learn_bash_from_session_data 1.0.6 → 1.0.8

package/scripts/quiz_generator.py

@@ -729,20 +729,44 @@ def _generate_distractor_flags(cmd: str, correct_flag: str, count: int = 3) -> l
 
 
 def _generate_distractor_descriptions(correct_desc: str, count: int = 3) -> list[str]:
-    """Generate plausible wrong descriptions."""
+    """Generate plausible wrong descriptions using command-level descriptions for length parity."""
     distractors = []
 
-    # Collect all descriptions from merged sources
-    all_descriptions = []
-    for cmd in _get_all_flagged_commands():
-        all_descriptions.extend(_get_flags_for_cmd(cmd).values())
-
-    # Remove duplicates and the correct answer
-    all_descriptions = list(set(all_descriptions))
-    all_descriptions = [d for d in all_descriptions if d.lower() != correct_desc.lower()]
-
-    random.shuffle(all_descriptions)
-    return all_descriptions[:count]
+    # First: collect command-level descriptions from COMMAND_DB (similar length to correct answer)
+    cmd_descriptions = []
+    for cmd_name in COMMAND_DB:
+        cmd_info = COMMAND_DB[cmd_name]
+        desc = cmd_info.get('description', '')
+        if desc and desc.lower() != correct_desc.lower():
+            # Truncate very long descriptions to similar length as correct answer
+            max_len = max(len(correct_desc) + 40, 80)
+            if len(desc) > max_len:
+                desc = desc[:max_len].rsplit(' ', 1)[0] + '...'
+            cmd_descriptions.append(desc)
+
+    if cmd_descriptions:
+        random.shuffle(cmd_descriptions)
+        distractors.extend(cmd_descriptions[:count])
+
+    # Fallback: use flag descriptions if not enough command descriptions
+    if len(distractors) < count:
+        all_flag_descs = []
+        for cmd in _get_all_flagged_commands():
+            all_flag_descs.extend(_get_flags_for_cmd(cmd).values())
+        all_flag_descs = list(set(all_flag_descs))
+        all_flag_descs = [d for d in all_flag_descs if d.lower() != correct_desc.lower()]
+        random.shuffle(all_flag_descs)
+        distractors.extend(all_flag_descs[:count - len(distractors)])
+
+    # Remove duplicates
+    seen = set()
+    unique = []
+    for d in distractors:
+        dl = d.lower()
+        if dl not in seen:
+            seen.add(dl)
+            unique.append(d)
+    return unique[:count]
 
 
 def generate_what_does_quiz(
@@ -793,9 +817,27 @@ def generate_what_does_quiz(
         for rel_cmd in related_cmds[:3 - len(distractor_descriptions)]:
             distractor_descriptions.append(f"Runs {rel_cmd} to process files")
 
-    # Ensure we have exactly 3 distractors
+    # Ensure we have exactly 3 distractors with plausible alternatives
+    fallback_actions = [
+        f"List directory contents with detailed file information",
+        f"Search recursively through files for matching patterns",
+        f"Display or modify file permissions and ownership",
+        f"Compress or archive files for storage and transfer",
+        f"Monitor system processes and resource usage",
+        f"Download files from a remote server or URL",
+        f"Edit configuration files in the default text editor",
+        f"Install or update packages from the package manager",
+    ]
+    random.shuffle(fallback_actions)
+    fb_idx = 0
     while len(distractor_descriptions) < 3:
-        distractor_descriptions.append(f"Performs an unrelated {base_cmd} operation")
+        if fb_idx < len(fallback_actions):
+            fallback = fallback_actions[fb_idx]
+            if fallback.lower() != correct_desc.lower():
+                distractor_descriptions.append(fallback)
+            fb_idx += 1
+        else:
+            distractor_descriptions.append(f"Run a system utility to process input data")
 
     # Create options (shuffle positions)
     options = []
@@ -929,19 +971,20 @@ def generate_build_command_quiz(
     parsed = _parse_command(cmd_string)
     base_cmd = parsed["base"]
 
-    # Create the correct command structure
-    correct_components = [base_cmd] + parsed["flags"] + parsed["args"]
-    correct_answer = " ".join(correct_components)
+    # Use the original command string as correct answer (preserves flag-argument ordering)
+    correct_answer = cmd_string.strip()
 
-    # Generate wrong arrangements
+    # Generate wrong arrangements using parsed components
+    all_parts = [base_cmd] + parsed["flags"] + parsed["args"]
     distractors = []
 
-    # Distractor 1: Wrong order
-    if len(correct_components) > 2:
-        wrong_order = correct_components.copy()
+    # Distractor 1: Wrong order of components
+    if len(all_parts) > 2:
+        wrong_order = all_parts.copy()
         random.shuffle(wrong_order)
-        if wrong_order != correct_components:
-            distractors.append(" ".join(wrong_order))
+        wrong_str = " ".join(wrong_order)
+        if wrong_str != correct_answer:
+            distractors.append(wrong_str)
 
     # Distractor 2: Missing flag
     if parsed["flags"]:
@@ -961,15 +1004,31 @@ def generate_build_command_quiz(
         wrong_cmd = [related[0]] + parsed["flags"] + parsed["args"]
         distractors.append(" ".join(wrong_cmd))
 
-    # Ensure we have exactly 3 distractors
+    # Ensure we have exactly 3 distractors with plausible alternatives
+    # Use real flags from the knowledge base as fallback distractors
+    all_cmd_flags = list(_get_flags_for_cmd(base_cmd).keys())
+    random.shuffle(all_cmd_flags)
+    fb_flag_idx = 0
     while len(distractors) < 3:
-        # Add a clearly wrong option
-        distractors.append(f"{base_cmd} --invalid-option")
+        if fb_flag_idx < len(all_cmd_flags):
+            fallback_flag = all_cmd_flags[fb_flag_idx]
+            fallback_cmd = f"{base_cmd} {fallback_flag} {' '.join(parsed['args'])}"
+            if fallback_cmd.strip() != correct_answer:
+                distractors.append(fallback_cmd.strip())
+            fb_flag_idx += 1
+        else:
+            # Use a related command as last resort
+            related = _get_related_commands(base_cmd)
+            if related:
+                rel = related[len(distractors) % len(related)]
+                distractors.append(f"{rel} {' '.join(parsed['flags'])} {' '.join(parsed['args'])}".strip())
+            else:
+                distractors.append(f"{base_cmd} {' '.join(parsed['args'])}".strip())
 
     # Remove duplicates and correct answer from distractors
     distractors = list(set(d for d in distractors if d != correct_answer))[:3]
     while len(distractors) < 3:
-        distractors.append(f"{base_cmd} --wrong-flag")
+        distractors.append(f"{base_cmd} {' '.join(parsed['args'])}".strip())
 
     # Create options
     all_answers = [correct_answer] + distractors[:3]
@@ -1203,7 +1262,13 @@ def generate_quiz_set(
     target_build = max(1, int(count * 0.2))
     target_spot_diff = max(1, int(count * 0.15))
 
-    used_commands = set()
+    # Track used commands per quiz type to avoid repeating the same command
+    used_per_type = {
+        QuizType.WHAT_DOES: set(),
+        QuizType.WHICH_FLAG: set(),
+        QuizType.BUILD_COMMAND: set(),
+        QuizType.SPOT_DIFFERENCE: set(),
+    }
 
     # Generate "What does this do?" questions
     random.shuffle(weighted_commands)
@@ -1211,38 +1276,47 @@ def generate_quiz_set(
         if len([q for q in questions if q.quiz_type == QuizType.WHAT_DOES]) >= target_what_does:
             break
         cmd_id = cmd.get("command", "")
-        if cmd_id not in used_commands:
+        if cmd_id not in used_per_type[QuizType.WHAT_DOES]:
             q = generate_what_does_quiz(cmd)
             questions.append(q)
-            used_commands.add(cmd_id)
+            used_per_type[QuizType.WHAT_DOES].add(cmd_id)
 
     # Generate "Which flag?" questions
     random.shuffle(weighted_commands)
     for cmd in weighted_commands:
         if len([q for q in questions if q.quiz_type == QuizType.WHICH_FLAG]) >= target_which_flag:
             break
-        q = generate_which_flag_quiz(cmd)
-        if q:
-            questions.append(q)
+        cmd_id = cmd.get("command", "")
+        if cmd_id not in used_per_type[QuizType.WHICH_FLAG]:
+            q = generate_which_flag_quiz(cmd)
+            if q:
+                questions.append(q)
+                used_per_type[QuizType.WHICH_FLAG].add(cmd_id)
 
     # Generate "Build the command" questions
     random.shuffle(weighted_commands)
     for cmd in weighted_commands:
         if len([q for q in questions if q.quiz_type == QuizType.BUILD_COMMAND]) >= target_build:
             break
-        q = generate_build_command_quiz(cmd)
-        questions.append(q)
+        cmd_id = cmd.get("command", "")
+        if cmd_id not in used_per_type[QuizType.BUILD_COMMAND]:
+            q = generate_build_command_quiz(cmd)
+            questions.append(q)
+            used_per_type[QuizType.BUILD_COMMAND].add(cmd_id)
 
     # Generate "Spot the difference" questions
     random.shuffle(weighted_commands)
     for cmd in weighted_commands:
         if len([q for q in questions if q.quiz_type == QuizType.SPOT_DIFFERENCE]) >= target_spot_diff:
             break
-        variant = _create_similar_command_variant(cmd)
-        if variant:
-            q = generate_spot_difference_quiz(cmd, variant)
-            if q:
-                questions.append(q)
+        cmd_id = cmd.get("command", "")
+        if cmd_id not in used_per_type[QuizType.SPOT_DIFFERENCE]:
+            variant = _create_similar_command_variant(cmd)
+            if variant:
+                q = generate_spot_difference_quiz(cmd, variant)
+                if q:
+                    questions.append(q)
+                    used_per_type[QuizType.SPOT_DIFFERENCE].add(cmd_id)
 
     # Shuffle final questions
     random.shuffle(questions)