196 lines
7.5 KiB
Python
196 lines
7.5 KiB
Python
#!/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 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 dailyver(date = datetime.datetime.today()):
|
|
"""Returns versions present on the latest daily build."""
|
|
url = Base.dailyurl(date)['x86_64']
|
|
# If no daily releases has been provided yet, return empty
|
|
if url == '-':
|
|
return []
|
|
|
|
# Rerun the page parsing, this time to find out the versions built
|
|
b = etree.HTML(urllib.request.urlopen(url).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 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(f"//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
|
|
|
|
@staticmethod
|
|
def collectedbuilds(query):
|
|
"""Creates a list of Builds based on each namedver found."""
|
|
retval = []
|
|
a = Base.namedver(query)
|
|
if isinstance(a, list) and len(a) > 1:
|
|
retval = [ Build(query, version) for version in Base.namedver(query) ]
|
|
else:
|
|
retval.append(RemoteBuild(query))
|
|
|
|
return retval
|
|
|
|
|
|
class RemoteBuild(object):
|
|
|
|
def __init__(self, query, version = None):
|
|
"""Should simplify the single builded version."""
|
|
self.query = query
|
|
self.version = ''
|
|
self.basedirurl = { 'x86': '-', 'x86_64': '-' }
|
|
|
|
if not '.' in self.query:
|
|
# Named version.
|
|
# Let's check if a specific version was requested.
|
|
if version:
|
|
self.version = version
|
|
else:
|
|
# In case it was not requested, we will carry on the generic
|
|
# namedver() query.
|
|
# If the results are more than one, we'll take the latest (since we are requested to provide a single build).
|
|
a = Base.namedver(self.query)
|
|
|
|
if isinstance(a, list) and len(a) == 0:
|
|
# No results from the query - let's return default values
|
|
return
|
|
|
|
if len(a) == 1:
|
|
# version is a single one.
|
|
self.version = a[0]
|
|
else:
|
|
# In this case, we will select the latest release.
|
|
self.version = sorted(a)[-1]
|
|
else:
|
|
# In case of numbered queries, put it as initial version
|
|
self.version = self.query
|
|
|
|
if len(self.version.split('.')) < 4:
|
|
# If not 4 dotted, let's search for the 4 dotted version
|
|
self.version = Base.fullversion(self.version)
|
|
|
|
self.basedirurl = Base.urlfromqueryandver(self.query, self.version)
|
|
|
|
def todict(self):
|
|
return {
|
|
'query': self.query,
|
|
'version': self.version,
|
|
'basedirurl': self.basedirurl
|
|
}
|
|
|
|
def __str__(self):
|
|
return f"""query: {self.query}
|
|
version: {self.version}
|
|
x86: {self.basedirurl['x86']}
|
|
x86_64: {self.basedirurl['x86_64']}"""
|