#!/usr/bin/env python3
# pylint: disable=C0103,C0114,C0115,C0116,C0209,R0912,R0914,R0915,W0125,W0621,exec-used
######################################################################

import argparse
import glob
import multiprocessing
import os
import re
import subprocess
import sys

RealPath = os.path.dirname(os.path.realpath(__file__))
Exclude_Branch_Regexps = []
Exclude_Line_Regexps = []
Remove_Gcda_Regexps = []
Remove_Sources = []
Source_Globs = []

if 'VERILATOR_ROOT' not in os.environ:
    os.environ['VERILATOR_ROOT'] = os.getcwd()

######################################################################


def test():
    if not os.path.exists("nodist/code_coverage.dat"):
        sys.exit("%Error: Run code_coverage from the top of the verilator kit")
    exec(open("./nodist/code_coverage.dat", "r", encoding="utf8").read())  # pylint: disable=consider-using-with

    if Args.stage_enabled[0]:
        ci_fold_start("distclean")
        print("Stage 0: distclean")
        run("make distclean || true")
        ci_fold_end()

    if Args.stage_enabled[1]:
        ci_fold_start("configure")
        print("Stage 1: configure (coverage on)")
        run("autoconf")
        run("./configure --enable-longtests --enable-coverage CXX=g++")
        ci_fold_end()

    if Args.stage_enabled[2]:
        ci_fold_start("build")
        print("Stage 2: build")
        nproc = multiprocessing.cpu_count()
        run("make -k -j " + str(nproc) + " VERILATOR_NO_OPT_BUILD=1")
        # The optimized versions will not collect good coverage, overwrite them
        run("cp bin/verilator_bin_dbg bin/verilator_bin")
        run("cp bin/verilator_coverage_bin_dbg bin/verilator_coverage_bin")
        ci_fold_end()

    if Args.stage_enabled[3]:
        ci_fold_start("test")
        print("Stage 3: make tests (with coverage on)")
        if not Args.tests:
            if not Args.scenarios or re.match('dist', Args.scenarios):
                run("make examples VERILATOR_NO_OPT_BUILD=1")
            run("make test_regress VERILATOR_NO_OPT_BUILD=1" +
                (" SCENARIOS='" + Args.scenarios + "'" if Args.scenarios else "") +
                (" DRIVER_HASHSET='--hashset=" + Args.hashset + "'" if Args.hashset else "") +
                ('' if Args.stop else ' || true'))
        else:
            for test in Args.tests:
                if not os.path.exists(test) and os.path.exists("test_regress/t/" + test):
                    test = "test_regress/t/" + test
                run(test)
        ci_fold_end()

    cc_dir = "nodist/obj_dir/coverage"
    if Args.stage_enabled[4]:
        ci_fold_start("gcno")
        print("Stage 4: Create gcno files under " + cc_dir)
        os.makedirs(cc_dir, exist_ok=True)
        os.makedirs(cc_dir + "/info", exist_ok=True)

        with subprocess.Popen("find . -print | grep .gcda", shell=True,
                              stdout=subprocess.PIPE) as sp:
            datout = sp.stdout.read()

        dats = {}
        for dat in datout.splitlines():
            dat = dat.decode('utf-8')
            dats[dat] = 1
        for dat in sorted(dats.keys()):
            gcno = re.sub(r'\.gcda$', '.gcno', dat)
            for regexp in Remove_Gcda_Regexps:
                if re.search(regexp, dat):
                    # Remove .gcda/.gcno for files we don't care about before we slowly
                    # read them
                    unlink_ok(dat)
                    unlink_ok(gcno)
                    del dats[dat]
                    break

        with subprocess.Popen("find . -print | grep .gcno", shell=True,
                              stdout=subprocess.PIPE) as sp:
            datout = sp.stdout.read()

        gcnos = {}
        for gcno in datout.splitlines():
            gcno = gcno.decode('utf-8')
            gbase = re.sub(r'.*/', '', gcno, count=1)
            gcnos[gbase] = os.path.abspath(gcno)
        # We need a matching .gcno for every .gcda, try to find a matching file elsewhere
        for dat in sorted(dats):
            gcno = re.sub(r'\.gcda$', '.gcno', dat)
            gbase = re.sub(r'.*/', '', gcno, count=1)
            if not os.path.exists(gcno):
                if gbase in gcnos:
                    os.symlink(gcnos[gbase], gcno)
                else:
                    print("MISSING .gcno for a .gcda: " + gcno, file=sys.stderr)
        ci_fold_end()

    if Args.stage_enabled[5]:
        ci_fold_start("fastcov")
        # Must run in root directory to find all files
        os.makedirs(cc_dir, exist_ok=True)
        run(RealPath + "/fastcov.py -b -c src/obj_dbg -X --lcov" +
            #            " --exclude /usr --exclude test_regress" + " -o " + cc_dir +
            " -o " + cc_dir + "/app_total.info")
        # For debug to convert single .gcna/.gcno in a directory to cov.info:
        #    lcov -c -d . -o cov.info
        ci_fold_end()

    if Args.stage_enabled[6]:
        ci_fold_start("clone")
        # No control file to override single lines, so replicate the sources
        # Also lets us see the insertion markers in the HTML source res
        print("Stage 6: Clone sources under " + cc_dir)
        clone_sources(cc_dir)
        ci_fold_end()

    if Args.stage_enabled[11]:
        ci_fold_start("dirs")
        print("Stage 11: Cleanup paths")
        cleanup_abs_paths_info(cc_dir, cc_dir + "/app_total.info", cc_dir + "/app_total.info")
        ci_fold_end()

    if Args.stage_enabled[12]:
        ci_fold_start("filter")
        print("Stage 12: Filter processed source files")
        inc = ''
        for globf in Source_Globs:
            for infile in glob.glob(globf):
                inc += " '" + infile + "'"
        exc = ''
        for globf in Remove_Sources:
            # Fastcov does exact match not globbing at present
            # Lcov requires whole path match so needs the glob
            globf = re.sub(r'^\*', '', globf)
            globf = re.sub(r'\*$', '', globf)
            exc += " '" + globf + "'"
        if inc != '':
            inc = "--include " + inc
        if exc != '':
            exc = "--exclude " + exc
        run("cd " + cc_dir + " ; " + RealPath + "/fastcov.py -C app_total.info " + inc + " " +
            exc + " -x --lcov -o app_total_f.info")
        ci_fold_end()

    if Args.stage_enabled[17]:
        ci_fold_start("report")
        print("Stage 17: Create HTML")
        run("cd " + cc_dir + " ; genhtml app_total_f.info --demangle-cpp" +
            " --rc lcov_branch_coverage=1 --rc genhtml_hi_limit=100 --output-directory html")
        ci_fold_end()

    if Args.stage_enabled[18]:
        ci_fold_start("upload")
        print("Stage 18: Upload")
        # curl -Os https://cli.codecov.io/latest/linux/codecov ; sudo chmod +x codecov
        # --disable-search does not seem to work
        # -d with false directory does not seem to work
        # So, remove gcno files before calling codecov
        upload_dir = "nodist/obj_dir/upload"
        os.makedirs(upload_dir, exist_ok=True)
        cmd = "ci/codecov -v upload-process -Z" + " -f " + cc_dir + "/app_total.info )"
        print("print: Not running:")
        print("  export CODECOV_TOKEN=<hidden>")
        print("  find . -name '*.gcno' -exec rm {} \\;")
        print("  " + cmd)
        ci_fold_end()

    if Args.stage_enabled[19]:
        print("*-* All Finished *-*")
        print("")
        print("* See report in " + cc_dir + "/html/index.html")
        print("* Remember to make distclean && ./configure before working on non-coverage")


def clone_sources(cc_dir):
    excluded_lines = 0
    excluded_br_lines = 0
    for globf in Source_Globs:
        for infile in glob.glob(globf):
            if re.match(r'^/', infile):
                sys.exit("%Error: source globs should be relative not absolute filenames, " +
                         infile)
            outfile = cc_dir + "/" + infile
            outpath = re.sub(r'/[^/]*$', '', outfile, count=1)
            os.makedirs(outpath, exist_ok=True)
            with open(infile, "r", encoding="utf8") as fh:
                with open(outfile, "w", encoding="utf8") as ofh:
                    lineno = 0
                    for line in fh:
                        lineno += 1
                        line = line.rstrip()
                        done = False
                        if re.search(r'LCOV_EXCL_LINE', line):
                            line += " LCOV_EXCL_BR_LINE"
                            done = True
                        elif re.search(r'LCOV_EXCL_START', line):
                            line += " LCOV_EXCL_BR_START"
                            done = True
                        elif re.search(r'LCOV_EXCL_STOP', line):
                            line += " LCOV_EXCL_BR_STOP"
                            done = True

                        for regexp in Exclude_Line_Regexps:
                            if done:
                                break
                            if re.search(regexp, line):
                                # print("%s:%d: %s" % (infile, lineno, line)
                                line += "  //code_coverage: // LCOV_EXCL_LINE LCOV_EXCL_BR_LINE"
                                excluded_lines += 1
                                excluded_br_lines += 1
                                done = True

                        for regexp in Exclude_Branch_Regexps:
                            if done:
                                break
                            if re.search(regexp, line):
                                # print("%s:%d: %s" % (infile, lineno, line)
                                line += "  //code_coverage: // LCOV_EXCL_BR_LINE"
                                excluded_br_lines += 1
                                done = True

                        ofh.write(line + "\n")
    print("Number of source lines automatically LCOV_EXCL_LINE'ed: %d" % excluded_lines)
    print("Number of source lines automatically LCOV_EXCL_BR_LINE'ed: %d" % excluded_br_lines)


def cleanup_abs_paths_info(cc_dir, infile, outfile):
    lines = []
    with open(infile, "r", encoding="utf8") as fh:
        for line in fh:
            if re.search(r'^SF:', line) and not re.search(r'^SF:/usr/', line):
                line = re.sub(os.environ['VERILATOR_ROOT'] + '/', '', line, count=1)
                line = re.sub(cc_dir + '/', '', line, count=1)
                line = re.sub(r'^SF:.*?/include/', 'SF:include/', line, count=1)
                line = re.sub(r'^SF:.*?/src/', 'SF:src/', line, count=1)
                line = re.sub(r'^SF:.*?/test_regress/', 'SF:test_regress/', line, count=1)
                line = re.sub(r'obj_dbg/verilog.y$', 'verilog.y', line)
                # print("Remaining SF: "+line)
            lines.append(line)

    with open(outfile, "w", encoding="utf8") as ofh:
        for line in lines:
            ofh.write(line)


def cleanup_abs_paths_json(cc_dir, infile, outfile):
    # Handcrafted cleanup, alternative would be to deserialize/serialize JSON,
    # but this is much faster
    lines = []
    with open(infile, "r", encoding="utf8") as fh:
        for line in fh:
            line = re.sub('"' + os.environ['VERILATOR_ROOT'] + '/', '"', line)
            line = re.sub('"' + cc_dir + '/', '"', line)
            line = re.sub(r'obj_dbg/verilog.y$', 'verilog.y', line)
            lines.append(line)

    with open(outfile, "w", encoding="utf8") as ofh:
        for line in lines:
            ofh.write(line)


######################################################################
# .dat file callbacks


def exclude_branch_regexp(*regexps):
    Exclude_Branch_Regexps.extend(regexps)


def exclude_line_regexp(*regexps):
    Exclude_Line_Regexps.extend(regexps)


def remove_gcda_regexp(*regexps):
    Remove_Gcda_Regexps.extend(regexps)


def remove_source(*sources):
    Remove_Sources.extend(sources)


def source_globs(*dirs):
    Source_Globs.extend(dirs)


#######################################################################


def run(command):
    # run a system command, check errors
    print("\t%s" % command)
    status = subprocess.call(command, shell=True)
    if status < 0:
        raise RuntimeError("%Error: Command failed " + command + ", stopped")


def unlink_ok(filename):
    try:
        os.unlink(filename)
    except OSError:
        pass


def ci_fold_start(action):
    print("::group::" + action, flush=True)


def ci_fold_end():
    print("::endgroup::\n", flush=True)


#######################################################################
#######################################################################

parser = argparse.ArgumentParser(
    allow_abbrev=False,
    formatter_class=argparse.RawDescriptionHelpFormatter,
    description="""code_coverage builds Verilator with C++ coverage support and runs
tests with coverage enabled. This will rebuild the current object
files. Run as:

  cd $VERILATOR_ROOT
  nodist/code_coverage""",
    epilog="""Copyright 2019-2025 by Wilson Snyder. This program is free software; you
can redistribute it and/or modify it under the terms of either the GNU
Lesser General Public License Version 3 or the Perl Artistic License
Version 2.0.

SPDX-License-Identifier: LGPL-3.0-only OR Artistic-2.0""")

parser.add_argument('--debug', action='store_true', help='enable debug')
parser.add_argument('--hashset',
                    action='store',
                    help='pass test hashset onto driver.py test harness')
parser.add_argument('--scenarios',
                    action='store',
                    help='pass test scenarios onto driver.py test harness')
parser.add_argument('--stages',
                    '--stage',
                    action='store',
                    help='runs a specific stage or range of stages (see the script)')
parser.add_argument(
    '--tests',
    '--test',
    action='append',
    default=[],
    help='Instead of normal regressions, run the specified test(s), may be used multiple times')
parser.add_argument('--no-stop',
                    dest='stop',
                    action='store_false',
                    help='do not stop collecting data if tests fail')

parser.set_defaults(stop=True)
Args = parser.parse_args()

if True:
    start = 0
    end = 99
    Args.stage_enabled = {}
    if Args.stages:
        match_one = re.match(r'^(\d+)$', Args.stages)
        match_range = re.match(r'^(\d+)-(\d+)$', Args.stages)
        match_to = re.match(r'^-(\d+)$', Args.stages)
        match_from = re.match(r'^(\d+)-$', Args.stages)
        if match_one:
            start = end = int(match_one.group(1))
        elif match_range:
            start = int(match_range.group(1))
            end = int(match_range.group(2))
        elif match_to:
            end = int(match_to.group(1))
        elif match_from:
            start = int(match_from.group(1))
        else:
            sys.exit("%Error: --stages not understood: " + Args.stages)
    for n in range(0, 100):
        Args.stage_enabled[n] = False
    for n in range(start, end + 1):
        Args.stage_enabled[n] = True

test()

######################################################################
# Local Variables:
# compile-command: "cd .. ; nodist/code_coverage "
# End: