src/core/import_processor.py
author ymh <ymh.work@gmail.com>
Mon, 24 Jun 2013 00:38:29 +0200
changeset 28 5918a9d353d0
parent 0 4095911a7830
child 33 61c3ffd94f11
permissions -rw-r--r--
Correction in importing csv.
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
0
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
     1
# -*- coding: utf-8 -*-
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
     2
'''
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
     3
Created on Jun 10, 2013
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
     4
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
     5
@author: ymh
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
     6
'''
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
     7
from .models import Term
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
     8
from .rdf_models import graph
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
     9
from dateutil import parser
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    10
import re
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    11
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    12
class ImportProcessor(object):
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    13
    
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    14
    def __init__(self, field):
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    15
        self.field = field
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    16
        
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    17
    def process(self, obj,  value):
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    18
        return {}
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    19
    
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    20
    
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    21
class CharFieldProcessor(ImportProcessor):
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    22
    
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    23
    def process(self, obj, value):
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    24
        setattr(obj, self.field, value)
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    25
        return {}
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    26
28
5918a9d353d0 Correction in importing csv.
ymh <ymh.work@gmail.com>
parents: 0
diff changeset
    27
class TrimCharFieldProcessor(CharFieldProcessor):
5918a9d353d0 Correction in importing csv.
ymh <ymh.work@gmail.com>
parents: 0
diff changeset
    28
    
5918a9d353d0 Correction in importing csv.
ymh <ymh.work@gmail.com>
parents: 0
diff changeset
    29
    def process(self, obj, value):
5918a9d353d0 Correction in importing csv.
ymh <ymh.work@gmail.com>
parents: 0
diff changeset
    30
        return super(TrimCharFieldProcessor, self).process(obj, value.strip())
5918a9d353d0 Correction in importing csv.
ymh <ymh.work@gmail.com>
parents: 0
diff changeset
    31
0
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    32
class BooleanFieldProcessor(ImportProcessor):
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    33
    
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    34
    def process(self, obj, value):
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    35
        setattr(obj, self.field, value and value.strip().lower() in ['oui', '1', 't', 'yes', 'y', 'o'])
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    36
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    37
class DateFieldProcessor(ImportProcessor):
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    38
    
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    39
    def process(self, obj, value):
28
5918a9d353d0 Correction in importing csv.
ymh <ymh.work@gmail.com>
parents: 0
diff changeset
    40
        setattr(obj, self.field, parser.parse(value) if value else None)
0
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    41
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    42
class TermProcessor(ImportProcessor):
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    43
    
28
5918a9d353d0 Correction in importing csv.
ymh <ymh.work@gmail.com>
parents: 0
diff changeset
    44
    def __init__(self, field, context, notice_term_klass, re_split = r"[\;\,\:\(\)]", re_sub = "\(.+?\)"):
0
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    45
        ImportProcessor.__init__(self, field)
28
5918a9d353d0 Correction in importing csv.
ymh <ymh.work@gmail.com>
parents: 0
diff changeset
    46
        self.re_split = re.compile(re_split)
5918a9d353d0 Correction in importing csv.
ymh <ymh.work@gmail.com>
parents: 0
diff changeset
    47
        self.re_sub = re.compile(re_sub) if re_sub else None
0
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    48
        self.context = context
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    49
        self.notice_term_klass = notice_term_klass
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    50
    
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    51
    def build_notice_term(self, token, obj):
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    52
        uri = graph.get_uri_for_term(token, context=self.context)
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    53
        if not uri:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    54
            return None
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    55
        term_obj = Term.objects.get(uri=uri)
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    56
        if not self.notice_term_klass.objects.filter(notice=obj,term=term_obj).exists():
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    57
            return self.notice_term_klass(notice=obj,term=term_obj)            
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    58
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    59
    def process(self, obj, value):
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    60
        res = {}
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    61
        #remove everything between ()
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    62
        value = getattr(obj, self.field)
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    63
        if self.re_sub:
28
5918a9d353d0 Correction in importing csv.
ymh <ymh.work@gmail.com>
parents: 0
diff changeset
    64
            value = self.re_sub.sub("", value)
5918a9d353d0 Correction in importing csv.
ymh <ymh.work@gmail.com>
parents: 0
diff changeset
    65
        for token in self.re_split.split(value):
0
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    66
            token = token.strip()
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    67
            nt = self.build_notice_term(token, obj)
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    68
            if nt is not None:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    69
                res.setdefault(self.notice_term_klass,[]).append(nt)
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    70
        return res