python code containing href

from pathlib import Path
import re

# Directory containing the PHP files
ROOT_DIRECTORY = Path(r"C:\Users\Mayank\Documents\GitHub\ERP")

# Output text file
OUTPUT_FILE = ROOT_DIRECTORY / "php_search_results.txt"

# Exclude files when any folder in their path contains these terms
EXCLUDED_FOLDER_TERMS = ("tcpdf", "phpqrcode")

# Case-insensitive search patterns
SEARCH_PATTERNS = {
    "href and location on same line": re.compile(
        r"(?=.*\bhref\b)(?=.*\blocation\b)",
        re.IGNORECASE,
    ),
    "window.location.href": re.compile(
        r"\bwindow\s*\.\s*location\s*\.\s*href\b",
        re.IGNORECASE,
    ),
    "Location:": re.compile(
        r"\blocation\s*:",
        re.IGNORECASE,
    ),
}


def is_excluded(file_path: Path) -> bool:
    """Return True when a folder in the path contains an excluded term."""
    relative_parts = file_path.relative_to(ROOT_DIRECTORY).parts[:-1]

    return any(
        excluded_term in folder_name.lower()
        for folder_name in relative_parts
        for excluded_term in EXCLUDED_FOLDER_TERMS
    )


def find_matching_patterns(line: str) -> list[str]:
    """Return the names of all patterns matched by a line."""
    return [
        pattern_name
        for pattern_name, pattern in SEARCH_PATTERNS.items()
        if pattern.search(line)
    ]


def search_php_files() -> None:
    if not ROOT_DIRECTORY.exists():
        raise FileNotFoundError(
            f"Directory does not exist: {ROOT_DIRECTORY}"
        )

    matched_files = 0
    matched_lines = 0
    scanned_files = 0

    with OUTPUT_FILE.open("w", encoding="utf-8") as output:
        output.write(f"Search directory: {ROOT_DIRECTORY}\n")
        output.write(
            "Excluded folder terms: "
            + ", ".join(EXCLUDED_FOLDER_TERMS)
            + "\n"
        )
        output.write("=" * 100 + "\n\n")

        for php_file in ROOT_DIRECTORY.rglob("*.php"):
            if not php_file.is_file() or is_excluded(php_file):
                continue

            scanned_files += 1
            file_matches = []

            try:
                # errors="replace" prevents one badly encoded file
                # from stopping the entire search.
                with php_file.open(
                    "r",
                    encoding="utf-8",
                    errors="replace",
                ) as source_file:
                    for line_number, line in enumerate(source_file, start=1):
                        matched_patterns = find_matching_patterns(line)

                        if matched_patterns:
                            file_matches.append(
                                (
                                    line_number,
                                    line.rstrip("\r\n"),
                                    matched_patterns,
                                )
                            )

            except OSError as error:
                output.write(f"ERROR READING: {php_file}\n")
                output.write(f"{error}\n")
                output.write("-" * 100 + "\n\n")
                continue

            if not file_matches:
                continue

            matched_files += 1
            matched_lines += len(file_matches)

            output.write(f"File: {php_file}\n")
            output.write(
                f"Relative path: {php_file.relative_to(ROOT_DIRECTORY)}\n"
            )
            output.write("-" * 100 + "\n")

            for line_number, line_text, matched_patterns in file_matches:
                output.write(
                    f"Line {line_number} "
                    f"[{', '.join(matched_patterns)}]\n"
                )
                output.write(f"{line_text}\n\n")

            output.write("=" * 100 + "\n\n")

        output.write("\nSUMMARY\n")
        output.write("-" * 100 + "\n")
        output.write(f"PHP files scanned: {scanned_files}\n")
        output.write(f"Files containing matches: {matched_files}\n")
        output.write(f"Matching lines: {matched_lines}\n")

    print("Search completed.")
    print(f"PHP files scanned: {scanned_files}")
    print(f"Files containing matches: {matched_files}")
    print(f"Matching lines: {matched_lines}")
    print(f"Results exported to: {OUTPUT_FILE}")


if __name__ == "__main__":
    try:
        search_php_files()
    except Exception as error:
        print(f"Search failed: {error}")