gnatsparse.py

bugzilla的安装文件

                self.fields["bug_severity"] = "major"
        self.fields["bug_severity"] = SqlQuote(self.fields["bug_severity"])
        self.fields["keywords"] = 0
        if keywordids.has_key(self.fields["Class"]):
            self.fields["keywords"] = self.fields["Class"]
            self.fields["keywordid"] = keywordids[self.fields["Class"]]
            self.fields["keywords"] = SqlQuote(self.fields["keywords"])
        self.fields["priority"] = "P1"
        if self.fields.has_key("Severity") and self.fields.has_key("Priority"):
            severity = self.fields["Severity"]
            priority = self.fields["Priority"]
            if severity == "critical":
                if priority == "high":
                    self.fields["priority"] = "P1"
                else:
                    self.fields["priority"] = "P2"
            elif severity == "serious":
                if priority == "low":
                    self.fields["priority"] = "P4"
                else:
                    self.fields["priority"] = "P3"
            else:
                if priority == "high":
                    self.fields["priority"] = "P4"
                else:
                    self.fields["priority"] = "P5"
        self.fields["priority"] = SqlQuote(self.fields["priority"])
        state = self.fields["State"]
        if (state == "open" or state == "analyzed") and self.fields["userid"] != 3:
            self.fields["bug_status"] = "ASSIGNED"
            self.fields["resolution"] = ""
        elif state == "feedback":
            self.fields["bug_status"] = "WAITING"
            self.fields["resolution"] = ""
        elif state == "closed":
            self.fields["bug_status"] = "CLOSED"
            if self.fields.has_key("Class"):
                theclass = self.fields["Class"]
                if theclass.find("duplicate") != -1:
                    self.fields["resolution"]="DUPLICATE"
                elif theclass.find("mistaken") != -1:
                    self.fields["resolution"]="INVALID"                    
                else:
                    self.fields["resolution"]="FIXED"
            else:
                self.fields["resolution"]="FIXED"
        elif state == "suspended":
            self.fields["bug_status"] = "SUSPENDED"
            self.fields["resolution"] = ""
        elif state == "analyzed" and self.fields["userid"] == 3:
            self.fields["bug_status"] = "NEW"
            self.fields["resolution"] = ""
        else:
            self.fields["bug_status"] = "UNCONFIRMED"
            self.fields["resolution"] = ""
        self.fields["bug_status"] = SqlQuote(self.fields["bug_status"])
        self.fields["resolution"] = SqlQuote(self.fields["resolution"])
        self.fields["creation_ts"] = ""
        if self.fields.has_key("Arrival-Date") and self.fields["Arrival-Date"] != "":
            self.fields["creation_ts"] = unixdate2datetime(self.fields["Arrival-Date"])
        self.fields["creation_ts"] = SqlQuote(self.fields["creation_ts"])
        self.fields["delta_ts"] = ""
        if self.fields.has_key("Audit-Trail"):
            result = lastdatere.findall(self.fields["Audit-Trail"])
            result.reverse()
            if len(result) > 0:
                self.fields["delta_ts"] = unixdate2timestamp(result[0])
        if self.fields["delta_ts"] == "":
            if self.fields.has_key("Arrival-Date") and self.fields["Arrival-Date"] != "":
                self.fields["delta_ts"] = unixdate2timestamp(self.fields["Arrival-Date"])
        self.fields["delta_ts"] = SqlQuote(self.fields["delta_ts"])
        self.fields["short_desc"] = SqlQuote(self.fields["Synopsis"])
        if self.fields.has_key("Reply-To") and self.fields["Reply-To"] != "":
            self.fields["reporter"] = get_userid(self.fields["Reply-To"])
        elif self.fields.has_key("Mail-Header"):
            result = re.search(r"""From .*?([\w.]+@[\w.]+)""", self.fields["Mail-Header"])
            if result:
                self.fields["reporter"] = get_userid(result.group(1))
            else:
                self.fields["reporter"] = get_userid(gnats_username)
        else:
            self.fields["reporter"] = get_userid(gnats_username)
        long_desc = self.fields["Description"]
        long_desc2 = ""
        for field in ["Release", "Environment", "How-To-Repeat"]:
            if self.fields.has_key(field) and self.fields[field] != "":
                long_desc += ("\n\n%s:\n" % field) + self.fields[field]
        if self.fields.has_key("Fix") and self.fields["Fix"] != "":
            long_desc2 = "Fix:\n" + self.fields["Fix"]
        if self.need_unformatted  == 1 and self.fields["Unformatted"] != "":
            long_desc += "\n\nUnformatted:\n" + self.fields["Unformatted"]
        if long_desc != "":
            self.long_descs.append((self.bug_id, self.fields["reporter"],
                                    self.fields["creation_ts"],
                                    SqlQuote(long_desc)))
        if long_desc2 != "":
            self.long_descs.append((self.bug_id, self.fields["reporter"],
                                    self.fields["creation_ts"],
                                    SqlQuote(long_desc2)))
        for field in ["gcchost", "gccbuild", "gcctarget"]:
            self.fields[field] = SqlQuote(self.fields[field])
        self.fields["version"] = ""
        if self.fields["Release"] != "":
            self.fields["version"] = convert_gccver_to_ver (self.fields["Release"])
        self.fields["version"] = SqlQuote(self.fields["version"])
        self.fields["product"] = SqlQuote("gcc")
        self.fields["component"] = "invalid"
        if self.fields.has_key("Category"):
            self.fields["component"] = self.fields["Category"]
        self.fields["component"] = SqlQuote(self.fields["component"])
        self.fields["target_milestone"] = "---"
        if self.fields["version"].find("3.4") != -1:
            self.fields["target_milestone"] = "3.4"
        self.fields["target_milestone"] = SqlQuote(self.fields["target_milestone"])
        if self.fields["userid"] == 2:
            self.fields["userid"] = "\'NULL\'"

class GNATSbug(object):
    """ Represents a single GNATS PR """
    def __init__(self, filename):
        self.attachments = []
        self.has_unformatted_attach = 0
        fp = open (filename)
        self.fields = self.parse_pr(fp.xreadlines())
        self.bug_id = int(self.fields["Number"])
        if self.fields.has_key("Unformatted"):
            self.find_gnatsweb_attachments()
        if self.fields.has_key("How-To-Repeat"):
            self.find_regular_attachments("How-To-Repeat")
        if self.fields.has_key("Fix"):
            self.find_regular_attachments("Fix")

    def get_attacher(fields):
        if fields.has_key("Reply-To") and fields["Reply-To"] != "":
            return get_userid(fields["Reply-To"])
        else:
            result = None
            if fields.has_key("Mail-Header"):
                result = re.search(r"""From .*?([\w.]+\@[\w.]+)""",
                                   fields["Mail-Header"])
            if result is not None:
                reporter = get_userid(result.group(1))
            else:
                reporter = get_userid(gnats_username)
    get_attacher = staticmethod(get_attacher)
    def find_regular_attachments(self, which):
        fields = self.fields
        while re.search("^begin [0-7]{3}", fields[which],
                        re.DOTALL | re.MULTILINE):
            outfp = cStringIO.StringIO()
            infp = cStringIO.StringIO(fields[which])
            filename, start, end = specialuu.decode(infp, outfp, quiet=0)
            fields[which]=fields[which].replace(fields[which][start:end],
                                                "See attachments for %s\n" % filename)
            self.attachments.append((filename, outfp.getvalue(),
                                     self.get_attacher(fields)))

    def decode_gnatsweb_attachment(self, attachment):
        result = re.split(r"""\n\n""", attachment, 1)
        if len(result) == 1:
            return -1
        envelope, body = result
        envelope = uselessre.split(envelope)
        envelope.pop(0)
        # Turn the list of key, value into a dict of key => value
        attachinfo = dict([(envelope[i], envelope[i+1]) for i in xrange(0,len(envelope),2)])
        for x in attachinfo.keys():
            attachinfo[x] = attachinfo[x].rstrip()
        if (attachinfo.has_key("Content-Type") == 0) or \
           (attachinfo.has_key("Content-Disposition") == 0):
            raise ValueError, "Unable to parse file attachment"
        result = dispositionre.search(attachinfo["Content-Disposition"])
        filename = result.group(2)
        filename = re.sub(".*/","", filename)
        filename = re.sub(".*\\\\","", filename)
        attachinfo["filename"]=filename
        result = re.search("""(\S+);.*""", attachinfo["Content-Type"])
        if result is not None:
            attachinfo["Content-Type"] = result.group(1)
        if attachinfo.has_key("Content-Transfer-Encoding"):
            if attachinfo["Content-Transfer-Encoding"] == "base64":
                attachinfo["data"] = base64.decodestring(body)
        else:
            attachinfo["data"]=body

        return (attachinfo["filename"], attachinfo["data"],
                self.get_attacher(self.fields))

    def find_gnatsweb_attachments(self):
        fields = self.fields
        attachments = re.split(attachment_delimiter, fields["Unformatted"])
        fields["Unformatted"] = attachments.pop(0)
        for attachment in attachments:
            result = self.decode_gnatsweb_attachment (attachment)
            if result != -1:
                self.attachments.append(result)
            self.has_unformatted_attach = 1
    def parse_pr(lines):
        #fields = {"envelope":[]}
        fields = {"envelope":array.array("c")}
        hdrmulti = "envelope"
        for line in lines:
            line = line.rstrip('\n')
            line += '\n'
            result = gnatfieldre.search(line)
            if result is None:
                if hdrmulti != "":
                    if fields.has_key(hdrmulti):
                        #fields[hdrmulti].append(line)
                        fields[hdrmulti].fromstring(line)
                    else:
                        #fields[hdrmulti] = [line]
                        fields[hdrmulti] = array.array("c", line)
                continue
            hdr, arg = result.groups()
            ghdr = "*not valid*"
            result = fieldnamere.search(hdr)
            if result != None:
                ghdr = result.groups()[0]
            if ghdr in fieldnames:
                if multilinefields.has_key(ghdr):
                    hdrmulti = ghdr
                    #fields[ghdr] = [""]
                    fields[ghdr] = array.array("c")
                else:
                    hdrmulti = ""
                    #fields[ghdr] = [arg]
                    fields[ghdr] = array.array("c", arg)
            elif hdrmulti != "":
                #fields[hdrmulti].append(line)
                fields[hdrmulti].fromstring(line)
            if hdrmulti == "envelope" and \
               (hdr == "Reply-To" or hdr == "From" \
                or hdr == "X-GNATS-Notify"):
                arg = fix_email_addrs(arg)
                #fields[hdr] = [arg]
                fields[hdr] = array.array("c", arg)
	if fields.has_key("Reply-To") and len(fields["Reply-To"]) > 0:
            fields["Reply-To"] = fields["Reply-To"]
        else:
            fields["Reply-To"] = fields["From"]
        if fields.has_key("From"):
            del fields["From"]
        if fields.has_key("X-GNATS-Notify") == 0:
            fields["X-GNATS-Notify"] = array.array("c")
            #fields["X-GNATS-Notify"] = ""
        for x in fields.keys():
            fields[x] = fields[x].tostring()
            #fields[x] = "".join(fields[x])            
        for x in fields.keys():
            if multilinefields.has_key(x):
                fields[x] = fields[x].rstrip()

        return fields
    parse_pr = staticmethod(parse_pr)
load_index("%s/gnats-adm/index" % gnats_db_dir)
load_categories("%s/gnats-adm/categories" % gnats_db_dir)
load_responsible("%s/gnats-adm/responsible" % gnats_db_dir)
get_userid(gnats_username)
get_userid(unassigned_username)
for x in pr_list:
    print "Processing %s..." % x
    a = GNATSbug ("%s/%s" % (gnats_db_dir, x))
    b = Bugzillabug(a)
write_non_bug_tables()
outfile.close()