re-common 10.0.39__py3-none-any.whl → 10.0.41__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (221) hide show
  1. re_common/baselibrary/__init__.py +4 -4
  2. re_common/baselibrary/baseabs/__init__.py +6 -6
  3. re_common/baselibrary/baseabs/baseabs.py +26 -26
  4. re_common/baselibrary/database/mbuilder.py +132 -132
  5. re_common/baselibrary/database/moudle.py +93 -93
  6. re_common/baselibrary/database/msqlite3.py +194 -194
  7. re_common/baselibrary/database/mysql.py +169 -169
  8. re_common/baselibrary/database/sql_factory.py +26 -26
  9. re_common/baselibrary/mthread/MThreadingRun.py +486 -486
  10. re_common/baselibrary/mthread/MThreadingRunEvent.py +349 -349
  11. re_common/baselibrary/mthread/__init__.py +2 -2
  12. re_common/baselibrary/mthread/mythreading.py +695 -695
  13. re_common/baselibrary/pakge_other/socks.py +404 -404
  14. re_common/baselibrary/readconfig/config_factory.py +18 -18
  15. re_common/baselibrary/readconfig/ini_config.py +317 -317
  16. re_common/baselibrary/readconfig/toml_config.py +49 -49
  17. re_common/baselibrary/temporary/envdata.py +36 -36
  18. re_common/baselibrary/tools/all_requests/aiohttp_request.py +118 -118
  19. re_common/baselibrary/tools/all_requests/httpx_requet.py +102 -102
  20. re_common/baselibrary/tools/all_requests/mrequest.py +412 -412
  21. re_common/baselibrary/tools/all_requests/requests_request.py +81 -81
  22. re_common/baselibrary/tools/batch_compre/bijiao_batch.py +31 -31
  23. re_common/baselibrary/tools/contrast_db3.py +123 -123
  24. re_common/baselibrary/tools/copy_file.py +39 -39
  25. re_common/baselibrary/tools/db3_2_sizedb3.py +102 -102
  26. re_common/baselibrary/tools/foreachgz.py +39 -39
  27. re_common/baselibrary/tools/get_attr.py +10 -10
  28. re_common/baselibrary/tools/image_to_pdf.py +61 -61
  29. re_common/baselibrary/tools/java_code_deal.py +139 -139
  30. re_common/baselibrary/tools/javacode.py +79 -79
  31. re_common/baselibrary/tools/mdb_db3.py +48 -48
  32. re_common/baselibrary/tools/merge_file.py +171 -171
  33. re_common/baselibrary/tools/merge_gz_file.py +165 -165
  34. re_common/baselibrary/tools/mhdfstools/down_hdfs_files.py +42 -42
  35. re_common/baselibrary/tools/mhdfstools/hdfst.py +42 -42
  36. re_common/baselibrary/tools/mhdfstools/up_hdfs_files.py +38 -38
  37. re_common/baselibrary/tools/mongo_tools.py +50 -50
  38. re_common/baselibrary/tools/move_file.py +170 -170
  39. re_common/baselibrary/tools/move_mongo/mongo_table_to_file.py +63 -63
  40. re_common/baselibrary/tools/move_mongo/move_mongo_table.py +354 -354
  41. re_common/baselibrary/tools/move_mongo/use_mttf.py +18 -18
  42. re_common/baselibrary/tools/move_mongo/use_mv.py +93 -93
  43. re_common/baselibrary/tools/mpandas/mpandasreadexcel.py +125 -125
  44. re_common/baselibrary/tools/mpandas/pandas_visualization.py +7 -7
  45. re_common/baselibrary/tools/myparsel.py +104 -104
  46. re_common/baselibrary/tools/rename_dir_file.py +37 -37
  47. re_common/baselibrary/tools/sequoiadb_utils.py +398 -398
  48. re_common/baselibrary/tools/split_line_to_many.py +25 -25
  49. re_common/baselibrary/tools/stringtodicts.py +33 -33
  50. re_common/baselibrary/tools/workwechant_bot.py +84 -84
  51. re_common/baselibrary/utils/baseaiohttp.py +296 -296
  52. re_common/baselibrary/utils/baseaiomysql.py +87 -87
  53. re_common/baselibrary/utils/baseallstep.py +191 -191
  54. re_common/baselibrary/utils/baseavro.py +19 -19
  55. re_common/baselibrary/utils/baseboto3.py +291 -291
  56. re_common/baselibrary/utils/basecsv.py +32 -32
  57. re_common/baselibrary/utils/basedict.py +133 -133
  58. re_common/baselibrary/utils/basedir.py +241 -241
  59. re_common/baselibrary/utils/baseencode.py +351 -351
  60. re_common/baselibrary/utils/baseencoding.py +28 -28
  61. re_common/baselibrary/utils/baseesdsl.py +86 -86
  62. re_common/baselibrary/utils/baseexcel.py +264 -264
  63. re_common/baselibrary/utils/baseexcept.py +109 -109
  64. re_common/baselibrary/utils/basefile.py +654 -654
  65. re_common/baselibrary/utils/baseftp.py +214 -214
  66. re_common/baselibrary/utils/basegzip.py +60 -60
  67. re_common/baselibrary/utils/basehdfs.py +135 -135
  68. re_common/baselibrary/utils/basehttpx.py +268 -268
  69. re_common/baselibrary/utils/baseip.py +87 -87
  70. re_common/baselibrary/utils/basejson.py +2 -2
  71. re_common/baselibrary/utils/baselist.py +32 -32
  72. re_common/baselibrary/utils/basemotor.py +190 -190
  73. re_common/baselibrary/utils/basemssql.py +98 -98
  74. re_common/baselibrary/utils/baseodbc.py +113 -113
  75. re_common/baselibrary/utils/basepandas.py +302 -302
  76. re_common/baselibrary/utils/basepeewee.py +11 -11
  77. re_common/baselibrary/utils/basepika.py +180 -180
  78. re_common/baselibrary/utils/basepydash.py +143 -143
  79. re_common/baselibrary/utils/basepymongo.py +230 -230
  80. re_common/baselibrary/utils/basequeue.py +22 -22
  81. re_common/baselibrary/utils/baserar.py +57 -57
  82. re_common/baselibrary/utils/baserequest.py +279 -279
  83. re_common/baselibrary/utils/baseset.py +8 -8
  84. re_common/baselibrary/utils/basesmb.py +403 -403
  85. re_common/baselibrary/utils/basestring.py +382 -382
  86. re_common/baselibrary/utils/basetime.py +320 -320
  87. re_common/baselibrary/utils/baseurl.py +121 -121
  88. re_common/baselibrary/utils/basezip.py +57 -57
  89. re_common/baselibrary/utils/core/__init__.py +7 -7
  90. re_common/baselibrary/utils/core/bottomutils.py +18 -18
  91. re_common/baselibrary/utils/core/mdeprecated.py +327 -327
  92. re_common/baselibrary/utils/core/mlamada.py +16 -16
  93. re_common/baselibrary/utils/core/msginfo.py +25 -25
  94. re_common/baselibrary/utils/core/requests_core.py +103 -103
  95. re_common/baselibrary/utils/fateadm.py +429 -429
  96. re_common/baselibrary/utils/importfun.py +123 -123
  97. re_common/baselibrary/utils/mfaker.py +57 -57
  98. re_common/baselibrary/utils/my_abc/__init__.py +3 -3
  99. re_common/baselibrary/utils/my_abc/better_abc.py +32 -32
  100. re_common/baselibrary/utils/mylogger.py +414 -414
  101. re_common/baselibrary/utils/myredisclient.py +861 -861
  102. re_common/baselibrary/utils/pipupgrade.py +21 -21
  103. re_common/baselibrary/utils/ringlist.py +85 -85
  104. re_common/baselibrary/utils/version_compare.py +36 -36
  105. re_common/baselibrary/utils/ydmhttp.py +126 -126
  106. re_common/facade/lazy_import.py +11 -11
  107. re_common/facade/loggerfacade.py +25 -25
  108. re_common/facade/mysqlfacade.py +467 -467
  109. re_common/facade/now.py +31 -31
  110. re_common/facade/sqlite3facade.py +257 -257
  111. re_common/facade/use/mq_use_facade.py +83 -83
  112. re_common/facade/use/proxy_use_facade.py +19 -19
  113. re_common/libtest/base_dict_test.py +19 -19
  114. re_common/libtest/baseavro_test.py +13 -13
  115. re_common/libtest/basefile_test.py +14 -14
  116. re_common/libtest/basemssql_test.py +77 -77
  117. re_common/libtest/baseodbc_test.py +7 -7
  118. re_common/libtest/basepandas_test.py +38 -38
  119. re_common/libtest/get_attr_test/get_attr_test_settings.py +14 -14
  120. re_common/libtest/get_attr_test/settings.py +54 -54
  121. re_common/libtest/idencode_test.py +53 -53
  122. re_common/libtest/iniconfig_test.py +35 -35
  123. re_common/libtest/ip_test.py +34 -34
  124. re_common/libtest/merge_file_test.py +20 -20
  125. re_common/libtest/mfaker_test.py +8 -8
  126. re_common/libtest/mm3_test.py +31 -31
  127. re_common/libtest/mylogger_test.py +88 -88
  128. re_common/libtest/myparsel_test.py +27 -27
  129. re_common/libtest/mysql_test.py +151 -151
  130. re_common/libtest/pymongo_test.py +21 -21
  131. re_common/libtest/split_test.py +11 -11
  132. re_common/libtest/sqlite3_merge_test.py +5 -5
  133. re_common/libtest/sqlite3_test.py +34 -34
  134. re_common/libtest/tomlconfig_test.py +30 -30
  135. re_common/libtest/use_tools_test/__init__.py +2 -2
  136. re_common/libtest/user/__init__.py +4 -4
  137. re_common/studio/__init__.py +4 -4
  138. re_common/studio/assignment_expressions.py +36 -36
  139. re_common/studio/mydash/test1.py +18 -18
  140. re_common/studio/pydashstudio/first.py +9 -9
  141. re_common/studio/streamlitstudio/first_app.py +65 -65
  142. re_common/studio/streamlitstudio/uber_pickups.py +23 -23
  143. re_common/studio/test.py +18 -18
  144. re_common/v2/baselibrary/business_utils/BusinessStringUtil.py +235 -220
  145. re_common/v2/baselibrary/business_utils/baseencodeid.py +100 -100
  146. re_common/v2/baselibrary/business_utils/full_doi_path.py +116 -116
  147. re_common/v2/baselibrary/business_utils/rel_tools.py +6 -6
  148. re_common/v2/baselibrary/decorators/utils.py +59 -59
  149. re_common/v2/baselibrary/helpers/search_packge/NearestNeighbors_test.py +105 -105
  150. re_common/v2/baselibrary/helpers/search_packge/fit_text_match.py +253 -253
  151. re_common/v2/baselibrary/helpers/search_packge/scikit_learn_text_matcher.py +260 -260
  152. re_common/v2/baselibrary/helpers/search_packge/test.py +1 -1
  153. re_common/v2/baselibrary/s3object/baseboto3.py +230 -230
  154. re_common/v2/baselibrary/tools/WeChatRobot.py +95 -95
  155. re_common/v2/baselibrary/tools/ac_ahocorasick.py +75 -75
  156. re_common/v2/baselibrary/tools/concurrency.py +35 -35
  157. re_common/v2/baselibrary/tools/data_processer/base.py +53 -53
  158. re_common/v2/baselibrary/tools/data_processer/data_processer.py +497 -508
  159. re_common/v2/baselibrary/tools/data_processer/data_reader.py +187 -187
  160. re_common/v2/baselibrary/tools/data_processer/data_writer.py +38 -38
  161. re_common/v2/baselibrary/tools/dict_tools.py +44 -44
  162. re_common/v2/baselibrary/tools/dolphinscheduler.py +187 -187
  163. re_common/v2/baselibrary/tools/hdfs_base_processor.py +204 -204
  164. re_common/v2/baselibrary/tools/hdfs_bulk_processor.py +67 -67
  165. re_common/v2/baselibrary/tools/hdfs_data_processer.py +338 -338
  166. re_common/v2/baselibrary/tools/hdfs_line_processor.py +74 -74
  167. re_common/v2/baselibrary/tools/list_tools.py +69 -69
  168. re_common/v2/baselibrary/tools/resume_tracker.py +94 -94
  169. re_common/v2/baselibrary/tools/search_hash_tools.py +54 -54
  170. re_common/v2/baselibrary/tools/text_matcher.py +326 -326
  171. re_common/v2/baselibrary/tools/tree_processor/__init__.py +0 -0
  172. re_common/v2/baselibrary/tools/tree_processor/builder.py +25 -0
  173. re_common/v2/baselibrary/tools/tree_processor/node.py +13 -0
  174. re_common/v2/baselibrary/tools/unionfind_tools.py +60 -60
  175. re_common/v2/baselibrary/utils/BusinessStringUtil.py +196 -196
  176. re_common/v2/baselibrary/utils/api_net_utils.py +270 -270
  177. re_common/v2/baselibrary/utils/author_smi.py +361 -361
  178. re_common/v2/baselibrary/utils/base_string_similarity.py +158 -158
  179. re_common/v2/baselibrary/utils/basedict.py +37 -37
  180. re_common/v2/baselibrary/utils/basehdfs.py +163 -163
  181. re_common/v2/baselibrary/utils/basepika.py +180 -180
  182. re_common/v2/baselibrary/utils/basetime.py +94 -77
  183. re_common/v2/baselibrary/utils/db.py +174 -156
  184. re_common/v2/baselibrary/utils/elasticsearch.py +46 -0
  185. re_common/v2/baselibrary/utils/json_cls.py +16 -16
  186. re_common/v2/baselibrary/utils/mq.py +83 -83
  187. re_common/v2/baselibrary/utils/n_ary_expression_tree.py +243 -243
  188. re_common/v2/baselibrary/utils/string_bool.py +187 -186
  189. re_common/v2/baselibrary/utils/string_clear.py +246 -246
  190. re_common/v2/baselibrary/utils/string_smi.py +18 -18
  191. re_common/v2/baselibrary/utils/stringutils.py +312 -271
  192. re_common/vip/base_step_process.py +11 -11
  193. re_common/vip/baseencodeid.py +90 -90
  194. re_common/vip/changetaskname.py +28 -28
  195. re_common/vip/core_var.py +24 -24
  196. re_common/vip/mmh3Hash.py +89 -89
  197. re_common/vip/proxy/allproxys.py +127 -127
  198. re_common/vip/proxy/allproxys_thread.py +159 -159
  199. re_common/vip/proxy/cnki_proxy.py +153 -153
  200. re_common/vip/proxy/kuaidaili.py +87 -87
  201. re_common/vip/proxy/proxy_all.py +113 -113
  202. re_common/vip/proxy/update_kuaidaili_0.py +42 -42
  203. re_common/vip/proxy/wanfang_proxy.py +152 -152
  204. re_common/vip/proxy/wp_proxy_all.py +181 -181
  205. re_common/vip/read_rawid_to_txt.py +91 -91
  206. re_common/vip/title/__init__.py +5 -5
  207. re_common/vip/title/transform/TransformBookTitleToZt.py +125 -125
  208. re_common/vip/title/transform/TransformConferenceTitleToZt.py +139 -139
  209. re_common/vip/title/transform/TransformCstadTitleToZt.py +195 -195
  210. re_common/vip/title/transform/TransformJournalTitleToZt.py +203 -203
  211. re_common/vip/title/transform/TransformPatentTitleToZt.py +132 -132
  212. re_common/vip/title/transform/TransformRegulationTitleToZt.py +114 -114
  213. re_common/vip/title/transform/TransformStandardTitleToZt.py +135 -135
  214. re_common/vip/title/transform/TransformThesisTitleToZt.py +135 -135
  215. re_common/vip/title/transform/__init__.py +10 -10
  216. {re_common-10.0.39.dist-info → re_common-10.0.41.dist-info}/LICENSE +201 -201
  217. {re_common-10.0.39.dist-info → re_common-10.0.41.dist-info}/METADATA +16 -16
  218. re_common-10.0.41.dist-info/RECORD +252 -0
  219. {re_common-10.0.39.dist-info → re_common-10.0.41.dist-info}/WHEEL +1 -1
  220. re_common-10.0.39.dist-info/RECORD +0 -248
  221. {re_common-10.0.39.dist-info → re_common-10.0.41.dist-info}/top_level.txt +0 -0
@@ -1,125 +1,125 @@
1
- # -*- coding:utf-8 -*-
2
- # @Time : 2021/12/19 18:16
3
- # @Author: suhong
4
- # @File : TransformBookTitleToZt.py
5
- # @Function : 图书a层转智图
6
- from re_common.baselibrary.utils.basetime import BaseTime
7
-
8
- from re_common.facade.mysqlfacade import MysqlUtiles
9
-
10
-
11
- class TransformBookTitleToZt():
12
- def __init__(self):
13
- self.fields = [
14
- "lngid",
15
- "rawid",
16
- "title",
17
- "title_alternative",
18
- "title_sub",
19
- "title_edition",
20
- "title_series",
21
- "identifier_eisbn",
22
- "identifier_pisbn",
23
- "identifier_doi",
24
- "creator",
25
- "creator_en",
26
- "creator_bio",
27
- "creator_institution",
28
- "publisher",
29
- "date",
30
- "description",
31
- "description_en",
32
- "description_unit",
33
- "subject",
34
- "subject_en",
35
- "subject_clc",
36
- "subject_esc",
37
- "page",
38
- "beginpage",
39
- "endpage",
40
- "pagecount",
41
- "date_created",
42
- "rawtype",
43
- "folio_size",
44
- "price",
45
- "pub_place",
46
- "language",
47
- "country",
48
- "type",
49
- "provider",
50
- "provider_url",
51
- "provider_id",
52
- "medium",
53
- "batch",
54
- "is_deprecated"
55
- ]
56
- self.zt_providermap = dict()
57
- self.mysqlutils = MysqlUtiles(None, None, builder="MysqlBuilderForDicts", dicts={
58
- "host": "192.168.31.24",
59
- "user": "root",
60
- "passwd": "vipdatacenter",
61
- "db": "data_warehouse_sql",
62
- "port": "3306",
63
- "chartset": "utf8mb4",
64
- })
65
- self.get_zt_provider()
66
-
67
- def get_zt_provider(self):
68
- rows = self.mysqlutils.SelectFromDB(
69
- "select sub_db_id,provider from a_transform_task where source_type = '1' and out_type = 'zt'")
70
- for row in rows[1]:
71
- self.zt_providermap[row[0]] = row[1]
72
-
73
- def transform(self, titleMap):
74
- transMap = dict()
75
- sub_db_id = titleMap.get("sub_db_id", "")
76
- transMap["lngid"] = titleMap.get("lngid", "")
77
- transMap["rawid"] = titleMap.get("rawid", "")
78
- transMap["title"] = titleMap.get("title", "")
79
- transMap["title_alternative"] = titleMap.get("title_alt", "")
80
- transMap["title_series"] = titleMap.get("title_series", "")
81
- transMap["title_edition"] = titleMap.get("revision", "")
82
- transMap["identifier_eisbn"] = titleMap.get("eisbn", "")
83
- transMap["identifier_pisbn"] = titleMap.get("isbn", "")
84
- transMap["identifier_doi"] = titleMap.get("doi", "")
85
- transMap["creator"] = titleMap.get("author", "")
86
- transMap["creator_en"] = titleMap.get("author_alt", "")
87
- transMap["creator_institution"] = titleMap.get("organ", "")
88
- transMap["creator_bio"] = titleMap.get("author_intro", "")
89
- transMap["publisher"] = titleMap.get("publisher", "")
90
- transMap["date"] = titleMap.get("pub_year", "")
91
- transMap["description"] = titleMap.get("abstract", "")
92
- transMap["description_en"] = titleMap.get("abstract_alt", "")
93
- transMap["description_unit"] = titleMap.get("catalog", "")
94
- transMap["subject"] = titleMap.get("keyword", "")
95
- transMap["subject_en"] = titleMap.get("keyword_alt", "")
96
- transMap["subject_clc"] = titleMap.get("clc_no", "")
97
- transMap["subject_esc"] = titleMap.get("subject_edu", "")
98
- transMap["page"] = titleMap.get("page_info", "")
99
- transMap["beginpage"] = titleMap.get("begin_page", "")
100
- transMap["endpage"] = titleMap.get("end_page", "")
101
- transMap["pagecount"] = titleMap.get("page_cnt", "")
102
- transMap["date_created"] = titleMap.get("pub_date", "")
103
- transMap["rawtype"] = titleMap.get("raw_type", "")
104
- transMap["folio_size"] = titleMap.get("book_size", "")
105
- transMap["price"] = titleMap.get("price", "")
106
- transMap["pub_place"] = titleMap.get("pub_place", "")
107
- # transMap["provider"] = titleMap.get("zt_provider", "")
108
- # if transMap["provider"] == "":
109
- transMap["provider"] = self.zt_providermap[sub_db_id]
110
- transMap["provider_url"] = transMap["provider"] + "@" + titleMap.get("provider_url")
111
- transMap["provider_id"] = transMap["provider"] + "@" + titleMap.get("rawid")
112
- transMap["batch"] = BaseTime().get_beijin_date_strins("%Y%m%d") + "00"
113
- transMap["type"] = titleMap.get("source_type", "")
114
- transMap["medium"] = "2"
115
- transMap["country"] = titleMap.get("country", "")
116
- transMap["language"] = titleMap.get("language", "")
117
- transMap["is_deprecated"] = titleMap.get("is_deprecated", "")
118
-
119
- for field in self.fields:
120
- if field not in transMap.keys():
121
- transMap[field] = ""
122
- return transMap
123
-
124
- if __name__ == '__main__':
125
- t = TransformBookTitleToZt()
1
+ # -*- coding:utf-8 -*-
2
+ # @Time : 2021/12/19 18:16
3
+ # @Author: suhong
4
+ # @File : TransformBookTitleToZt.py
5
+ # @Function : 图书a层转智图
6
+ from re_common.baselibrary.utils.basetime import BaseTime
7
+
8
+ from re_common.facade.mysqlfacade import MysqlUtiles
9
+
10
+
11
+ class TransformBookTitleToZt():
12
+ def __init__(self):
13
+ self.fields = [
14
+ "lngid",
15
+ "rawid",
16
+ "title",
17
+ "title_alternative",
18
+ "title_sub",
19
+ "title_edition",
20
+ "title_series",
21
+ "identifier_eisbn",
22
+ "identifier_pisbn",
23
+ "identifier_doi",
24
+ "creator",
25
+ "creator_en",
26
+ "creator_bio",
27
+ "creator_institution",
28
+ "publisher",
29
+ "date",
30
+ "description",
31
+ "description_en",
32
+ "description_unit",
33
+ "subject",
34
+ "subject_en",
35
+ "subject_clc",
36
+ "subject_esc",
37
+ "page",
38
+ "beginpage",
39
+ "endpage",
40
+ "pagecount",
41
+ "date_created",
42
+ "rawtype",
43
+ "folio_size",
44
+ "price",
45
+ "pub_place",
46
+ "language",
47
+ "country",
48
+ "type",
49
+ "provider",
50
+ "provider_url",
51
+ "provider_id",
52
+ "medium",
53
+ "batch",
54
+ "is_deprecated"
55
+ ]
56
+ self.zt_providermap = dict()
57
+ self.mysqlutils = MysqlUtiles(None, None, builder="MysqlBuilderForDicts", dicts={
58
+ "host": "192.168.31.24",
59
+ "user": "root",
60
+ "passwd": "vipdatacenter",
61
+ "db": "data_warehouse_sql",
62
+ "port": "3306",
63
+ "chartset": "utf8mb4",
64
+ })
65
+ self.get_zt_provider()
66
+
67
+ def get_zt_provider(self):
68
+ rows = self.mysqlutils.SelectFromDB(
69
+ "select sub_db_id,provider from a_transform_task where source_type = '1' and out_type = 'zt'")
70
+ for row in rows[1]:
71
+ self.zt_providermap[row[0]] = row[1]
72
+
73
+ def transform(self, titleMap):
74
+ transMap = dict()
75
+ sub_db_id = titleMap.get("sub_db_id", "")
76
+ transMap["lngid"] = titleMap.get("lngid", "")
77
+ transMap["rawid"] = titleMap.get("rawid", "")
78
+ transMap["title"] = titleMap.get("title", "")
79
+ transMap["title_alternative"] = titleMap.get("title_alt", "")
80
+ transMap["title_series"] = titleMap.get("title_series", "")
81
+ transMap["title_edition"] = titleMap.get("revision", "")
82
+ transMap["identifier_eisbn"] = titleMap.get("eisbn", "")
83
+ transMap["identifier_pisbn"] = titleMap.get("isbn", "")
84
+ transMap["identifier_doi"] = titleMap.get("doi", "")
85
+ transMap["creator"] = titleMap.get("author", "")
86
+ transMap["creator_en"] = titleMap.get("author_alt", "")
87
+ transMap["creator_institution"] = titleMap.get("organ", "")
88
+ transMap["creator_bio"] = titleMap.get("author_intro", "")
89
+ transMap["publisher"] = titleMap.get("publisher", "")
90
+ transMap["date"] = titleMap.get("pub_year", "")
91
+ transMap["description"] = titleMap.get("abstract", "")
92
+ transMap["description_en"] = titleMap.get("abstract_alt", "")
93
+ transMap["description_unit"] = titleMap.get("catalog", "")
94
+ transMap["subject"] = titleMap.get("keyword", "")
95
+ transMap["subject_en"] = titleMap.get("keyword_alt", "")
96
+ transMap["subject_clc"] = titleMap.get("clc_no", "")
97
+ transMap["subject_esc"] = titleMap.get("subject_edu", "")
98
+ transMap["page"] = titleMap.get("page_info", "")
99
+ transMap["beginpage"] = titleMap.get("begin_page", "")
100
+ transMap["endpage"] = titleMap.get("end_page", "")
101
+ transMap["pagecount"] = titleMap.get("page_cnt", "")
102
+ transMap["date_created"] = titleMap.get("pub_date", "")
103
+ transMap["rawtype"] = titleMap.get("raw_type", "")
104
+ transMap["folio_size"] = titleMap.get("book_size", "")
105
+ transMap["price"] = titleMap.get("price", "")
106
+ transMap["pub_place"] = titleMap.get("pub_place", "")
107
+ # transMap["provider"] = titleMap.get("zt_provider", "")
108
+ # if transMap["provider"] == "":
109
+ transMap["provider"] = self.zt_providermap[sub_db_id]
110
+ transMap["provider_url"] = transMap["provider"] + "@" + titleMap.get("provider_url")
111
+ transMap["provider_id"] = transMap["provider"] + "@" + titleMap.get("rawid")
112
+ transMap["batch"] = BaseTime().get_beijin_date_strins("%Y%m%d") + "00"
113
+ transMap["type"] = titleMap.get("source_type", "")
114
+ transMap["medium"] = "2"
115
+ transMap["country"] = titleMap.get("country", "")
116
+ transMap["language"] = titleMap.get("language", "")
117
+ transMap["is_deprecated"] = titleMap.get("is_deprecated", "")
118
+
119
+ for field in self.fields:
120
+ if field not in transMap.keys():
121
+ transMap[field] = ""
122
+ return transMap
123
+
124
+ if __name__ == '__main__':
125
+ t = TransformBookTitleToZt()
@@ -1,139 +1,139 @@
1
- # -*- coding:utf-8 -*-
2
- # @Time : 2021/12/15 14:09
3
- # @Author: suhong
4
- # @File : TransformConferenceTitleToZt.py
5
- # @Function :转换会议a层到智图代码
6
- from re_common.baselibrary.utils.basetime import BaseTime
7
-
8
- from re_common.facade.mysqlfacade import MysqlUtiles
9
-
10
-
11
- class TransformConferenceTitleToZt():
12
- def __init__(self):
13
- # 初始化fields
14
- self.fields = [
15
- "lngid",
16
- "rawid",
17
- "title",
18
- "title_alternative",
19
- "title_series",
20
- "title_edition",
21
- "applicant",
22
- "identifier_doi",
23
- "creator",
24
- "creator_en",
25
- "creator_institution",
26
- "creator_bio",
27
- "creator_drafting",
28
- "creator_release",
29
- "source",
30
- "source_en",
31
- "source_institution",
32
- "publisher",
33
- "date",
34
- "description",
35
- "description_en",
36
- "description_fund",
37
- "subject",
38
- "subject_en",
39
- "page",
40
- "beginpage",
41
- "endpage",
42
- "jumppage",
43
- "pagecount",
44
- "subject_clc",
45
- "subject_esc",
46
- "date_created",
47
- "pub_place",
48
- "language",
49
- "country",
50
- "type",
51
- "provider",
52
- "provider_url",
53
- "provider_id",
54
- "medium",
55
- "batch",
56
- "is_deprecated",
57
- "if_pdf_fulltext",
58
- "if_html_fulltext"
59
- ]
60
- self.zt_providermap = dict()
61
- self.mysqlutils = MysqlUtiles(None, None, builder="MysqlBuilderForDicts", dicts={
62
- "host": "192.168.31.24",
63
- "user": "root",
64
- "passwd": "vipdatacenter",
65
- "db": "data_warehouse_sql",
66
- "port": "3306",
67
- "chartset": "utf8mb4",
68
- })
69
- self.get_zt_provider()
70
-
71
- def get_zt_provider(self):
72
- rows = self.mysqlutils.SelectFromDB(
73
- "select sub_db_id,provider from a_transform_task where source_type = '6' and out_type = 'zt'")
74
- for row in rows[1]:
75
- self.zt_providermap[row[0]] = row[1]
76
-
77
- def transform(self, titleMap):
78
- transMap = dict()
79
- sub_db_id = titleMap.get("sub_db_id", "")
80
- transMap["lngid"] = titleMap.get("lngid", "")
81
- transMap["rawid"] = titleMap.get("rawid", "")
82
- transMap["title"] = titleMap.get("title", "")
83
- transMap["title_alternative"] = titleMap.get("title_alt", "")
84
- transMap["title_series"] = titleMap.get("meeting_record_name", "")
85
- transMap["title_edition"] = titleMap.get("revision", "")
86
- transMap["applicant"] = titleMap.get("applicant", "")
87
- transMap["identifier_doi"] = titleMap.get("doi", "")
88
- transMap["creator"] = titleMap.get("author", "")
89
- transMap["creator_en"] = titleMap.get("author_alt", "")
90
- transMap["creator_institution"] = titleMap.get("organ", "")
91
- transMap["creator_bio"] = titleMap.get("author_intro", "")
92
- transMap["creator_drafting"] = titleMap.get("society", "")
93
- transMap["creator_release"] = titleMap.get("host_organ", "")
94
- transMap["source"] = titleMap.get("meeting_name", "")
95
- transMap["source_en"] = titleMap.get("meeting_name_alt", "")
96
- transMap["source_institution"] = titleMap.get("meeting_place", "")
97
- transMap["publisher"] = titleMap.get("publisher", "")
98
- transMap["date"] = titleMap.get("pub_year", "")
99
- transMap["description"] = titleMap.get("abstract", "")
100
- transMap["description_en"] = titleMap.get("abstract_alt", "")
101
- transMap["description_fund"] = titleMap.get("fund", "")
102
- transMap["subject"] = titleMap.get("keyword", "")
103
- transMap["subject_en"] = titleMap.get("keyword_alt", "")
104
- transMap["page"] = titleMap.get("page_info", "")
105
- transMap["beginpage"] = titleMap.get("begin_page", "")
106
- transMap["endpage"] = titleMap.get("end_page", "")
107
- transMap["jumppage"] = titleMap.get("jump_page", "")
108
- transMap["pagecount"] = titleMap.get("page_cnt", "")
109
- transMap["subject_clc"] = titleMap.get("clc_no", "")
110
- transMap["subject_esc"] = titleMap.get("subject_edu", "")
111
- transMap["date_created"] = titleMap.get("pub_date", "")
112
- transMap["pub_place"] = titleMap.get("pub_place", "")
113
- # transMap["provider"] = titleMap.get("zt_provider", "")
114
- # if transMap["provider"] == "":
115
- transMap["provider"] = self.zt_providermap[sub_db_id]
116
- transMap["provider_url"] = transMap["provider"] + "@" + titleMap.get("provider_url")
117
- transMap["provider_id"] = transMap["provider"] + "@" + titleMap.get("rawid")
118
- transMap["batch"] = BaseTime().get_beijin_date_strins("%Y%m%d") + "00"
119
- transMap["type"] = titleMap.get("source_type", "")
120
- transMap["medium"] = "2"
121
- transMap["country"] = titleMap.get("country", "")
122
- transMap["language"] = titleMap.get("language", "")
123
- fulltext_type = titleMap.get("fulltext_type", "")
124
- transMap["if_pdf_fulltext"] = "0"
125
- transMap["if_html_fulltext"] = "0"
126
- if "pdf" in fulltext_type:
127
- transMap["if_pdf_fulltext"] = "1"
128
- if "html" in fulltext_type:
129
- transMap["if_html_fulltext"] = "1"
130
- transMap["is_deprecated"] = titleMap.get("is_deprecated", "")
131
-
132
- for field in self.fields:
133
- if field not in transMap.keys():
134
- transMap[field] = ""
135
- return transMap
136
-
137
-
138
- if __name__ == '__main__':
139
- t = TransformConferenceTitleToZt()
1
+ # -*- coding:utf-8 -*-
2
+ # @Time : 2021/12/15 14:09
3
+ # @Author: suhong
4
+ # @File : TransformConferenceTitleToZt.py
5
+ # @Function :转换会议a层到智图代码
6
+ from re_common.baselibrary.utils.basetime import BaseTime
7
+
8
+ from re_common.facade.mysqlfacade import MysqlUtiles
9
+
10
+
11
+ class TransformConferenceTitleToZt():
12
+ def __init__(self):
13
+ # 初始化fields
14
+ self.fields = [
15
+ "lngid",
16
+ "rawid",
17
+ "title",
18
+ "title_alternative",
19
+ "title_series",
20
+ "title_edition",
21
+ "applicant",
22
+ "identifier_doi",
23
+ "creator",
24
+ "creator_en",
25
+ "creator_institution",
26
+ "creator_bio",
27
+ "creator_drafting",
28
+ "creator_release",
29
+ "source",
30
+ "source_en",
31
+ "source_institution",
32
+ "publisher",
33
+ "date",
34
+ "description",
35
+ "description_en",
36
+ "description_fund",
37
+ "subject",
38
+ "subject_en",
39
+ "page",
40
+ "beginpage",
41
+ "endpage",
42
+ "jumppage",
43
+ "pagecount",
44
+ "subject_clc",
45
+ "subject_esc",
46
+ "date_created",
47
+ "pub_place",
48
+ "language",
49
+ "country",
50
+ "type",
51
+ "provider",
52
+ "provider_url",
53
+ "provider_id",
54
+ "medium",
55
+ "batch",
56
+ "is_deprecated",
57
+ "if_pdf_fulltext",
58
+ "if_html_fulltext"
59
+ ]
60
+ self.zt_providermap = dict()
61
+ self.mysqlutils = MysqlUtiles(None, None, builder="MysqlBuilderForDicts", dicts={
62
+ "host": "192.168.31.24",
63
+ "user": "root",
64
+ "passwd": "vipdatacenter",
65
+ "db": "data_warehouse_sql",
66
+ "port": "3306",
67
+ "chartset": "utf8mb4",
68
+ })
69
+ self.get_zt_provider()
70
+
71
+ def get_zt_provider(self):
72
+ rows = self.mysqlutils.SelectFromDB(
73
+ "select sub_db_id,provider from a_transform_task where source_type = '6' and out_type = 'zt'")
74
+ for row in rows[1]:
75
+ self.zt_providermap[row[0]] = row[1]
76
+
77
+ def transform(self, titleMap):
78
+ transMap = dict()
79
+ sub_db_id = titleMap.get("sub_db_id", "")
80
+ transMap["lngid"] = titleMap.get("lngid", "")
81
+ transMap["rawid"] = titleMap.get("rawid", "")
82
+ transMap["title"] = titleMap.get("title", "")
83
+ transMap["title_alternative"] = titleMap.get("title_alt", "")
84
+ transMap["title_series"] = titleMap.get("meeting_record_name", "")
85
+ transMap["title_edition"] = titleMap.get("revision", "")
86
+ transMap["applicant"] = titleMap.get("applicant", "")
87
+ transMap["identifier_doi"] = titleMap.get("doi", "")
88
+ transMap["creator"] = titleMap.get("author", "")
89
+ transMap["creator_en"] = titleMap.get("author_alt", "")
90
+ transMap["creator_institution"] = titleMap.get("organ", "")
91
+ transMap["creator_bio"] = titleMap.get("author_intro", "")
92
+ transMap["creator_drafting"] = titleMap.get("society", "")
93
+ transMap["creator_release"] = titleMap.get("host_organ", "")
94
+ transMap["source"] = titleMap.get("meeting_name", "")
95
+ transMap["source_en"] = titleMap.get("meeting_name_alt", "")
96
+ transMap["source_institution"] = titleMap.get("meeting_place", "")
97
+ transMap["publisher"] = titleMap.get("publisher", "")
98
+ transMap["date"] = titleMap.get("pub_year", "")
99
+ transMap["description"] = titleMap.get("abstract", "")
100
+ transMap["description_en"] = titleMap.get("abstract_alt", "")
101
+ transMap["description_fund"] = titleMap.get("fund", "")
102
+ transMap["subject"] = titleMap.get("keyword", "")
103
+ transMap["subject_en"] = titleMap.get("keyword_alt", "")
104
+ transMap["page"] = titleMap.get("page_info", "")
105
+ transMap["beginpage"] = titleMap.get("begin_page", "")
106
+ transMap["endpage"] = titleMap.get("end_page", "")
107
+ transMap["jumppage"] = titleMap.get("jump_page", "")
108
+ transMap["pagecount"] = titleMap.get("page_cnt", "")
109
+ transMap["subject_clc"] = titleMap.get("clc_no", "")
110
+ transMap["subject_esc"] = titleMap.get("subject_edu", "")
111
+ transMap["date_created"] = titleMap.get("pub_date", "")
112
+ transMap["pub_place"] = titleMap.get("pub_place", "")
113
+ # transMap["provider"] = titleMap.get("zt_provider", "")
114
+ # if transMap["provider"] == "":
115
+ transMap["provider"] = self.zt_providermap[sub_db_id]
116
+ transMap["provider_url"] = transMap["provider"] + "@" + titleMap.get("provider_url")
117
+ transMap["provider_id"] = transMap["provider"] + "@" + titleMap.get("rawid")
118
+ transMap["batch"] = BaseTime().get_beijin_date_strins("%Y%m%d") + "00"
119
+ transMap["type"] = titleMap.get("source_type", "")
120
+ transMap["medium"] = "2"
121
+ transMap["country"] = titleMap.get("country", "")
122
+ transMap["language"] = titleMap.get("language", "")
123
+ fulltext_type = titleMap.get("fulltext_type", "")
124
+ transMap["if_pdf_fulltext"] = "0"
125
+ transMap["if_html_fulltext"] = "0"
126
+ if "pdf" in fulltext_type:
127
+ transMap["if_pdf_fulltext"] = "1"
128
+ if "html" in fulltext_type:
129
+ transMap["if_html_fulltext"] = "1"
130
+ transMap["is_deprecated"] = titleMap.get("is_deprecated", "")
131
+
132
+ for field in self.fields:
133
+ if field not in transMap.keys():
134
+ transMap[field] = ""
135
+ return transMap
136
+
137
+
138
+ if __name__ == '__main__':
139
+ t = TransformConferenceTitleToZt()