# vim: set fileencoding=utf-8 :
#
# (C) 2011,2015,2017 Guido Günther <agx@sigxcpu.org>
#    This program is free software; you can redistribute it and/or modify
#    it under the terms of the GNU General Public License as published by
#    the Free Software Foundation; either version 2 of the License, or
#    (at your option) any later version.
#
#    This program is distributed in the hope that it will be useful,
#    but WITHOUT ANY WARRANTY; without even the implied warranty of
#    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
#    GNU General Public License for more details.
#
#    You should have received a copy of the GNU General Public License
#    along with this program; if not, please see
#    <http://www.gnu.org/licenses/>
"""Handle Patches and Patch Series"""

import collections
import os
import re
import tempfile
from gbp.errors import GbpError
from gbp.git.repository import GitRepository

VALID_DEP3_ENDS = re.compile(r'(?:---|\*\*\*|Index:)[ \t][^ \t]|^diff -|^---')


class Patch(object):
    """
    A patch in a L{PatchSeries}

    @ivar path: path to the patch
    @type path: string
    @ivar topic: the topic of the patch (the directory component)
    @type topic: string
    @ivar strip: path components to strip (think patch -p<strip>)
    @type strip: integer
    @ivar info: Information retrieved from a RFC822 style patch header
    @type info: C{dict} with C{str} keys and values
    @ivar long_desc: the long description of the patch
    """
    patch_exts = ['diff', 'patch']

    def __init__(self, path, topic=None, strip=None):
        self.path = path
        self.topic = topic
        self.strip = strip
        self.info = None
        self.long_desc = None

    def __repr__(self):
        repr = "<gbp.patch_series.Patch path='%s' " % self.path
        if self.topic:
            repr += "topic='%s' " % self.topic
        if self.strip is not None:
            repr += "strip=%d " % self.strip
        repr += ">"
        return repr

    def _read_info(self):
        self._read_git_mailinfo()

    def _read_git_mailinfo(self):
        """
        Read patch information into a structured form

        using I{git mailinfo}
        """
        self.info = {}
        self.long_desc = ''

        body = tempfile.NamedTemporaryFile(prefix='gbp_')

        # No patch yet, file name information only
        if not os.path.exists(self.path):
            return
        # The patch description might contain UTF-8 while the actual patch is ascii.
        # To unconfuse git-mailinfo stop at the patch separator
        toparse = []
        for line in open(self.path, 'rb'):
            if line == b'---\n':
                break
            toparse.append(line)

        input = b''.join(toparse)
        if input.strip():
            out, err, ret = GitRepository.git_inout(command='mailinfo',
                                                    args=['-k', body.name, '/dev/null'],
                                                    input=input,
                                                    extra_env=None,
                                                    cwd=None,
                                                    capture_stderr=True)
            if ret != 0:
                raise GbpError("Failed to read patch header of '%s': %s" %
                               (self.path, err))
        else:
            out = b''

        # Header
        for line in out.decode().split('\n'):
            if ':' in line:
                rfc_header, value = line.split(" ", 1)
                header = rfc_header[:-1].lower()
                self.info[header] = value.strip()
        # Body
        try:
            self.long_desc = "".join([l.decode("utf-8", "backslashreplace") for l in body])
        except (IOError, UnicodeDecodeError) as msg:
            raise GbpError("Failed to read patch header of '%s': %s" %
                           (self.path, msg))
        finally:
            body.close()
            if os.path.exists(body.name):
                os.unlink(body.name)

    def _get_subject_from_filename(self):
        """
        Determine the patch's subject based on the its filename

        >>> p = Patch('debian/patches/foo.patch')
        >>> p._get_subject_from_filename()
        'foo'
        >>> Patch('foo.patch')._get_subject_from_filename()
        'foo'
        >>> Patch('debian/patches/foo.bar')._get_subject_from_filename()
        'foo.bar'
        >>> p = Patch('debian/patches/foo')
        >>> p._get_subject_from_filename()
        'foo'
        >>> Patch('0123-foo.patch')._get_subject_from_filename()
        'foo'
        >>> Patch('0123.patch')._get_subject_from_filename()
        '0123'
        >>> Patch('0123-foo-0123.patch')._get_subject_from_filename()
        'foo-0123'

        @return: the patch's subject
        @rtype: C{str}
        """
        subject = os.path.basename(self.path)
        # Strip of .diff or .patch from patch name
        try:
            base, ext = subject.rsplit('.', 1)
            if ext in self.patch_exts:
                subject = base
        except ValueError:
                pass  # No ext so keep subject as is
        return subject.lstrip('0123456789-') or subject

    def _get_info_field(self, key, get_val=None):
        """
        Return the key I{key} from the info C{dict}
        or use val if I{key} is not a valid key.

        Fill self.info if not already done.

        @param key: key to fetch
        @type key: C{str}
        @param get_val: alternate value if key is not in info dict
        @type get_val: C{()->str}
        """
        if self.info is None:
            self._read_info()

        if key in self.info:
            return self.info[key]
        else:
            return get_val() if get_val else None

    @property
    def subject(self):
        """
        The patch's subject, either from the patch header or from the filename.
        """
        return self._get_info_field('subject', self._get_subject_from_filename)

    @property
    def author(self):
        """The patch's author"""
        return self._get_info_field('author')

    @property
    def email(self):
        """The patch author's email address"""
        return self._get_info_field('email')

    @property
    def date(self):
        """The patch's modification time"""
        return self._get_info_field('date')


class Dep3Patch(Patch):
    def _read_info(self):
        super(Dep3Patch, self)._read_info()
        self._check_dep3()

    def _dep3_get_value(self, lines):
        value = []
        for line in lines:
            if line.startswith(' '):
                line = line[1:]
                if line == '.\n':
                    line = line[1:]
            else:
                line = line.split(':', 1)[1].lstrip()
            value.append(line)
        return ''.join(value)

    def _dep3_to_info(self, headers):
        """
        Process the ordered dict generated by check_dep3 and add the
        information to self.info
        """

        def add_author(lines):
            if 'email' in self.info and 'author' in self.info:
                return 0
            value = self._dep3_get_value(lines).strip()
            m = re.match('(.*)<([^<>]+)>', value)
            if m:
                value = m.group(1).strip()
                self.info['email'] = m.group(2)
            self.info['author'] = value
            return 1

        def add_subject(lines, long_desc, changes):
            if 'subject' in self.info:
                return long_desc, 0
            value = self._dep3_get_value(lines).lstrip()
            if '\n' in value:
                value, description = value.split('\n', 1)
                # prepend the continuation lines
                long_desc = description + long_desc
            self.info['subject'] = value
            return long_desc, changes + 1

        def add_date(lines):
            if 'date' in self.info:
                return 0
            self.info['date'] = self._dep3_get_value(lines).strip()
            return 1

        changes = 0
        pseudo_headers = ''
        long_desc = self._dep3_get_value(headers.get('long_desc', list()))

        for k, v in headers.items():
            if k in ('author', 'from'):
                changes += add_author(v)
            elif k in ('subject', 'description'):
                long_desc, changes = add_subject(v, long_desc, changes)
            elif k in ['date']:
                changes += add_date(v)
            elif k == 'long_desc':
                pass
            elif k in (
                'content-transfer-encoding',
                'content-type',
                'mime-version',
            ):
                # These can appear in `git format-patch` or `gbp pq export`
                # output. They are not semantically significant: they're
                # part of the encoding of the patch as an email, rather
                # than real patch metadata.
                pass
            else:
                pseudo_headers += ''.join(v)
                changes += 1
        if changes:
            self.long_desc = (pseudo_headers +
                              ('\n' if pseudo_headers else '') +
                              long_desc + self.long_desc)

    def _check_dep3(self):
        """
        Read DEP3 patch information into a structured form
        """
        if not os.path.exists(self.path):
            return

        # patch_header logic from quilt plus any line starting with ---
        # which is the dep3 stop processing and the git separation between the
        # header and diff stat
        headers = collections.OrderedDict()
        current = 'long_desc'
        with open(self.path, errors='replace') as file:
            for line in file:
                if VALID_DEP3_ENDS.search(line):
                    break

                if line.startswith(' '):
                    # continuation
                    headers.setdefault(current, list()).append(line)
                elif ':' in line:
                    current = line.split(':', 1)[0].lower()
                    headers.setdefault(current, list()).append(line)
                else:
                    # end of paragraph or not a header, read_info already left
                    # everything else in the long_desc, nothing else to do
                    break
        self._dep3_to_info(headers)


class PatchSeries(list):
    """
    A series of L{Patch}es as read from a quilt series file).
    """
    comment_re = re.compile(r'\s+#.*$')
    level_re = re.compile(r'-p(?P<level>[0-9]+)')

    @classmethod
    def read_series_file(cls, seriesfile):
        """Read a series file into L{Patch} objects"""
        patch_dir = os.path.dirname(seriesfile)

        if not os.path.exists(seriesfile):
            return []

        try:
            s = open(seriesfile)
        except Exception as err:
            raise GbpError("Cannot open series file: %s" % err)

        queue = cls._read_series(s, patch_dir)
        s.close()
        return queue

    @classmethod
    def _read_series(cls, series, patch_dir):
        """
        Read patch series

        >>> PatchSeries._read_series(['a/b',
        ...                           'a -p1 # comment',
        ...                           'a/b -p2'], '.')
        ... # doctest:+NORMALIZE_WHITESPACE
        [<gbp.patch_series.Patch path='./a/b' topic='a' >,
         <gbp.patch_series.Patch path='./a' strip=1 >,
         <gbp.patch_series.Patch path='./a/b' topic='a' strip=2 >]

        >>> PatchSeries._read_series(['# foo', 'a/b', '', '# bar'], '.')
        [<gbp.patch_series.Patch path='./a/b' topic='a' >]

        @param series: series of patches in quilt format
        @type series: iterable of strings
        @param patch_dir: path prefix to prepend to each patch path
        @type patch_dir: string
        """
        queue = PatchSeries()
        for line in series:
            try:
                if line[0] in ['\n', '#']:
                    continue
            except IndexError:
                continue  # ignore empty lines
            queue.append(cls._parse_line(line, patch_dir))
        return queue

    @staticmethod
    def _get_topic(line):
        """
        Get the topic from the patch's path

        >>> PatchSeries._get_topic("a/b c")
        'a'
        >>> PatchSeries._get_topic("asdf")
        >>> PatchSeries._get_topic("/asdf")
        """
        topic = os.path.dirname(line)
        if topic in ['', '/']:
            topic = None
        return topic

    @classmethod
    def _strip_comment(cls, line):
        """
        Strip a comment from a series file line

        >>> PatchSeries._strip_comment("does/not matter")
        'does/not matter'
        >>> PatchSeries._strip_comment("remove/the  # comment # text")
        'remove/the'
        >>> PatchSeries._strip_comment("leave/level/intact -p1 # comment # text")
        'leave/level/intact -p1'
        """
        return re.sub(cls.comment_re, '', line)

    @classmethod
    def _split_strip(cls, line):
        """
        Separate the -p<num> option from the patch name

        >>> PatchSeries._split_strip("asdf -p1")
        ('asdf', 1)
        >>> PatchSeries._split_strip("a/nice/patch")
        ('a/nice/patch', None)
        >>> PatchSeries._split_strip("asdf foo")
        ('asdf foo', None)
        """
        patch = line
        strip = None

        split = line.rsplit(None, 1)
        if len(split) > 1:
            m = cls.level_re.match(split[1])
            if m:
                patch = split[0]
                strip = int(m.group('level'))

        return (patch, strip)

    @classmethod
    def _parse_line(cls, line, patch_dir):
        """
        Parse a single line from a series file

        >>> PatchSeries._parse_line("a/b -p1", '/tmp/patches')
        <gbp.patch_series.Patch path='/tmp/patches/a/b' topic='a' strip=1 >
        >>> PatchSeries._parse_line("a/b", '.')
        <gbp.patch_series.Patch path='./a/b' topic='a' >
        """
        line = cls._strip_comment(line.rstrip())
        topic = cls._get_topic(line)
        (patch, split) = cls._split_strip(line)
        return Dep3Patch(os.path.join(patch_dir, patch), topic, split)