Level 2 exam — Python · GUIDINEE.COM

Exam — fresh questions

12 original questions: recall (25%) + application (50%) + arbitration (25%) on Data structures & OOP.
Answer everything, then click Check: your score and explanations will appear.

2.1 · Recall

1. Which structure should you choose to map a customer code to its info?

A string, because everything is text A tuple, because it is frozen A dictionary: key → value pairs, direct access by key A set, because it deduplicates

The dict is the directory: access by key, .get() with no KeyError, iteration with .items().

2.1 · Application

2. Nabil writes "copie = ventes" then modifies "copie". What happens?

The program automatically raises KeyError The list is sorted automatically "ventes" stays intact: = always copies Both names share the same object: "ventes" is modified too

= shares, it does not copy: .copy() flat, deepcopy() for nested data.

2.1 · Arbitration

3. A colleague looks up each customer by scanning a 50,000-row list every time. Nabil's verdict?

To index: dict by customer code for direct access, list kept for order To replace with separate variables To keep: scanning is always fastest To replace with hand-sorted tuples

Looking up by identifier = indexed dict; the list stays for order and iteration.

2.2 · Recall

4. Why prefer f-strings over concatenation with +?

They close files automatically They are more readable, avoid manual conversions and offer built-in formatting (:.2f) They run without a Python interpreter They replace JSON files

f"Total : {total:.2f} €": variable, computation and format in one readable line.

2.2 · Application

5. Nabil reads an export with broken accents and a sometimes locked file. What does he fix?

He concatenates paths with + without pathlib He opens in "w" mode to read He removes all accents from the file He opens with with open(..., encoding="utf-8"): auto closing + intact accents

with closes by itself even on error; utf-8 encoding keeps accents; "w" overwrites, it does not read.

2.2 · Application

6. Nabil's summary must survive a restart and stay exchangeable. Which format?

A JSON file via json.dump/load: structured, universal, re-read as dicts and lists An unsaved global variable A file name with no content A print in the terminal

JSON = persistent lists and dicts, readable by other tools; the csv module for spreadsheets.

2.3 · Recall

7. What does "self" mean in a method?

Only the parent class The object itself: access to ITS attributes, passed automatically at the call The current file on disk The module imported at the top of the file

self is written in the definition, never at the call: each object keeps its own state.

2.3 · Application

8. Nabil creates VenteRemise(Vente) with a rate and super().init(). What did he do?

Deleting the parent class A copy-paste of Vente with no link An is-a inheritance: clean specialization, parent reused via super() A composition with no inheritance

Inheritance for is-a, kept shallow; composition (has) otherwise.

2.3 · Arbitration

9. A colleague proposes 6 inheritance levels and dicts + global functions "to go fast". Verdict?

To replace with global variables To accept: more levels = better design To accept for small scripts only To refuse: unreadable hierarchy, scattered data — targeted classes, inheritance ≤ 3, useful dunders

Data + behaviors in a class, measured inheritance, __str__/__eq__ and @dataclass where suitable.

2.4 · Application

10. Nabil converts input with float(). How to make it reliable?

A narrow try + except ValueError with retry; the rest propagates A silent except: pass over the whole program A try framing the whole import at once No check: input is always valid

Precise risky part framed, targeted type, real handling — never a bare except.

2.4 · Application

11. A negative amount arrives in Vente(...). Where to reject it?

Nowhere: accept it and fix later With raise ValueError from __init__: fail early and loudly, at the source With a silent print without raising By masking with except: pass

raise signals the violated invariant at the source; finally cleans up, else continues if OK.

2.4 · Arbitration

12. A colleague wants a general "except: pass" to "no longer see errors". Nabil's verdict?

To replace with deleting the tests To accept: no more visible errors = more reliability To refuse: it buries bugs — targeted excepts, real handling, logged unexpected that propagates To accept by adding a second pass

Silencing buries live bugs: only catch what you know how to handle.

What's next?

Score ≥ 80%: move to Level 3 — modules, pip/venv and advanced functions, on your solid data and object foundations.
Score 60-79%: review the missed sections (2.1 containers/copies, 2.2 f-strings/with/JSON, 2.3 classes/inheritance/dunders, 2.4 targeted try/raise) and redo the 48h challenges.
Score < 60%: rebuild Nabil's project step by step (sales as a list of dicts, report in f-strings + JSON, Vente classes, hardened import), then retake the test.
In all cases: keep the reflexes — right container for the right use, f-strings + with/encoding, first-class objects, targeted excepts with no silencing.