#!/usr/bin/python # Copyright (C) 2006 Peter Poeml. All rights reserved. # This program is free software; it may be used, copied, modified # and distributed under the terms of the GNU General Public Licence, # either version 2, or (at your option) any later version. __version__ = '0.8' import os import sys import urllib2 from urllib import pathname2url, quote_plus from urlparse import urlunsplit import cElementTree as ET from cStringIO import StringIO import shutil netloc = 'api.opensuse.org' scheme = 'http' BUFSIZE = 1024*1024 store = '.osc' exclude_stuff = [store, '.svn', 'CVS', '.git'] new_project_templ = """\ Short title of NewProject This project aims at providing some foo and bar. It also does some weird stuff. """ new_package_templ = """\ Title of New Package LONG DESCRIPTION GOES HERE """ class File: """represent a file, including its metadata""" def __init__(self, name, md5, size, mtime): self.name = name self.md5 = md5 self.size = size self.mtime = mtime def __str__(self): return self.name class Project: """represent a project directory, holding packages""" def __init__(self, dir): self.dir = dir self.absdir = os.path.abspath(dir) self.name = store_read_project(self.dir) self.pacs_available = meta_get_packagelist(self.name) self.pacs_have = [ i for i in os.listdir(self.dir) if i in self.pacs_available ] self.pacs_missing = [ i for i in self.pacs_available if i not in self.pacs_have ] def checkout_missing_pacs(self): for pac in self.pacs_missing: print 'checking out new package %s' % pac olddir = os.getcwd() os.chdir(os.pardir) checkout_package(self.name, pac) os.chdir(olddir) def __str__(self): r = [] r.append('*****************************************************') r.append('Project %s (dir=%s, absdir=%s)' % (self.name, self.dir, self.absdir)) r.append('have pacs:\n%s' % ', '.join(self.pacs_have)) r.append('missing pacs:\n%s' % ', '.join(self.pacs_missing)) r.append('*****************************************************') return '\n'.join(r) class Package: """represent a package (its directory) and read/keep/write its metadata""" def __init__(self, workingdir): self.dir = workingdir self.absdir = os.path.abspath(self.dir) self.storedir = os.path.join(self.dir, store) check_store_version(self.dir) self.prjname = store_read_project(self.dir) self.name = store_read_package(self.dir) files_tree = read_filemeta(self.dir) files_tree_root = files_tree.getroot() self.rev = files_tree_root.get('rev') self.filenamelist = [] self.filelist = [] for node in files_tree_root.findall('entry'): try: f = File(node.get('name'), node.get('md5'), int(node.get('size')), int(node.get('mtime'))) except: # okay, a very old version of _files, which didn't contain any metadata yet... f = File(node.get('name'), '', 0, 0) self.filelist.append(f) self.filenamelist.append(f.name) self.to_be_deleted = read_tobedeleted(self.dir) self.in_conflict = read_inconflict(self.dir) self.todo = [] self.todo_send = [] self.todo_delete = [] # gather unversioned files (the ones not listed in _meta) self.filenamelist_unvers = [ i for i in os.listdir(self.dir) if i not in exclude_stuff if i not in self.filenamelist ] def addfile(self, n): st = os.stat(os.path.join(self.dir, n)) f = File(n, None, st.st_size, st.st_mtime) self.filelist.append(f) self.filenamelist.append(n) self.filenamelist_unvers.remove(n) shutil.copy2(os.path.join(self.dir, n), os.path.join(self.storedir, n)) def delete_localfile(self, n): try: os.unlink(os.path.join(self.dir, n)) except: pass try: os.unlink(os.path.join(self.storedir, n)) except: pass def put_on_deletelist(self, n): if n not in self.to_be_deleted: self.to_be_deleted.append(n) def put_on_conflictlist(self, n): if n not in self.in_conflict: self.in_conflict.append(n) def clear_from_conflictlist(self, n): """delete an entry from the file, and remove the file if it would be empty""" if n in self.in_conflict: filename = os.path.join(self.dir, n) storefilename = os.path.join(self.storedir, n) myfilename = os.path.join(self.dir, n + '.mine') upfilename = os.path.join(self.dir, n + '.r' + self.rev) try: os.unlink(myfilename) # the working copy may be updated, so the .r* ending may be obsolete... # then we don't care os.unlink(upfilename) except: pass self.in_conflict.remove(n) self.write_conflictlist() def write_deletelist(self): if len(self.to_be_deleted) == 0: try: os.unlink(os.path.join(self.storedir, '_to_be_deleted')) except: pass else: fname = os.path.join(self.storedir, '_to_be_deleted') f = open(fname, 'w') f.write('\n'.join(self.to_be_deleted)) f.write('\n') f.close() def delete_source_file(self, n): import othermethods u = makeurl(['source', self.prjname, self.name, pathname2url(n)]) othermethods.delfile(u, n, username, password) self.delete_localfile(n) def put_source_file(self, n): import othermethods # escaping '+' in the URL path (note: not in the URL query string) is # only a workaround for ruby on rails, which swallows it otherwise u = makeurl(['source', self.prjname, self.name, pathname2url(n)]) othermethods.putfile(u, username, password, file = os.path.join(self.dir, n)) shutil.copy2(os.path.join(self.dir, n), os.path.join(self.storedir, n)) def write_conflictlist(self): if len(self.in_conflict) == 0: os.unlink(os.path.join(self.storedir, '_in_conflict')) else: fname = os.path.join(self.storedir, '_in_conflict') f = open(fname, 'w') f.write('\n'.join(self.in_conflict)) f.write('\n') f.close() def updatefile(self, n): filename = os.path.join(self.dir, n) storefilename = os.path.join(self.storedir, n) mtime = self.findfilebyname(n).mtime get_source_file(self.prjname, self.name, n, targetfilename=filename) os.utime(filename, (-1, mtime)) shutil.copy2(filename, storefilename) def mergefile(self, n): filename = os.path.join(self.dir, n) storefilename = os.path.join(self.storedir, n) myfilename = os.path.join(self.dir, n + '.mine') upfilename = os.path.join(self.dir, n + '.r' + self.rev) os.rename(filename, myfilename) mtime = self.findfilebyname(n).mtime get_source_file(self.prjname, self.name, n, targetfilename=upfilename) os.utime(upfilename, (-1, mtime)) if binary_file(myfilename) or binary_file(upfilename): # don't try merging shutil.copy2(upfilename, filename) shutil.copy2(upfilename, storefilename) self.in_conflict.append(n) self.write_conflictlist() return 'C' else: # try merging ret = os.system('diff3 -m -E %s %s %s > %s' \ % (myfilename, storefilename, upfilename, filename)) if ret == 0: # merge was successful... clean up os.rename(upfilename, filename) shutil.copy2(filename, storefilename) os.unlink(myfilename) return 'G' else: # unsuccessful merge self.in_conflict.append(n) self.write_conflictlist() return 'C' def update_filesmeta(self): meta = ''.join(show_files_meta(self.prjname, self.name)) f = open(os.path.join(self.storedir, '_files'), 'w') f.write(meta) f.close() def update_pacmeta(self): meta = ''.join(show_package_meta(self.prjname, self.name)) f = open(os.path.join(self.storedir, '_meta'), 'w') f.write(meta) f.close() def findfilebyname(self, n): for i in self.filelist: if i.name == n: return i def status(self, n): """ status can be: file storefile file present STATUS exists exists in _files x x - 'A' x x x ' ' if digest differs: 'M' and if in conflicts file: 'C' x - - '?' x - x 'D' and listed in _to_be_deleted - x x '!' - x - 'D' (when file in working copy is already deleted) - - x 'F' (new in repo, but not yet in working copy) - - - NOT DEFINED """ known_by_meta = False exists = False exists_in_store = False if n in self.filenamelist: known_by_meta = True if os.path.exists(os.path.join(self.dir, n)): exists = True if os.path.exists(os.path.join(self.storedir, n)): exists_in_store = True if exists and not exists_in_store and known_by_meta: state = 'D' elif n in self.to_be_deleted: state = 'D' elif n in self.in_conflict: state = 'C' elif exists and exists_in_store and known_by_meta: #print self.findfilebyname(n) if dgst(os.path.join(self.dir, n)) != self.findfilebyname(n).md5: state = 'M' else: state = ' ' elif exists and not exists_in_store and not known_by_meta: state = '?' elif exists and exists_in_store and not known_by_meta: state = 'A' elif not exists and exists_in_store and known_by_meta: state = '!' elif not exists and not exists_in_store and known_by_meta: state = 'F' elif not exists and exists_in_store and not known_by_meta: state = 'D' elif not exists and not exists_in_store and not known_by_meta: print '%s: not exists and not exists_in_store and not nown_by_meta' % n print 'this code path should never be reached!' sys.exit(1) return state def merge(self, otherpac): self.todo += otherpac.todo def __str__(self): r = """ name: %s prjname: %s workingdir: %s localfilelist: %s rev: %s 'todo' files: %s """ % (self.name, self.prjname, self.dir, '\n '.join(self.filenamelist), self.rev, self.todo) return r def read_meta_from_spec(self): specfile = os.path.join(self.dir, self.name + '.spec') name, summary, descr = read_meta_from_spec(specfile) if name != self.name: print 'name from spec does not match name of package... this is probably a problem' sys.exit(1) self.summary = summary self.descr = descr def update_pac_meta(self, template=new_package_templ): import othermethods import tempfile (fd, filename) = tempfile.mkstemp(prefix = 'osc_editmeta.', suffix = '.xml', dir = '/tmp') try: u = makeurl(['source', self.prjname, self.name, '_meta']) m = urllib2.urlopen(u).readlines() except urllib2.HTTPError, e: if e.code == 404: print 'package does not exist yet... creating it' m = template % (pac, username) else: print 'error getting package meta for project \'%s\' package \'%s\':' % (prj, pac) print e sys.exit(1) f = os.fdopen(fd, 'w') f.write(''.join(m)) f.close() tree = ET.parse(filename) tree.find('title').text = self.summary tree.find('description').text = ''.join(self.descr) tree.write(filename) # FIXME: escape stuff for xml print '*' * 36, 'old', '*' * 36 print ''.join(m) print '*' * 36, 'new', '*' * 36 tree.write(sys.stdout) print '*' * 72 # FIXME: for testing... # open the new description in $EDITOR instead? repl = raw_input('Write? (y/N) ') if repl == 'y': print 'Sending meta data...', u = makeurl(['source', self.prjname, self.name, '_meta']) othermethods.putfile(u, username, password, file=filename) print 'Done.' else: print 'discarding', filename os.unlink(filename) def is_project_dir(d): if os.path.exists(os.path.join(d, store, '_project')) and not \ os.path.exists(os.path.join(d, store, '_package')): return True else: return False def findpacs(files): pacs = [] for f in files: if f in exclude_stuff: break p = filedir_to_pac(f) known = None for i in pacs: if i.name == p.name: known = i break if known: i.merge(p) else: pacs.append(p) return pacs def read_filemeta(dir): return ET.parse(os.path.join(dir, store, '_files')) def read_tobedeleted(dir): r = [] fname = os.path.join(dir, store, '_to_be_deleted') if os.path.exists(fname): r = [ line.strip() for line in open(fname) ] return r def read_inconflict(dir): r = [] fname = os.path.join(dir, store, '_in_conflict') if os.path.exists(fname): r = [ line.strip() for line in open(fname) ] return r def parseargs(list_of_args): if list_of_args: return list_of_args else: return [ os.curdir ] def filedir_to_pac(f): if os.path.isdir(f): wd = f p = Package(wd) elif os.path.isfile(f): wd = os.path.dirname(f) if wd == '': wd = os.curdir p = Package(wd) p.todo = [ os.path.basename(f) ] else: wd = os.path.dirname(f) if wd == '': wd = os.curdir p = Package(wd) p.todo = [ os.path.basename(f) ] #else: # print # print 'error: %s is neither a valid file or directory' % f # sys.exit(1) return p def statfrmt(statusletter, filename): return '%s %s' % (statusletter, filename) def makeurl(l): """given a list of path compoments, construct a complete URL""" return urlunsplit((scheme, netloc, '/'.join(l), '', '')) def urlopen(url, data=None): """wrapper around urllib2.urlopen for error handling""" try: # adding data to the request makes it a POST fd = urllib2.urlopen(url, data=data) except urllib2.HTTPError, e: print >>sys.stderr, 'Error: can\'t get \'%s\'' % url print >>sys.stderr, e if e.code == 500: print >>sys.stderr, '\nDebugging output follows.\nurl:\n%s\nresponse:\n%s' % (url, e.read()) sys.exit(1) return fd def readauth(): """look for the credentials. If there aren't any, ask and store them""" # # try .netrc first # # the needed entry in .netrc looks like this: # machine api.opensuse.org login your_login password your_pass # but it is not able for credentials containing spaces import netrc global username, password try: info = netrc.netrc() username, account, password = info.authenticators(netloc) return username, password except (IOError, TypeError, netrc.NetrcParseError): pass # # try .oscrc next # import ConfigParser conffile = os.path.expanduser('~/.oscrc') if os.path.exists(conffile): config = ConfigParser.ConfigParser() config.read(conffile) username = config.get(netloc, 'user') password = config.get(netloc, 'pass') return username, password # # create .oscrc # import getpass print >>sys.stderr, \ """your user account / password are not configured yet. You will be asked for them below, and they will be stored in %s for later use. """ % conffile username = raw_input('Username: ') password = getpass.getpass() fd = open(conffile, 'w') os.chmod(conffile, 0600) print >>fd, """[general] # Downloaded packages are cached here. Must be writable by you. #packagecachedir: /var/tmp/osbuild-packagecache # Wrapper to call build as root (sudo, su -, ...) #su-wrapper: su -c # rootdir to setup the chroot environment #build-root: /var/tmp/build-root [%s] user: %s pass: %s """ % (netloc, username, password) fd.close() return username, password def init_basicauth(): username, password = readauth() passmgr = urllib2.HTTPPasswordMgrWithDefaultRealm() # this creates a password manager passmgr.add_password(None, netloc, username, password) # because we have put None at the start it will always # use this username/password combination for urls # for which `netloc` is a super-url authhandler = urllib2.HTTPBasicAuthHandler(passmgr) # create the AuthHandler opener = urllib2.build_opener(authhandler) opener.addheaders = [('User-agent', 'osc/%s' % __version__)] urllib2.install_opener(opener) # All calls to urllib2.urlopen will now use our handler # Make sure not to include the protocol in with the URL, or # HTTPPasswordMgrWithDefaultRealm will be very confused. # You must (of course) use it when fetching the page though. def init_package_dir(project, package, dir): if not os.path.isdir(store): os.mkdir(store) os.chdir(store) f = open('_project', 'w') f.write(project + '\n') f.close f = open('_package', 'w') f.write(package + '\n') f.close f = open('_files', 'w') f.write(''.join(show_files_meta(project, package))) f.close() f = open('_osclib_version', 'w') f.write(__version__ + '\n') f.close() os.chdir(os.pardir) return def check_store_version(dir): versionfile = os.path.join(dir, store, '_osclib_version') try: v = open(versionfile).read().strip() except: v = '' if v == '': print 'error: "%s" is not an osc working copy' % dir sys.exit(1) if v != __version__: if v in ['0.2', '0.3', '0.4', '0.5', '0.6', '0.7']: # version is fine, no migration needed f = open(versionfile, 'w') f.write(__version__ + '\n') f.close() return print print 'the osc metadata of your working copy "%s"' % dir print 'has the wrong version (%s), should be %s' % (v, __version__) print 'please do a fresh checkout' print sys.exit(1) def meta_get_packagelist(prj): u = makeurl(['source', prj]) try: f = urllib2.urlopen(u) except urllib2.HTTPError, e: if e.code == 404: print 'project \'%s\' does not exist' % prj sys.exit(1) else: print e print 'url: \'%s\'' % u sys.exit(1) tree = ET.parse(f) root = tree.getroot() return [ node.get('name') for node in root.findall('entry') ] def meta_get_filelist(prj, package): u = makeurl(['source', prj, package]) try: f = urllib2.urlopen(u) except urllib2.HTTPError, e: print >>sys.stderr, 'error getting filelist for project \'%s\' package \'%s\'' % (prj, package) print >>sys.stderr, e sys.exit(1) root = ET.parse(f).getroot() return [ node.get('name') for node in root ] def get_slash_source(): u = makeurl(['source']) try: root = ET.parse(urllib2.urlopen(u)).getroot() except urllib2.HTTPError, e: print >>sys.stderr, 'error getting /source' print >>sys.stderr, e sys.exit(1) return sorted([ node.get('name') for node in root ]) def show_project_meta(prj): try: url = makeurl(['source', prj, '_meta']) f = urllib2.urlopen(url) except urllib2.HTTPError, e: print >>sys.stderr, 'error getting meta for project \'%s\'' % prj print >>sys.stderr, e if e.code == 500: print >>sys.stderr, '\nDebugging output follows.\nurl:\n%s\nresponse:\n%s' % (url, e.read()) sys.exit(1) return f.readlines() def show_package_meta(prj, pac): try: url = makeurl(['source', prj, pac, '_meta']) f = urllib2.urlopen(url) except urllib2.HTTPError, e: print >>sys.stderr, 'error getting meta for project \'%s\' package \'%s\'' % (prj, pac) print >>sys.stderr, e if e.code == 500: print >>sys.stderr, '\nDebugging output follows.\nurl:\n%s\nresponse:\n%s' % (url, e.read()) sys.exit(1) return f.readlines() def edit_meta(prj, pac, template=new_package_templ, change_is_required=True): import othermethods import tempfile (fd, filename) = tempfile.mkstemp(prefix = 'osc_editmeta.', suffix = '.xml', dir = '/tmp') if pac: # package meta u = makeurl(['source', prj, pac, '_meta']) try: m = urllib2.urlopen(u).readlines() except urllib2.HTTPError, e: if e.code == 404: m = template % (pac, username) else: print 'error getting package meta for project \'%s\' package \'%s\':' % (prj, pac) print e sys.exit(1) else: # project meta u = makeurl(['source', prj, '_meta']) try: m = urllib2.urlopen(u).readlines() except urllib2.HTTPError, e: if e.code == 404: m = new_project_templ % (prj, username) else: print 'error getting package meta for project \'%s\':' % prj print e sys.exit(1) f = os.fdopen(fd, 'w') f.write(''.join(m)) f.close() timestamp = os.path.getmtime(filename) editor = os.getenv('EDITOR', default='vim') os.system('%s %s' % (editor, filename)) if change_is_required == True and os.path.getmtime(filename) == timestamp: print 'File unchanged. Not saving.' os.unlink(filename) else: print 'Sending meta data...', othermethods.putfile(u, username, password, file=filename) os.unlink(filename) print 'Done.' def show_files_meta(prj, pac): f = urlopen(makeurl(['source', prj, pac])) return f.readlines() def show_upstream_rev(prj, pac): m = show_files_meta(prj, pac) return ET.parse(StringIO(''.join(m))).getroot().get('rev') def read_meta_from_spec(specfile): """read Name, Summary and %description from spec file""" if not os.path.isfile(specfile): print 'file \'%s\' is not a readable file' % specfile return None lines = open(specfile).readlines() for line in lines: if line.startswith('Name:'): name = line.split(':')[1].strip() break for line in lines: if line.startswith('Summary:'): summary = line.split(':')[1].strip() break descr = [] start = lines.index('%description\n') + 1 for line in lines[start:]: if line.startswith('%'): break descr.append(line) return name, summary, descr def get_user_id(user): u = makeurl(['person', quote_plus(user)]) try: f = urllib2.urlopen(u) return ''.join(f.readlines()) except urllib2.HTTPError: print 'user \'%s\' not found' % user return None def get_source_file(prj, package, filename, targetfilename=None): u = makeurl(['source', prj, package, pathname2url(filename)]) f = urlopen(u) o = open(targetfilename or filename, 'w') while 1: buf = f.read(BUFSIZE) if not buf: break o.write(buf) o.close() def dgst(file): #if not os.path.exists(file): #return None import md5 s = md5.new() f = open(file, 'r') while 1: buf = f.read(BUFSIZE) if not buf: break s.update(buf) return s.hexdigest() def binary(s): """return true if a string is binary data using diff's heuristic""" if s and '\0' in s[:4096]: return True return False def binary_file(fn): """read 4096 bytes from a file named fn, and call binary() on the data""" return binary(open(fn, 'r').read(4096)) def get_source_file_diff(dir, filename, rev): import difflib file1 = os.path.join(dir, store, filename) # stored original file2 = os.path.join(dir, filename) # working copy f1 = open(file1, 'r') s1 = f1.read() f1.close() f2 = open(file2, 'r') s2 = f2.read() f2.close() if binary(s1) or binary (s2): d = ['Binary file %s has changed\n' % filename] else: d = difflib.unified_diff(\ s1.splitlines(1), \ s2.splitlines(1), \ fromfile = '%s (revision %s)' % (filename, rev), \ tofile = '%s (working copy)' % filename) return ''.join(d) def make_dir(project, package): #print "creating directory '%s'" % project if not os.path.exists(project): print statfrmt('A', project) os.mkdir(project) os.mkdir(os.path.join(project, store)) f = open(os.path.join(project, store, '_project'), 'w') f.write(project + '\n') f.close() #print "creating directory '%s/%s'" % (project, package) if not os.path.exists(os.path.join(project, package)): print statfrmt('A', '%s/%s' % (project, package)) os.mkdir(os.path.join(project, package)) os.mkdir(os.path.join(project, package, store)) return(os.path.join(project, package)) def checkout_package(project, package): olddir = os.getcwd() os.chdir(make_dir(project, package)) init_package_dir(project, package, store) p = Package(os.curdir) for filename in p.filenamelist: p.updatefile(filename) print 'A ', os.path.join(project, package, filename) os.chdir(olddir) def link_pac(src_project, src_package, dst_project, dst_package): """ create a linked package - "src" is the original package - "dst" is the "link" package that we are creating here """ import othermethods import tempfile src_meta = show_package_meta(src_project, src_package) # replace package name and username # using a string buffer # and create the package tree = ET.parse(StringIO(''.join(src_meta))) root = tree.getroot() root.set('name', '%s') tree.find('person').set('userid', '%s') buf = StringIO() tree.write(buf) src_meta = buf.getvalue() edit_meta(dst_project, dst_package, template=src_meta, change_is_required=False) # create the _link file # but first, make sure not to overwrite an existing one if '_link' in meta_get_filelist(dst_project, dst_package): print print '_link file already exists...! Aborting' sys.exit(1) print 'Creating _link...', link_template = """\ """ % (src_project, src_package) u = makeurl(['source', dst_project, dst_package, '_link']) othermethods.putfile(u, username, password, strbuf = link_template) print 'Done.' def delete_package(prj, pac): import othermethods u = makeurl(['source', prj, pac]) othermethods.delfile(u, pac, username, password) def get_platforms(): f = urlopen(makeurl(['platform'])) tree = ET.parse(f) r = [ node.get('name') for node in tree.getroot() ] r.sort() return r def get_platforms_of_project(prj): f = show_project_meta(prj) tree = ET.parse(StringIO(''.join(f))) r = [ node.get('name') for node in tree.findall('repository')] return r def get_repos_of_project(prj): f = show_project_meta(prj) tree = ET.parse(StringIO(''.join(f))) repo_line_templ = '%-15s %-10s' r = [] for node in tree.findall('repository'): for node2 in node.findall('arch'): r.append(repo_line_templ % (node.get('name'), node2.text)) return r def show_results_meta(prj, package, platform): u = makeurl(['result', prj, platform, package, 'result']) f = urlopen(u) return f.readlines() def get_results(prj, package, platform): #print '----------------------------------------' r = [] #result_line_templ = '%(prj)-15s %(pac)-15s %(rep)-15s %(arch)-10s %(status)s' result_line_templ = '%(rep)-15s %(arch)-10s %(status)s' f = show_results_meta(prj, package, platform) tree = ET.parse(StringIO(''.join(f))) root = tree.getroot() rmap = {} rmap['prj'] = root.get('project') rmap['pac'] = root.get('package') rmap['rep'] = root.get('repository') for node in root.findall('archresult'): rmap['arch'] = node.get('arch') statusnode = node.find('status') rmap['status'] = statusnode.get('code') if rmap['status'] in ['expansion error', 'broken']: rmap['status'] += ': ' + statusnode.find('summary').text if rmap['status'] == 'failed': rmap['status'] += ': %s://%s' % (scheme, netloc) + \ '/result/%(prj)s/%(rep)s/%(pac)s/%(arch)s/log' % rmap r.append(result_line_templ % rmap) return r def get_log(prj, package, platform, arch, offset): u = makeurl(['result', prj, platform, package, arch, 'log?nostream=1&start=%s' % offset]) f = urlopen(u) return f.read() def get_buildinfo(prj, package, platform, arch, specfile=None): # http://api.opensuse.org/rpm/Subversion/Apache_SuSE_Linux_10.1/i586/subversion/buildinfo u = makeurl(['rpm', prj, platform, arch, package, 'buildinfo']) f = urlopen(u, data=specfile) return f.read() def get_buildconfig(prj, package, platform, arch): # http://api.opensuse.org/rpm///_repository//_buildconfig u = makeurl(['rpm', prj, platform, '_repository', arch, '_buildconfig']) f = urlopen(u) return f.read() def get_buildhistory(prj, package, platform, arch): u = makeurl(['rpm', prj, platform, arch, package, 'history']) f = urlopen(u) return f.readlines() def cmd_rebuild(prj, package): u = makeurl(['source', prj, package, '?cmd=rebuild']) try: # adding data to the request makes it a POST f = urllib2.urlopen(u, data=' ') except urllib2.HTTPError, e: print >>sys.stderr, 'could not trigger rebuild for project \'%s\' package \'%s\'' % (prj, package) print >>sys.stderr, u print >>sys.stderr, e sys.exit(1) root = ET.parse(f).getroot() #code = root.get('code') return root.find('summary').text def store_read_project(dir): p = open(os.path.join(dir, store, '_project')).readlines()[0].strip() return p def store_read_package(dir): p = open(os.path.join(dir, store, '_package')).readlines()[0].strip() return p def get_osc_version(): return __version__