loappimage-helpers/loaih/versions.py

#!/usr/bin/env python
# encoding: utf-8

import urllib.request
from lxml import etree
from packaging.version import parse as parse_version
import datetime

class Definitions(object):
    DOWNLOADPAGE = "https://www.libreoffice.org/download/download/"
    ARCHIVE = "https://downloadarchive.documentfoundation.org/libreoffice/old/"
    RELEASE = "https://download.documentfoundation.org/libreoffice/stable/"
    DAILY = "https://dev-builds.libreoffice.org/daily/master/Linux-rpm_deb-x86_64@tb87-TDF/"
    PRERELEASE = "https://dev-builds.libreoffice.org/pre-releases/deb/x86_64/"

    SELECTORS = {
        'still': {
            'URL': DOWNLOADPAGE,
            'xpath': '(//span[@class="dl_version_number"])[last()]/text()'
        },
        'fresh': {
            'URL': DOWNLOADPAGE,
            'xpath': '(//span[@class="dl_version_number"])[1]/text()'
        },
        'prerelease': {
            'URL': DOWNLOADPAGE,
            'xpath': '//p[@class="lead_libre"][last()]/following-sibling::ul[last()]/li/a/text()'
        },
        'daily': {
            'URL': DAILY,
            'xpath': '//td/a'
        }
    }

class Base(object):
    # Class for static methods which might be useful even outside the build
    # scripts.

    @staticmethod
    def dailyver(date = datetime.datetime.today()):
        """Returns versions present on the latest daily build."""
        url = Base.dailyurl(date)['x86_64']
        # Since it is possible that a daily build is not yet provided...
        if url == '-':
            return []

        # Rerun the page parsing, this time to find out the versions built
        
        b = etree.HTML(urllib.request.urlopen(Base.dailyurl(date)['x86_64']).read()).xpath("//td/a[contains(text(), '_deb.tar.gz')]/text()")
        # This should have returned the main package for a version, but can
        # have returned multiple ones, so let's treat it as a list
        return [ x.split('_')[1] for x in b ]

    @staticmethod
    def dailyurl(date = datetime.datetime.today()):
        """Returns the URL for the latest valid daily build."""
        # As per other parts of the build, we need to maintain an URL also for
        # x86 versions that it isn't really provided.
        # As such, the return value must be a dictionary

        # Get the anchor for today's builds
        a = etree.HTML(urllib.request.urlopen(Definitions.DAILY).read()).xpath("//td/a[contains(text(), '" + date.strftime('%Y-%m-%d') + "')]/text()")
        if len(a) == 0:
            # No results found, no version found, let's return a
            return { 'x86': '-', 'x86_64': '-' }

        # On the contrary, more than a version is found. let's order the
        # list and get the latest item
        return { 'x86': '-', 'x86_64': Definitions.SELECTORS['daily']['URL'] + sorted(a)[-1] }

    @staticmethod
    def namedver(query):
        """Gets the version for a specific named version."""

        if query == 'daily' or query == 'yesterday':
            # Daily needs double parsing for the same result to apply.
            # We first select today's build anchor:
            date = datetime.datetime.today()
            if query == 'yesterday':
                # Use yesterdays' date for testing purposes.
                date += datetime.timedelta(days=-1)
            return Base.dailyver(date)

        # In case the query isn't for daily
        return etree.HTML(urllib.request.urlopen(Definitions.SELECTORS[query]['URL']).read()).xpath(Definitions.SELECTORS[query]['xpath'])

    @staticmethod
    def fullversion(version):
        """Get latest full version from Archive based on partial version."""
        versionlist = etree.HTML(urllib.request.urlopen(Definitions.ARCHIVE).read()).xpath("//td/a[starts-with(text(), '" + version + "')]/text()")
        cleanlist = sorted([ x.strip('/') for x in versionlist ])

        # Sorting, then returning the last version
        return cleanlist[-1]

    @staticmethod
    def urlfromqueryandver(query, version):
        """Returns the fetching URL based on the queried version and the numeric version of it."""
        # This has the purpose to simplify and explain how the releases are
        # layed out.

        # If the query tells about daily or 'yesterday' (for testing purposes),
        # we might ignore versions and return the value coming from dailyurl:
        if query == 'daily':
            return Base.dailyurl()
        if query == 'yesterday':
            date = datetime.datetime.today() + datetime.timedelta(days=-1)
            return Base.dailyurl(date)

        # All other versions will be taken from Archive, as such we need a full
        # version.

        # If the version has only 2 points in it (or splits into three parts by '.'), that's not a full version and we will call the getlatestver() function
        fullversion = version
        if len(version.split('.')) == 3:
            fullversion = Base.fullversion(version)

        # So the final URL is the Archive one, plus the full versions, plus a
        # final '/deb/' - and an arch subfolder
        baseurl = Definitions.ARCHIVE + fullversion + '/deb/'
        retval = {}

        # x86 binaries are not anymore offered after 6.3.0.
        if parse_version(version) < parse_version('6.3.0'):
            retval['x86'] = baseurl + 'x86/'
        else:
            retval['x86'] = '-'
        
        retval['x86_64'] = baseurl + 'x86_64/'
    
        return retval


class Build(object):

    def __init__(self, query):
        """Should simplify the single builded version."""
        self.query = query
        self.version = ''
        self.basedirurl = {}

        # For simplification process, let's before see if there's a named query
        # (it shouldn't)
        if not '.' in self.query:
            # Named version. Let's get the result of the query.
            a = Base.namedver(self.query)

            if len(a) != 1:
                raise Exception("Queried version for build does not return a single result. Please unfold manually. Exiting.")
        
            # So if it is here, the version is one.
            self.query = a[0]

        if len(self.query.split('.')) == 4:
            self.version = self.query
        else:
            # If not 4 dotted, let's search for the 4 dotted version
            self.version = Base.fullversion(self.query)
        
        self.basedirurl = Base.urlfromqueryandver(self.query, self.version)