
for path, (sz, files, folders, mod) in sorted(top_level_sizes.items()):
    print(f'{sz/1024/1024/1024:>8.2f} {files:>8} {folders:>7}  {mod:<19} {path}')

print(f'\n=== SourceSafe artifacts ({len(ss_hits)} matches, top 50) ===')
ss_hits.sort(reverse=True)
seen_paths = set()
for mod, sz, p in ss_hits[:50]:
    # Dedupe by leaf path (snapshots often duplicate)
    leaf_key = p.lower().split('\\estimating archive')[-1] if 'estimating archive' in p.lower() else p.lower()
    print(f'{mod}  {sz:>8}b  {p}')