#!/usr/bin/python3

# This script checks the code for namespace related issues
# Note: This script is not perfect. It can not parse all LMMS files,
#       and does not contain a complete C++ parser. If you encounter
#       difficulties with this tests, it could be a fault of your
#       changes, but it could also be a fault of this script.

import re
import subprocess
import sys
from pathlib import Path
from typing import NamedTuple


# types

class BlockType:
    pass


IF_MACRO = BlockType()
HEADER_GUARD = BlockType()
CODE_BLOCK = BlockType()
EXTERN = BlockType()


class Expectation(NamedTuple):
    type: BlockType
    statement: str  # expected end statement
    name: str  # expected name in comment


# global variables

errors = 0


# functions

def caption(my_str):
    print(f'\n# {my_str}\n')


def error(where, match, my_str):
    global errors
    errors += 1
    if match:
        line = match.string[:match.start()].count('\n') + 1  # first line is 1
        where = f'{where}:{line}'
    print(f'Error: {where}: {my_str}')


if not Path('.gitmodules').is_file():
    print('You need to call this script from the LMMS top directory')
    exit(1)

result = subprocess.run(['git', 'ls-files', '*.[ch]', '*.[ch]pp', ':!tests/*'],
                        capture_output=True, text=True, check=True)

known_no_namespace_lmms = {
    # main.cpp
    'src/core/main.cpp',
    # nothing to set under a namespace
    'include/debug.h',
    'include/versioninfo.h',
    'plugins/CarlaBase/CarlaConfig/config.h',
    'plugins/CarlaBase/DummyCarla.cpp',
    # unclear why it has no namespace
    'plugins/ZynAddSubFx/RemoteZynAddSubFx.cpp',
}

exclude_files = re.compile(
    # not ours:
    'include/(aeffectx|fenv|ladspa).h|'
    'plugins/LadspaEffect/(calf|caps|cmt|swh|tap)/|'
    'plugins/MidiExport/MidiFile.hpp|'
    'plugins/ReverbSC/[a-z]|'
    'plugins/Sf2Player/fluidsynthshims.h|'
    '/portsmf/|'
    # only forward to headers that are not ours:
    'plugins/ZynAddSubFx/ThreadShims.h'
)

files = [Path(f) for f in result.stdout.splitlines() if not exclude_files.search(f)]

# Debug argument
if len(sys.argv) > 1:
    files = [Path(arg) for arg in sys.argv[1:]]

statement_pattern = re.compile(
    # Capture comments first to prevent them from matching any other regex
    #  Include next line if line ends with backslash
    r'/[*](.|\n)*?[*]/|//(.*\\\n)*.*|'

    # Macro with <30 lines and no other #if inside needs no end comment
    #  Match here to prevent from matching next regex
    #  With a (?!negative lookahead) we can allow all # except the ones followed by "if"
    r'^ *# *(?P<short_macro>ifn?def)(?=(([^#\n]|#(?!if|endif))*\n){,30} *# *endif)|'
    # Macro followed by name or comment
    #  With (?P<name>...) you can do a backreference to \name later
    r'^ *# *(?P<named_macro>ifn?def|endif)(( *// *| +)(?P<macro_name>\w+))?|'
    # Other macros where we don't want the argument
    r'^ *# *(?P<macro>include|if|el(se|if))|'

    # Namespace that contains no other braces needs no end comment
    #  With a (?=lookahead) we can let the "namespace" part be eaten by the parser
    #  but save the braces and their content for later so they can be matched again
    r'^ *namespace *[\w:]*\s*(?={[^{}]*})|'
    # Start of named namespace, extern "C" or just a opening brace
    r'(^ *(namespace +(?P<namespace>[\w:]+)|(?P<extern>extern *"C"))\s*)?(?P<opening_brace>{)|'
    # End of namespace including comment, or just a closing brace
    r'(?P<closing_brace>})( *// *namespace +(?P<namespace_end>[\w:]+))?'

    # In all the regexes above match both tab and space when a space is used
    r''.replace(' ', r'[\t ]'),
    # Make ^ match on every line, not just beginning of file
    re.MULTILINE)

# Comments and whitespace followed by header guard
header_guard_pattern = re.compile(r'^(/[*](.|\n)*?[*]/|//.*|\s)*#\s*(ifndef|pragma\s+once)')

# Namespace lmms
namespace_pattern = re.compile(r'^\s*namespace\s+lmms', re.MULTILINE)

#
# the real code
#

caption('namespace checks')

for cur_file in files:
    if cur_file.is_file():
        cur_text = cur_file.read_text(errors='replace')

        if cur_file.as_posix() not in known_no_namespace_lmms:
            namespace_pattern.search(cur_text) or error(cur_file, None, f'File has no namespace lmms')

        header_guard = str(cur_file).endswith('.h')
        expectations = []  # type: list[Expectation]

        if header_guard:
            if not header_guard_pattern.match(cur_text):
                error(cur_file, None, 'First statement should be header guard')
                header_guard = False

        for m in statement_pattern.finditer(cur_text):
            # Find the matched regex group
            statement = m.group('opening_brace') or m.group('closing_brace')
            if not statement:
                statement = '#' + (m.group('macro') or m.group('named_macro') or m.group('short_macro') or '')
            if not statement:
                continue

            # Start statements
            if statement == '{':
                etype = EXTERN if m.group('extern') else CODE_BLOCK
                expectations.append(Expectation(etype, '}', m.group('namespace')))
            elif statement.startswith('#if'):
                etype = HEADER_GUARD if header_guard else IF_MACRO
                expectations.append(Expectation(etype, '#endif', m.group('macro_name')))
                header_guard = False

            # End statements
            elif statement == '#endif' or statement.startswith('#el') or statement == '}':
                if not expectations:
                    error(cur_file, m, f'Unexpected {statement}')
                    break
                if statement.startswith('#el') and expectations[-1].statement == '#endif':
                    # After an #else or #elif the comment is no longer mandatory, since it's hard to define
                    expectations[-1] = Expectation(IF_MACRO, '#endif', '')
                    # Don't pop the expectations, we are still waiting for the #endif
                    continue
                exp = expectations.pop()
                name = m.group('macro_name') or m.group('namespace_end') or ''
                if statement != exp.statement:
                    error(cur_file, m, f'Expected {exp.statement} before {statement}')
                    break
                # Require no end comment for header guard
                elif exp.type is HEADER_GUARD and not name:
                    continue
                elif exp.name and name != exp.name:
                    comment = 'namespace ' if exp.type is CODE_BLOCK else ''
                    error(cur_file, m, f'Missing comment // {comment}{exp.name}')

            # Extra checks
            elif statement == '#include':
                if any(True for e in expectations if e.type is CODE_BLOCK):
                    error(cur_file, m, '#include inside a code block')

        else:
            # Leftover expected statements
            for exp in reversed(expectations):
                error(cur_file, None, f'Expected {exp.statement} before end of file')

caption('summary')

print(f'{str(errors)} errors.')
exit(1 if errors > 0 else 0)