debug_labeling_visualizer.py

import os
import sys

# Ensure src can be imported
sys.path.append(os.getcwd())

from src.data_process.dataset_builder import DatasetBuilder
from src.config import config
from src.core.constants import EXT_TO_LANG


def debug_specific_file(target_file_name):
    """
    Scans the input project, finds the specific file, and prints how it is labeled.
    """
    builder = DatasetBuilder()
    input_root = config.project.input_root

    # Find file
    target_path = None
    for root, dirs, files in os.walk(input_root):
        if target_file_name in files:
            target_path = os.path.join(root, target_file_name)
            break

    if not target_path:
        print(f"File {target_file_name} not found in {input_root}")
        return

    print(f"=== Debugging Labeling Logic for: {target_file_name} ===")

    # Read and Parse
    content = builder.file_manager.read_file(target_path)
    tree = builder.parser.parse_code(content, target_path)
    functions = builder.parser.extract_functions(tree, target_path)

    ext = os.path.splitext(target_path)[1]
    lang = EXT_TO_LANG.get(ext)

    for func in functions:
        print(f"\n--- Function: {func['name']} ---")
        lines = func['code'].split('\n')

        i = 0
        line_display_idx = 0

        while i < len(lines):
            line_content = lines[i]

            # 1. Skip empty lines (Simulation of DatasetBuilder logic)
            if not line_content.strip():
                print(f"{line_display_idx+1:03d} [SKIP] {line_content}")
                i += 1
                continue

            # 2. Skip existing Log lines (They are removed from input, serve as labels for prev line)
            if builder.is_log_line(line_content, lang):
                print(f"{line_display_idx+1:03d} [LOG ] {line_content}  <-- This line is REMOVED from input")
                i += 1
                continue

            # 3. Determine Label for current line
            # Look ahead to see if next valid line is a log
            label = 0
            j = i + 1
            has_log_after = False
            while j < len(lines):
                next_line = lines[j]
                if not next_line.strip():  # Skip empty lookahead
                    j += 1
                    continue
                if builder.is_log_line(next_line, lang):
                    has_log_after = True
                break

            label_str = "🔴 1 (INSERT HERE)" if has_log_after else "⚪ 0"
            print(f"{line_display_idx+1:03d} [{label_str}] {line_content}")

            i += 1
            line_display_idx += 1


if __name__ == "__main__":
    # Replace with a filename you know has logs in your dataset
    # e.g., "Axis2BackEndServerController.java"
    target_file = "Axis2BackEndServerController.java"
    debug_specific_file(target_file)