diff options
Diffstat (limited to 'misc/xml/be-xml-to-mbox')
| -rwxr-xr-x | misc/xml/be-xml-to-mbox | 123 |
1 files changed, 68 insertions, 55 deletions
diff --git a/misc/xml/be-xml-to-mbox b/misc/xml/be-xml-to-mbox index 0151792..48454f9 100755 --- a/misc/xml/be-xml-to-mbox +++ b/misc/xml/be-xml-to-mbox @@ -27,23 +27,20 @@ Messages begin with a a From_ line, followed by RFC 822 email, followed by a blank line. """ -#from mailbox import mbox, Message # the mailbox people really want an on-disk copy +# from mailbox import mbox, Message # the mailbox people really want an on-disk copy import email.utils from libbe.util.encoding import get_output_encoding from libbe.util.utility import str_to_time as rfc2822_to_gmtime_integer from time import asctime, gmtime import types -try: # import core module, Python >= 2.5 - from xml.etree import ElementTree -except ImportError: # look for non-core module - from elementtree import ElementTree +from xml.etree import ElementTree from xml.sax.saxutils import unescape - DEFAULT_DOMAIN = "invalid.com" DEFAULT_EMAIL = "dummy@" + DEFAULT_DOMAIN DEFAULT_ENCODING = get_output_encoding() + def rfc2822_to_asctime(rfc2822_string): """Convert an RFC 2822-fomatted string into a asctime string. >>> rfc2822_to_asctime("Thu, 01 Jan 1970 00:00:00 +0000") @@ -53,27 +50,32 @@ def rfc2822_to_asctime(rfc2822_string): return asctime(gmtime(0)) return asctime(gmtime(rfc2822_to_gmtime_integer(rfc2822_string))) -class LimitedAttrDict (dict): + +class LimitedAttrDict(dict): """ Dict with error checking, to avoid invalid bug/comment fields. """ - _attrs = [] # override with list of valid attribute names + _attrs = [] # override with list of valid attribute names + def __init__(self, **kwargs): dict.__init__(self) - for key,value in kwargs.items(): + for key, value in kwargs.items(): self[key] = value + def __setitem__(self, key, item): self._validate_key(key) dict.__setitem__(self, key, item) + def _validate_key(self, key): if key in self._attrs: return elif type(key) not in types.StringTypes: - raise TypeError, "Invalid attribute type %s for '%s'" % (type(key), key) + raise TypeError("Invalid attribute type %s for '%s'" % (type(key), key)) else: - raise ValueError, "Invalid attribute name '%s'" % key + raise ValueError("Invalid attribute name '%s'" % key) + -class Bug (LimitedAttrDict): +class Bug(LimitedAttrDict): _attrs = [u"uuid", u"short-name", u"severity", @@ -85,32 +87,34 @@ class Bug (LimitedAttrDict): u"summary", u"comments", u"extra-strings"] + def print_to_mbox(self): if "creator" in self: # otherwise, probably a `be show` uuid-only bug to avoid # root comments. - name,addr = email.utils.parseaddr(self["creator"]) - print "From %s %s" % (addr, rfc2822_to_asctime(self["created"])) - print "Message-id: <%s@%s>" % (self["uuid"], DEFAULT_DOMAIN) - print "Date: %s" % self["created"] - print "From: %s" % self["creator"] - print "Content-Type: %s; charset=%s" \ - % ("text/plain", DEFAULT_ENCODING) - print "Content-Transfer-Encoding: 8bit" - print "Subject: %s: %s" % (self["short-name"], self["summary"]) + name, addr = email.utils.parseaddr(self["creator"]) + print("From %s %s" % (addr, rfc2822_to_asctime(self["created"]))) + print("Message-id: <%s@%s>" % (self["uuid"], DEFAULT_DOMAIN)) + print("Date: %s" % self["created"]) + print("From: %s" % self["creator"]) + print("Content-Type: %s; charset=%s" + % ("text/plain", DEFAULT_ENCODING)) + print("Content-Transfer-Encoding: 8bit") + print("Subject: %s: %s" % (self["short-name"], self["summary"])) if "extra-strings" in self: for estr in self["extra-strings"]: - print "X-Extra-String: %s" % estr - print "" - print self["summary"] - print "" + print("X-Extra-String: %s" % estr) + print() + print(self["summary"]) + print() if "comments" in self: for comment in self["comments"]: - comment.print_to_mbox(self) + comment.print_to_mbox(self) + def init_from_etree(self, element): assert element.tag == "bug", element.tag for field in element.getchildren(): - text = unescape(unicode(field.text).decode("unicode_escape").strip()) + text = unescape(bytes(field.text).decode("unicode_escape").strip()) if field.tag == "comment": comm = Comment() comm.init_from_etree(field) @@ -126,12 +130,14 @@ class Bug (LimitedAttrDict): else: self[field.tag] = text + def wrap_id(id): if "@" not in id: return "<%s@%s>" % (id, DEFAULT_DOMAIN) return id -class Comment (LimitedAttrDict): + +class Comment(LimitedAttrDict): _attrs = [u"uuid", u"alt-id", u"short-name", @@ -141,49 +147,54 @@ class Comment (LimitedAttrDict): u"content-type", u"body", u"extra-strings"] + def print_to_mbox(self, bug=None): - if bug == None: + if bug is None: bug = Bug() bug[u"uuid"] = u"no-uuid" - name,addr = email.utils.parseaddr(self["author"]) - print "From %s %s" % (addr, rfc2822_to_asctime(self["date"])) - if "uuid" in self: id = self["uuid"] - elif "alt-id" in self: id = self["alt-id"] - else: id = None - if id != None: - print "Message-id: %s" % wrap_id(id) + name, addr = email.utils.parseaddr(self["author"]) + print("From %s %s" % (addr, rfc2822_to_asctime(self["date"]))) + if "uuid" in self: + id = self["uuid"] + elif "alt-id" in self: + id = self["alt-id"] + else: + id = None + if id is not None: + print("Message-id: %s" % wrap_id(id)) if "alt-id" in self: - print "Alt-id: %s" % wrap_id(self["alt-id"]) - print "Date: %s" % self["date"] - print "From: %s" % self["author"] + print("Alt-id: %s" % wrap_id(self["alt-id"])) + print("Date: %s" % self["date"]) + print("From: %s" % self["author"]) subject = "" if "short-name" in self: - subject += self["short-name"]+u": " + subject += self["short-name"] + u": " if "summary" in bug: subject += bug["summary"] else: subject += u"no-subject" - print "Subject: %s" % subject + print("Subject: %s" % subject) if "in-reply-to" not in self.keys(): self["in-reply-to"] = bug["uuid"] - print "In-Reply-To: %s" % wrap_id(self["in-reply-to"]) + print("In-Reply-To: %s" % wrap_id(self["in-reply-to"])) if "extra-strings" in self: for estr in self["extra-strings"]: - print "X-Extra-String: %s" % estr + print("X-Extra-String: %s" % estr) if self["content-type"].startswith("text/"): - print "Content-Transfer-Encoding: 8bit" - print "Content-Type: %s; charset=%s" \ - % (self["content-type"], DEFAULT_ENCODING) + print("Content-Transfer-Encoding: 8bit") + print("Content-Type: %s; charset=%s" + % (self["content-type"], DEFAULT_ENCODING)) else: - print "Content-Transfer-Encoding: base64" - print "Content-Type: %s;" % (self["content-type"]) - print "" - print self["body"] - print "" + print("Content-Transfer-Encoding: base64") + print("Content-Type: %s;" % (self["content-type"])) + print() + print(self["body"]) + print() + def init_from_etree(self, element): assert element.tag == "comment", element.tag for field in element.getchildren(): - text = unescape(unicode(field.text).decode("unicode_escape").strip()) + text = unescape(bytes(field.text).decode("unicode_escape").strip()) if field.tag == "extra-string": if "extra-strings" in self: self["extra-strings"].append(text) @@ -191,9 +202,10 @@ class Comment (LimitedAttrDict): self["extra-strings"] = [text] else: if field.tag == "body": - text+="\n" + text += "\n" self[field.tag] = text + def print_to_mbox(element): if element.tag == "bug": b = Bug() @@ -207,14 +219,15 @@ def print_to_mbox(element): for elt in element.getchildren(): print_to_mbox(elt) + if __name__ == "__main__": import codecs import sys - + sys.stdin = codecs.getreader(DEFAULT_ENCODING)(sys.stdin) sys.stdout = codecs.getwriter(DEFAULT_ENCODING)(sys.stdout) - if len(sys.argv) == 1: # no filename given, use stdin + if len(sys.argv) == 1: # no filename given, use stdin xml_unicode = sys.stdin.read() else: xml_unicode = codecs.open(sys.argv[1], "r", DEFAULT_ENCODING).read() |