opensextant.extractors.poli

 1"""
 2These "Patterns of Life" are strictly examples to illustrate more general reg-ex patterns
 3 (that is more general than the coordinate and date/time patterns).
 4
 5 The main objective here is to show how beyond basic regex matches, we can add important business logic to the
 6 entity extraction process.
 7
 8"""
 9
10from opensextant.FlexPat import PatternMatch, RegexPatternManager
11
12
13class TelephoneNumber(PatternMatch):
14    def __init__(self, *args, **kwargs):
15        PatternMatch.__init__(self, *args, **kwargs)
16        self.case = PatternMatch.UPPER_CASE
17
18    def normalize(self):
19        PatternMatch.normalize(self)
20        print("TBD - normalize phone")
21
22
23class MACAddress(PatternMatch):
24    def __init__(self, *args, **kwargs):
25        PatternMatch.__init__(self, *args, **kwargs)
26        self.case = PatternMatch.UPPER_CASE
27
28    def normalize(self):
29        PatternMatch.normalize(self)
30        print("TBD - normalize MAC address")
31
32
33class Money(PatternMatch):
34    def __init__(self, *args, **kwargs):
35        PatternMatch.__init__(self, *args, **kwargs)
36        self.case = PatternMatch.LOWER_CASE
37
38    def normalize(self):
39        PatternMatch.normalize(self)
40        print("TBD - normalize Money")
41
42
43class PatternsOfLifeManager(RegexPatternManager):
44    #
45    # Demonstration.  PatternsOfLifeManager is a custom RegexPatternManager
46    # that shows how to apply FlexPat to extracting common things like currency amounts, MAC addresses
47    # and telephone numbers.
48    #
49    def __init__(self, cfg):
50        """
51        Call as
52            mgr = PatternsOfLifeManager("poli_patterns.cfg")
53            patternsApp = PatternExtractor( mgr )
54
55            test_results = patternsApp.default_tests()
56            real_results = patternsApp.extract( ".... text blob..." )
57
58        :param cfg: patterns config file.
59        """
60        RegexPatternManager.__init__(self, cfg, debug=True, testing=True)
61
62    def _initialize(self):
63        #
64        # This Class registry maps the existing Java classes (in the config file) to Python variations here.
65        self.match_class_registry = {
66            "org.opensextant.extractors.poli.data.TelephoneNumber":
67                "opensextant.extractors.poli.TelephoneNumber",
68
69            "org.opensextant.extractors.poli.data.MACAddress":
70                "opensextant.extractors.poli.MACAddress",
71
72            "org.opensextant.extractors.poli.data.Money":
73                "opensextant.extractors.poli.Money",
74
75            "org.opensextant.extractors.poli.data.EmailAddress":
76                "opensextant.extractors.poli.EmailAddress"
77        }
78        RegexPatternManager._initialize(self)