#!/usr/bin/env python3
#
# Copyright (c) 2015 Jon Turney
#
# Permission is hereby granted, free of charge, to any person obtaining a copy
# of this software and associated documentation files (the "Software"), to deal
# in the Software without restriction, including without limitation the rights
# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
# copies of the Software, and to permit persons to whom the Software is
# furnished to do so, subject to the following conditions:
#
# The above copyright notice and this permission notice shall be included in
# all copies or substantial portions of the Software.
#
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
# THE SOFTWARE.
#

#
# parser for .hint files
#

import argparse
import re
from collections import OrderedDict

try:
    import license_expression
except ModuleNotFoundError:
    licensing = None
else:
    # reach inside license_expression to add custom license ids we permit
    json = license_expression.get_license_index()
    extra_licenses = [
        'Linux-man-pages-copyleft',  # requires SPDX license-list 3.15
        'OFSFDL',                    # "Old FSF documentation license"
        'Public-Domain',
    ]
    for l in extra_licenses:
        if not any(j["spdx_license_key"] == l for j in json):
            json.append({"spdx_license_key": l})
    licensing = license_expression.build_spdx_licensing(json)

# types of key:
# 'multilineval' - always have a value, which may be multiline
# 'val'          - always have a value
# 'optval'       - may have an empty value
# 'noval'        - always have an empty value
keytypes = ['multilineval', 'val', 'optval', 'noval']

# kinds of hint file, and their allowed keys
pvr, override, spvr = range(3)

hintkeys = {}

commonkeys = {
    'ldesc': 'multilineval',
    'category': 'val',
    'sdesc': 'val',
    'test': 'noval',   # mark the package as a test version
    'version': 'val',  # version override
    'disable-check': 'val',
    'notes': 'val',    # tool notes; not significant to calm itself
}

hintkeys[pvr] = commonkeys.copy()
hintkeys[pvr].update({
    'message': 'multilineval',
    'external-source': 'val',
    'requires': 'optval',
    'obsoletes': 'optval',
    'provides': 'val',
    'conflicts': 'val',
})

hintkeys[spvr] = commonkeys.copy()
hintkeys[spvr].update({
    'skip': 'noval',   # in all spvr hints, but ignored
    'homepage': 'val',
    'build-depends': 'optval',
    'license': 'val',
})

hintkeys[override] = {
    'keep': 'val',
    'keep-count': 'val',
    'keep-count-test': 'val',
    'keep-days': 'val',
    'keep-superseded-test': 'noval',
    'disable-check': 'val',
    'replace-versions': 'val',
    'noretain': 'val',
}

# valid categories
categories = ['accessibility',
              'admin',
              'archive',
              'audio',  # debian uses 'sound'
              'base',
              'comm',
              'database',
              'debug',
              'devel',
              'doc',
              'editors',
              'fonts',
              'games',
              'gnome',
              'graphics',
              'interpreters',
              'kde',
              'libs',
              'lua',
              'lxde',
              'mail',
              'mate',
              'math',
              'net',
              'ocaml',
              'office',
              'perl',
              'php',
              'publishing',
              'python',
              'ruby',
              'scheme',
              'science',
              'security',
              'shells',
              'source',  # added to all source packages created by deduplicator to ensure they have a category
              'sugar',
              'system',
              'tcl',
              'text',
              'utils',
              'video',
              'virtual',
              'web',
              'x11',
              'xfce',
              '_obsolete',
              ]


#
# A simple lexer to handle multi-line quoted values
#
# Historically, a multi-line quote is terminated only by a quote at the end of
# the line, and embedded quotes are transformed to single quotes.  So there is
# no escaping of embedded quotes, and no way to represent one.
#
# XXX: Fix the few packages which use embedded quotes, then we can switch this
# to a simpler character by character lexer, which just reads until next
# newline, and next quote when we encounter a quote.
#
def item_lexer(c):
    i = -1
    lines = c.splitlines()

    while i < len(lines) - 1:
        i = i + 1
        o = lines[i]

        # discard lines starting with '#'
        if o.startswith('#'):
            continue

        o = o.strip()

        # discard empty lines
        if not o:
            continue

        # line containing quoted text
        if o.count('"') == 2:
            yield (i, o, None)
            continue

        # if the line contains an opening quote
        if '"' in o:
            # continue reading lines till closing quote
            while i < len(lines) - 1:
                i = i + 1
                # multi-line quoted text preserves any leading space used for
                # indentation, but removes any trailing space
                o = o + '\n' + lines[i].rstrip()
                # multi-line quoted text is only terminated by a quote at the
                # end of the line
                if o.endswith('"'):
                    yield (i, o, None)
                    break
            else:
                yield (i, o, "unterminated quote")

            continue

        # an unquoted line
        yield (i, o, None)


def split_trim_sort_join(hint, splitchar, joinchar=None):
    if joinchar is None:
        joinchar = splitchar + ' '

    return joinchar.join(sorted([s.strip() for s in hint.split(splitchar)]))


# parse the file |fn| as a .hint file of kind |kind|
def hint_file_parse(fn, kind, strict=False):
    hints = OrderedDict()
    errors = []
    warnings = []

    assert (kind in hintkeys) or (kind is None)

    with open(fn, 'rb') as f:
        c = f.read()

        # validate that .hint file is UTF-8 encoded
        try:
            c = c.decode('utf-8')

            # parse as key:value items
            for (i, item, error) in item_lexer(c):

                if (error):
                    errors.append('%s at line %d' % (error, i))

                if (item.count('"') != 0) and (item.count('"') != 2):
                    errors.append('double-quote within double-quotes at line %d (hint files have no escape character)' % (i))

                # key:value
                match = re.match(r'^([^:\s]+):\s*(.*)$', item, re.DOTALL)
                if match:
                    key = match.group(1)
                    value = match.group(2)

                    if kind is not None:
                        if key not in hintkeys[kind]:
                            errors.append('unknown key %s at line %d' % (key, i))
                            continue
                        valtype = hintkeys[kind][key]

                        # check if the key occurs more than once
                        if key in hints:
                            errors.append('duplicate key %s' % (key))

                        # check the value meets any key-specific constraints
                        if (valtype == 'val') and (len(value) == 0):
                            errors.append('%s has empty value' % (key))

                        if (valtype == 'noval') and (len(value) != 0):
                            errors.append("%s has non-empty value '%s'" % (key, value))

                        # only 'ldesc' and 'message' are allowed a multi-line value
                        if (valtype != 'multilineval') and (len(value.splitlines()) > 1):
                            errors.append("key %s has multi-line value" % (key))

                    # validate all categories are in the category list (case-insensitively)
                    if key == 'category':
                        for c in value.split():
                            if c.lower() not in categories:
                                errors.append("unknown category '%s'" % (c))

                    if key in ['sdesc', 'ldesc']:
                        # verify that value for ldesc or sdesc is quoted (genini
                        # forces this)
                        if not (value.startswith('"') and value.endswith('"')):
                            errors.append("%s value '%s' should be quoted" % (key, value))

                        # warn about and fix common typos in ldesc/sdesc
                        value, msg = typofix(value)
                        if msg:
                            warnings.append("%s in %s" % (','.join(msg), key))

                    # if sdesc ends with a '.', warn and fix it
                    if key == 'sdesc':
                        if re.search(r'\."$', value):
                            warnings.append("sdesc ends with '.'")
                            value = re.sub(r'\."$', '"', value)

                    # if sdesc contains '  ', warn and fix it
                    if key == 'sdesc':
                        if '  ' in value:
                            warnings.append("sdesc contains '  '")
                            value = value.replace('  ', ' ')

                    # message must have an id and some text
                    if key == 'message':
                        if not re.match(r'(\S+)\s+(\S.*)', value):
                            errors.append('message value must have id and text')

                    # license must be a valid spdx license expression
                    if key == 'license' and licensing:
                        # first, normalize whitespace in license, so we don't
                        # get spurious normalization warnings over whitespace
                        value = re.sub(r'\s+', ' ', value)

                        try:
                            licensing.parse(value, strict=True)
                            le = licensing.validate(value, strict=True)
                        except license_expression.ExpressionParseError as e:
                            errors.append('errors parsing license expression: %s' % (e))
                        except license_expression.ExpressionError as e:
                            errors.append('errors validating license expression: %s' % (e))
                        else:
                            if not le.normalized_expression:
                                warnings.append('errors in license expression: %s' % (le.errors))
                            elif le.original_expression != le.normalized_expression:
                                warnings.append("license expression: '%s' normalizes to '%s'" % (value, le.normalized_expression))

                    # warn if value starts with a quote followed by whitespace
                    if re.match(r'^"[ \t]+', value):
                        warnings.append('value for key %s starts with quoted whitespace' % (key))

                    # store the key:value
                    hints[key] = value
                else:
                    errors.append("unknown construct '%s' at line %d" % (item, i))

            if ('skip' in hints) and (len(hints) == 1):
                errors.append("hint only contains skip: key, please update to cygport >= 0.22.0")

            # for the pvr kind, 'category' and 'sdesc' must be present
            # (genini also requires 'requires' but that seems wrong)
            # for the spvr kind, 'homepage' must be present for new packages
            if (kind == pvr) or (kind == spvr):
                mandatory = ['category', 'sdesc']
                if (kind == spvr) and strict:
                    mandatory.append('homepage')

                for k in mandatory:
                    if k not in hints:
                        errors.append("required key '%s' missing" % (k))

                suggested = []
                if (kind == spvr) and strict:
                    suggested.append('license')

                for k in suggested:
                    if k not in hints:
                        warnings.append("key '%s' missing" % (k))

            # warn if ldesc and sdesc seem transposed
            #
            # (Unfortunately we can't be totally strict about this, as some
            # packages like to repeat the basic description in ldesc in every
            # subpackage, but add to sdesc to distinguish the subpackages)
            if 'ldesc' in hints:
                if len(hints['sdesc']) > 2 * len(hints['ldesc']):
                    warnings.append('sdesc is much longer than ldesc')

            # sort these hints, as differences in ordering are uninteresting
            if 'build-depends' in hints:
                if ',' in hints['build-depends']:
                    hints['build-depends'] = split_trim_sort_join(hints['build-depends'], ',')
                else:
                    hints['build-depends'] = split_trim_sort_join(hints['build-depends'], None, ', ')

            if 'obsoletes' in hints:
                # obsoletes is specified as comma separated, but cygport writes it space separated at the moment...
                if ',' in hints['obsoletes']:
                    hints['obsoletes'] = split_trim_sort_join(hints['obsoletes'], ',')
                else:
                    hints['obsoletes'] = split_trim_sort_join(hints['obsoletes'], None, ', ')

            if 'replace-versions' in hints:
                hints['replace-versions'] = split_trim_sort_join(hints['replace-versions'], None, ' ')

        except UnicodeDecodeError:
            errors.append('invalid UTF-8')

    if errors:
        hints['parse-errors'] = errors

    if warnings:
        hints['parse-warnings'] = warnings

    return hints


# write hints |hints| to file |fn|
def hint_file_write(fn, hints):
    with open(fn, 'w') as f:
        for k, v in hints.items():
            print("%s: %s" % (k, v), file=f)


#
# words that Cygwin package maintainers apparently can't spell correctly
#

words = [
    (' accomodates ', ' accommodates '),
    (' consistant ', ' consistent '),
    (' examing ', ' examining '),
    (' extremly ', ' extremely '),
    (' interm ', ' interim '),
    (' procesors ', ' processors '),
    (' utilitzed ', ' utilized '),
    (' utilties ', ' utilities '),
]


def typofix(v):
    msg = []

    for (wrong, right) in words:
        if wrong in v:
            v = v.replace(wrong, right)
            msg.append('%s -> %s' % (wrong.strip(), right.strip()))

    return v, msg


#
#
#

def main(args):
    status = 0

    for fn in args.files:
        hints = hint_file_parse(fn, spvr if fn.endswith('src.hint') else pvr)

        if args.verbose >= 1:
            print(hints)

        if 'parse-warnings' in hints:
            if args.verbose > 0:
                for l in hints['parse-warnings']:
                    print('%s: %s' % (fn, l))
            status = 1

        if 'parse-errors' in hints:
            for l in hints['parse-errors']:
                print('%s: %s' % (fn, l))
            status = 255

    return status


#
#
#

if __name__ == "__main__":
    parser = argparse.ArgumentParser(description='.hint file validator')
    parser.add_argument('files', nargs='*', metavar='filename', help='list of files')
    parser.add_argument('-v', '--verbose', action='count', dest='verbose', help='verbose output', default=0)

    (args) = parser.parse_args()

    exit(main(args))