OA: Email Checksum + Verification-String Lookup

SDE II Early-Career HackerRank OA, problem 1 of 2 (90 min total). Implementation-heavy: slice an email body into k-character chunks, map each chunk to a checksum integer using a fixed letter→digit table, then index into a verification string (with modular wrap) to produce the final verification token. Framed as Gmail's anti-spam check. Requirements Inputs: an email content string, a verification string, and an integer chunk size k. For each non-overlapping k-character window of email: Map each character to an integer using a fixed table: a–z → 1–25 (lowercase only, per the report), and :, /, . → 26, 27, 28 respectively. Sum the mapped values over the window — this is the window's checksum. Use that checksum to index into verification. If the checksum exceeds len(verification), take it modulo the length. Append the character found at that index to the output verify string. Return the assembled verify string. Time budget: candidate aimed to finish in 5–10 minutes so the second (much larger) OA problem has the bulk of the remaining time. Notes Pure implementation, no algorithmic trick — clean lookup tables and one pass. Characters outside the documented table (uppercase, digits, other punctuation) were not specified in the prompt. Decide on a sensible default (skip / treat as 0 / raise) and call it out in a comment; reviewers tend to accept any explicit choice. Handle the tail window when len(email) % k != 0: the prompt as recovered doesn't pin this down — either pad with a 0-contribution sentinel or stop early. Ask in chat; both are accepted. The mod wrap on the verification string is the only easy-to-miss spec — idx = checksum % len(verification) is intended even when checksum < len(verification) (handle the zero edge consistently). Preparation Write the lookup table once with {ch: i+1 for i, ch in enumerate(string.ascii_lowercase)} plus three special-char overrides; saves time vs hand-typing 28 entries. Practice slicing email[i:i+k] in a loop where i strides by k — common off-by-one source. Run a 2-minute sanity check end-to-end on a toy input before moving to problem 2; the OA gives only 2 visible test cases, no hidden cases run during the session.