CC-Tweaked/tools/parse-reports.py

#!/usr/bin/env python3

# SPDX-FileCopyrightText: 2021 The CC: Tweaked Developers
#
# SPDX-License-Identifier: MPL-2.0

"""
Parse reports generated by Gradle and convert them into GitHub annotations.

See https://github.com/actions/toolkit/blob/master/docs/problem-matchers.md and
https://github.com/actions/toolkit/blob/master/docs/commands.md.
"""

import dataclasses
import os.path
import pathlib
import re
import xml.etree.ElementTree as ET

LUA_ERROR_LOCATION = re.compile(r"^\s+(/[\w./-]+):(\d+):", re.MULTILINE)
JAVA_LUA_ERROR_LOCATION = re.compile(r"^java.lang.IllegalStateException: (/[\w./-]+):(\d+):")
JAVA_ERROR_LOCATION = re.compile(r"^\tat ([\w.]+)\.[\w]+\([\w.]+:(\d+)\)$", re.MULTILINE)
ERROR_MESSAGE = re.compile(r"(.*)\nstack traceback:", re.DOTALL)

SPACES = re.compile(r"\s+")

SOURCE_LOCATIONS = [
    "src/main/java",
    "src/main/resources/data/computercraft/lua",
    "src/test/java",
    "src/test/resources",
]

PROJECT_LOCATIONS = [
    "projects/core-api",
    "projects/core",
    "projects/common-api",
    "projects/common",
    "projects/fabric-api",
    "projects/fabric",
    "projects/forge-api",
    "projects/forge",
]


def find_file(path: str) -> str | None:
    while len(path) > 0 and path[0] == "/":
        path = path[1:]

    for project in PROJECT_LOCATIONS:
        for source_dir in SOURCE_LOCATIONS:
            child_path = os.path.join(project, source_dir, path)
            if os.path.exists(child_path):
                return child_path

    return None


def find_location(message: str) -> tuple[str, str] | None:
    location = LUA_ERROR_LOCATION.search(message)
    if location:
        file = find_file(location[1])
        if file:
            return file, location[2]

    location = JAVA_LUA_ERROR_LOCATION.search(message)
    if location:
        file = find_file(location[1])
        if file:
            return file, location[2]

    for location in JAVA_ERROR_LOCATION.findall(message):
        file = find_file(location[0].replace(".", "/") + ".java")
        if file:
            return file, location[1]

    return None


_OTR_CORE = "https://schemas.opentest4j.org/reporting/core/0.2.0"
_OTR_EVENTS = "https://schemas.opentest4j.org/reporting/events/0.2.0"
_OTR_JAVA = "https://schemas.opentest4j.org/reporting/java/0.2.0"


def _parse_junit_file(path: pathlib.Path):
    tests: dict[str, str] = {}
    tree = ET.parse(path).getroot()
    for event in tree:
        if event.tag == f"{{{_OTR_CORE}}}infrastructure":
            continue
        elif event.tag == f"{{{_OTR_EVENTS}}}started":
            test_id: str = event.get("id")
            test_name: str = event.get("name")
            if (parent_id := event.get("parentId")) is not None:
                test_name = tests[parent_id] + " ⊳ " + test_name

            tests[test_id] = test_name
        elif event.tag == f"{{{_OTR_EVENTS}}}finished":
            test_id: str = event.get("id")
            test_name = tests[test_id]

            result = event.find(f"./{{{_OTR_CORE}}}result")
            if result is None:
                continue

            if result.get("status") in ("SUCCESSFUL", "ABORTED"):
                continue

            reason = result.find(f"./{{{_OTR_CORE}}}reason")
            throwable = result.find(f"./{{{_OTR_JAVA}}}throwable")

            message = reason.text if reason is not None else ""
            full_message = throwable.text if throwable is not None else ""

            location = find_location(full_message)
            if (match := ERROR_MESSAGE.match(message)) is not None:
                error = match.group(1)
            else:
                error = message
            error = SPACES.sub(' ', error)

            if location:
                print(f"## {location[0]}:{location[1]}: {test_name} failed: {error}")
            else:
                print(f"::error::{test_name} failed: {error}")

            if len(full_message) > 0:
                print("::group::Full error message")
                print(full_message)
                print("::endgroup::")
        else:
            print(f"Unknown tag: {event.tag}")



def parse_junit() -> None:
    """
    Scrape JUnit test reports for errors. We determine the location from the Lua
    or Java stacktrace.
    """
    print("::add-matcher::.github/matchers/junit.json")

    for project in PROJECT_LOCATIONS:
        for path in pathlib.Path(os.path.join(project, "build/test-results")).glob("*/open-test-report.xml"):
            _parse_junit_file(path)

        for path in pathlib.Path(os.path.join(project, "build/test-results")).glob("run*.xml"):
            _parse_junit_file(path)

    print("::remove-matcher owner=junit::")


def _parse_checkstyle(path: pathlib.Path):
    for file in ET.parse(path).getroot():
        for error in file:
            filename = os.path.relpath(file.attrib["name"])

            attrib = error.attrib
            print(
                f"{attrib['severity']} {filename}:{attrib['line']}:{attrib.get('column', 1)}: {SPACES.sub(' ', attrib['message'])}"
            )


def parse_checkstyle() -> None:
    """
    Scrape JUnit test reports for errors. We determine the location from the Lua
    or Java stacktrace.
    """
    print("::add-matcher::.github/matchers/checkstyle.json")

    for project in PROJECT_LOCATIONS:
        for path in pathlib.Path(os.path.join(project, "build/reports/checkstyle/")).glob("*.xml"):
            _parse_checkstyle(path)

    print("::remove-matcher owner=checkstyle::")


if __name__ == "__main__":
    parse_junit()
    parse_checkstyle()