Programming Assessment
Question 1
Write a function `parse_config(text: str) -> tuple[dict[str, str], int]` that parses a tiny config l
anguage and returns:
- a dictionary of key→value (strings), and
- a checksum: XOR of the Unicode code points of all characters of the canonicalized values only (not
keys, whitespace, or comments).
Rules:
- Each non-empty logical line is `key = value`.
- Lines may end with a trailing backslash `\` to continue onto the next line.
- Leading/trailing whitespace around keys and values is ignored after handling continuations.
- `#` begins a comment unless inside double quotes.
- Values can be:
* unquoted text (stop before unescaped `#`), or
* double-quoted strings with escapes: `\"`, `\\`, `\n`, `\t`, `\xHH` (two hex digits).
- Duplicate keys keep the last occurrence.
- Invalid lines must raise `ValueError` mentioning the 1-based line number.
Canonicalization:
- For checksum, use the final unescaped value only.
- Do not include quotes.
Example:
```
user = "Ada Lovelace" # pioneer
greeting="Hello,\nworld" \"!" # continuation
path = C:\Users\Public
hex = "\x41\x42" # -> "AB"
note=unquoted # comment starts here
```
Question 2
Design a plugin system:
- Define a base class with attributes `name` and `priority`, and an abstract `handle` method.
- Subclasses auto-register unless marked abstract.
- A pipeline function executes plugins in descending priority order. Each may mutate the command or
return `None` to stop the pipeline.
- A context manager `mute_plugins(*names)` temporarily disables selected plugins.
Demo scenario:
- `AuthPlugin` adds `authenticated=True`.
- `AuditPlugin` appends `seen:<name>` to an audit log list.
- `KillSwitch` stops processing if the command contains `kill=1`.
Question 3
Implement a function that interprets a 1D numeric array of length R*C as an R×C torus (wrap-around i
n both directions).
For each cell, return the sum of values in its Chebyshev-≤k neighborhood.
Constraints:
- No explicit element-wise loops; use vectorized operations.
- Complexity should be near O(R*C).
- Preserve numeric dtype where possible.
- `k=0` should return the original array.
Quick check:
For R=3, C=5, k=1, with values `0..14`, the result at position (0,0) should equal the sum of its 3×3
toroidal neighborhood.
Question 4
You are given two sets of data:
A) Sales, each with an ID, SKU, timestamp, and quantity.
B) Promotions with half-open intervals [start, end) containing price, campaign, and priority.
Tasks:
1) For each sale, attach the applicable promotion where start ≤ ts < end and SKU matches. If multipl
e apply, select the one with highest priority; break ties by the latest start. If none apply, mark p
rice as missing.
2) For each SKU, produce a daily snapshot of the effective price at 00:00:00Z across the date range
covered. Apply the same precedence rules as above.
3) Compute revenue per sale as `qty * matched price`. If no price, revenue is missing.
Quick checks:
- A sale at 01:50Z with two overlapping promos should pick the higher-priority one.
- A sale at 02:10Z should select the later valid promo.
- The daily snapshot at midnight should reflect the correct interval-based price.
