1

"""

     2

    Full-text searching of logs

     3

"""

     4

     5

import datetime, calendar, pytz

     6

     7

import HyperEstraier as hype

     8

     9

import log_line

    10

    11

class LogSearchIndex (object) :

    12

    """

    13

        An index on the logs for a group of channels.

    14

    15

        This uses Hyper Estraier to handle searching, whereby each log line is a document (yes, I have a powerful server).

    16

    17

        These log documents have the following attributes:

    18

            @uri        - channel/date/line

    19

            @channel    - channel id

    20

            @type       - the LogType id

    21

            @timestamp  - UTC timestamp

    22

            @source     - nickname

    23

    24

        Each document then has a single line of data, which is the log message itself

    25

    """

    26

    27

    def __init__ (self, path, mode='r') :

    28

        """

    29

            Open the database, with the given mode:

    30

                r       - read-only

    31

                w       - read-write, create if not exists

    32

                a       - read-write, do not create

    33

                *       - read-write, truncate and create new

    34

        """

    35

    36

        # mapping of { mode -> flags }

    37

        mode_to_flag = {

    38

            'r':    hype.Database.DBREADER,

    39

            'w':    hype.Database.DBREADER | hype.Database.DBWRITER | hype.Database.DBCREAT,

    40

            'a':    hype.Database.DBREADER | hype.Database.DBWRITER | hype.Database.DBCREAT,

    41

            '*':    hype.Database.DBREADER | hype.Database.DBWRITER | hype.Database.DBCREAT | hype.Database.DBTRUNC,

    42

        }

    43

    44

        # look up flags

    45

        flags = mode_to_flag[mode]

    46

    47

        # make instance

    48

        self.db = hype.Database()

    49

    50

        # open

    51

        if not self.db.open(path, flags) :

    52

            raise Exception("Index open failed: %s, mode=%s, flags=%#06x: %s" % (path, mode, flags, self.db.err_msg(self.db.error())))

    53

    54

    def insert (self, channel, lines) :

    55

        """

    56

            Adds a sequence of LogLines from the given LogChannel to the index

    57

        """

    58

    59

        # validate the LogChannel

    60

        assert channel.name

    61

    62

        # iterate

    63

        for line in lines :

    64

            # validate the LogLine

    65

            assert line.offset

    66

            assert line.timestamp

    67

    68

            # create new document

    69

            doc = hype.Document()

    70

    71

            # line date

    72

            date = line.timestamp.date()

    73

    74

            # convert to UTC timestamp

    75

            utc_timestamp = calendar.timegm(line.timestamp.utctimetuple())

    76

    77

            # ensure that it's not 1900

    78

            assert date.year != 1900

    79

    80

            # add URI

    81

            doc.add_attr('@uri',        "%s/%s/%d" % (channel.id, date.strftime('%Y-%m-%d'), line.offset))

    82

    83

            # add channel id

    84

            doc.add_attr('@channel',    channel.id)

    85

    86

            # add type

    87

            doc.add_attr('@type',       str(line.type))

    88

    89

            # add UTC timestamp

    90

            doc.add_attr('@timestamp',  str(utc_timestamp))

    91

    92

            # add source attribute?

    93

            if line.source :

    94

                doc.add_attr('@source', str(line.source))

    95

    96

            # add data text

    97

            doc.add_text(line.data.encode('utf8'))

    98

    99

            # put

   100

            # XXX: what does this flag mean?

   101

            if not self.db.put_doc(doc, hype.Database.PDCLEAN) :

   102

                raise Exeception("Index put_doc failed")

   103

   104

    def search_cond (self, cond) :

   105

        """

   106

            Search using a raw hype.Condition

   107

        """

   108

   109

        # execute search, unused 'flags' arg stays zero

   110

        results = self.db.search(cond, 0)

   111

   112

        # iterate over the document IDs

   113

        for doc_id in results :

   114

            # load document, this throws an exception...

   115

            # option constants are hype.Database.GDNOATTR/GDNOTEXT

   116

            doc = self.db.get_doc(doc_id, 0)

   117

   118

            # load the attributes/text

   119

            channel_id  = doc.attr('@channel')

   120

            type        = int(doc.attr('@type'))

   121

            timestamp   = datetime.datetime.fromtimestamp(int(doc.attr('@timestamp')), pytz.utc)

   122

            source      = doc.attr('@source')

   123

            data        = doc.cat_texts().decode('utf8')

   124

   132

   139

        """

   140

   141

        # build condition

   142

        cond = hype.Condition()

   163

   167

   170

   180