预处理规则.json 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322
  1. {
  2. "_id" : ObjectId("5b7e54a3b0084b39647cd7e6"),
  3. "pattern" : "(?<replace>单\\s?位)",
  4. "replacement" : "单位",
  5. "priority" : NumberInt(3),
  6. "is_active" : true
  7. }
  8. {
  9. "_id" : ObjectId("5b7e54a3b0084b39647cd7ea"),
  10. "pattern" : "(?<replace>[【】])",
  11. "replacement" : "#space#",
  12. "priority" : NumberInt(7),
  13. "is_active" : true
  14. }
  15. {
  16. "_id" : ObjectId("5b7e54a4b0084b39647cd7f5"),
  17. "pattern" : "(?<replace>第(&[a-z]{1,6};|[<][^>]*[>]|\\s){0,10}2(?=(&[a-z]{1,6};|[<][^>]*[>]|\\s){0,10}[\\u4E00-\\u9FA5]))",
  18. "replacement" : "第二",
  19. "priority" : NumberInt(18),
  20. "is_active" : true
  21. }
  22. {
  23. "_id" : ObjectId("5b7e54a4b0084b39647cd7f6"),
  24. "pattern" : "(?<replace>第(三|3|叁)标段)",
  25. "replacement" : "第三标段",
  26. "priority" : NumberInt(19),
  27. "is_active" : true
  28. }
  29. {
  30. "_id" : ObjectId("5b7e54a4b0084b39647cd7f7"),
  31. "pattern" : "(?<replace>第(&[a-z]{1,6};|[<][^>]*[>]|\\s){0,10}3(?=(&[a-z]{1,6};|[<][^>]*[>]|\\s){0,10}[\\u4E00-\\u9FA5]))",
  32. "replacement" : "第三",
  33. "priority" : NumberInt(20),
  34. "is_active" : true
  35. }
  36. {
  37. "_id" : ObjectId("5b7e54a4b0084b39647cd7f8"),
  38. "pattern" : "(?<replace>第(&[a-z]{1,6};|[<][^>]*[>]|\\s){0,10}1(?=(&[a-z]{1,6};|[<][^>]*[>]|\\s){0,10}[\\u4E00-\\u9FA5]))",
  39. "replacement" : "第一",
  40. "priority" : NumberInt(21),
  41. "is_active" : true
  42. }
  43. {
  44. "_id" : ObjectId("5b7e54a4b0084b39647cd7f9"),
  45. "pattern" : "(?<replace>第(一|1|壹)标段)",
  46. "replacement" : "第一标段",
  47. "priority" : NumberInt(22),
  48. "is_active" : true
  49. }
  50. {
  51. "_id" : ObjectId("5b7e54a4b0084b39647cd7fa"),
  52. "pattern" : "(?<replace>(商品|物[品资料]|产品|采购品|设备|器材|家具|标的物?)(名称|内容|明细|类别|描述))",
  53. "replacement" : "货物名称",
  54. "priority" : NumberInt(23),
  55. "is_active" : true
  56. }
  57. {
  58. "_id" : ObjectId("5b7e54a4b0084b39647cd7fb"),
  59. "pattern" : "(?<replace>截止时间)",
  60. "replacement" : "结束时间",
  61. "priority" : NumberInt(24),
  62. "is_active" : true
  63. }
  64. {
  65. "_id" : ObjectId("5b7e54a4b0084b39647cd7fc"),
  66. "pattern" : "(?<replace>开标(时间|日期))",
  67. "replacement" : "开标日期",
  68. "priority" : NumberInt(25),
  69. "is_active" : true
  70. }
  71. {
  72. "_id" : ObjectId("5b7e54a4b0084b39647cd7fd"),
  73. "pattern" : "(?<replace>起始时间)",
  74. "replacement" : "开始时间",
  75. "priority" : NumberInt(26),
  76. "is_active" : true
  77. }
  78. {
  79. "_id" : ObjectId("5b7e54a4b0084b39647cd7fe"),
  80. "pattern" : "(?<replace>批复文号)",
  81. "replacement" : "批准文件",
  82. "priority" : NumberInt(27),
  83. "is_active" : true
  84. }
  85. {
  86. "_id" : ObjectId("5b7e54a4b0084b39647cd7ff"),
  87. "pattern" : "(?<replace>(评标|评审|定标)(地点|地址))",
  88. "replacement" : "评标地点",
  89. "priority" : NumberInt(28),
  90. "is_active" : true
  91. }
  92. {
  93. "_id" : ObjectId("5b7e54a4b0084b39647cd800"),
  94. "pattern" : "(?<replace>评审时间|评审日期|定标日期|定标时间)",
  95. "replacement" : "评标日期",
  96. "priority" : NumberInt(29),
  97. "is_active" : true
  98. }
  99. {
  100. "_id" : ObjectId("5b7e54a5b0084b39647cd802"),
  101. "pattern" : "(?<replace>((入场|进场)登记号)|((谈判|采购|招标|磋商|竞争性谈判)?文件编号))",
  102. "replacement" : "项目编号",
  103. "priority" : NumberInt(31),
  104. "is_active" : true
  105. }
  106. {
  107. "_id" : ObjectId("5b7e54a5b0084b39647cd803"),
  108. "pattern" : "(?<replace>招标项目编号|合同编号|项目(标段)编号|招标代码|中标公示备案编号|标书编号|招标公告代码|标段(包)编号|公告编号)",
  109. "replacement" : "项目编号",
  110. "priority" : NumberInt(32),
  111. "is_active" : true
  112. }
  113. {
  114. "_id" : ObjectId("5b7e54a5b0084b39647cd804"),
  115. "pattern" : "(?<replace>(招标|项目|工程|采购|交易|中标|询价(单)?|竞[标价]|标段(包)?)(项目|公示|公告)?(单|编|标)(号|码))",
  116. "replacement" : "项目编号",
  117. "priority" : NumberInt(33),
  118. "is_active" : true
  119. }
  120. {
  121. "_id" : ObjectId("5b7e54a5b0084b39647cd805"),
  122. "pattern" : "(?<replace>工程地点|工程地址|项目地点|项目地址|项目所在地区|项目实施地点|项目坐落地址|实施地点|建设地点|建设地址)",
  123. "replacement" : "项目地点",
  124. "priority" : NumberInt(34),
  125. "is_active" : true
  126. }
  127. {
  128. "_id" : ObjectId("5b7e54a5b0084b39647cd806"),
  129. "pattern" : "(?<replace>((工程名称)|(招标|中标)名称(?=[^(,;。\\s)]{0,30}(项目|工程))))",
  130. "replacement" : "项目名称",
  131. "priority" : NumberInt(35),
  132. "is_active" : true
  133. }
  134. {
  135. "_id" : ObjectId("5b7e54a5b0084b39647cd807"),
  136. "pattern" : "(?<replace>(?<!代理|联系)(招标|采购)\\s?(商|方|单位|机构|部门|公司))",
  137. "replacement" : "招标单位",
  138. "priority" : NumberInt(36),
  139. "is_active" : true
  140. }
  141. {
  142. "_id" : ObjectId("5b7e54a5b0084b39647cd808"),
  143. "pattern" : "(?<replace>(项目|工程)?(建设|实施|发布)\\s?(单位|机构|方))",
  144. "replacement" : "招标单位",
  145. "priority" : NumberInt(37),
  146. "is_active" : true
  147. }
  148. {
  149. "_id" : ObjectId("5b7e54a5b0084b39647cd809"),
  150. "pattern" : "(?<replace>(项目业主|业主|需方|比选人|询价人|采购组织|拍卖企业|甲方|需求公司))",
  151. "replacement" : "招标单位",
  152. "priority" : NumberInt(38),
  153. "is_active" : true
  154. }
  155. {
  156. "_id" : ObjectId("5b7e54a5b0084b39647cd80a"),
  157. "pattern" : "(?<replace>((?<!代理|联系))(采购单位|项目业主|招标单位|建设单位|采购单位|竞价单位|委托单位))",
  158. "replacement" : "招标单位",
  159. "priority" : NumberInt(39),
  160. "is_active" : true
  161. }
  162. {
  163. "_id" : ObjectId("5b7e54a5b0084b39647cd80b"),
  164. "pattern" : "(?<replace>招标机构[((]招标机构[))]\\s?)",
  165. "replacement" : "招标单位",
  166. "priority" : NumberInt(40),
  167. "is_active" : true
  168. }
  169. {
  170. "_id" : ObjectId("5b7e54a5b0084b39647cd80c"),
  171. "pattern" : "(?<replace>采购联系人)",
  172. "replacement" : "招标联系人",
  173. "priority" : NumberInt(41),
  174. "is_active" : true
  175. }
  176. {
  177. "_id" : ObjectId("5b7e54a5b0084b39647cd80d"),
  178. "pattern" : "(?<replace>(采购|工程)项目)",
  179. "replacement" : "招标项目",
  180. "priority" : NumberInt(42),
  181. "is_active" : true
  182. }
  183. {
  184. "_id" : ObjectId("5b7e54a5b0084b39647cd80e"),
  185. "pattern" : "(?<replace>(?<=(\\d日|\\d\\d:\\d\\d))-(?=20\\d{2}))",
  186. "replacement" : "至",
  187. "priority" : NumberInt(43),
  188. "is_active" : true
  189. }
  190. {
  191. "_id" : ObjectId("5b7e54a5b0084b39647cd80f"),
  192. "pattern" : "(?<replace>中标名称(?=[^(,;。\\s)]{0,30}公司))",
  193. "replacement" : "中标人",
  194. "priority" : NumberInt(44),
  195. "is_active" : true
  196. }
  197. {
  198. "_id" : ObjectId("5b7e54a5b0084b39647cd810"),
  199. "pattern" : "(?<replace>(供应|承接|(成交|中(签|标|选)(候选)?))\\s?(人|单位|(厂)?商|使用权[人]?))",
  200. "replacement" : "中标人",
  201. "priority" : NumberInt(45),
  202. "is_active" : true
  203. }
  204. {
  205. "_id" : ObjectId("5b7e54a3b0084b39647cd7e4"),
  206. "pattern" : "(?<replace>全(\\s)?称)",
  207. "replacement" : "名称",
  208. "priority" : NumberInt(1),
  209. "is_active" : true
  210. }
  211. {
  212. "_id" : ObjectId("5b7e54a3b0084b39647cd7e5"),
  213. "pattern" : "(?<replace>联系人[^<\\u4E00-\\u9FA5]{0,5}?[::])",
  214. "replacement" : "联系人:",
  215. "priority" : NumberInt(2),
  216. "is_active" : true
  217. }
  218. {
  219. "_id" : ObjectId("5b7e54a3b0084b39647cd7e7"),
  220. "pattern" : "(?<replace>联系(方式|电话|号码))",
  221. "replacement" : "电话",
  222. "priority" : NumberInt(4),
  223. "is_active" : true
  224. }
  225. {
  226. "_id" : ObjectId("5b7e54a3b0084b39647cd7e8"),
  227. "pattern" : "(?<replace>人民币|[\\u00a5\\uffe5])",
  228. "replacement" : "¥",
  229. "priority" : NumberInt(5),
  230. "is_active" : true
  231. }
  232. {
  233. "_id" : ObjectId("5b7e54a3b0084b39647cd7e9"),
  234. "pattern" : "(?<replace>手机|(?<!\\w)tel(?!\\w))",
  235. "replacement" : "#space#电话",
  236. "priority" : NumberInt(6),
  237. "is_active" : true
  238. }
  239. {
  240. "_id" : ObjectId("5b7e54a4b0084b39647cd7eb"),
  241. "pattern" : "\\d{4}[\\x00-\\xff]*年[\\x00-\\xff]*\\d{1,2}[\\x00-\\xff]*月[\\x00-\\xff]*\\d{1,2}[\\x00-\\xff]*日[\\x00-\\xff]*?(?<replace>招标附件[\\s\\S]+|附件[\\s\\S]+)",
  242. "replacement" : "#space#",
  243. "priority" : NumberInt(8),
  244. "is_active" : true
  245. }
  246. {
  247. "_id" : ObjectId("5b7e54a4b0084b39647cd7ec"),
  248. "pattern" : "(?<replace>[((](盖|公)章[))]|[((]签字[))])",
  249. "replacement" : "#space#",
  250. "priority" : NumberInt(9),
  251. "is_active" : true
  252. }
  253. {
  254. "_id" : ObjectId("5b7e54a4b0084b39647cd7ed"),
  255. "pattern" : "(?<replace>(?<=[^\\x00-\\xff ][a-zA-Z0-9,.()])[:])",
  256. "replacement" : ":",
  257. "priority" : NumberInt(10),
  258. "is_active" : true
  259. }
  260. {
  261. "_id" : ObjectId("5b7e54a4b0084b39647cd7ee"),
  262. "pattern" : "(?<replace>(标包|子包|包组))",
  263. "replacement" : "标段",
  264. "priority" : NumberInt(11),
  265. "is_active" : true
  266. }
  267. {
  268. "_id" : ObjectId("5b7e54a4b0084b39647cd7ef"),
  269. "pattern" : "(?<replace>(招标|采购)?\\s?代理(招标|采购)?\\s?(商|单位|机构))",
  270. "replacement" : "代理机构",
  271. "priority" : NumberInt(12),
  272. "is_active" : true
  273. }
  274. {
  275. "_id" : ObjectId("5b7e54a4b0084b39647cd7f0"),
  276. "pattern" : "(?<replace>(招标|采购)\\s?服务\\s?(单位|机构|商))",
  277. "replacement" : "代理机构",
  278. "priority" : NumberInt(13),
  279. "is_active" : true
  280. }
  281. {
  282. "_id" : ObjectId("5b7e54a4b0084b39647cd7f1"),
  283. "pattern" : "(?<replace>(招标|采购)\\s?代理\\s?(?!公司|有限公司|中心|处))",
  284. "replacement" : "代理机构",
  285. "priority" : NumberInt(14),
  286. "is_active" : true
  287. }
  288. {
  289. "_id" : ObjectId("5b7e54a4b0084b39647cd7f2"),
  290. "pattern" : "(?<replace>招标代理人|采购代理机构|(?<![\\u4E00-\\u9FA5])招标代理公司|招标代理机构|招标代理单位|(?<=[^\\u4E00-\\u9FA5])招标代理(?!有限公司)|集中采购机构|招标服务单位|采购机构|招标组织单位|招标机构|招标[((]?采购[))]?机构|(?<![\\u4E00-\\u9FA5])代理公司|招标组织机构|采购代理)",
  291. "replacement" : "代理机构",
  292. "priority" : NumberInt(15),
  293. "is_active" : true
  294. }
  295. {
  296. "_id" : ObjectId("5b7e54a4b0084b39647cd7f3"),
  297. "pattern" : "(?<replace>代理机构[((]代理机构[))]\\s?)",
  298. "replacement" : "代理机构",
  299. "priority" : NumberInt(16),
  300. "is_active" : true
  301. }
  302. {
  303. "_id" : ObjectId("5b7e54a4b0084b39647cd7f4"),
  304. "pattern" : "(?<replace>第(二|2|贰)标段)",
  305. "replacement" : "第二标段",
  306. "priority" : NumberInt(17),
  307. "is_active" : true
  308. }
  309. {
  310. "_id" : ObjectId("5b7e54a4b0084b39647cd801"),
  311. "pattern" : "(?<replace>评标委员会|评标委员会成员|评审委员会成员|评委会成员|评审专家|项目评审小组|审查委员会|评审小组成员)",
  312. "replacement" : "评标委员会成员",
  313. "priority" : NumberInt(30),
  314. "is_active" : true
  315. }
  316. {
  317. "_id" : ObjectId("5b7e54a5b0084b39647cd811"),
  318. "pattern" : "(?<replace>(中标供应商|乙方))",
  319. "replacement" : "中标人",
  320. "priority" : NumberInt(46),
  321. "is_active" : true
  322. }