Find Duplicate SAP User Identities Across Every System

Identify cases where the same person holds different User Name (BNAME) values,
across one or many SAP systems/clients. The same User Name repeated across
systems is OK and is not a duplicate by itself. Two user IDs belong to the same
person if they are linked by ANY shared identity signal — email, person number,
or normalized name — not by name alone and not by one signal suppressing another.

Use these source columns:
- Full name        (person name)
- User Name        (BNAME)
- System           (SID)
- Client           (MANDT)
- Email            (user email from USR21 -> ADR6, matched on PERSNUMBER +
                    ADDRNUMBER, primary address only. Customer/vendor emails that
                    also live in ADR6 are out of scope and must not be attached.)
- Person Number    (USR21.PERSNUMBER — assigned per system)
- User Type        (USR02.USTYP)

Logic:

1. Clean inputs:
   - Normalize Full name:
     - Convert to uppercase
     - Remove titles (Mr., Mrs., Ms., Miss, Dr.)
     - Fold accents to plain letters (JOSÉ -> JOSE)
     - If the name contains a comma, treat it as "Last, First" and reorder to
       First Last before splitting
     - Remove punctuation, special characters, digits, underscores, extra spaces
     - Split the cleaned name into words
   - Normalize Email: lowercase and trim. Treat blank, or any value without "@",
     as missing.
   - Normalize Person Number: trim. Treat blank or all-zeros as missing.

2. Derive:
   - First Name Normalized = first word of cleaned Full name
   - Last Name Normalized  = last word of cleaned Full name
   - Name Key First Last   = First Name Normalized + space + Last Name Normalized

3. Exclude records where:
   - Full name, Email, and Person Number are ALL missing (nothing to match on)
   - User Type is a non-person account: USTYP in B (System), C (Communication),
     S (Service), L (Reference). Keep only A (Dialog) and any custom
     dialog-equivalent types the landscape defines.
   - Full name equals User Name AND Email is missing AND Person Number is missing
     (a bare account ID with no stronger signal)
   - For the name signal only: cleaned Full name has fewer than two tokens.
     Records that still carry an Email or Person Number remain eligible for the
     email / person-number signals.

4. Link identities — build groups from THREE independent signals, then merge any
   groups that share a user (connected components). A user ID is linked to
   another when they share:
   - the same normalized Email, OR
   - the same System + Client + Person Number, OR
   - the same Name Key First Last.
   Every user ID that ends up in the same connected group is treated as the same
   person. No signal is a "fallback" — all three are evaluated for every record.

5. Assign Match Basis PER USER ID = the strongest signal that actually links that
   user to at least one other user in its group, ranked:
   Email (highest) > Person Number > Name (lowest).
   - Email  : linked to another ID by a shared user email — high confidence.
   - Person Number : linked by a shared per-system person number — high confidence.
   - Name   : linked ONLY by a shared name key (person numbers differ or are
     absent) — LOW confidence, must be reviewed by a human before any action,
     because two different real people can share a name.

6. Keep only groups that contain more than one distinct User Name.
   Output one row per group + User Name combination.

Required output columns, in this exact order:
- Person Key
- Match Basis
- Name Key First Last
- First Name Normalized
- Last Name Normalized
- Email
- User Name
- Rows For User ID
- Systems

Column definitions:
- Person Key: readable identifier of the group — the shared email if the group is
  email-linked, else System|Client|PersonNumber if person-number-linked, else the
  Name Key First Last
- Match Basis: Email, Person Number, or Name — the strongest signal linking THIS
  user to its group, and therefore its confidence (see step 5)
- Name Key First Last: normalized first + last name key
- First Name Normalized: first token from cleaned Full name
- Last Name Normalized: last token from cleaned Full name
- Email: normalized user email for that User Name, blank if none
- User Name: distinct user ID (BNAME)
- Rows For User ID: number of source rows for that group and User Name combination
- Systems: comma-separated list of distinct System values for that group and
  User Name combination

Sort the output by:
1. Match Basis (Email, then Person Number, then Name)
2. Person Key ascending
3. User Name ascending

Important:
- Do not create summary tabs, charts, pivots, or extra sheets.
- Do not include same-user-ID-only duplicates unless the same person has multiple
  distinct User Name values.
- Match Basis = Name rows are the review bucket: lower confidence, human check
  before any deprovisioning or license action.
- Email must be the user's own address (USR21 -> ADR6 on PERSNUMBER + ADDRNUMBER),
  never a customer or vendor email that happens to share the ADR6 table.
- Save the final file as Distinct_User_IDs_By_Name.xlsx.