2 HCE project, Python bindings, Distributed Tasks Manager application. 3 Event objects definitions. 7 @author Oleksii <developers.hce@gmail.com> 8 @author madk <developers.hce@gmail.com> 9 @link: http://hierarchical-cluster-engine.com/ 10 @copyright: Copyright © 2013-2014 IOIX Ukraine 11 @license: http://hierarchical-cluster-engine.com/license/ 16 FETCHER_TIME_LIMIT_MAX = 100
17 CONNECTION_TIMEOUT = 1.0
19 MAX_HTTP_REDIRECTS_LIMIT = 5
20 MAX_HTTP_SIZE_UNLIMIT = 0
22 MAX_HTML_REDIRECTS_LIMIT = 1
27 RTC_FINALIZER_APP_NAME =
"rtc-finalizer" 28 RTC_PREPROCESSOR_APP_NAME =
"rtc-preprocessor" 31 pubdateFeedNames = [
"pubdate",
"published",
"pubDate",
"published_parsed",
"updated_parsed"]
32 pubdateRssFeedHeaderName =
"X-pubdateRssFeed" 33 rssFeedUrlHeaderName =
"X-feed_url" 34 baseUrlHeaderName =
"X-base_url" 41 REDIRECT_HTTP_CODES = [301, 302, 303, 304]
42 REDIRECT_HEADER_FIELDS_FOR_REMOVE = [
'referer',
'content-type',
'Location',
'cookie']
45 charsetDetectorMap = {
46 'win-1251':
'windows-1251',
53 'ascii':
'646, us-ascii',
54 'big5':
'big5-tw, csbig5',
55 'big5hkscs':
'big5-hkscs, hkscs',
56 'cp037':
'IBM037, IBM039',
57 'cp424':
'EBCDIC-CP-HE, IBM424',
58 'cp437':
'437, IBM437',
59 'cp500':
'EBCDIC-CP-BE, EBCDIC-CP-CH, IBM500',
63 'cp850':
'850, IBM850',
64 'cp852':
'852, IBM852',
65 'cp855':
'855, IBM855',
67 'cp857':
'857, IBM857',
68 'cp858':
'858, IBM858',
69 'cp860':
'860, IBM860',
70 'cp861':
'861, CP-IS, IBM861',
71 'cp862':
'862, IBM862',
72 'cp863':
'863, IBM863',
74 'cp865':
'865, IBM865',
75 'cp866':
'866, IBM866',
76 'cp869':
'869, CP-GR, IBM869',
79 'cp932':
'932, ms932, mskanji, ms-kanji',
80 'cp949':
'949, ms949, uhc',
85 'cp1250':
'windows-1250',
86 'cp1251':
'windows-1251',
87 'cp1252':
'windows-1252',
88 'cp1253':
'windows-1253',
89 'cp1254':
'windows-1254',
90 'cp1255':
'windows-1255',
91 'cp1256':
'windows-1256',
92 'cp1257':
'windows-1257',
93 'cp1258':
'windows-1258',
94 'euc_jp':
'eucjp, ujis, u-jis',
95 'euc_jis_2004':
'jisx0213, eucjis2004',
96 'euc_jisx0213':
'eucjisx0213',
97 'euc_kr':
'euckr, korean, ksc5601, ks_c-5601, ks_c-5601-1987, ksx1001, ks_x-1001',
98 'gb2312':
'chinese, csiso58gb231280, euc- cn, euccn, eucgb2312-cn, gb2312-1980, gb2312-80, iso- ir-58',
99 'gbk':
'936, cp936, ms936',
100 'gb18030':
'gb18030-2000',
101 'hz':
'hzgb, hz-gb, hz-gb-2312',
102 'iso2022_jp':
'csiso2022jp, iso2022jp, iso-2022-jp',
103 'iso2022_jp_1':
'iso2022jp-1, iso-2022-jp-1',
104 'iso2022_jp_2':
'iso2022jp-2, iso-2022-jp-2',
105 'iso2022_jp_2004':
'iso2022jp-2004, iso-2022-jp-2004',
106 'iso2022_jp_3':
'iso2022jp-3, iso-2022-jp-3',
107 'iso2022_jp_ext':
'iso2022jp-ext, iso-2022-jp-ext',
108 'iso2022_kr':
'csiso2022kr, iso2022kr, iso-2022-kr',
109 'latin_1':
'iso-8859-1, iso8859-1, 8859, cp819, latin, latin1, L1',
110 'iso8859_2':
'iso-8859-2, latin2, L2',
111 'iso8859_3':
'iso-8859-3, latin3, L3',
112 'iso8859_4':
'iso-8859-4, latin4, L4',
113 'iso8859_5':
'iso-8859-5, cyrillic',
114 'iso8859_6':
'iso-8859-6, arabic',
115 'iso8859_7':
'iso-8859-7, greek, greek8',
116 'iso8859_8':
'iso-8859-8, hebrew',
117 'iso8859_9':
'iso-8859-9, latin5, L5',
118 'iso8859_10':
'iso-8859-10, latin6, L6',
119 'iso8859_11':
'iso-8859-11, thai',
120 'iso8859_13':
'iso-8859-13, latin7, L7',
121 'iso8859_14':
'iso-8859-14, latin8, L8',
122 'iso8859_15':
'iso-8859-15, latin9, L9',
123 'iso8859_16':
'iso-8859-16, latin10, L10',
124 'johab':
'cp1361, ms1361',
127 'mac_cyrillic':
'maccyrillic',
128 'mac_greek':
'macgreek',
129 'mac_iceland':
'maciceland',
130 'mac_latin2':
'maclatin2, maccentraleurope',
131 'mac_roman':
'macroman',
132 'mac_turkish':
'macturkish',
133 'ptcp154':
'csptcp154, pt154, cp154, cyrillic-asian',
134 'shift_jis':
'csshiftjis, shiftjis, sjis, s_jis',
135 'shift_jis_2004':
'shiftjis2004, sjis_2004, sjis2004',
136 'shift_jisx0213':
'shiftjisx0213, sjisx0213, s_jisx0213',
137 'utf_32':
'U32, utf32',
138 'utf_32_be':
'UTF-32BE',
139 'utf_32_le':
'UTF-32LE',
140 'utf_16':
'U16, utf16',
141 'utf_16_be':
'UTF-16BE',
142 'utf_16_le':
'UTF-16LE',
143 'utf_7':
'U7, unicode-1-1-utf-7',
144 'utf_8':
'U8, UTF, utf8',