WebSVN – SmartDukaan – Diff – /prototype/ScrapyDatastore/src/datastore/DataCodeAccessor.py



key_dict = {'INFIBEAM_DOMAINNAME' : "infibeam",
    'INFIBEAM_URL' : "http://www.infibeam.com/Mobiles/search?page=",
    'INFIBEAM_HOMEPAGE' : "www.infibeam.com",
    'INFIBEAM_REFERER' : "www.google.com/search",
    'INFIBEAM_VATPLUSTAX' : 0,
    'INFIBEAM_REMOVELIST' : "Rs.;,;-;/;Rs",
    'INFIBEAM_XPATH1' : '//ul[@class="srch_result portrait"]/li',
    'INFIBEAM_XPATH2' : './/p/span[@class="title"]/text()',
    'INFIBEAM_XPATH3' : './/p/span[@class="price"]/text()',
    'INDIAPLAZA_DOMAINNAME' : "indiaplaza",
    'INDIAPLAZA_URL' : "http://www.indiaplaza.in/mobile-phones-Mobiles-1.htm?PageNo=",

    'INDIAPLAZA_URL1' : "http://www.indiaplaza.in",
    'INDIAPLAZA_XPATH1' : '//tr/td/table[@id="browsesku"]',
    'INDIAPLAZA_XPATH2' : './/div[@class="skuimg"]/a/@title',
    'INDIAPLAZA_XPATH3' : './/div[@class="skuimg"]/a/@href',
    'INDIAPLAZA_DOMAINNAME1' : "indiaplaza1",
    'INDIAPLAZA_REMOVELIST' : "Rs.;,;-;/;Rs",
    'INDIAPLAZA_XPATH4' : './/div[@class="finDetHdr"]/h1/text()',
    'INDIAPLAZA_XPATH5' : './/div[@class="priceArea"]/span[1]/text()',
    'INDIAPLAZA_XPATH6' : './/div[@class="priceArea"]/div[@class="row"][2]/text()',
    'INDIAPLAZA_XPATH7' : './/div[@class="priceArea"]/div[@class="row"][2]/span/text()',
    'INDIAPLAZA_XPATH8' : './/div[@class="priceArea"]/div[@class="row"][3]/text()',

    'UNIVERCELL_VAR1' : ";",
    'UNIVERCELL_VAR2' : "?",
    'UNIVERCELL_VAR3' : "populate",
    'UNIVERCELL_VAR4' : "rePopulate",
    'UNIVERCELL_DOMAINNAME1' : "univercell1",
    'UNIVERCELL_REMOVELIST' : "Rs.;,;-;/;Rs",
    'UNIVERCELL_XPATH4' : '//td[@class="gray-border"]',
    'UNIVERCELL_XPATH5' : './/tr[2]/td/a/text()',
    'UNIVERCELL_XPATH6' : './/tr[3]/th/label/text()',
    'UNIVERCELL_VATPLUSTAX' : 0,
    'BABUCHAK_DOMAINNAME' : "babuchak",

    'BABUCHAK_XPATH4' : '//td[@class="mod-item-body-title"]/a/@href',
    'BABUCHAK_DOMAINNAME2' : "babuchak2",
    'BABUCHAK_XPATH5' : '//td[@class="text-header"]/text()',
    'BABUCHAK_XPATH6' : '//td[@class="xl63"]//strong/span/text()',
    'BABUCHAK_XPATH7' : '//td[@class="mod-item-body-title"]/b/text()',
    'BABUCHAK_REMOVELIST' : "Rs.;,;-;/;Rs",
    'NAAPTOL_DOMAINNAME' : "naaptol",
    'NAAPTOL_URL' : "http://www.naaptol.com/sitemap.xml",
    'NAAPTOL_HOMEPAGE' : "http://www.naaptol.com",
    'NAAPTOL_REFERER' : "http://www.google.com",
    'NAAPTOL_XPATH1' : '//url/loc/text()',
    'NAAPTOL_CHKLIST1' : ["mobile_phones/pdas_and_smartphones" ,"mobile_phones/gsm_handsets" ,"mobile_phones/cdma_handsets"],
    'NAAPTOL_DOMAINNAME1' : "naaptol1",
    'NAAPTOL_DOMAINNAME2' : "naaptol2",
    'NAAPTOL_CHKLIST2' : ['price'],
    'NAAPTOL_PART' : "features",
    'NAAPTOL_REMOVELIST' : "Rs.;,",
    'NAAPTOL_XPATH2' : '//table[@class ="ProductDetails"]//td[@class="Price"]/span/text()',
    'NAAPTOL_XPATH3' : '//table[@class ="ProductDetails"]//td[@class="Price"]/span/script/text()',
    'NAAPTOL_XPATH4' : '//div[@id="OnlineSellers"]//div[@class="ProductResultHead"]//div[@class="headingstyle"]/text()',
    'NAAPTOL_XPATH5' : '//div[@id="onSellerContents"]//td[@class="price"]',
    'NAAPTOL_XPATH6' : '//div[@id="onSellerContents"]//tr[@class="DottedBorder"]/td/a[@id="storeInfoPop',

Subversion Repositories SmartDukaan

(root)/prototype/ScrapyDatastore/src/datastore/DataCodeAccessor.py – Rev 274 → 287

Rev 274		Rev 287
Line 10...		Line 10...
10	`key_dict = {'INFIBEAM_DOMAINNAME' : "infibeam",`	10	`key_dict = {'INFIBEAM_DOMAINNAME' : "infibeam",`
11	`'INFIBEAM_URL' : "http://www.infibeam.com/Mobiles/search?page=",`	11	`'INFIBEAM_URL' : "http://www.infibeam.com/Mobiles/search?page=",`
12	`'INFIBEAM_HOMEPAGE' : "www.infibeam.com",`	12	`'INFIBEAM_HOMEPAGE' : "www.infibeam.com",`
13	`'INFIBEAM_REFERER' : "www.google.com/search",`	13	`'INFIBEAM_REFERER' : "www.google.com/search",`
14	`'INFIBEAM_VATPLUSTAX' : 0,`	14	`'INFIBEAM_VATPLUSTAX' : 0,`
15	`'INFIBEAM_REMOVELIST' : ["Rs.",",","-","/","Rs"],`	15	`'INFIBEAM_REMOVELIST' : "Rs.;,;-;/;Rs",`
16	`'INFIBEAM_XPATH1' : '//ul[@class="srch_result portrait"]/li',`	16	`'INFIBEAM_XPATH1' : '//ul[@class="srch_result portrait"]/li',`
17	`'INFIBEAM_XPATH2' : './/p/span[@class="title"]/text()',`	17	`'INFIBEAM_XPATH2' : './/p/span[@class="title"]/text()',`
18	`'INFIBEAM_XPATH3' : './/p/span[@class="price"]/text()',`	18	`'INFIBEAM_XPATH3' : './/p/span[@class="price"]/text()',`
19	`'INDIAPLAZA_DOMAINNAME' : "indiaplaza",`	19	`'INDIAPLAZA_DOMAINNAME' : "indiaplaza",`
20	`'INDIAPLAZA_URL' : "http://www.indiaplaza.in/mobile-phones-Mobiles-1.htm?PageNo=",`	20	`'INDIAPLAZA_URL' : "http://www.indiaplaza.in/mobile-phones-Mobiles-1.htm?PageNo=",`
Line 23...		Line 23...
23	`'INDIAPLAZA_URL1' : "http://www.indiaplaza.in",`	23	`'INDIAPLAZA_URL1' : "http://www.indiaplaza.in",`
24	`'INDIAPLAZA_XPATH1' : '//tr/td/table[@id="browsesku"]',`	24	`'INDIAPLAZA_XPATH1' : '//tr/td/table[@id="browsesku"]',`
25	`'INDIAPLAZA_XPATH2' : './/div[@class="skuimg"]/a/@title',`	25	`'INDIAPLAZA_XPATH2' : './/div[@class="skuimg"]/a/@title',`
26	`'INDIAPLAZA_XPATH3' : './/div[@class="skuimg"]/a/@href',`	26	`'INDIAPLAZA_XPATH3' : './/div[@class="skuimg"]/a/@href',`
27	`'INDIAPLAZA_DOMAINNAME1' : "indiaplaza1",`	27	`'INDIAPLAZA_DOMAINNAME1' : "indiaplaza1",`
28	`'INDIAPLAZA_REMOVELIST' : ["Rs.","Rs",",","-","/"],`	28	`'INDIAPLAZA_REMOVELIST' : "Rs.;,;-;/;Rs",`
29	`'INDIAPLAZA_XPATH4' : './/div[@class="finDetHdr"]/h1/text()',`	29	`'INDIAPLAZA_XPATH4' : './/div[@class="finDetHdr"]/h1/text()',`
30	`'INDIAPLAZA_XPATH5' : './/div[@class="priceArea"]/span[1]/text()',`	30	`'INDIAPLAZA_XPATH5' : './/div[@class="priceArea"]/span[1]/text()',`
31	`'INDIAPLAZA_XPATH6' : './/div[@class="priceArea"]/div[@class="row"][2]/text()',`	31	`'INDIAPLAZA_XPATH6' : './/div[@class="priceArea"]/div[@class="row"][2]/text()',`
32	`'INDIAPLAZA_XPATH7' : './/div[@class="priceArea"]/div[@class="row"][2]/span/text()',`	32	`'INDIAPLAZA_XPATH7' : './/div[@class="priceArea"]/div[@class="row"][2]/span/text()',`
33	`'INDIAPLAZA_XPATH8' : './/div[@class="priceArea"]/div[@class="row"][3]/text()',`	33	`'INDIAPLAZA_XPATH8' : './/div[@class="priceArea"]/div[@class="row"][3]/text()',`
Line 56...		Line 56...
56	`'UNIVERCELL_VAR1' : ";",`	56	`'UNIVERCELL_VAR1' : ";",`
57	`'UNIVERCELL_VAR2' : "?",`	57	`'UNIVERCELL_VAR2' : "?",`
58	`'UNIVERCELL_VAR3' : "populate",`	58	`'UNIVERCELL_VAR3' : "populate",`
59	`'UNIVERCELL_VAR4' : "rePopulate",`	59	`'UNIVERCELL_VAR4' : "rePopulate",`
60	`'UNIVERCELL_DOMAINNAME1' : "univercell1",`	60	`'UNIVERCELL_DOMAINNAME1' : "univercell1",`
61	`'UNIVERCELL_REMOVELIST' : ["Rs",",","-","/"],`	61	`'UNIVERCELL_REMOVELIST' : "Rs.;,;-;/;Rs",`
62	`'UNIVERCELL_XPATH4' : '//td[@class="gray-border"]',`	62	`'UNIVERCELL_XPATH4' : '//td[@class="gray-border"]',`
63	`'UNIVERCELL_XPATH5' : './/tr[2]/td/a/text()',`	63	`'UNIVERCELL_XPATH5' : './/tr[2]/td/a/text()',`
64	`'UNIVERCELL_XPATH6' : './/tr[3]/th/label/text()',`	64	`'UNIVERCELL_XPATH6' : './/tr[3]/th/label/text()',`
65	`'UNIVERCELL_VATPLUSTAX' : 0,`	65	`'UNIVERCELL_VATPLUSTAX' : 0,`
66	`'BABUCHAK_DOMAINNAME' : "babuchak",`	66	`'BABUCHAK_DOMAINNAME' : "babuchak",`
Line 77...		Line 77...
77	`'BABUCHAK_XPATH4' : '//td[@class="mod-item-body-title"]/a/@href',`	77	`'BABUCHAK_XPATH4' : '//td[@class="mod-item-body-title"]/a/@href',`
78	`'BABUCHAK_DOMAINNAME2' : "babuchak2",`	78	`'BABUCHAK_DOMAINNAME2' : "babuchak2",`
79	`'BABUCHAK_XPATH5' : '//td[@class="text-header"]/text()',`	79	`'BABUCHAK_XPATH5' : '//td[@class="text-header"]/text()',`
80	`'BABUCHAK_XPATH6' : '//td[@class="xl63"]//strong/span/text()',`	80	`'BABUCHAK_XPATH6' : '//td[@class="xl63"]//strong/span/text()',`
81	`'BABUCHAK_XPATH7' : '//td[@class="mod-item-body-title"]/b/text()',`	81	`'BABUCHAK_XPATH7' : '//td[@class="mod-item-body-title"]/b/text()',`
82	`'BABUCHAK_REMOVELIST' : ["Rs.","Rs",",","-","/"],`	82	`'BABUCHAK_REMOVELIST' : "Rs.;,;-;/;Rs",`
83	`'NAAPTOL_DOMAINNAME' : "naaptol",`	83	`'NAAPTOL_DOMAINNAME' : "naaptol",`
84	`'NAAPTOL_URL' : "http://www.naaptol.com/sitemap.xml",`	84	`'NAAPTOL_URL' : "http://www.naaptol.com/sitemap.xml",`
85	`'NAAPTOL_HOMEPAGE' : "http://www.naaptol.com",`	85	`'NAAPTOL_HOMEPAGE' : "http://www.naaptol.com",`
86	`'NAAPTOL_REFERER' : "http://www.google.com",`	86	`'NAAPTOL_REFERER' : "http://www.google.com",`
87	`'NAAPTOL_XPATH1' : '//url/loc/text()',`	87	`'NAAPTOL_XPATH1' : '//url/loc/text()',`
88	`'NAAPTOL_CHKLIST1' : ["mobile_phones/pdas_and_smartphones" ,"mobile_phones/gsm_handsets" ,"mobile_phones/cdma_handsets"],`	88	`'NAAPTOL_CHKLIST1' : ["mobile_phones/pdas_and_smartphones" ,"mobile_phones/gsm_handsets" ,"mobile_phones/cdma_handsets"],`
89	`'NAAPTOL_DOMAINNAME1' : "naaptol1",`	89	`'NAAPTOL_DOMAINNAME1' : "naaptol1",`
90	`'NAAPTOL_DOMAINNAME2' : "naaptol2",`	90	`'NAAPTOL_DOMAINNAME2' : "naaptol2",`
91	`'NAAPTOL_CHKLIST2' : ['price'],`	91	`'NAAPTOL_CHKLIST2' : ['price'],`
92	`'NAAPTOL_PART' : "features",`	92	`'NAAPTOL_PART' : "features",`
93	`'NAAPTOL_REMOVELIST' : ["Rs.",","],`	93	`'NAAPTOL_REMOVELIST' : "Rs.;,",`
94	`'NAAPTOL_XPATH2' : '//table[@class ="ProductDetails"]//td[@class="Price"]/span/text()',`	94	`'NAAPTOL_XPATH2' : '//table[@class ="ProductDetails"]//td[@class="Price"]/span/text()',`
95	`'NAAPTOL_XPATH3' : '//table[@class ="ProductDetails"]//td[@class="Price"]/span/script/text()',`	95	`'NAAPTOL_XPATH3' : '//table[@class ="ProductDetails"]//td[@class="Price"]/span/script/text()',`
96	`'NAAPTOL_XPATH4' : '//div[@id="OnlineSellers"]//div[@class="ProductResultHead"]//div[@class="headingstyle"]/text()',`	96	`'NAAPTOL_XPATH4' : '//div[@id="OnlineSellers"]//div[@class="ProductResultHead"]//div[@class="headingstyle"]/text()',`
97	`'NAAPTOL_XPATH5' : '//div[@id="onSellerContents"]//td[@class="price"]',`	97	`'NAAPTOL_XPATH5' : '//div[@id="onSellerContents"]//td[@class="price"]',`
98	`'NAAPTOL_XPATH6' : '//div[@id="onSellerContents"]//tr[@class="DottedBorder"]/td/a[@id="storeInfoPop',`	98	`'NAAPTOL_XPATH6' : '//div[@id="onSellerContents"]//tr[@class="DottedBorder"]/td/a[@id="storeInfoPop',`