#!/usr/bin/env python

import os
import sys
import re
import urllib
from tempfile import mkstemp
import tarfile
import gzip
import logging
import shutil
import subprocess

from gentoolkit.query import Query
from BeautifulSoup import BeautifulSoup, SoupStrainer


logger = logging.getLogger(__name__)


# From portage-janitor
def guess_indent_values(before):
    rindent = -1
    indent = -1
    tab = False

    def guess_for_tags(tags):
        for tag in tags:
            for i in [0, 2, 4, 6, 8, 12, 16]:
                if '\n%s<%s' % (' ' * i, tag) in before:
                    return i, False
            for i in [0, 1, 2]:
                if '\n%s<%s' % ('\t' * i, tag) in before:
                    return i, True
        return -1, False

    rindent, tab = guess_for_tags(
        ['herd', 'maintainer', 'longdescription', 'use', 'upstream']
    )
    if rindent == -1:
        rindent = 2
    rindent_str = ('\t' if tab else ' ') * rindent
    indent, tab = guess_for_tags(['watch', 'name', 'email'])
    if indent == -1:
        indent = rindent * 2 if rindent else 4
        if rindent and rindent_str == '\t':
            tab = True
    indent_str = ('\t' if tab else ' ') * indent
    return rindent_str, indent_str


def get_watch_data(package):
    deb_url, deb_type = get_deb_url(package.name)
    if deb_type == "source":
        return handle_source(deb_url)
    if deb_type == "diff":
        return handle_diff(deb_url)


def handle_diff(deb_url):
    _, temp_deb = mkstemp()

    logger.info("  Downloading debian diff %s...", deb_url)
    urllib.urlretrieve(deb_url, temp_deb)

    watch_data = ""

    fp = gzip.open(temp_deb, 'rb')
    for line in fp:
        if re.match(r"\+\+\+ .+?/debian/watch", line):
            fp.readline()  # diff lines, don't care
            cur_line = fp.readline()
            while cur_line.startswith("+"):
                watch_data += cur_line[1:]
                cur_line = fp.readline()
    fp.close()

    os.unlink(temp_deb)

    return watch_data


def handle_source(deb_url):
    _, temp_deb = mkstemp()
    temp_dir = os.path.dirname(temp_deb)

    logger.info("  Downloading debian source %s...", deb_url)
    urllib.urlretrieve(deb_url, temp_deb)
    tar = tarfile.open(temp_deb)

    watch_data = None

    try:
        tar.extract("debian/watch", temp_dir)
    except KeyError:
        pass
    else:
        debian_path = os.path.join(temp_dir, "debian")
        watch_path = os.path.join(debian_path, "watch")
        watch_data = open(os.path.join(watch_path)).read()
        shutil.rmtree(debian_path)

    os.unlink(temp_deb)

    return watch_data


def get_deb_url(name):
    deb_url = None
    deb_type = None

    while not deb_url:
        url = "http://packages.debian.org/source/unstable/%s" % name
        opened = urllib.urlopen(url)

        content = opened.read()

        for link in BeautifulSoup(content, parseOnlyThese=SoupStrainer("a")):
            if re.match("[^\s]+\.debian\.tar\.gz", link.text):
                deb_url = link["href"]
                deb_type = "source"
                break
            if re.match("[^\s]+\.diff\.gz", link.text):
                deb_url = link["href"]
                deb_type = "diff"
                break

        if not deb_url:
            logger.error("  Cannot get package from %s" % url)
            name = raw_input("  Package name in Debian: ")

    return deb_url, deb_type


def patch_metadata(metadata_path, watch_data, diff=False):
    watch_data = "\n".join([line for line in watch_data.split("\n")
                            if not line.startswith("#")])  # comments
    watch_data = watch_data.replace("\\\n", "")  # remove backslashes
    watch_data = " ".join(watch_data.split())  # remove extra spaces and \n

    result = re.match(
        r'(version=\d+?) (?:opts=(?:"([^"]+?)"|([^\s]+?)) )?(.*)', watch_data
    )

    version, attrs_quote, attrs, url = result.groups()
    attrs = attrs_quote or attrs

    if attrs:
        attrs = [x.replace('=', '="') + '"' for x in attrs.split(",")]
        attrs = " ".join(attrs)

    with open(metadata_path) as fp:
        original = fp.read()
    rindent, indent = guess_indent_values(original)

    data = original

    logger.info("  Patching metadata file")

    if attrs:
        watch_tag = '%s<watch %s %s>%s</watch>' % (indent, version, attrs, url)
    else:
        watch_tag = '%s<watch %s>%s</watch>' % (indent, version, url)

    if '<upstream>' in data:
        data = data.replace('<upstream>', '<upstream>\n%s' % watch_tag, 1)
    else:
        rep = '%s<upstream>\n%s\n%s</upstream>\n</pkgmetadata>' % \
            (rindent, watch_tag, rindent)
        data = data.replace('</pkgmetadata>', rep, 1)

    if not diff:
        return data
    else:
        _, data_path = mkstemp()
        with open(data_path, "w") as f:
            f.write(data)
        sub = subprocess.Popen(["diff", metadata_path, data_path],
                               stdout=subprocess.PIPE)
        content = sub.stdout.read()
        os.unlink(data_path)
        return content


def process_package(query, diff=False):
    matches = Query(query).smart_find(
            in_installed=True,
            in_porttree=True,
            in_overlay=True,
            include_masked=True,
            show_progress=False,
            no_matches_fatal=False,
    )

    if len(matches) == 0:
        logger.error("  Package not found")
        return None

    matches = sorted(matches)
    package = matches.pop()
    if '9999' in package.version and len(matches) > 0:
        package = matches.pop()

    metadata_path = package.metadata.metadata_path
    watch_data = get_watch_data(package)
    if watch_data is None:
        logger.error("  No watch file found")
    else:
        return patch_metadata(metadata_path, watch_data, diff=diff)


def main():
    import optparse
    p = optparse.OptionParser(
        usage="usage: %prog <package> [<package> [...]]",
    )
    p.add_option('-d', '--diff', action="store_true", dest="diff",
                 default=False,
                 help="Outputs a diff")
    opts, packages = p.parse_args()

    logging.basicConfig(stream=sys.stderr, level=logging.INFO,
                        format='%(message)s')

    for package in packages:
        logger.info("Processing %s..." % package)
        print process_package(package, opts.diff)

if __name__ == "__main__":
    main()