Subversion Repositories SmartDukaan

Rev

Details | Last modification | View Log | RSS feed

Rev Author Line No. Line
1336 rajveer 1
from datastore.DataAccessor import DataHelper
2
import os
3
import sys
4
import subprocess
5
from mail import *
6
 
7
"""
8
Documentation for script consetup.py
9
This is the first script called by test.py, this file is created because before importing
10
datasore, its PYTHONPATH must be set which is done in test.py, so now we can import it.
11
Task of this script is to run the spiders for individual suppliers and call the modules to
12
generate respective csv files.
13
In the code, for each supplier a separate section is made, in that section last part is for 
14
csv file generation, rest of the part is for spiders. 
15
initxy method is called so as to make database visible in this scope
16
@param  path to the folder in which spider-projects reside (:/home/gaurav/code) but start with pathsep
17
"""
18
 
19
 
20
path =  sys.argv[1]
21
ps = os.pathsep
22
ds = os.sep
23
path0 = path[1:len(path)]+ds+"ScrapyDatastore"+ds+"src"
24
path1 = path+ds+"ExtraFunctions"+ds+"src"
25
path2 = path+ds+"filterforname"+ds+"src"
26
path3 = path0+path1+path2
27
os.environ['PYTHONPATH'] = path3
28
 
29
da = DataHelper()
30
da.initxy()
31
da.add_newcrawler()
32
attach = ""
33
 
34
# infibeam
35
path4 = path
36
path4 = path4.replace(ps,"")
37
path4 = path4+ds+"runinfibeam.py"
38
subprocess.call(["python", path4, path])
39
 
40
 
41
phones = da.get_all_infibeam_data()
42
if not phones:
43
    mail("ashish.singh@shop2020.in","Bug","There is some mistake in parameters for infibeam spider. This mail is automatically generated via python script",attach)
44
 
45
path3 = str(os.getenv("PYTHONPATH"))
46
path4 = path+ds+"data2csv"+ds+"src"
47
path3 = path3 + path4
48
os.putenv("PYTHONPATH",path3)
49
path4 = path4.replace(ps,"")
50
path4 = path4+ds+"data2csv"+ds+"testfilteredinfibeam.py"
51
subprocess.call(["python", path4])
52
 
53
 
54
#babuchak
55
path3 = str(os.getenv("PYTHONPATH"))
56
path4 = path+ds+"babuchakScrapypass1"+ds+"src"
57
path3 = path3 + path4
58
os.putenv("PYTHONPATH",path3)
59
path4 = path4.replace(ps,"")
60
path4 = path4 + ds+"scrapy-ctl.py"
61
subprocess.call(["python", path4, "crawl", "babuchak"])
62
 
63
urls = da.get_allbabuchakurls()
64
if not urls:
65
    mail("ashish.singh@shop2020.in","Bug","There is some mistake in parameters for first babuchak spider. This mail is automatically generated via python script",attach)
66
 
67
path3 = str(os.getenv("PYTHONPATH"))
68
path4 = path + ds+"babuchakScrapypass2"+ds+"src"
69
path3 = path3 + path4
70
os.putenv("PYTHONPATH",path3)
71
path4 = path4.replace(ps,"")
72
path4 = path4 + ds+"scrapy-ctl.py"
73
subprocess.call(["python", path4, "crawl", "babuchak1"])
74
 
75
purls = da.get_allbabuchakphoneurls()
76
if urls and not purls:
77
    mail("ashish.singh@shop2020.in","Bug","There is some mistake in parameters for second babuchak spider. This mail is automatically generated via python script",attach)
78
 
79
path3 = str(os.getenv("PYTHONPATH"))
80
path4 = path + ds+"babuchakScrapypass3"+ds+"src"
81
path3 = path3 + path4
82
os.putenv("PYTHONPATH",path3)
83
path4 = path4.replace(ps,"")
84
path4 = path4 + ds+"scrapy-ctl.py"
85
subprocess.call(["python", path4, "crawl", "babuchak2"])
86
 
87
phones = da.get_allbabuchakphones()
88
if urls and purls and not phones:
89
    mail("ashish.singh@shop2020.in","Bug","There is some mistake in parameters for third babuchak spider. This mail is automatically generated via python script",attach)
90
 
91
path3 = str(os.getenv("PYTHONPATH"))
92
path4 = path + ds+"data2csv"+ds+"src"
93
path3 = path3 + path4
94
os.putenv("PYTHONPATH",path3)
95
path4 = path4.replace(ps,"")
96
path4 = path4 + ds+"data2csv"+ds+"testbabuchak.py"
97
subprocess.call(["python", path4])
98
 
99
 
100
 
101
#indiaplaza
102
path4 = path
103
path4 = path4.replace(ps,"")
104
path4 = path4+ds+"runindiaplaza.py"
105
subprocess.call(["python", path4, path])
106
 
107
purls = da.get_all_ipbasic()
108
if not purls:
109
    mail("ashish.singh@shop2020.in","Bug","There is some mistake in parameters for first indiaplaza spider. This mail is automatically generated via python script",attach)
110
 
111
path3 = str(os.getenv("PYTHONPATH"))
112
path4 = path + ds+"indiaplazascapypass2"+ds+"src"
113
path3 = path3 + path4
114
os.putenv("PYTHONPATH",path3)
115
path4 = path4.replace(ps,"")
116
path4 = path4 + ds+"scrapy-ctl.py"
117
subprocess.call(["python", path4, "crawl", "indiaplaza1"])
118
 
119
 
120
phones = da.get_all_indiaplaza_phones()
121
if purls and not phones:
122
    mail("ashish.singh@shop2020.in","Bug","There is some mistake in parameters for second indiaplaza spider. This mail is automatically generated via python script",attach)
123
 
124
path3 = str(os.getenv("PYTHONPATH"))
125
path4 = path + ds+"data2csv"+ds+"src"
126
path3 = path3 + path4
127
os.putenv("PYTHONPATH",path3)
128
path4 = path4.replace(ps,"")
129
path4 = path4 + ds+"data2csv"+ds+"testfilterindiaplaza.py"
130
subprocess.call(["python", path4])
131
 
132
#This section commented out by ashish as it was running in infinite loop
133
#naaptol
134
#path3 = str(os.getenv("PYTHONPATH"))
135
#path4 = path + ds+"naaptolpass1"+ds+"src"
136
#path3 = path3 + path4
137
#os.putenv("PYTHONPATH",path3)
138
#path4 = path4.replace(ps,"")
139
#path4 = path4 + ds+"scrapy-ctl.py"
140
#subprocess.call(["python", path4, "crawl", "naaptol"])
141
 
142
#urls = da.get_allnaaptolurls()
143
#if not urls:
144
#	mail("ashish.singh@shop2020.in","Bug","There is some mistake in parameters for first naaptol spider. This mail is automatically generated via python script",attach)
145
 
146
#path3 = str(os.getenv("PYTHONPATH"))
147
#path4 = path + ds+"naaptolpass2"+ds+"src"
148
#path3 = path3 + path4
149
#os.putenv("PYTHONPATH",path3)
150
#path4 = path4.replace(ps,"")
151
#path4 = path4 + ds+"scrapy-ctl.py"
152
#subprocess.call(["python", path4, "crawl", "naaptol1"])
153
 
154
#path3 = str(os.getenv("PYTHONPATH"))
155
#path4 = path + ds+"naaptolpass3"+ds+"src"
156
#path3 = path3 + path4
157
#os.putenv("PYTHONPATH",path3)
158
#path4 = path4.replace(ps,"")
159
#path4 = path4 + ds+"scrapy-ctl.py"
160
#subprocess.call(["python", path4, "crawl", "naaptol2"])
161
 
162
#phones = da.get_allnaaptolphones()
163
#if urls and not phones:
164
#	mail("ashish.singh@shop2020.in","Bug","There is some mistake in parameters for second or third naaptol spider. This mail is automatically generated via python script",attach)
165
 
166
#path3 = str(os.getenv("PYTHONPATH"))
167
#path4 = path + ds+"data2csv"+ds+"src"
168
#path3 = path3 + path4
169
#os.putenv("PYTHONPATH",path3)
170
#path4 = path4.replace(ps,"")
171
#path4 = path4 + ds+"data2csv"+ds+"testnaaptol.py"
172
#subprocess.call(["python", path4])
173
 
174
 
175
#univercell
176
path3 = str(os.getenv("PYTHONPATH"))
177
path4 = path + ds+"universalScrapypass1"+ds+"src"
178
path3 = path3 + path4
179
os.putenv("PYTHONPATH",path3)
180
path4 = path4.replace(ps,"")
181
path4 = path4 + ds+"scrapy-ctl.py"
182
subprocess.call(["python", path4, "crawl", "univercell"])
183
 
184
vendors = da.get_all_univervendors()
185
if not vendors:
186
	mail("ashish.singh@shop2020.in","Bug","There is some mistake in parameters for first univercell spider. This mail is automatically generated via python script",attach)
187
 
188
path3 = str(os.getenv("PYTHONPATH"))
189
path4 = path + ds+"univercellScrapypass2"+ds+"src"
190
path3 = path3 + path4
191
os.putenv("PYTHONPATH",path3)
192
path4 = path4.replace(ps,"")
193
path4 = path4 + ds+"scrapy-ctl.py"
194
subprocess.call(["python", path4, "crawl", "univercell1"])
195
 
196
phones = da.get_all_univercell_phones()
197
if vendors and not phones:
198
	mail("ashish.singh@shop2020.in","Bug","There is some mistake in parameters for second univercell spider. This mail is automatically generated via python script",attach)
199
 
200
path3 = str(os.getenv("PYTHONPATH"))
201
path4 = path + ds+"data2csv"+ds+"src"
202
path3 = path3 + path4
203
os.putenv("PYTHONPATH",path3)
204
path4 = path4.replace(ps,"")
205
path4 = path4 + ds+"data2csv"+ds+"testfilterunivercell.py"
206
subprocess.call(["python", path4])
207
 
208
 
209
#mobilestore
210
path4 = path
211
path4 = path4.replace(ps,"")
212
path4 = path4+ds+"runmobstore.py"
213
subprocess.call(["python", path4, path])
214
 
215
phones = da.get_allmobstorephones_new()
216
if not phones:
217
	mail("ashish.singh@shop2020.in","Bug","There is some mistake in parameters for mobilestore spider. This mail is automatically generated via python script",attach)
218
 
219
 
220
path3 = str(os.getenv("PYTHONPATH"))
221
path4 = path + ds+"data2csv"+ds+"src"
222
path3 = path3 + path4
223
os.putenv("PYTHONPATH",path3)
224
path4 = path4.replace(ps,"")
225
path4 = path4 + ds+"data2csv"+ds+"new_mobilestore.py"
226
subprocess.call(["python", path4])
227
 
228
 
229
#call indexing
230
path3 = str(os.getenv("PYTHONPATH"))
231
path4 = path + ds+"Lucene"+ds+"src"
232
path3 = path3 + path4
233
os.putenv("PYTHONPATH",path3)
234
path4 = path4.replace(ps,"")
235
path4 = path4 + ds+"Searching"+ds+"Searchbymodel.py"
236
subprocess.call(["python", path4])
237