| 13828 |
kshitij.so |
1 |
import pysolr
|
| 14186 |
kshitij.so |
2 |
import pymongo
|
| 13828 |
kshitij.so |
3 |
import json
|
| 14186 |
kshitij.so |
4 |
import os
|
| 14243 |
kshitij.so |
5 |
import optparse
|
| 14245 |
kshitij.so |
6 |
import shutil
|
| 18252 |
kshitij.so |
7 |
import traceback
|
| 20285 |
kshitij.so |
8 |
from dtr.utils.utils import CATEGORY_MAP, SUB_CATEGORY_MAP
|
| 13828 |
kshitij.so |
9 |
|
|
|
10 |
con=None
|
|
|
11 |
|
| 17790 |
kshitij.so |
12 |
SOURCE_MAP = {1:'AMAZON',2:'FLIPKART',3:'SNAPDEAL',4:'SAHOLIC',5:"SHOPCLUES.COM",6:"PAYTM.COM",7:"HOMESHOP18.COM"}
|
| 13828 |
kshitij.so |
13 |
xstr = lambda s: s or ""
|
|
|
14 |
|
| 14243 |
kshitij.so |
15 |
parser = optparse.OptionParser()
|
|
|
16 |
parser.add_option("-m", "--m", dest="mongoHost",
|
|
|
17 |
default="localhost",
|
|
|
18 |
type="string", help="The HOST where the mongo server is running",
|
|
|
19 |
metavar="HOST")
|
| 20285 |
kshitij.so |
20 |
parser.add_option("-s", "--s", dest="solrPath",
|
|
|
21 |
default="http://localhost:8080/solr/",
|
|
|
22 |
type="string", help="Complete solr path",
|
| 14248 |
kshitij.so |
23 |
metavar="HOST")
|
| 14243 |
kshitij.so |
24 |
|
|
|
25 |
(options, args) = parser.parse_args()
|
|
|
26 |
|
| 15067 |
kshitij.so |
27 |
synonymsMap = {}
|
|
|
28 |
modelNameSynonymMap = {}
|
| 14243 |
kshitij.so |
29 |
|
| 15067 |
kshitij.so |
30 |
|
| 13828 |
kshitij.so |
31 |
class __SkuInfo:
|
|
|
32 |
|
| 20285 |
kshitij.so |
33 |
def __init__(self, id, ids, brand, model_name, category_id, subCategoryId,thumbnail, title, brand_synonyms, model_name_synonyms, source_product_names, \
|
|
|
34 |
category, subCategory):
|
| 13828 |
kshitij.so |
35 |
self.id = id
|
| 20285 |
kshitij.so |
36 |
self.ids = ids
|
| 13828 |
kshitij.so |
37 |
self.brand = brand
|
|
|
38 |
self.model_name = model_name
|
|
|
39 |
self.category_id = category_id
|
| 20285 |
kshitij.so |
40 |
self.subCategoryId = subCategoryId
|
|
|
41 |
self.thumbnail = thumbnail
|
| 13828 |
kshitij.so |
42 |
self.title= title
|
| 15067 |
kshitij.so |
43 |
self.brand_synonyms = brand_synonyms
|
|
|
44 |
self.model_name_synonyms = model_name_synonyms
|
| 20285 |
kshitij.so |
45 |
self.source_product_names = source_product_names
|
|
|
46 |
self.category = category
|
|
|
47 |
self.subCategory = subCategory
|
| 13828 |
kshitij.so |
48 |
|
|
|
49 |
|
| 20285 |
kshitij.so |
50 |
solr = pysolr.Solr(options.solrPath, timeout=10)
|
| 13828 |
kshitij.so |
51 |
|
|
|
52 |
def pushData():
|
| 15067 |
kshitij.so |
53 |
populateBrandSynonymMap()
|
| 15068 |
kshitij.so |
54 |
populateModelNameSynonyms()
|
| 13828 |
kshitij.so |
55 |
l = []
|
| 17790 |
kshitij.so |
56 |
exception = []
|
| 14186 |
kshitij.so |
57 |
catalogMap = {}
|
| 17791 |
kshitij.so |
58 |
items = get_mongo_connection(options.mongoHost).Catalog.MasterData.find({'source_id' : { "$in": SOURCE_MAP.keys() } })
|
| 13828 |
kshitij.so |
59 |
print items.count()
|
|
|
60 |
for item in items:
|
| 17790 |
kshitij.so |
61 |
try:
|
|
|
62 |
title = xstr(item['brand'])+" "+xstr(item['model_name'])
|
|
|
63 |
except:
|
|
|
64 |
exception.append(item['_id'])
|
|
|
65 |
continue
|
| 17996 |
kshitij.so |
66 |
|
| 20285 |
kshitij.so |
67 |
if catalogMap.has_key(int(item['skuBundleId'])):
|
|
|
68 |
skuInfo = catalogMap.get(int(item['skuBundleId']))
|
|
|
69 |
skuInfo.source_product_names.append(item['source_product_name'].strip())
|
|
|
70 |
skuInfo.ids.append(int(item['_id']))
|
|
|
71 |
else:
|
|
|
72 |
syn_brand = synonymsMap.get((item['brand']).upper())
|
|
|
73 |
model_syn = modelNameSynonymMap.get(item['skuBundleId'])
|
|
|
74 |
if len(str(item['thumbnail']).strip()) == 0:
|
|
|
75 |
item['thumbnail'] = "http://api.profittill.com/img/no_image_available.png"
|
| 20286 |
kshitij.so |
76 |
try:
|
|
|
77 |
skuInfo = __SkuInfo(str(item['skuBundleId']), [int(item['_id'])], str(item['brand']).strip(), str(item['model_name']).strip(), \
|
| 20285 |
kshitij.so |
78 |
int(item['category_id']), int(item['subCategoryId']), item['thumbnail'].strip(), title, syn_brand, model_syn, [item['source_product_name'].strip()], \
|
|
|
79 |
CATEGORY_MAP.get(int(item['category_id'])), SUB_CATEGORY_MAP.get(int(item['subCategoryId'])))
|
| 20286 |
kshitij.so |
80 |
except:
|
|
|
81 |
print "Exception in item ",item['_id']
|
| 20285 |
kshitij.so |
82 |
|
|
|
83 |
catalogMap[int(item['skuBundleId'])] = skuInfo
|
|
|
84 |
|
|
|
85 |
for data in catalogMap.values():
|
|
|
86 |
l.append(data.__dict__)
|
|
|
87 |
solr.add(l)
|
| 14186 |
kshitij.so |
88 |
|
| 17790 |
kshitij.so |
89 |
|
|
|
90 |
for x in exception:
|
|
|
91 |
print x
|
| 13828 |
kshitij.so |
92 |
|
| 15067 |
kshitij.so |
93 |
def populateBrandSynonymMap():
|
|
|
94 |
global synonymsMap
|
|
|
95 |
brand_synonyms = get_mongo_connection(options.mongoHost).Catalog.BrandSynonyms.find()
|
|
|
96 |
for synonym in brand_synonyms:
|
|
|
97 |
synonymsMap[synonym['brand'].upper()] = synonym['synonyms']
|
|
|
98 |
|
|
|
99 |
def populateModelNameSynonyms():
|
|
|
100 |
global modelNameSynonymMap
|
|
|
101 |
model_synonyms = get_mongo_connection(options.mongoHost).Catalog.ModelNameSynonyms.find()
|
|
|
102 |
for synonym in model_synonyms:
|
|
|
103 |
modelNameSynonymMap[synonym['skuBundleId']] = synonym['synonyms']
|
|
|
104 |
|
| 13828 |
kshitij.so |
105 |
def get_mongo_connection(host='localhost', port=27017):
|
|
|
106 |
global con
|
|
|
107 |
if con is None:
|
|
|
108 |
print "Establishing connection %s host and port %d" %(host,port)
|
|
|
109 |
try:
|
|
|
110 |
con = pymongo.MongoClient(host, port)
|
|
|
111 |
except Exception, e:
|
|
|
112 |
print e
|
|
|
113 |
return None
|
|
|
114 |
return con
|
| 14186 |
kshitij.so |
115 |
|
| 13828 |
kshitij.so |
116 |
if __name__=='__main__':
|
| 17790 |
kshitij.so |
117 |
pushData()
|