Lesson 26. Compare our learning parser with qazaq-ir

Why this matters

Two people inspect a bicycle: a learner names three familiar parts, while a workshop also records each part’s condition. Compare only the work they have in common. Our lesson 11 recognizes two preset forms; qazaq-ir is a separate Rust project that returns JSON with a root, suffix chain, and analysis status. It is not a source for club schedules.

Before the code

The program always shows the learning analysis мектептерімізде → мектеп + тер + іміз + де. The second experiment is optional. sys.argv is the list of command words: its first item names the program, and the next may be a path to the qazaq-ir executable. Path(...).is_file() checks that the file exists. subprocess.run starts the external program with an argument list rather than a shell. capture_output=True stores its output, text=True decodes it as text, and check=False lets us inspect failure ourselves. timeout=10 limits waiting, and returncode reports success. json.loads reads its JSON, and tokens is its list of analyzed words. The first experiment works without Rust installed.

qazaq-ir source project.

From the project root, enter the step folder and run the program:

cd course/kazakh-ai/step-26
python3 compare.py

On Windows, replace python3 with py.

import json
import subprocess
import sys
from pathlib import Path

word = "мектептерімізде"
ours = {"root": "мектеп", "parts": ["тер", "іміз", "де"]}
print("Оқу үлгісі:", word, "→", ours["root"], ours["parts"])
if len(sys.argv) == 1:
    print("qazaq-ir: қосымша салыстыру іске қосылмады")
else:
    binary = Path(sys.argv[1])
    if not binary.is_file():
        print("qazaq-ir: бағдарлама файлы табылмады")
    else:
        run = subprocess.run([str(binary), "analyze", "--format", "compact", word],
                             text=True, capture_output=True, timeout=10, check=False)
        if run.returncode != 0:
            print("qazaq-ir: іске қосу қатесі")
        else:
            result = json.loads(run.stdout)
            for token in result["tokens"]:
                print("qazaq-ir:", token["surface"], "→", token["root"],
                      token["analysis_status"])

Step files.

How the program works

With no second argument, the program prints qazaq-ir: қосымша салыстыру іске қосылмады. For an optional comparison, build the project with cargo build --release -p qazaq-ir-cli and pass the resulting target/release/qazaq-ir path (.exe on Windows). Installing Rust is a separate task; the Python course does not require it. A local check with qazaq-ir 0.31.0 returned root: мектеп and analysis_status: partial for this form: the root was found, but a full parse was not confirmed. Another version may differ. Compare root and analysis_status in the project JSON. Agreement on one word does not establish which system is “better AI,” nor does it verify a timetable fact or its source.

Support map

One word → learning parse → optional qazaq-ir CLI → compare root and status → state both limits.

Recall and check

Hide the code and recall the two lines available without Rust: Оқу үлгісі: мектептерімізде → мектеп ['тер', 'іміз', 'де'] and the skipped-comparison message. Hint: follow len(sys.argv) == 1. Then pass a nonexistent file path and find бағдарлама файлы табылмады. Common mistake: treating a matching root as proof of a schedule. Word analysis and fact checking are different stages.