#!/usr/bin/env python3
# SPDX-License-Identifier: Apache-2.0 OR GPL-2.0-or-later
#
# Copyright 2016 Jeremy Kerr <jk@ozlabs.org>

import os.path
import re
import sys
import string
import json
import argparse
import subprocess
from pyparsing import Regex, Literal, Word, Combine, OneOrMore, QuotedString, \
         lineno

json_params = {
    'indent': 1,
    'sort_keys': True,
    'separators': (',', ': '),
}

def create_parser():
    # Match a C-style comment starting with two *s
    comment = Regex(r'/\*\*(?P<content>.*?)\*/', re.DOTALL)

    # Match an @fwts-<tag> annotation (within the comment), plus the proceeding
    # text
    annotation = Regex(r'@fwts-(?P<tag>\w+)\W+(?P<text>.*?)(?=@fwts-|\Z)',
                re.DOTALL)

    # Match the following prlog() call
    log_call = (((Literal("prerror") + Literal('(').suppress()) |
                 (Literal("prlog")   + Literal('(').suppress() +
                  Word(string.ascii_letters + string.digits + '_') +
                  Literal(',').suppress())) +
                Combine(OneOrMore(QuotedString('"')), adjacent=False) +
                (Literal(')') | Literal(',')).suppress()
               )

    pattern = comment + log_call
    pattern.setWhitespaceChars(string.whitespace + '\n')

    def comment_action(tok):
        patterns = {}
        for result in annotation.scanString(tok['content']):
            patterns.update(result[0][0])
        return patterns

    def annotation_action(tok):
        return {
            tok['tag']: cleanup_content(tok['text'])
        }

    comment.setParseAction(comment_action)
    annotation.setParseAction(annotation_action)
    pattern.parseWithTabs()

    return pattern

def find_sources(dirname):
    sources = []

    def is_source(fname):
        return fname.endswith('.c')

    for directory, dirnames, filenames in os.walk(dirname):
        sources.extend([ os.path.join(directory, fname)
                         for fname in filenames if is_source(fname) ])
    return sources

def cleanup_content(content):
    comment_prefix_re = re.compile(r'^\s*\*\s*', re.MULTILINE)
    whitespace_re = re.compile(r'\s+')

    content = comment_prefix_re.sub(' ', content)
    content = whitespace_re.sub(' ', content)
    return content.strip()

def warn(loc, message):
    print >>sys.stderr, 'WARNING:%s:%d: %s' % (loc[0], loc[1], message)

def log_level_to_fwts(level):
    level_map = {
        'PR_EMERG':     'LOG_LEVEL_CRITICAL',
        'PR_ALERT':     'LOG_LEVEL_CRITICAL',
        'PR_CRIT':      'LOG_LEVEL_CRITICAL',
        'PR_ERR':       'LOG_LEVEL_CRITICAL',
        'PR_WARNING':   'LOG_LEVEL_HIGH',
        'PR_NOTICE':    'LOG_LEVEL_MEDIUM',
        'PR_PRINTF':    'LOG_LEVEL_MEDIUM',
    }
    return level_map.get(level, 'LOG_LEVEL_LOW')

def message_to_pattern(loc, msg):
    """ Convert a C printf()-style template to a pattern suitable for fwts """

    # Somewhat-simplified match for a %-template
    template_re = re.compile(
            '%(?P<flag>[-#0 +]*)'
            '(?P<width>(?:[0-9]*|\*))?'
            '(?P<precision>\.*(?:[1-9][0-9]*|\*))?'
            '(?:hh|h|ll|l|L|j|z|t)?'
            '(?P<conversion>[a-zA-Z%])')
    global is_regex
    is_regex = False

    def expand_template(match):
        global is_regex
        c = match.group('conversion').lower()
        if c == '%':
            return '%'
        is_regex = True
        if c in ['d', 'i', 'u']:
            return '[0-9]+'
        elif c == 'o':
            return '[0-7]+'
        elif c == 'x':
            return '[0-9a-f]+'
        elif c == 'p':
            return '(0x[0-9a-f]+|nil)'
        elif c == 's':
            return '.*'
        else:
            warn(loc, "Unknown template conversion '%s'" % match.group(0))
            return '.*'

    escape_re = re.compile(r'\\(?P<char>.)', re.DOTALL)
    def expand_escape(match):
        global is_regex
        c = match.group('char')
        if c == 'n':
            return '\n'
        elif c in ['\\', '"']:
            return c
        else:
            warn(loc, "Unhandled escape sequence '%s'" % match.group(0))
            is_regex = True
            return '.'

    pattern = template_re.sub(expand_template, msg)
    pattern = escape_re.sub(expand_escape, pattern)
    pattern = pattern.strip()

    compare_mode = "string"
    if is_regex:
        compare_mode = "regex"

    return (compare_mode, pattern)

def parse_patterns(parser, fname, tag):
    patterns = []
    data = open(fname).read()
    i = 1
    for result in parser.scanString(data):
        (token, loc, _) = result
        if token[1] == 'prlog':
            (annotations, logfn, level, msg) = token
        else:
            (annotations, logfn, msg) = token
            level = 'PR_ERR'

        loc = (fname, lineno(loc, data))

        if logfn != 'prlog' and logfn != 'prerror':
            warn(loc, "unknown log output function '%s'" % logfn)

        compare_mode, pattern_str = message_to_pattern(loc, msg)

        pattern = {
            'log_level': log_level_to_fwts(level),
            'compare_mode': compare_mode,
            'pattern': pattern_str,
            'last_tag': tag,
        }

        pattern.update(annotations)

        if not 'label' in pattern:
            warn(loc, "missing label")
            pattern['label'] = '%s:%d' % (fname, i)
            i += 1

        if not 'advice' in pattern:
            warn(loc, "missing advice")

        allowed_data = ['compare_mode', 'log_level',
                        'pattern', 'advice', 'label', 'last_tag']
        extras = set(pattern.keys()) - set(allowed_data)
        if extras:
            warn(loc, "unknown pattern annotation: %s" %
                    ','.join([ "'%s'" % e for e in extras]))
            for e in extras:
                del pattern[e]

        patterns.append(pattern)

    return patterns

if __name__ == '__main__':
    argparser = argparse.ArgumentParser(
            description='Generate FWTS olog definitions from the skiboot '
                        'source tree')
    argparser.add_argument('directories', metavar='DIR', nargs='*',
            help='path to source files (default .)', default=['.'])
    argparser.add_argument('--output', '-o', metavar='FILE',
            type=argparse.FileType('w'), default=sys.stdout,
            help='output to FILE (default to stdout)', nargs='?')
    args = argparser.parse_args()

    sources = []
    for directory in args.directories:
        try:
            git_tag = subprocess.check_output(["git","-C", directory, "describe", "--abbrev=0" ], text=True)
        except:
            git_tag = "???"
        git_tag = git_tag.replace("\n", "")
        sources.extend([ (x, git_tag) for x in find_sources(directory)])

    parser = create_parser()
    patterns = []
    for source, tag in sources:
        patterns.extend(parse_patterns(parser, source, tag))

    data = {'olog_error_warning_patterns': patterns}

    args.output.write(json.dumps(data, **json_params) + '\n')