Skip to content

周伟奇 / bmw-ocr

Go to a project
Toggle navigation
Toggle navigation pinning
c8a54f58 by 周伟奇
Options

Merge branch 'feature/license' into feature/mssql

2 parents 540f8777 56281e38
Inline Side-by-side
Showing 7 changed files with 1641 additions and 233 deletions
.gitignore
...@@ -35,3 +35,4 @@ src/*.sh ...@@ -35,3 +35,4 @@ src/*.sh
35 35
36 test* 36 test*
37 ocr_test.py 37 ocr_test.py
38 ocr_process.py
...\ No newline at end of file ...\ No newline at end of file
......
src/apps/doc/consts.py
1 import copy
2
1 PAGE_DEFAULT = 1 3 PAGE_DEFAULT = 1
2 PAGE_SIZE_DEFAULT = 10 4 PAGE_SIZE_DEFAULT = 10
3 5
...@@ -53,15 +55,19 @@ TRANS_MAP = { ...@@ -53,15 +55,19 @@ TRANS_MAP = {
53 'L': "1", 55 'L': "1",
54 56
55 'A': "4", 57 'A': "4",
58
56 's': "5", 59 's': "5",
57 'S': "5", 60 'S': "5",
61
58 'b': "6", 62 'b': "6",
63
59 'g': "9", 64 'g': "9",
60 'E': "9", 65 'E': "9",
66
61 'B': "13", 67 'B': "13",
62 } 68 }
63 TRANS = str.maketrans(TRANS_MAP) 69 TRANS = str.maketrans(TRANS_MAP)
64 ERROR_CHARS = {'.', ':', '：', '•', '·'} 70 ERROR_CHARS = {'.', '。', ':', '：', '•', '·', ',', '，'}
65 SKIP_IMG_SHEET_NAME = '未处理图片' 71 SKIP_IMG_SHEET_NAME = '未处理图片'
66 SKIP_IMG_SHEET_HEADER = ('页码', '序号') 72 SKIP_IMG_SHEET_HEADER = ('页码', '序号')
67 73
...@@ -70,12 +76,34 @@ UNKNOWN_CARD = '未知卡号' ...@@ -70,12 +76,34 @@ UNKNOWN_CARD = '未知卡号'
70 UNKNOWN_ROLE = '未知户名' 76 UNKNOWN_ROLE = '未知户名'
71 DATE_FORMAT = ['%Y年%m月%d日', '%Y/%m/%d', '%Y-%m-%d', '%Y%m%d'] 77 DATE_FORMAT = ['%Y年%m月%d日', '%Y/%m/%d', '%Y-%m-%d', '%Y%m%d']
72 78
73 AMOUNT_COL_TITLE_SET = {"交易金额", "金额", "收入/支出金额", "发生额"}
74 OVERAGE_COL_TITLE_SET = {"账户余额", "余额"}
75 PROOF_COL_TITLE = '核对结果' 79 PROOF_COL_TITLE = '核对结果'
76 PROOF_RES = ('对', '错') 80 PROOF_RES = ('对', '错')
77 META_SHEET_TITLE = '关键信息提取和展示' 81 META_SHEET_TITLE = '关键信息提取和展示'
78 82
83 SUMMARY_KEY = 'summary_col'
84 DATE_KEY = 'date_col'
85 AMOUNT_KEY = 'amount_col'
86 OVER_KEY = 'over_col'
87 IMCOME_KEY = 'income_col'
88 OUTLAY_KEY = 'outlay_col'
89 BORROW_KEY = 'borrow_col'
90 MIN_ROW_KEY = 'min_row'
91 FIND_COUNT_KEY = 'find_count'
92 FIND_COL_KEY = 'find_col'
93 HEADER_KEY = 'header'
94
95 KEY_LIST = [SUMMARY_KEY, DATE_KEY, OVER_KEY, BORROW_KEY, AMOUNT_KEY, IMCOME_KEY, OUTLAY_KEY]
96
97 CLASSIFY_MAP = {
98 SUMMARY_KEY: 5,
99 DATE_KEY: 0,
100 AMOUNT_KEY: 2,
101 OVER_KEY: 3,
102 IMCOME_KEY: 11,
103 OUTLAY_KEY: 12,
104 BORROW_KEY: 10,
105 }
106
79 FIXED_HEADERS = ('记账日期', '记账时间', '金额', '余额', '交易名称', '附言', '对方账户名', '对方卡号/账号', 107 FIXED_HEADERS = ('记账日期', '记账时间', '金额', '余额', '交易名称', '附言', '对方账户名', '对方卡号/账号',
80 '对方开户行', '核对结果', '借贷', '收入', '支出') 108 '对方开户行', '核对结果', '借贷', '收入', '支出')
81 FIXED_COL_AMOUNT = len(FIXED_HEADERS) 109 FIXED_COL_AMOUNT = len(FIXED_HEADERS)
...@@ -103,36 +131,60 @@ OUTLAY_HEADERS_SET = {'支出金额', '支出', '支取金额（借）', '支取 ...@@ -103,36 +131,60 @@ OUTLAY_HEADERS_SET = {'支出金额', '支出', '支取金额（借）', '支取
103 131
104 # ------------------普通打印-全格线-------------------------------------------------------------------------------------- 132 # ------------------普通打印-全格线--------------------------------------------------------------------------------------
105 HEADERS_MAPPING = {} 133 HEADERS_MAPPING = {}
134
135 # 借贷
136 HEADERS_MAPPING.update(
137 {
138 '借贷': BORROW_KEY,
139 '借贷状态': BORROW_KEY,
140 '收/支': BORROW_KEY,
141 }
142 )
143
144 # 收入
145 HEADERS_MAPPING.update(
146 {
147 '收入金额': IMCOME_KEY,
148 '收入': IMCOME_KEY,
149 '存入': IMCOME_KEY,
150 '存入金额（贷）': IMCOME_KEY,
151 '存入金额(贷)': IMCOME_KEY,
152 }
153 )
154
155 # 支出
156 HEADERS_MAPPING.update(
157 {
158 '支出金额': OUTLAY_KEY,
159 '支出': OUTLAY_KEY,
160 '支取金额（借）': OUTLAY_KEY,
161 '支取金额(借)': OUTLAY_KEY,
162 }
163 )
164
165
106 # 横版-表格-中国银行（不规则） 166 # 横版-表格-中国银行（不规则）
107 HEADERS_MAPPING.update( 167 HEADERS_MAPPING.update(
108 { 168 {
109 '记账日期': BASE_HEADERS_MAPPING['记账日期'], 169 '记账日期': DATE_KEY,
110 '记账时间': BASE_HEADERS_MAPPING['记账时间'], 170 '金额': AMOUNT_KEY,
111 '金额': BASE_HEADERS_MAPPING['金额'], 171 '余额': OVER_KEY,
112 '余额': BASE_HEADERS_MAPPING['余额'], 172 '附言': SUMMARY_KEY,
113 '交易名称': BASE_HEADERS_MAPPING['交易名称'],
114 '附言': BASE_HEADERS_MAPPING['附言'],
115 '对方账户名': BASE_HEADERS_MAPPING['对方账户名'],
116 '对方卡号/账号': BASE_HEADERS_MAPPING['对方卡号/账号'],
117 '对方开户行': BASE_HEADERS_MAPPING['对方开户行'],
118 } 173 }
119 ) 174 )
120 # 横版-表格-农业银行-中国农业银行个人账户明细 175 # 横版-表格-农业银行-中国农业银行个人账户明细
121 HEADERS_MAPPING.update( 176 HEADERS_MAPPING.update(
122 { 177 {
123 '交易日期': BASE_HEADERS_MAPPING['记账日期'], 178 '交易日期': DATE_KEY,
124 # '存入': BASE_HEADERS_MAPPING['金额'], 179 # '存入': AMOUNT_KEY,
125 '对方账号': BASE_HEADERS_MAPPING['对方卡号/账号'], 180 '摘要': SUMMARY_KEY,
126 '对方名称': BASE_HEADERS_MAPPING['对方账户名'],
127 '摘要': BASE_HEADERS_MAPPING['附言'],
128 } 181 }
129 ) 182 )
130 # 横版-表格-北京银行 183 # 横版-表格-北京银行
131 HEADERS_MAPPING.update( 184 HEADERS_MAPPING.update(
132 { 185 {
133 '业务摘要': BASE_HEADERS_MAPPING['附言'], 186 '业务摘要': SUMMARY_KEY,
134 '发生额': BASE_HEADERS_MAPPING['金额'], 187 '发生额': AMOUNT_KEY,
135 '对方户名': BASE_HEADERS_MAPPING['对方账户名'],
136 } 188 }
137 ) 189 )
138 # 横版-表格-工商银行 借记卡账户历史明细清单 190 # 横版-表格-工商银行 借记卡账户历史明细清单
...@@ -142,8 +194,8 @@ HEADERS_MAPPING.update( ...@@ -142,8 +194,8 @@ HEADERS_MAPPING.update(
142 # 工商银行历史明细（申请单号：20042501303039397888） 194 # 工商银行历史明细（申请单号：20042501303039397888）
143 HEADERS_MAPPING.update( 195 HEADERS_MAPPING.update(
144 { 196 {
145 '收入/支出金额': BASE_HEADERS_MAPPING['金额'], 197 '收入/支出金额': AMOUNT_KEY,
146 '工作日期': BASE_HEADERS_MAPPING['记账日期'], 198 '工作日期': DATE_KEY,
147 } 199 }
148 ) 200 )
149 201
...@@ -153,26 +205,23 @@ HEADERS_MAPPING.update( ...@@ -153,26 +205,23 @@ HEADERS_MAPPING.update(
153 # 竖版-表格-建设银行-个人活期账户交易明细 CH-B005832604 (2) 205 # 竖版-表格-建设银行-个人活期账户交易明细 CH-B005832604 (2)
154 HEADERS_MAPPING.update( 206 HEADERS_MAPPING.update(
155 { 207 {
156 '交易金额': BASE_HEADERS_MAPPING['金额'], 208 '交易金额': AMOUNT_KEY,
157 '账户余额': BASE_HEADERS_MAPPING['余额'], 209 '账户余额': OVER_KEY,
158 '对方账号与户名': BASE_HEADERS_MAPPING['对方卡号/账号'],
159 } 210 }
160 ) 211 )
161 # 微信 212 # 微信
162 HEADERS_MAPPING.update( 213 HEADERS_MAPPING.update(
163 { 214 {
164 '交易时间': BASE_HEADERS_MAPPING['记账时间'], 215 '交易类型': SUMMARY_KEY,
165 '交易类型': BASE_HEADERS_MAPPING['附言'], 216 '金额(元)': AMOUNT_KEY,
166 '金额(元)': BASE_HEADERS_MAPPING['金额'], 217 '金额（元）': AMOUNT_KEY,
167 '金额（元）': BASE_HEADERS_MAPPING['金额'],
168 '交易对方': BASE_HEADERS_MAPPING['对方账户名'],
169 } 218 }
170 ) 219 )
171 # 支付宝 220 # 支付宝
172 HEADERS_MAPPING.update( 221 HEADERS_MAPPING.update(
173 { 222 {
174 '时间': BASE_HEADERS_MAPPING['记账日期'], 223 '时间': DATE_KEY,
175 '名称/备注': BASE_HEADERS_MAPPING['附言'], 224 '名称/备注': SUMMARY_KEY,
176 } 225 }
177 ) 226 )
178 227
...@@ -182,33 +231,28 @@ HEADERS_MAPPING.update( ...@@ -182,33 +231,28 @@ HEADERS_MAPPING.update(
182 # 竖版-无表格-农业银行CH-B008805428 231 # 竖版-无表格-农业银行CH-B008805428
183 HEADERS_MAPPING.update( 232 HEADERS_MAPPING.update(
184 { 233 {
185 '摘要/附言': BASE_HEADERS_MAPPING['附言'], 234 '摘要/附言': SUMMARY_KEY,
186 '交易地点/对方账号和户名': BASE_HEADERS_MAPPING['对方卡号/账号'],
187 } 235 }
188 ) 236 )
189 # 农业银行-窄页 237 # 农业银行-窄页
190 HEADERS_MAPPING.update( 238
191 {
192 '交易对手账号': BASE_HEADERS_MAPPING['对方卡号/账号'],
193 }
194 )
195 # 竖版-特殊-农商行 239 # 竖版-特殊-农商行
196 HEADERS_MAPPING.update( 240 HEADERS_MAPPING.update(
197 { 241 {
198 '交易发生额': BASE_HEADERS_MAPPING['金额'], 242 '交易发生额': AMOUNT_KEY,
199 } 243 }
200 ) 244 )
201 # 横版-特殊-中信银行-账户交易明细 245 # 横版-特殊-中信银行-账户交易明细
202 HEADERS_MAPPING.update( 246 HEADERS_MAPPING.update(
203 { 247 {
204 '对方银行': BASE_HEADERS_MAPPING['对方开户行'], 248 '交易摘要': SUMMARY_KEY,
205 '交易摘要': BASE_HEADERS_MAPPING['附言'],
206 } 249 }
207 ) 250 )
208 # 平安电子账单 251 # 平安电子账单
209 HEADERS_MAPPING.update( 252 HEADERS_MAPPING.update(
210 { 253 {
211 '借贷发生额(借:-贷:+)': BASE_HEADERS_MAPPING['金额'], 254 '借贷发生额(借:-贷:+)': AMOUNT_KEY,
255 '借贷发生额(借:-贷:+）': AMOUNT_KEY,
212 } 256 }
213 ) 257 )
214 258
...@@ -218,7 +262,7 @@ HEADERS_MAPPING.update( ...@@ -218,7 +262,7 @@ HEADERS_MAPPING.update(
218 # 竖版-无表格-招商银行账户历史交易明细表 262 # 竖版-无表格-招商银行账户历史交易明细表
219 HEADERS_MAPPING.update( 263 HEADERS_MAPPING.update(
220 { 264 {
221 '联机余额': BASE_HEADERS_MAPPING['余额'], 265 '联机余额': OVER_KEY,
222 } 266 }
223 ) 267 )
224 # 竖版-无表格-邮储银行-账户对账单 含有对手方户名 对手方账户 268 # 竖版-无表格-邮储银行-账户对账单 含有对手方户名 对手方账户
...@@ -226,28 +270,23 @@ HEADERS_MAPPING.update( ...@@ -226,28 +270,23 @@ HEADERS_MAPPING.update(
226 # 竖版-无表格-邮储银行-电子章 邮储银行 账户对账单 270 # 竖版-无表格-邮储银行-电子章 邮储银行 账户对账单
227 HEADERS_MAPPING.update( 271 HEADERS_MAPPING.update(
228 { 272 {
229 '交易金额(元)': BASE_HEADERS_MAPPING['金额'], 273 '交易金额(元)': AMOUNT_KEY,
230 '交易金额（元）': BASE_HEADERS_MAPPING['金额'], 274 '交易金额（元）': AMOUNT_KEY,
231 '账户余额（元）': BASE_HEADERS_MAPPING['余额'], 275 '账户余额（元）': OVER_KEY,
232 '账户余额(元)': BASE_HEADERS_MAPPING['余额'], 276 '账户余额(元)': OVER_KEY,
233 '对手方户名': BASE_HEADERS_MAPPING['对方账户名'],
234 '对手方账户': BASE_HEADERS_MAPPING['对方卡号/账号'],
235 } 277 }
236 ) 278 )
237 # 横版-无表格-广发银行-账户交易历史 --> 已废弃 279 # 横版-无表格-广发银行-账户交易历史 --> 已废弃
238 # 竖版-无表格-广发银行-账户交易历史 --> 已废弃 280 # 竖版-无表格-广发银行-账户交易历史 --> 已废弃
239 HEADERS_MAPPING.update( 281 HEADERS_MAPPING.update(
240 { 282 {
241 '会计日期': BASE_HEADERS_MAPPING['记账日期'], 283 '会计日期': DATE_KEY,
242 '对手户名': BASE_HEADERS_MAPPING['对方账户名'],
243 '对手账号': BASE_HEADERS_MAPPING['对方卡号/账号'],
244 } 284 }
245 ) 285 )
246 # 招行电子账单 TODO 有英文，需测试 286 # 招行电子账单 TODO 有英文，需测试
247 HEADERS_MAPPING.update( 287 HEADERS_MAPPING.update(
248 { 288 {
249 '对手信息': BASE_HEADERS_MAPPING['对方账户名'], 289 '摘要代码': SUMMARY_KEY,
250 '摘要代码': BASE_HEADERS_MAPPING['附言'],
251 } 290 }
252 ) 291 )
253 # 横版-无表格-民生银行-中国民生银行个人账户对账单（客户卡号） 292 # 横版-无表格-民生银行-中国民生银行个人账户对账单（客户卡号）
...@@ -255,46 +294,36 @@ HEADERS_MAPPING.update( ...@@ -255,46 +294,36 @@ HEADERS_MAPPING.update(
255 # 横版-无表格-民生银行 294 # 横版-无表格-民生银行
256 HEADERS_MAPPING.update( 295 HEADERS_MAPPING.update(
257 { 296 {
258 '摘要信息': BASE_HEADERS_MAPPING['附言'], 297 '摘要信息': SUMMARY_KEY,
259 '对方行名': BASE_HEADERS_MAPPING['对方开户行'],
260 } 298 }
261 ) 299 )
262 # 竖版-无表格-农业银行整数 300 # 竖版-无表格-农业银行整数
263 # 竖版-无表格-农业银行-中国农业银行银行卡交易明细清单 301 # 竖版-无表格-农业银行-中国农业银行银行卡交易明细清单
264 HEADERS_MAPPING.update( 302
265 {
266 '对方账号和户名': BASE_HEADERS_MAPPING['对方卡号/账号'],
267 }
268 )
269 # 竖版-无表格-农业银行-中国农业银行银行卡活期存折交易明细清单.pdf 303 # 竖版-无表格-农业银行-中国农业银行银行卡活期存折交易明细清单.pdf
270 # 竖版-无表格-农业银行-扩张.pdf 304 # 竖版-无表格-农业银行-扩张.pdf
271 # 竖版-无表格-农业银行-缩进.pdf 305 # 竖版-无表格-农业银行-缩进.pdf
272 HEADERS_MAPPING.update( 306 HEADERS_MAPPING.update(
273 { 307 {
274 '日期': BASE_HEADERS_MAPPING['记账日期'], 308 '日期': DATE_KEY,
275 '短摘要': BASE_HEADERS_MAPPING['附言'], 309 '短摘要': SUMMARY_KEY,
276 '本次余额': BASE_HEADERS_MAPPING['余额'], 310 '本次余额': OVER_KEY,
277 } 311 }
278 ) 312 )
279 # 竖版-无表格-农业银行-无标题(对手帐号) 313 # 竖版-无表格-农业银行-无标题(对手帐号)
280 HEADERS_MAPPING.update( 314 HEADERS_MAPPING.update(
281 { 315 {
282 '交易后余额': BASE_HEADERS_MAPPING['余额'], 316 '交易后余额': OVER_KEY,
283 '对手帐号': BASE_HEADERS_MAPPING['对方卡号/账号'],
284 } 317 }
285 ) 318 )
286 # 竖版-无表格-农商行（非常规） 319 # 竖版-无表格-农商行（非常规）
287 HEADERS_MAPPING.update( 320 HEADERS_MAPPING.update(
288 { 321 {
289 '交易说明': BASE_HEADERS_MAPPING['附言'], 322 '交易说明': SUMMARY_KEY,
290 } 323 }
291 ) 324 )
292 # 竖版-无表格-工商银行 抬头三行 活期历史明细清单 325 # 竖版-无表格-工商银行 抬头三行 活期历史明细清单
293 HEADERS_MAPPING.update( 326
294 {
295 '对方账户': BASE_HEADERS_MAPPING['对方卡号/账号'],
296 }
297 )
298 327
299 # -----------针式打印-全格线-------------------------------------------------------------------------------------------- 328 # -----------针式打印-全格线--------------------------------------------------------------------------------------------
300 # 竖版-表格-建设银行-中国建设银行活期账户交易明细 329 # 竖版-表格-建设银行-中国建设银行活期账户交易明细
...@@ -302,25 +331,19 @@ HEADERS_MAPPING.update( ...@@ -302,25 +331,19 @@ HEADERS_MAPPING.update(
302 # 竖版-表格-建设银行-对私活期账户明细- (1).pdf 331 # 竖版-表格-建设银行-对私活期账户明细- (1).pdf
303 HEADERS_MAPPING.update( 332 HEADERS_MAPPING.update(
304 { 333 {
305 '帐户余额': BASE_HEADERS_MAPPING['余额'], 334 '帐户余额': OVER_KEY,
306 '对方帐户名称': BASE_HEADERS_MAPPING['对方账户名'],
307 } 335 }
308 ) 336 )
309 # 竖版-特殊-交通银行 零售客户交易清单 5000以上交易记录 337 # 竖版-特殊-交通银行 零售客户交易清单 5000以上交易记录
310 HEADERS_MAPPING.update( 338 HEADERS_MAPPING.update(
311 { 339 {
312 '交易日期 记账日期': BASE_HEADERS_MAPPING['记账日期'], 340 '交易日期 记账日期': DATE_KEY,
313 } 341 }
314 ) 342 )
315 343
316 # ----------针式打印-部分格线------------------------------------------------------------------------------------------ 344 # ----------针式打印-部分格线------------------------------------------------------------------------------------------
317 # 竖版-特殊-邮储银行-一本通绿卡通交易明细(客户) 345 # 竖版-特殊-邮储银行-一本通绿卡通交易明细(客户)
318 # 竖版-特殊-邮储银行-账户交易明细(客户) 346 # 竖版-特殊-邮储银行-账户交易明细(客户)
319 HEADERS_MAPPING.update(
320 {
321 '对方账号/卡号/汇票号': BASE_HEADERS_MAPPING['对方卡号/账号'],
322 }
323 )
324 347
325 # -------------------------------------------------------------------------------------------------------------------- 348 # --------------------------------------------------------------------------------------------------------------------
326 349
...@@ -432,63 +455,6 @@ HEADERS_MAPPING.update( ...@@ -432,63 +455,6 @@ HEADERS_MAPPING.update(
432 455
433 OTHER_TUPLE = (None, None, None, None, None, None, None, None, None, None, None, None, None) 456 OTHER_TUPLE = (None, None, None, None, None, None, None, None, None, None, None, None, None)
434 457
435 # {
436 # "0":"其他",
437 # "1":"普通打印-全表格-中国农业银行",
438 # "2":"普通打印-全表格-中国银行",
439 # "3":"普通打印-全表格-北京银行",
440 # "4":"普通打印-全表格-工商银行",
441 # "5":"普通打印-全表格-建设银行",
442 # "6":"普通打印-全表格-微信账单",
443 # "7":"普通打印-全表格-支付宝账单",
444 # "8":"普通打印-无格线-中国邮政储蓄银行",
445
446 # "9":"普通打印-无格线-交通银行",
447 # "10":"普通打印-无格线-农业银行整数",
448 # "11":"普通打印-无格线-农业银行银行活期扩张缩进",
449 # "12":"普通打印-无格线-招商银行",
450 # "13":"普通打印-无格线-招行电子账单",
451 # "14":"普通打印-无格线-民生银行",
452
453 # "15":"普通打印-部分格线-横版-中信银行",
454 # "16":"普通打印-部分格线-竖版-中国农业银行分账户窄页",
455 # "17":"普通打印-部分格线-竖版-农业银行",
456 # "18":"普通打印-部分格线-竖版-农业银行银行卡交易明细",
457 # "19":"普通打印-部分格线-竖版-平安电子账单",
458
459 # "20":"针式打印-全格线-建设银行",
460 # "21":"针式打印-部分格线-竖版-邮储银行账户交易",
461 # "22":"针式打印-部分格线-邮储银行一本通绿卡"
462 # }
463
464 # CLASSIFY_LIST = [
465 # ('其他', OTHER_TUPLE),
466 # ('农业银行', (1, None, 3, 5, None, 8, 7, 6, None, None, None, None, None)),
467 # ('中国银行', (1, 2, 4, 5, 6, 9, 10, 11, 12, None, None, None, None)),
468 # ('北京银行', (1, None, 4, 5, None, 2, 6, 7, None, None, 3, None, None)),
469 # ('工商银行', (1, None, 9, 10, None, 7, None, None, None, None, None, None, None)),
470 # ('建设银行', (None, None, None, None, None, 2, None, None, None, None, None, None, None)),
471 # ('微信', (2, None, 6, None, None, 3, 7, None, None, None, None, None, None)),
472 # ('支付宝', (2, None, None, 6, None, 3, None, None, None, None, None, 4, 5)),
473 #
474 # ('交通银行', (1, None, 5, 6, None, 3, None, None, None, None, 4, None, None)),
475 # ('农业银行', (1, None, 3, None, None, 2, None, 4, None, None, None, None, None)),
476 # ('农业银行', (1, 2, None, None, None, None, None, None, None, None, None, None, None)),
477 # ('招商银行', (1, None, 3, 4, None, 6, None, None, None, None, None, None, None)),
478 # ('招商银行电子版', (1, None, 3, 4, None, 5, 6, None, None, None, None, None, None)),
479 # ('民生银行', (None, None, 5, 6, None, None, 7, None, 8, None, None, None, None)),
480 #
481 # ('中信银行', (1, None, None, 5, None, 2, 6, 7, 8, None, None, 3, 4)),
482 # ('农业银行', (1, None, 3, 4, None, 2, None, 5, None, None, None, None, None)),
483 # ('农业银行', (1, None, 3, 4, None, 2, None, 5, None, None, None, None, None)),
484 # ('农业银行', (1, None, 3, 4, None, 2, None, 5, None, None, None, None, None)),
485 # ('平安电子账单', (2, None, 5, 6, None, 4, None, None, None, None, None, None, None)),
486 #
487 # ('建设银行', (None, None, None, None, None, None, None, None, None, None, None, None, None)),
488 # ('邮储银行', (2, None, 5, 6, None, 4, None, 7, None, None, None, None, None)),
489 # ('邮储银行', (2, None, 5, 6, None, 4, None, 7, None, None, None, None, None)),
490 # ]
491
492 # "4":"普通打印-全表格-中国银行", 458 # "4":"普通打印-全表格-中国银行",
493 # "5":"普通打印-全表格-农业银行-10列", 459 # "5":"普通打印-全表格-农业银行-10列",
494 # "6":"普通打印-全表格-农业银行-10列-1", 460 # "6":"普通打印-全表格-农业银行-10列-1",
...@@ -563,6 +529,49 @@ CLASSIFY_LIST = [ ...@@ -563,6 +529,49 @@ CLASSIFY_LIST = [
563 ('其他', OTHER_TUPLE), 529 ('其他', OTHER_TUPLE),
564 ] 530 ]
565 531
532 CLASSIFY_HEADER_LIST = [
533 OTHER_TUPLE,
534 OTHER_TUPLE,
535 OTHER_TUPLE,
536 OTHER_TUPLE,
537 ('记账日期', '记账时间', '币别', '金额', '余额', '交易名称', '渠道', '网点名称', '附言', '对方账户名', '对方卡号/账号', '对方开户行'),
538 ('交易日期', '交易网点', '存入', '支出', '余额', '对方账号', '对方名称', '摘要', '渠道', '附言'),
539 ('序号', '日期', '摘要', '交易金额', '余额', '对方账号', '对方名称', '交易地点', '渠道', '附言'),
540 ('交易日期', '摘要', '交易金额', '余额', '交易渠道', '交易网点', '对方账号', '对方名称', '附言'),
541 ('交易日期', '业务摘要', '收/支', '发生额', '余额', '对方户名', '对方账号', '交易渠道'),
542 ('交易日期', '账号', '储种', '序号', '币种', '钞汇', '摘要', '地区', '收入/支出金额', '余额', '渠道'),
543 ('交易日期', '账号', '储种', '序号', '币种', '钞汇', '摘要', '地区', '收入/支出金额', '余额', '对方户名', '对方账号', '渠道'),
544 (None, '摘要', '交易日期', '交易金额', '账户余额', '商户/网点号及其名称', '对方账号与户名'),
545 ('交易单号', '交易时间', '交易类型', '收/支/其他', '交易方式', '金额(元)', '交易对方', '商户单号'),
546 ('流水号', '时间', '名称/备注', '收入', '支出', '账户余额', '资金渠道'),
547
548 ('交易日期', '记账日期', '交易地点', '交易类型', '借贷状态', '交易金额', '余额'),
549 ('交易日期', '交易类型', '交易金额(元)', '账户余额(元)', '操作柜员'),
550 ('交易日期', '交易类型', '交易币种', '交易金额(元)', '账户余额(元)', '对手方户名', '对手方账户', '收支类型'),
551 ('日期', '时间', '日志号', '短摘要', '交易金额', '本次余额', '交易网点', '渠道', '附言'),
552 ('交易日期', '摘要/附言', '交易金额', '对方账号和户名'),
553 ('记账日期', '货币', '交易金额', '联机余额', '冲补账', '交易摘要'),
554 ('记账日期', '货币', '交易金额', '联机余额', '交易摘要', '对手信息'),
555 ('凭证类型', '凭证号码', '交易时间', '摘要', '交易金额', '账户余额', '现转标志', '交易渠道', '交易机构', '对方户名', '对方行名'),
556
557 ('交易日期', '交易摘要', '收入金额', '支出金额', '账户余额', '对方户名', '对方账号', '对方银行', '交易流水号'),
558 ('交易日期', '摘要/附言', '交易金额', '余额', '交易地点/对方账号和户名'),
559 ('日期', '地点', '摘要', '存入', '支出', '余额', '对方账号', '对方户名'),
560 ('日期', '摘要', '交易金额', '余额', '地点', '交易对手账号', '对方户名'),
561 ('序号', '交易日期', '交易网点', '摘要', '借贷发生额(借:-贷:+)', '账户余额'),
562 ('序号', '摘要', '币别', '钞汇', '交易日期', '交易金额', '账户余额', '交易地点附言', '对方账号与户名'),
563 OTHER_TUPLE,
564 OTHER_TUPLE,
565 OTHER_TUPLE,
566 OTHER_TUPLE,
567 OTHER_TUPLE,
568 OTHER_TUPLE,
569 OTHER_TUPLE,
570 ('序号', '交易日期', '交易渠道', '摘要', '交易金额', '账户余额', '对方账号/卡号/汇票号', '原子账号', '交易机构名称'),
571 ('序号', '交易日期', '交易渠道', '摘要', '交易金额', '账户余额', '对方账号/卡号/汇票号', '原子账号', '交易机构名称'),
572 OTHER_TUPLE,
573 ]
574
566 # ----------license相关------------------------------------------------------------------------------------------------ 575 # ----------license相关------------------------------------------------------------------------------------------------
567 576
568 # "0":"AVT Invioce", 577 # "0":"AVT Invioce",
...@@ -603,9 +612,9 @@ RP_FIELD_ORDER_1 = IC_FIELD_ORDER_1 ...@@ -603,9 +612,9 @@ RP_FIELD_ORDER_1 = IC_FIELD_ORDER_1
603 VAT_CN_NAME = 'VAT普票' 612 VAT_CN_NAME = 'VAT普票'
604 VAT_CLASSIFY = 0 613 VAT_CLASSIFY = 0
605 VAT_FIELD_ORDER = (('发票代码', '发票代码'), 614 VAT_FIELD_ORDER = (('发票代码', '发票代码'),
606 ('发票代码（开具）', '发票代码（开具）'), 615 ('发票代码_开具', '发票代码（开具）'),
607 ('发票号码', '发票号码'), 616 ('发票号码', '发票号码'),
608 ('发票号码（开具）', '发票号码（开具）'), 617 ('发票号码_开具', '发票号码（开具）'),
609 ('开票日期', '开票日期'), 618 ('开票日期', '开票日期'),
610 ('校验码', '校验码'), 619 ('校验码', '校验码'),
611 ('货物或应税劳务、服务名称', '货物或应税劳务、服务名称'), 620 ('货物或应税劳务、服务名称', '货物或应税劳务、服务名称'),
...@@ -622,7 +631,7 @@ VAT_FIELD_ORDER = (('发票代码', '发票代码'), ...@@ -622,7 +631,7 @@ VAT_FIELD_ORDER = (('发票代码', '发票代码'),
622 ('销方纳税人识别号', '销售方纳税人识别号'), 631 ('销方纳税人识别号', '销售方纳税人识别号'),
623 ('销方地址、电话', '销售方地址、电话'), 632 ('销方地址、电话', '销售方地址、电话'),
624 ('销方开户行及账号', '销售方开户行及账号'), 633 ('销方开户行及账号', '销售方开户行及账号'),
625 ('销售方：（章）', '销售方：（章）'), 634 ('下盖章', '销售方：（章）'),
626 ('备注', '备注'),) 635 ('备注', '备注'),)
627 # 机动车登记证书 636 # 机动车登记证书
628 MVC_CN_NAME = '机动车登记证书' 637 MVC_CN_NAME = '机动车登记证书'
...@@ -856,3 +865,11 @@ LICENSE_CLASSIFY_MAPPING = dict(LICENSE_ORDER) ...@@ -856,3 +865,11 @@ LICENSE_CLASSIFY_MAPPING = dict(LICENSE_ORDER)
856 OTHER_CLASSIFY_SET = {OTHER_CLASSIFY} 865 OTHER_CLASSIFY_SET = {OTHER_CLASSIFY}
857 LICENSE_CLASSIFY_SET_1 = {IC_CLASSIFY, VAT_CLASSIFY, MVC_CLASSIFY, MVI_CLASSIFY} 866 LICENSE_CLASSIFY_SET_1 = {IC_CLASSIFY, VAT_CLASSIFY, MVC_CLASSIFY, MVI_CLASSIFY}
858 LICENSE_CLASSIFY_SET_2 = {BL_CLASSIFY, UCI_CLASSIFY, EEP_CLASSIFY, DL_CLASSIFY, PP_CLASSIFY, BC_CLASSIFY} 867 LICENSE_CLASSIFY_SET_2 = {BL_CLASSIFY, UCI_CLASSIFY, EEP_CLASSIFY, DL_CLASSIFY, PP_CLASSIFY, BC_CLASSIFY}
868
869 WECHART_CLASSIFY = 12
870 WECHART_HEADERS_MAPPING = copy.deepcopy(HEADERS_MAPPING)
871 WECHART_HEADERS_MAPPING.update(
872 {
873 '交易时间': DATE_KEY,
874 }
875 )
......
src/apps/doc/consts_bak.py 0 → 100644
1 PAGE_DEFAULT = 1
2 PAGE_SIZE_DEFAULT = 10
3
4 FIXED_APPLICATION_ID_PREFIX = 'CH-S'
5
6 DOC_SCHEME_LIST = ['ACCEPTANCE', 'SETTLEMENT', 'CONTRACT MANAGEMENT']
7 DATA_SOURCE_LIST = ['POS', 'EAPP', 'ECONTRACT']
8
9 HIL_PREFIX = 'HIL'
10 AFC_PREFIX = 'AFC'
11 SPLIT_STR = '_'
12 BUSINESS_TYPE_LIST = [HIL_PREFIX, AFC_PREFIX]
13 HIL_SET = {'HIL', 'HIl', 'HiL', 'Hil', 'hIL', 'hIl', 'hiL', 'hil', 'CO00002'}
14
15 # -------EDMS相关---------------------------------------------------------------------------------------------------
16
17 SESSION_PREFIX = 'FHLSID'
18 CUSTOM_CLIENT = 'CustomClient'
19 FIXED_TOKEN = '00000000-0000-0000-0000-000000000000'
20 FIXED_FILE_SIZE = 0
21 DOWNLOAD_ACTION_TYPE = 'Downloaded'
22
23 DOC_SCHEMA_ID_FILL = {
24 'ACCEPTANCE': (1, 'DFE-AutoFilingScript'),
25 'SETTLEMENT': (20, 'DFE-AutoFilingScript'),
26 'CONTRACT MANAGEMENT': (86, 'Schema-Based')
27 }
28 BUSINESS_TYPE_DICT = {
29 HIL_PREFIX: 'CO00002',
30 AFC_PREFIX: 'CO00001'
31 }
32 DOC_SCHEMA_TYPE = 'ElectronicRecord'
33 APPLICATION_ID_META_FIELD_id = 1
34 DEALER_CODE_META_FIELD_id = 13
35 BUSINESS_TYPE_META_FIELD_id = 93
36 DEALER_CODE = 'ocr_situ_group'
37
38 RETRY_TIMES = 3
39
40 # ---------银行流水模板相关--------------------------------------------------------------------------------------------
41
42 TRANS_MAP = {
43 'C': "0",
44 'c': "0",
45 '(': "0",
46 'o': "0",
47 'O': "0",
48 'D': "0",
49
50 '[': "1",
51 ']': "1",
52 'l': "1",
53 'L': "1",
54
55 'A': "4",
56
57 's': "5",
58 'S': "5",
59
60 'b': "6",
61
62 'g': "9",
63 'E': "9",
64
65 'B': "13",
66 }
67 TRANS = str.maketrans(TRANS_MAP)
68 ERROR_CHARS = {'.', '。', ':', '：', '•', '·', ',', '，'}
69 SKIP_IMG_SHEET_NAME = '未处理图片'
70 SKIP_IMG_SHEET_HEADER = ('页码', '序号')
71
72 CARD_RATIO = 0.9
73 UNKNOWN_CARD = '未知卡号'
74 UNKNOWN_ROLE = '未知户名'
75 DATE_FORMAT = ['%Y年%m月%d日', '%Y/%m/%d', '%Y-%m-%d', '%Y%m%d']
76
77 PROOF_COL_TITLE = '核对结果'
78 PROOF_RES = ('对', '错')
79 META_SHEET_TITLE = '关键信息提取和展示'
80
81 FIXED_HEADERS = ('记账日期', '记账时间', '金额', '余额', '交易名称', '附言', '对方账户名', '对方卡号/账号',
82 '对方开户行', '核对结果', '借贷', '收入', '支出')
83 FIXED_COL_AMOUNT = len(FIXED_HEADERS)
84 BASE_HEADERS_MAPPING = {label: idx + 1 for idx, label in enumerate(FIXED_HEADERS)}
85 BORROW_HEADER_COL = BASE_HEADERS_MAPPING['借贷']
86 INCOME_HEADER_COL = BASE_HEADERS_MAPPING['收入']
87 OUTLAY_HEADER_COL = BASE_HEADERS_MAPPING['支出']
88 RESULT_HEADER_COL = BASE_HEADERS_MAPPING['核对结果']
89 BORROW_IDX = BORROW_HEADER_COL - 1
90 INCOME_IDX = INCOME_HEADER_COL - 1
91 OUTLAY_IDX = OUTLAY_HEADER_COL - 1
92 SUMMARY_IDX = FIXED_HEADERS.index('附言')
93 DATE_IDX = FIXED_HEADERS.index('记账日期')
94 AMOUNT_IDX = FIXED_HEADERS.index('金额')
95 OVER_IDX = FIXED_HEADERS.index('余额')
96 RESULT_IDX = FIXED_HEADERS.index('核对结果')
97 # '借贷': ('贷', '借'), # 竖版-无表格-广发银行
98 # '借贷状态': ('贷', '借'), # 竖版-特殊-交通银行
99 # '收/支': ('收入', '支出'), # 横版-表格-北京银行
100 BORROW_HEADERS_SET = {'借贷', '借贷状态', '收/支'}
101 BORROW_INCOME_SET = {'贷', '收入'}
102 BORROW_OUTLAY_SET = {'借', '支出'}
103 INCOME_HEADERS_SET = {'收入金额', '收入', '存入', '存入金额（贷）', '存入金额(贷)'}
104 OUTLAY_HEADERS_SET = {'支出金额', '支出', '支取金额（借）', '支取金额(借)'}
105
106 # ------------------普通打印-全格线--------------------------------------------------------------------------------------
107 HEADERS_MAPPING = {}
108 # 横版-表格-中国银行（不规则）
109 HEADERS_MAPPING.update(
110 {
111 '记账日期': BASE_HEADERS_MAPPING['记账日期'],
112 '记账时间': BASE_HEADERS_MAPPING['记账时间'],
113 '金额': BASE_HEADERS_MAPPING['金额'],
114 '余额': BASE_HEADERS_MAPPING['余额'],
115 '交易名称': BASE_HEADERS_MAPPING['交易名称'],
116 '附言': BASE_HEADERS_MAPPING['附言'],
117 '对方账户名': BASE_HEADERS_MAPPING['对方账户名'],
118 '对方卡号/账号': BASE_HEADERS_MAPPING['对方卡号/账号'],
119 '对方开户行': BASE_HEADERS_MAPPING['对方开户行'],
120 }
121 )
122 # 横版-表格-农业银行-中国农业银行个人账户明细
123 HEADERS_MAPPING.update(
124 {
125 '交易日期': BASE_HEADERS_MAPPING['记账日期'],
126 # '存入': BASE_HEADERS_MAPPING['金额'],
127 '对方账号': BASE_HEADERS_MAPPING['对方卡号/账号'],
128 '对方名称': BASE_HEADERS_MAPPING['对方账户名'],
129 '摘要': BASE_HEADERS_MAPPING['附言'],
130 }
131 )
132 # 横版-表格-北京银行
133 HEADERS_MAPPING.update(
134 {
135 '业务摘要': BASE_HEADERS_MAPPING['附言'],
136 '发生额': BASE_HEADERS_MAPPING['金额'],
137 '对方户名': BASE_HEADERS_MAPPING['对方账户名'],
138 }
139 )
140 # 横版-表格-工商银行 借记卡账户历史明细清单
141 # 横版-表格-工商银行-机打验证码 借记卡账户历史明细清单
142 # 横版-表格-工商银行CH-B008802400
143 # 横版-表格-工商银行 工资明细清单
144 # 工商银行历史明细（申请单号：20042501303039397888）
145 HEADERS_MAPPING.update(
146 {
147 '收入/支出金额': BASE_HEADERS_MAPPING['金额'],
148 '工作日期': BASE_HEADERS_MAPPING['记账日期'],
149 }
150 )
151
152 # 横版-表格-建设银行-个人活期账户交易明细
153 # 竖版-表格-建设银行-个人活期账户交易明细 CH-B005832604
154 # 竖版-表格-建设银行-工资账单CH-B008786812
155 # 竖版-表格-建设银行-个人活期账户交易明细 CH-B005832604 (2)
156 HEADERS_MAPPING.update(
157 {
158 '交易金额': BASE_HEADERS_MAPPING['金额'],
159 '账户余额': BASE_HEADERS_MAPPING['余额'],
160 '对方账号与户名': BASE_HEADERS_MAPPING['对方卡号/账号'],
161 }
162 )
163 # 微信
164 HEADERS_MAPPING.update(
165 {
166 '交易时间': BASE_HEADERS_MAPPING['记账时间'],
167 '交易类型': BASE_HEADERS_MAPPING['附言'],
168 '金额(元)': BASE_HEADERS_MAPPING['金额'],
169 '金额（元）': BASE_HEADERS_MAPPING['金额'],
170 '交易对方': BASE_HEADERS_MAPPING['对方账户名'],
171 }
172 )
173 # 支付宝
174 HEADERS_MAPPING.update(
175 {
176 '时间': BASE_HEADERS_MAPPING['记账日期'],
177 '名称/备注': BASE_HEADERS_MAPPING['附言'],
178 }
179 )
180
181 # ------------普通打印-部分格线-------------------------------------------------------------------------------------------
182
183 # 竖版-无表格-农业银行
184 # 竖版-无表格-农业银行CH-B008805428
185 HEADERS_MAPPING.update(
186 {
187 '摘要/附言': BASE_HEADERS_MAPPING['附言'],
188 '交易地点/对方账号和户名': BASE_HEADERS_MAPPING['对方卡号/账号'],
189 }
190 )
191 # 农业银行-窄页
192 HEADERS_MAPPING.update(
193 {
194 '交易对手账号': BASE_HEADERS_MAPPING['对方卡号/账号'],
195 }
196 )
197 # 竖版-特殊-农商行
198 HEADERS_MAPPING.update(
199 {
200 '交易发生额': BASE_HEADERS_MAPPING['金额'],
201 }
202 )
203 # 横版-特殊-中信银行-账户交易明细
204 HEADERS_MAPPING.update(
205 {
206 '对方银行': BASE_HEADERS_MAPPING['对方开户行'],
207 '交易摘要': BASE_HEADERS_MAPPING['附言'],
208 }
209 )
210 # 平安电子账单
211 HEADERS_MAPPING.update(
212 {
213 '借贷发生额(借:-贷:+)': BASE_HEADERS_MAPPING['金额'],
214 }
215 )
216
217 # ------------普通打印-无格线--------------------------------------------------------------------------------------------
218
219 # 竖版-无表格-招商银行(略歪）
220 # 竖版-无表格-招商银行账户历史交易明细表
221 HEADERS_MAPPING.update(
222 {
223 '联机余额': BASE_HEADERS_MAPPING['余额'],
224 }
225 )
226 # 竖版-无表格-邮储银行-账户对账单 含有对手方户名 对手方账户
227 # 竖版-无表格-邮储银行 账户对账单
228 # 竖版-无表格-邮储银行-电子章 邮储银行 账户对账单
229 HEADERS_MAPPING.update(
230 {
231 '交易金额(元)': BASE_HEADERS_MAPPING['金额'],
232 '交易金额（元）': BASE_HEADERS_MAPPING['金额'],
233 '账户余额（元）': BASE_HEADERS_MAPPING['余额'],
234 '账户余额(元)': BASE_HEADERS_MAPPING['余额'],
235 '对手方户名': BASE_HEADERS_MAPPING['对方账户名'],
236 '对手方账户': BASE_HEADERS_MAPPING['对方卡号/账号'],
237 }
238 )
239 # 横版-无表格-广发银行-账户交易历史 --> 已废弃
240 # 竖版-无表格-广发银行-账户交易历史 --> 已废弃
241 HEADERS_MAPPING.update(
242 {
243 '会计日期': BASE_HEADERS_MAPPING['记账日期'],
244 '对手户名': BASE_HEADERS_MAPPING['对方账户名'],
245 '对手账号': BASE_HEADERS_MAPPING['对方卡号/账号'],
246 }
247 )
248 # 招行电子账单 TODO 有英文，需测试
249 HEADERS_MAPPING.update(
250 {
251 '对手信息': BASE_HEADERS_MAPPING['对方账户名'],
252 '摘要代码': BASE_HEADERS_MAPPING['附言'],
253 }
254 )
255 # 横版-无表格-民生银行-中国民生银行个人账户对账单（客户卡号）
256 # 横版-无表格-民生银行-无标题(客户账户)
257 # 横版-无表格-民生银行
258 HEADERS_MAPPING.update(
259 {
260 '摘要信息': BASE_HEADERS_MAPPING['附言'],
261 '对方行名': BASE_HEADERS_MAPPING['对方开户行'],
262 }
263 )
264 # 竖版-无表格-农业银行整数
265 # 竖版-无表格-农业银行-中国农业银行银行卡交易明细清单
266 HEADERS_MAPPING.update(
267 {
268 '对方账号和户名': BASE_HEADERS_MAPPING['对方卡号/账号'],
269 }
270 )
271 # 竖版-无表格-农业银行-中国农业银行银行卡活期存折交易明细清单.pdf
272 # 竖版-无表格-农业银行-扩张.pdf
273 # 竖版-无表格-农业银行-缩进.pdf
274 HEADERS_MAPPING.update(
275 {
276 '日期': BASE_HEADERS_MAPPING['记账日期'],
277 '短摘要': BASE_HEADERS_MAPPING['附言'],
278 '本次余额': BASE_HEADERS_MAPPING['余额'],
279 }
280 )
281 # 竖版-无表格-农业银行-无标题(对手帐号)
282 HEADERS_MAPPING.update(
283 {
284 '交易后余额': BASE_HEADERS_MAPPING['余额'],
285 '对手帐号': BASE_HEADERS_MAPPING['对方卡号/账号'],
286 }
287 )
288 # 竖版-无表格-农商行（非常规）
289 HEADERS_MAPPING.update(
290 {
291 '交易说明': BASE_HEADERS_MAPPING['附言'],
292 }
293 )
294 # 竖版-无表格-工商银行 抬头三行 活期历史明细清单
295 HEADERS_MAPPING.update(
296 {
297 '对方账户': BASE_HEADERS_MAPPING['对方卡号/账号'],
298 }
299 )
300
301 # -----------针式打印-全格线--------------------------------------------------------------------------------------------
302 # 竖版-表格-建设银行-中国建设银行活期账户交易明细
303 # 竖版-表格-建设银行-中国建设银行活期账户明细清单
304 # 竖版-表格-建设银行-对私活期账户明细- (1).pdf
305 HEADERS_MAPPING.update(
306 {
307 '帐户余额': BASE_HEADERS_MAPPING['余额'],
308 '对方帐户名称': BASE_HEADERS_MAPPING['对方账户名'],
309 }
310 )
311 # 竖版-特殊-交通银行 零售客户交易清单 5000以上交易记录
312 HEADERS_MAPPING.update(
313 {
314 '交易日期 记账日期': BASE_HEADERS_MAPPING['记账日期'],
315 }
316 )
317
318 # ----------针式打印-部分格线------------------------------------------------------------------------------------------
319 # 竖版-特殊-邮储银行-一本通绿卡通交易明细(客户)
320 # 竖版-特殊-邮储银行-账户交易明细(客户)
321 HEADERS_MAPPING.update(
322 {
323 '对方账号/卡号/汇票号': BASE_HEADERS_MAPPING['对方卡号/账号'],
324 }
325 )
326
327 # --------------------------------------------------------------------------------------------------------------------
328
329 # ('记账日期', '记账时间', '金额', '余额', '交易名称', '附言', '对方账户名', '对方卡号/账号', '对方开户行', '核对结果', '借贷', '收入', '支出')
330 # CLASSIFY_LIST = [
331 # # --------------普通打印：全格线---------------------------------
332 # # 中国银行：记账日期 记账时间 币别 金额 余额 交易名称 渠道 网点名称 附言 对方账户名 对方卡号/账号 对方开户行
333 # ('中国银行', (1, 2, 4, 5, 6, 9, 10, 11, 12, None, None, None, None)), # 横版-表格-中国银行（不规则）
334 #
335 # # 农业银行：交易日期 交易网点 存入 支出 余额 对方账号 对方名称 摘要 渠道 附言
336 # ('农业银行-10', (1, None, None, 5, None, 8, 7, 6, None, None, None, 3, 4)), # 横版-表格-农业银行-中国农业银行个人账户明细
337 #
338 # # 农业银行：序号 日期 摘要 交易金额 余额 对方账号 对方名称 交易地点 渠道 附言
339 # ('农业银行-10-1', (2, None, 4, 5, None, 3, 7, 6, None, None, None, None, None)),
340 #
341 # # 农业银行：交易日期 摘要 交易金额 余额 交易渠道 交易网点 对方账号 对方名称 附言
342 # ('农业银行-9', (1, None, 3, 4, None, 2, 8, 7, None, None, None, None, None)),
343 #
344 # # 北京银行：交易日期 业务摘要 收/支 发生额 余额 对方户名 对方账号 交易渠道
345 # ('北京银行', (1, None, 4, 5, None, 2, 6, 7, None, None, 3, None, None)), # 横版-表格-北京银行
346 #
347 # # 工商银行：交易日期 账号 储种 序号 币种 钞汇 摘要 地区 收入/支出金额 余额 渠道
348 # ('工商银行', (1, None, 9, 10, None, 7, None, None, None, None, None, None, None)),
349 #
350 # # 工商银行：交易日期 账号 储种 序号 币种 钞汇 摘要 地区 收入/支出金额 余额 对方户名 对方账号 渠道
351 # ('工商银行-电子账单', (1, None, 9, 10, None, 7, 11, 12, None, None, None, None, None)),
352 #
353 # # 建设银行：空 摘要 交易日期 交易金额 账户余额 商户/网点号及其名称 对方账号与户名 --> 竖版-表格-建设银行
354 # # 序号 摘要 币别 钞汇 交易日期 交易金额 账户余额 交易地点附言 对方账号与户名 --> 横版-表格-建设银行
355 # ('建设银行-竖版', (3, None, 4, 5, None, 2, None, 7, None, None, None, None, None)),
356 # ('建设银行-横版', (5, None, 6, 7, None, 2, None, 9, None, None, None, None, None)),
357 #
358 # # 微信：交易单号 交易时间 交易类型 收/支/其他 交易方式 金额(元) 交易对方 商户单号
359 # ('微信', (2, None, 6, None, None, 3, 7, None, None, None, None, None, None)),
360 #
361 # # 支付宝：流水号 时间 名称/备注 收入 支出 账户余额 资金渠道
362 # ('支付宝', (2, None, None, 6, None, 3, None, None, None, None, None, 4, 5)),
363 #
364 # # -----------------普通打印：部分格线--------------------------------
365 #
366 # # 农业银行：交易日期 摘要/附言 交易金额 余额 交易地点/对方账号和户名
367 # ('农业银行-5', (1, None, 3, 4, None, 2, None, 5, None, None, None, None, None)),
368 #
369 # # 农业银行：日期 地点 摘要 存入 支出 余额 对方账号 对方户名
370 # ('农业银行-8', (1, None, None, 6, None, 3, 8, 7, None, None, None, 4, 5)),
371
372 # # 农业银行：日期 摘要 交易金额 余额 地点 交易对手账号 对方户名
373 # ('农业银行-窄页', (1, None, 3, 4, None, 2, 7, 6, None, None, None, None, None)),
374 #
375 # # 农商行：交易日期 交易发生额 账户余额 对方账号 对方户名 摘要 备注
376 # ('农商行', (1, None, 2, 3, None, 6, 5, 4, None, None, None, None, None)),
377 #
378 # # 中信银行：交易日期 交易摘要 收入金额 支出金额 账户余额 对方户名 对方账号 对方银行 交易流水号
379 # ('中信银行', (1, None, None, 5, None, 2, 6, 7, 8, None, None, 3, 4)),
380 #
381 # # 平安电子账单：序号 交易日期 交易网点 摘要 借贷发生额(借:-贷:+) 账户余额
382 # ('平安电子账单', (2, None, 5, 6, None, 4, None, None, None, None, None, None, None)),
383
384 # # 建设银行：序号 摘要 币别 钞汇 交易日期 交易金额 账户余额 交易地点附言 对方账号与户名
385 # ('建设银行-电子账单', (5, None, 6, 7, None, 2, None, 9, None, None, None, None, None)),
386 #
387 # # -----------------普通打印：无格线-------------------------------------
388 #
389 # # 招商银行：记账日期 货币 交易金额 联机余额 冲补账 交易摘要
390 # ('招商银行', (1, None, 3, 4, None, 6, None, None, None, None, None, None, None)),
391 #
392 # # 邮储银行：交易日期、交易类型 交易币种 交易金额(元) 账户余额(元) [对手方户名 对手方账户 收支类型] --> 竖版-无表格-邮储银行-账户对账单 含有对手方户名 对手方账户
393 # # 交易日期、交易类型 交易金额(元) 账户余额(元) 操作柜员 --> 竖版-无表格-邮储银行 账户对账单
394 # ('邮储银行-8', (1, None, 4, 5, None, 2, 6, 7, None, None, None, None, None)),
395 # ('邮储银行-5', (1, None, 3, 4, None, 2, None, None, None, None, None, None, None)),
396 #
397 # # 工商银行电子版：交易日期 账号 储种 序号 币种 妙汇 摘要 地区 收入/支出金额 余额 [对方户名 对方账号] 渠道
398 # ('工商银行电子版', (1, None, 9, 10, None, 7, None, None, None, None, None, None, None)),
399 #
400 # # 招商银行电子版：记账日期 货币 交易金额 联机余额 交易摘要 对手信息
401 # ('招商银行电子版', (1, None, 3, 4, None, 5, 6, None, None, None, None, None, None)),
402 #
403 # # 民生银行：凭证类型 凭证号码 摘要信息 交易时间 交易金额 账户余额 现转标志 交易渠道 交易机构 对方户名 对方行名 --> 横版-无表格-民生银行-中国民生银行个人账户对账单（客户卡号）
404 # # 凭证类型 凭证号码 交易时间 摘要 交易金额 账户余额 现转标志 交易渠道 交易机构 对方户名 对方行名 --> 横版-无表格-民生银行
405 # ('民生银行', (None, None, 5, 6, None, None, 7, None, 8, None, None, None, None)),
406 #
407 # # 农业银行：交易日期 摘要/附言 交易金额 对方账号和户名
408 # ('农业银行-整数', (1, None, 3, None, None, 2, None, 4, None, None, None, None, None)),
409 #
410 # # 农业银行：交易日期 摘要/附言 交易金额 余额 交易地点/对方账号和户名
411 # ('农业银行', (1, None, 3, 4, None, 2, None, 5, None, None, None, None, None)),
412 #
413 # # 农业银行：日期、时间、短摘要、交易金额、本次余额、交易网点、渠道、附言
414 # # 农业银行：日期、时间、日志号、短摘要、交易金额、本次余额、交易网点、渠道、附言
415 # ('农业银行', (1, 2, 4, 5, None, 3, None, None, None, None, None, None, None)),
416 # ('农业银行-扩张缩进', (1, 2, 5, 6, None, 4, None, None, None, None, None, None, None)),
417 #
418 # # 交通银行：交易日期 记账日期、交易地点、交易类型、借贷状态、交易金额、余额
419 # ('交通银行', (1, None, 5, 6, None, 3, None, None, None, None, 4, None, None)),
420 #
421 #
422 # # ================针式打印：有格线===================
423 #
424 # # 建设银行： 摘要、交易日期、交易金额、账户余额、商户/网点号及其名称、对方账号、对方户名 --> 竖版-表格-建设银行-中国建设银行活期账户明细清单
425 # # 交易日期、摘要、 币种、 钞汇、 交易金额、 帐户余额、对方账号、 对方帐户名称 --> 竖版-表格-建设银行-对私活期账户明细- (1)
426 # ('建设银行', (None, None, None, None, None, None, None, None, None, None, None, None, None)),
427 #
428 #
429 # # ================针式打印：无格线===================
430 #
431 # # 邮储银行：序号、交易日期、交易渠道、摘要、交易金额、账户余额、对方账号/卡号/汇票号、原子账号、交易机构名称
432 # ('邮储银行', (2, None, 5, 6, None, 4, None, 7, None, None, None, None, None)),
433 # ]
434
435 OTHER_TUPLE = (None, None, None, None, None, None, None, None, None, None, None, None, None)
436
437 # {
438 # "0":"其他",
439 # "1":"普通打印-全表格-中国农业银行",
440 # "2":"普通打印-全表格-中国银行",
441 # "3":"普通打印-全表格-北京银行",
442 # "4":"普通打印-全表格-工商银行",
443 # "5":"普通打印-全表格-建设银行",
444 # "6":"普通打印-全表格-微信账单",
445 # "7":"普通打印-全表格-支付宝账单",
446 # "8":"普通打印-无格线-中国邮政储蓄银行",
447
448 # "9":"普通打印-无格线-交通银行",
449 # "10":"普通打印-无格线-农业银行整数",
450 # "11":"普通打印-无格线-农业银行银行活期扩张缩进",
451 # "12":"普通打印-无格线-招商银行",
452 # "13":"普通打印-无格线-招行电子账单",
453 # "14":"普通打印-无格线-民生银行",
454
455 # "15":"普通打印-部分格线-横版-中信银行",
456 # "16":"普通打印-部分格线-竖版-中国农业银行分账户窄页",
457 # "17":"普通打印-部分格线-竖版-农业银行",
458 # "18":"普通打印-部分格线-竖版-农业银行银行卡交易明细",
459 # "19":"普通打印-部分格线-竖版-平安电子账单",
460
461 # "20":"针式打印-全格线-建设银行",
462 # "21":"针式打印-部分格线-竖版-邮储银行账户交易",
463 # "22":"针式打印-部分格线-邮储银行一本通绿卡"
464 # }
465
466 # CLASSIFY_LIST = [
467 # ('其他', OTHER_TUPLE),
468 # ('农业银行', (1, None, 3, 5, None, 8, 7, 6, None, None, None, None, None)),
469 # ('中国银行', (1, 2, 4, 5, 6, 9, 10, 11, 12, None, None, None, None)),
470 # ('北京银行', (1, None, 4, 5, None, 2, 6, 7, None, None, 3, None, None)),
471 # ('工商银行', (1, None, 9, 10, None, 7, None, None, None, None, None, None, None)),
472 # ('建设银行', (None, None, None, None, None, 2, None, None, None, None, None, None, None)),
473 # ('微信', (2, None, 6, None, None, 3, 7, None, None, None, None, None, None)),
474 # ('支付宝', (2, None, None, 6, None, 3, None, None, None, None, None, 4, 5)),
475 #
476 # ('交通银行', (1, None, 5, 6, None, 3, None, None, None, None, 4, None, None)),
477 # ('农业银行', (1, None, 3, None, None, 2, None, 4, None, None, None, None, None)),
478 # ('农业银行', (1, 2, None, None, None, None, None, None, None, None, None, None, None)),
479 # ('招商银行', (1, None, 3, 4, None, 6, None, None, None, None, None, None, None)),
480 # ('招商银行电子版', (1, None, 3, 4, None, 5, 6, None, None, None, None, None, None)),
481 # ('民生银行', (None, None, 5, 6, None, None, 7, None, 8, None, None, None, None)),
482 #
483 # ('中信银行', (1, None, None, 5, None, 2, 6, 7, 8, None, None, 3, 4)),
484 # ('农业银行', (1, None, 3, 4, None, 2, None, 5, None, None, None, None, None)),
485 # ('农业银行', (1, None, 3, 4, None, 2, None, 5, None, None, None, None, None)),
486 # ('农业银行', (1, None, 3, 4, None, 2, None, 5, None, None, None, None, None)),
487 # ('平安电子账单', (2, None, 5, 6, None, 4, None, None, None, None, None, None, None)),
488 #
489 # ('建设银行', (None, None, None, None, None, None, None, None, None, None, None, None, None)),
490 # ('邮储银行', (2, None, 5, 6, None, 4, None, 7, None, None, None, None, None)),
491 # ('邮储银行', (2, None, 5, 6, None, 4, None, 7, None, None, None, None, None)),
492 # ]
493
494 # "4":"普通打印-全表格-中国银行",
495 # "5":"普通打印-全表格-农业银行-10列",
496 # "6":"普通打印-全表格-农业银行-10列-1",
497 # "7":"普通打印-全表格-农业银行-9列",
498 # "8":"普通打印-全表格-北京银行",
499 # "9":"普通打印-全表格-工商银行",
500 # "10":"普通打印-全表格-工商银行-电子账单",
501 # "11":"普通打印-全表格-建设银行",
502 # "12":"普通打印-全表格-微信账单",
503 # "13":"普通打印-全表格-支付宝账单",
504
505 # "14":"普通打印-无格线-交通银行",
506 # "15":"普通打印-无格线-储蓄银行-5列",
507 # "16":"普通打印-无格线-储蓄银行-8列",
508 # "17":"普通打印-无格线-农业银行-扩张缩进",
509 # "18":"普通打印-无格线-农业银行-整数",
510 # "19":"普通打印-无格线-招商银行",
511 # "20":"普通打印-无格线-招商银行-电子账单",
512 # "21":"普通打印-无格线-民生银行",
513
514 # "22":"普通打印-部分格线-横版-中信银行",
515 # "23":"普通打印-部分格线-竖版-农业银行-5列",
516 # "24":"普通打印-部分格线-竖版-农业银行-8列",
517 # "25":"普通打印-部分格线-竖版-农业银行-窄页",
518 # "26":"普通打印-部分格线-竖版-平安电子账单",
519 # "27":"普通打印-部分格线-竖版-建设银行-电子账单",
520
521 # "34":"针式打印-全格线-建设银行",
522 # "35":"针式打印-部分格线-竖版-邮储银行",
523 # "36":"针式打印-部分格线-竖版-邮储银行-绿卡",
524
525 CLASSIFY_LIST = [
526 ('其他', OTHER_TUPLE),
527 ('其他', OTHER_TUPLE),
528 ('其他', OTHER_TUPLE),
529 ('其他', OTHER_TUPLE),
530 ('普通打印-全表格-中国银行', (1, 2, 4, 5, 6, 9, 10, 11, 12, None, None, None, None)),
531 ('普通打印-全表格-农业银行-10列', (1, None, None, 5, None, 8, 7, 6, None, None, None, 3, 4)),
532 ('普通打印-全表格-农业银行-10列-1', (2, None, 4, 5, None, 3, 7, 6, None, None, None, None, None)),
533 ('普通打印-全表格-农业银行-9列', (1, None, 3, 4, None, 2, 8, 7, None, None, None, None, None)),
534 ('普通打印-全表格-北京银行', (1, None, 4, 5, None, 2, 6, 7, None, None, 3, None, None)),
535 ('普通打印-全表格-工商银行', (1, None, 9, 10, None, 7, None, None, None, None, None, None, None)),
536 ('普通打印-全表格-工商银行-电子账单', (1, None, 9, 10, None, 7, 11, 12, None, None, None, None, None)),
537 ('普通打印-全表格-建设银行', (3, None, 4, 5, None, 2, None, 7, None, None, None, None, None)),
538 ('普通打印-全表格-微信账单', (2, None, 6, None, None, 3, 7, None, None, None, None, None, None)),
539 ('普通打印-全表格-支付宝账单', (2, None, None, 6, None, 3, None, None, None, None, None, 4, 5)),
540
541 ('普通打印-无格线-交通银行', (1, None, 5, 6, None, 3, None, None, None, None, 4, None, None)),
542 ('普通打印-无格线-储蓄银行-5列', (1, None, 3, 4, None, 2, None, None, None, None, None, None, None)),
543 ('普通打印-无格线-储蓄银行-8列', (1, None, 4, 5, None, 2, 6, 7, None, None, None, None, None)),
544 ('普通打印-无格线-农业银行-扩张缩进', (1, 2, 5, 6, None, 4, None, None, None, None, None, None, None)),
545 ('普通打印-无格线-农业银行-整数', (1, None, 3, None, None, 2, None, 4, None, None, None, None, None)),
546 ('普通打印-无格线-招商银行', (1, None, 3, 4, None, 6, None, None, None, None, None, None, None)),
547 ('普通打印-无格线-招商银行-电子账单', (1, None, 3, 4, None, 5, 6, None, None, None, None, None, None)),
548 ('普通打印-无格线-民生银行', (None, None, 5, 6, None, None, 7, None, 8, None, None, None, None)),
549
550 ('普通打印-部分格线-横版-中信银行', (1, None, None, 5, None, 2, 6, 7, 8, None, None, 3, 4)),
551 ('普通打印-部分格线-竖版-农业银行-5列', (1, None, 3, 4, None, 2, None, 5, None, None, None, None, None)),
552 ('普通打印-部分格线-竖版-农业银行-8列', (1, None, None, 6, None, 3, 8, 7, None, None, None, 4, 5)),
553 ('普通打印-部分格线-竖版-农业银行-窄页', (1, None, 3, 4, None, 2, 7, 6, None, None, None, None, None)),
554 ('普通打印-部分格线-竖版-平安电子账单', (2, None, 5, 6, None, 4, None, None, None, None, None, None, None)),
555 ('普通打印-部分格线-竖版-建设银行-电子账单', (5, None, 6, 7, None, 2, None, 9, None, None, None, None, None)),
556 ('其他', OTHER_TUPLE),
557 ('其他', OTHER_TUPLE),
558 ('其他', OTHER_TUPLE),
559 ('其他', OTHER_TUPLE),
560 ('其他', OTHER_TUPLE),
561 ('其他', OTHER_TUPLE),
562 ('针式打印-全格线-建设银行', OTHER_TUPLE),
563 ('针式打印-部分格线-竖版-邮储银行', (2, None, 5, 6, None, 4, None, 7, None, None, None, None, None)),
564 ('针式打印-部分格线-竖版-邮储银行-绿卡', (2, None, 5, 6, None, 4, None, 7, None, None, None, None, None)),
565 ('其他', OTHER_TUPLE),
566 ]
567
568 # ----------license相关------------------------------------------------------------------------------------------------
569
570 # "0":"AVT Invioce",
571 # "1":"二手车发票",
572 # "2":"其他",
573 # "3":"护照",
574 # "28":"机动车登记证",
575 # "29":"机动车销售统一发票",
576 # "30":"港澳通行证",
577 # "31":"营业执照",
578 # "32":"行驶证",
579 # "33":"身份证",
580 # "37":"银行卡"
581
582 # 其他
583 OTHER_CLASSIFY = 2
584
585 # 身份证
586 IC_CN_NAME = '身份证'
587 IC_CLASSIFY = 33
588 IC_FIELD_ORDER_0 = (('姓名', '姓名'),
589 ('公民身份号码', '公民身份号码'),
590 ('出生年月', '出生年月'),
591 ('住址', '住址'),
592 ('性别', '性别'),
593 ('民族', '民族'),)
594 IC_FIELD_ORDER_1 = (('有效期限', '有效期限'), ('签发机关', '签发机关'),)
595 # 居住证
596 RP_CN_NAME = '居住证'
597 RP_CLASSIFY = 10087
598 RP_FIELD_ORDER_0 = (('姓名', '姓名'),
599 ('公民身份号码', '公民身份号码'),
600 ('出生年月', '出生年月'),
601 ('住址', '住址'),
602 ('性别', '性别'),)
603 RP_FIELD_ORDER_1 = IC_FIELD_ORDER_1
604 # 增值税发票
605 VAT_CN_NAME = 'VAT普票'
606 VAT_CLASSIFY = 0
607 VAT_FIELD_ORDER = (('发票代码', '发票代码'),
608 ('发票代码（开具）', '发票代码（开具）'),
609 ('发票号码', '发票号码'),
610 ('发票号码（开具）', '发票号码（开具）'),
611 ('开票日期', '开票日期'),
612 ('校验码', '校验码'),
613 ('货物或应税劳务、服务名称', '货物或应税劳务、服务名称'),
614 ('金额合计', '开具金额合计（不含税）'),
615 ('税率', '税率'),
616 ('税额合计', '税额合计'),
617 ('价税合计小写', '价税合计（小写）'),
618 ('价税合计大写', '价税合计（大写）'),
619 ('购方名称', '购买方名称'),
620 ('购方纳税人识别号', '购买方纳税人识别号'),
621 ('购方地址、电话', '购买方地址、电话'),
622 ('购方开户行及账号', '购买方开户行及账号'),
623 ('销方名称', '销售方名称'),
624 ('销方纳税人识别号', '销售方纳税人识别号'),
625 ('销方地址、电话', '销售方地址、电话'),
626 ('销方开户行及账号', '销售方开户行及账号'),
627 ('销售方：（章）', '销售方：（章）'),
628 ('备注', '备注'),)
629 # 机动车登记证书
630 MVC_CN_NAME = '机动车登记证书'
631 MVC_CLASSIFY = 28
632 MVC_CLASSIFY_SE = 10086
633 MVC_FIELD_ORDER_1_2 = (('1.机动车所有人/身份证名称/号码', '机动车所有人/身份证明名称/号码'),
634 ('3.登记日期', '登记日期'),
635 ('9.车辆识别代号/车架号', '车辆识别代号/车架号'),
636 ('32.车辆出厂日期', '车辆出厂日期'),
637 ('34.发证日期', '发证日期'),
638 ('30.使用性质', '使用性质'),
639 ('31.车辆获得方式', '车辆获得方式'),
640 ('4.机动车登记编号', '机动车登记编号'),
641 ('空行占位', None),
642 ('5.车辆类型', '车辆类型'),
643 ('6.车辆品牌', '车辆品牌'),
644 ('7.车辆型号', '车辆型号'),
645 ('8.车身颜色', '车身颜色'),
646 ('10.国产/进口', '国产/进口'),
647 ('11.发动机号', '发动机号'),
648 ('12.发动机型号', '发动机型号'),
649 ('15.制造厂名称', '制造厂名称'),
650 ('2.登记机关', '登记机关'),
651 ('编号', '机动车登记证书编号'),)
652 MVC_FIELD_ORDER_3_4 = (
653 ('姓名/名称', '姓名/名称'),
654 ('身份证明名称/号码', '身份证明名称/号码'),
655 ('转移登记日期', '转移登记日期'),
656 )
657 MVC_SE_FIELD_ORDER_1_2 = (('9.车辆识别代号/车架号', '车辆识别代号/车架号'),
658 ('1.机动车所有人/身份证名称/号码', '机动车所有人/身份证明名称/号码'),
659 ('空行占位', None),
660 ('3.登记日期', '登记日期'),
661 ('32.车辆出厂日期', '车辆出厂日期'),
662 ('34.发证日期', '发证日期'),
663 ('30.使用性质', '使用性质'),
664 ('31.车辆获得方式', '车辆获得方式'),
665 ('5.车辆类型', '车辆类型'),
666 ('6.车辆品牌', '车辆品牌'),
667 ('7.车辆型号', '车辆型号'),
668 ('8.车身颜色', '车身颜色'),
669 ('10.国产/进口', '国产/进口'),
670 ('11.发动机号', '发动机号'),
671 ('12.发动机型号', '发动机型号'),
672 ('13.燃料种类', '燃料种类'),
673 ('14.排量/功率', '排量/功率'),
674 ('15.制造厂名称', '制造厂名称'),
675 ('16.转向形式', '转向形式'),
676 ('17.轮距', '轮距'),
677 ('18.轮胎数', '轮胎数'),
678 ('19.轮胎规格', '轮胎规格'),
679 ('20.钢板弹簧片数', '钢板弹簧片数'),
680 ('21.轴距', '轴距'),
681 ('22.轴数', '轴数'),
682 ('23.外廓尺寸', '外廓尺寸'),
683 ('24.货厢内部尺寸', '货厢内部尺寸'),
684 ('25.总质量', '总质量'),
685 ('26.核定载质量', '核定载质量'),
686 ('27.核定载客', '核定载客'),
687 ('28.准牵引总质量', '准牵引总质量'),
688 ('29.驾驶室载客', '驾驶室载客'),
689 ('2.登记机关', '登记机关'),
690 ('4.机动车登记编号', '机动车登记编号'),
691 ('编号', '机动车登记证书编号'),)
692 MVC_SE_FIELD_ORDER_3_4 = (
693 ('姓名/名称', '姓名/名称'),
694 ('身份证明名称/号码', '身份证明名称/号码'),
695 ('转移登记日期', '转移登记日期'),
696 )
697 # 机动车销售统一发票
698 MVI_CN_NAME = '机动车销售统一发票'
699 MVI_CLASSIFY = 29
700 MVI_FIELD_ORDER = (('发票代码', '发票代码'),
701 ('发票号码', '发票号码'),
702 ('开票日期', '开票日期'),
703 ('不含税价', '不含税价'),
704 ('发票类型', '发票联'),
705 ('购方名称', '购买方名称'),
706 ('购买方身份证号或组织机构代码', '购买方证件号码'),
707 ('纳税人识别号', '纳税人识别号'), # nodo
708 ('车辆识别代码', '车架号'),
709 ('价税合计小写', '价税合计小写'),
710 ('销方名称', '销货单位名称'),
711 ('增值税税额', '增值税税额'),
712 ('增值税税率', '增值税税率'), # nodo
713 ('发票章有无', '发票章有无'), # nodo 全国统一发票监制章 销售单位章
714 ('价税合计大写', '价税合计大写'), # nodo
715 ('', None),
716 ('发动机号码', '发动机号'),
717 ('车辆类型', '车辆类型'), # nodo
718 ('厂牌型号', '厂牌型号'), # nodo
719 ('产地', '产地'), # nodo
720 ('合格证号', '合格证号'), # nodo
721 ('进口证明书号', '进口证明书号'), # nodo
722 ('商检单号', '商检单号'), # nodo
723 ('电话', '电话'), # nodo
724 ('销方纳税人识别号', '销货方纳税人识别号'),
725 ('账号', '账号'), # nodo
726 ('地址', '地址'), # nodo
727 ('开户银行', '开户银行'), # nodo
728 ('主管税务机关及代码', '主管税务机关及代码'), # nodo
729 ('吨位', '吨位'), # nodo
730 ('限乘人数', '限乘人数'),) # nodo
731 IC_PID = VAT_PID = MVC_PID = MVI_PID = None
732
733 # 营业执照
734 BL_CN_NAME = '营业执照'
735 BL_CLASSIFY = 31
736 BL_PID = 41
737 BL_FIELD_ORDER = (('注册号', '统一社会信用代码'),
738 ('企业名称', '名称'),
739 ('企业类型', '类型'),
740 ('经营者姓名', '法定代表人'),
741 ('成立日期', '成立日期'),
742 ('营业期限', '营业期限'),
743 ('注册资本', '注册资本'),
744 ('地址', '住所'),
745 ('经营范围', '经营范围'),)
746 # 二手车发票
747 UCI_CN_NAME = '二手车发票'
748 UCI_CLASSIFY = 1
749 UCI_PID = 60
750 UCI_FIELD_ORDER = (('发票代码', '发票代码'),
751 ('发票号码', '发票号码'),
752 ('开票日期', '开票日期'),
753 ('车价合计', '车价合计小写'),
754 ('发票联', '发票联'),
755 ('购方单位', '买方单位/个人'),
756 ('购方号码', '买方单位代码/身份证号码'),
757 ('车架号码', '车架号'),
758 ('车价合计大写', '车价合计大写'),
759 ('二手车市场', '二手车市场'),
760 ('发票章有无', '发票章有无'),
761 ('空行占位', None),
762 ('车牌照号', '车牌照号'),
763 ('登记证号', '登记证号'),
764 ('购方地址', '买方单位/住址'),
765 ('车辆类型', '车辆类型'),
766 ('厂牌型号', '厂牌型号'),
767 ('车管所名称', '转入地车辆管理所名称'),
768 ('销方名称', '卖方单位/个人'),
769 ('销方号码', '卖方单位代码/身份证号码'),
770 ('销方地址', '卖方单位/个人住址'),)
771 # 港澳台通行证
772 EEP_CN_NAME = '港澳台通行证'
773 EEP_CLASSIFY = 30
774 EEP_PID = 1018
775 EEP_FIELD_ORDER = (('中文名', '姓名'), # 英文名
776 ('证件号码', '证件号码'),
777 ('签发次数', '换证次数（签发次数）'),
778 ('有效期限', '有效期限'),
779 ('出生日期', '出生日期'),
780 ('性别', '性别'),
781 ('签发机关', '签发机关'),
782 ('签发地点', '签发地点'),)
783 # 行驶证
784 DL_CN_NAME = '行驶证'
785 DL_CLASSIFY = 32
786 DL_PID = 5
787 DL_FIELD_ORDER_0 = (('号牌号码', '1 号牌号码'),
788 ('所有人', '3 所有人'),
789 ('使用性质', '5 使用性质'),
790 ('车辆识别代码', '7 车辆识别代号'),
791 ('注册日期', '9 注册日期'),
792 ('发证日期', '10 发证日期'),
793 ('车辆类型', '2 车辆类型'),
794 ('地址', '4 住址'),
795 ('品牌型号', '6 品牌型号'),
796 ('发动机号', '8 发动机号码'),)
797 DL_FIELD_ORDER_1 = (('号牌号码', '1 号牌号码'),
798 ('档案编号', '11 档案编号'),
799 ('核定载人数', '12 核定载人数'),
800 ('总质量', '13 总质量'),
801 ('整备质量', '14 整备质量'),
802 ('核定载质量', '15 核对载质量'),
803 ('外廓尺寸', '16 外廓尺寸'),
804 ('准牵引总质量', '17 准牵引总质量'),)
805 # 护照
806 PP_CN_NAME = '护照'
807 PP_CLASSIFY = 3
808 PP_PID = 8
809 PP_FIELD_ORDER = (('类型', '类型/Type'),
810 ('英文姓名', '姓名/Name'),
811 ('护照号码', '护照号码/Passport No'),
812 ('有效期至', '有效期至/Date of expiry'),
813 ('签发日期', '签发日期/Date of issue'),
814 ('国家码', '国家码/Country Code'),
815 ('性别', '性别/Sex'),
816 ('国籍', '国籍/Nationality'),
817 ('出生日期', '出生日期/Date of birth'),
818 ('出生地点', '出生地点/Place of birth'),
819 ('签发地点', '签发地点/Place of issue'),)
820 # 银行卡
821 BC_CN_NAME = '银行卡'
822 BC_CLASSIFY = 37
823 BC_PID = 4
824 # BC_FIELD = (('CardNum', '银行卡号'),
825 # ('BankName', '发卡行名称'),
826 # ('CardName', '银行卡名称'),
827 # ('BankCode', '发卡行代号'),
828 # ('CardType', '银行卡类型'),
829 # ('Date', '日期'))
830 BC_FIELD_ORDER = (('BankName', '发卡行名称'),
831 ('CardNum', '银行卡号'),
832 ('CardType', '银行卡类型'),)
833
834 SUCCESS_CODE_SET = {'0', 0}
835
836 FIELD_ORDER_MAP = {
837 IC_CLASSIFY: ('有效期限', IC_FIELD_ORDER_1, IC_FIELD_ORDER_0),
838 RP_CLASSIFY: ('有效期限', RP_FIELD_ORDER_1, RP_FIELD_ORDER_0),
839 DL_CLASSIFY: ('档案编号', DL_FIELD_ORDER_1, DL_FIELD_ORDER_0),
840 MVC_CLASSIFY: ('转移登记日期', MVC_FIELD_ORDER_3_4, MVC_FIELD_ORDER_1_2),
841 MVC_CLASSIFY_SE: ('转移登记日期', MVC_SE_FIELD_ORDER_3_4, MVC_SE_FIELD_ORDER_1_2)
842 }
843
844 LICENSE_ORDER = ((MVI_CLASSIFY, (MVI_PID, MVI_CN_NAME, MVI_FIELD_ORDER, False, False)),
845 (IC_CLASSIFY, (IC_PID, IC_CN_NAME, None, True, False)),
846 (RP_CLASSIFY, (None, RP_CN_NAME, None, True, False)),
847 (BC_CLASSIFY, (BC_PID, BC_CN_NAME, BC_FIELD_ORDER, False, False)),
848 (BL_CLASSIFY, (BL_PID, BL_CN_NAME, BL_FIELD_ORDER, False, False)),
849 (UCI_CLASSIFY, (UCI_PID, UCI_CN_NAME, UCI_FIELD_ORDER, False, False)),
850 (EEP_CLASSIFY, (EEP_PID, EEP_CN_NAME, EEP_FIELD_ORDER, False, False)),
851 (DL_CLASSIFY, (DL_PID, DL_CN_NAME, None, True, False)),
852 (PP_CLASSIFY, (PP_PID, PP_CN_NAME, PP_FIELD_ORDER, False, False)),
853 (MVC_CLASSIFY, (MVC_PID, MVC_CN_NAME, None, True, True)),
854 (VAT_CLASSIFY, (VAT_PID, VAT_CN_NAME, VAT_FIELD_ORDER, False, False)))
855
856 LICENSE_CLASSIFY_MAPPING = dict(LICENSE_ORDER)
857
858 OTHER_CLASSIFY_SET = {OTHER_CLASSIFY}
859 LICENSE_CLASSIFY_SET_1 = {IC_CLASSIFY, VAT_CLASSIFY, MVC_CLASSIFY, MVI_CLASSIFY}
860 LICENSE_CLASSIFY_SET_2 = {BL_CLASSIFY, UCI_CLASSIFY, EEP_CLASSIFY, DL_CLASSIFY, PP_CLASSIFY, BC_CLASSIFY}
src/apps/doc/management/commands/doc_ocr_process.py
...@@ -440,6 +440,19 @@ class Command(BaseCommand, LoggerMixin): ...@@ -440,6 +440,19 @@ class Command(BaseCommand, LoggerMixin):
440 merged_bs_summary = {} 440 merged_bs_summary = {}
441 card_num = 1 441 card_num = 1
442 for role_dict in unknown_summary.values(): 442 for role_dict in unknown_summary.values():
443 if len(role_dict) == 2 and consts.UNKNOWN_ROLE in role_dict:
444 summary_dict = role_dict.pop(consts.UNKNOWN_ROLE, {})
445 for summary in role_dict.values():
446 summary_dict['confidence'].extend(summary['confidence'])
447 summary_dict['role'] = summary['role']
448 summary_dict['code'].extend(summary['code'])
449 summary_dict['print_time'].extend(summary['print_time'])
450 summary_dict['start_date'].extend(summary['start_date'])
451 summary_dict['end_date'].extend(summary['end_date'])
452 summary_dict['sheet'].extend(summary['sheet'])
453 card = '{0}_{1}'.format(consts.UNKNOWN_CARD, card_num)
454 merged_bs_summary[card] = summary_dict
455 else:
443 for summary in role_dict.values(): 456 for summary in role_dict.values():
444 card = '{0}_{1}'.format(consts.UNKNOWN_CARD, card_num) 457 card = '{0}_{1}'.format(consts.UNKNOWN_CARD, card_num)
445 card_num += 1 458 card_num += 1
...@@ -460,6 +473,7 @@ class Command(BaseCommand, LoggerMixin): ...@@ -460,6 +473,7 @@ class Command(BaseCommand, LoggerMixin):
460 for role, summary in classify_summary.items(): 473 for role, summary in classify_summary.items():
461 if one_card or role in card_summary['role_set']: 474 if one_card or role in card_summary['role_set']:
462 merge_role.append(role) 475 merge_role.append(role)
476 card_summary['confidence'].extend(summary['confidence'])
463 card_summary['sheet'].extend(summary['sheet']) 477 card_summary['sheet'].extend(summary['sheet'])
464 card_summary['code'].extend(summary['code']) 478 card_summary['code'].extend(summary['code'])
465 card_summary['print_time'].extend(summary['print_time']) 479 card_summary['print_time'].extend(summary['print_time'])
......
src/apps/doc/mixins.py
...@@ -18,7 +18,7 @@ class DocHandler: ...@@ -18,7 +18,7 @@ class DocHandler:
18 18
19 def get_doc_list(self, doc_queryset, business_type): 19 def get_doc_list(self, doc_queryset, business_type):
20 for doc_dict in doc_queryset: 20 for doc_dict in doc_queryset:
21 if doc_dict['status'] != DocStatus.COMPLETE.value: 21 if doc_dict['status'] not in [DocStatus.COMPLETE.value, DocStatus.UPLOAD_FAILED.value]:
22 continue 22 continue
23 doc_id = doc_dict.get('id') 23 doc_id = doc_dict.get('id')
24 doc_dict['pdf_link'] = self.get_link(doc_id, business_type) 24 doc_dict['pdf_link'] = self.get_link(doc_id, business_type)
......
src/apps/doc/ocr/wb.py
...@@ -30,51 +30,118 @@ class BSWorkbook(Workbook): ...@@ -30,51 +30,118 @@ class BSWorkbook(Workbook):
30 self.MAX_MEAN = 31 30 self.MAX_MEAN = 31
31 31
32 @staticmethod 32 @staticmethod
33 def sheet_prune(ws, classify): 33 def header_collect(ws, sheet_header_info, header_info, max_column_list, classify):
34 ws.insert_cols(1, amount=consts.FIXED_COL_AMOUNT) 34 # sheet_header_info = {
35 moved_col_set = set() 35 # 'sheet_name': {
36 header_col_set = set() 36 # 'summary_col': 1,
37 # 根据第一行关键词排列 37 # 'date_col': 1,
38 for col in range(consts.FIXED_COL_AMOUNT + 1, ws.max_column + 1): 38 # 'amount_col': 1,
39 header_value = ws.cell(1, col).value 39 # 'over_col': 1,
40 # 'income_col': 1,
41 # 'outlay_col': 1,
42 # 'borrow_col': 1,
43 # 'min_row': 2,
44 # 'find_count': 3,
45 # 'find_col': {1},
46 # 'header': ('日期', '金额')
47 # }
48 # }
49
50 # header_info = {
51 # 'summary_col': {
52 # 5: 2,
53 # 3: 1,
54 # },
55 # 'date_col': {},
56 # 'amount_col': {},
57 # 'over_col': {},
58 # 'income_col': {},
59 # 'outlay_col': {},
60 # 'borrow_col': {},
61 # }
62
63 # 第一行关键词
64 find_count = 0
65 for first_row in ws.iter_rows(max_row=1, min_row=1, values_only=True):
66 sheet_header_info.setdefault(ws.title, {}).setdefault(consts.HEADER_KEY, first_row)
67 for idx, header_value in enumerate(first_row):
68 if classify == consts.WECHART_CLASSIFY:
69 header_col = consts.WECHART_HEADERS_MAPPING.get(header_value)
70 else:
40 header_col = consts.HEADERS_MAPPING.get(header_value) 71 header_col = consts.HEADERS_MAPPING.get(header_value)
41 if header_col is not None and header_col not in header_col_set: 72 if header_col is not None:
42 letter = get_column_letter(col) 73 find_count += 1
43 ws.move_range("{0}1:{0}{1}".format(letter, ws.max_row), cols=header_col - col) 74 sheet_header_info.setdefault(ws.title, {}).setdefault(header_col, idx)
44 moved_col_set.add(col) 75 find_col_set = sheet_header_info.setdefault(ws.title, {}).setdefault(consts.FIND_COL_KEY, set())
45 header_col_set.add(header_col) 76 find_col_set.add(idx)
46 elif header_value in consts.BORROW_HEADERS_SET: 77 col_count = header_info.setdefault(header_col, {}).get(idx)
47 letter = get_column_letter(col) 78 header_info.setdefault(header_col, {})[idx] = 1 if col_count is None else col_count+1
48 ws.move_range("{0}1:{0}{1}".format(letter, ws.max_row), cols=consts.BORROW_HEADER_COL - col) 79
49 moved_col_set.add(col) 80 sheet_header_info.setdefault(ws.title, {}).setdefault(consts.FIND_COUNT_KEY, find_count)
50 header_col_set.add(consts.BORROW_HEADER_COL) 81 min_row = 1 if find_count == 0 else 2
51 elif header_value in consts.INCOME_HEADERS_SET: 82 sheet_header_info.setdefault(ws.title, {}).setdefault(consts.MIN_ROW_KEY, min_row)
52 letter = get_column_letter(col) 83 max_column_list.append(ws.max_column)
53 ws.move_range("{0}1:{0}{1}".format(letter, ws.max_row), cols=consts.INCOME_HEADER_COL - col) 84
54 moved_col_set.add(col) 85 @staticmethod
55 header_col_set.add(consts.INCOME_HEADER_COL) 86 def header_statistics(sheet_header_info, header_info, classify):
56 elif header_value in consts.OUTLAY_HEADERS_SET: 87 # statistics_header_info = {
57 letter = get_column_letter(col) 88 # SUMMARY_KEY: 2,
58 ws.move_range("{0}1:{0}{1}".format(letter, ws.max_row), cols=consts.OUTLAY_HEADER_COL - col) 89 # DATE_KEY: 3,
59 moved_col_set.add(col) 90 # AMOUNT_KEY: 4,
60 header_col_set.add(consts.OUTLAY_HEADER_COL) 91 # OVER_KEY: 5,
61 92 # IMCOME_KEY: 6,
62 # 缺失表头再次查找 93 # OUTLAY_KEY: 7,
63 for header_col in range(1, consts.FIXED_COL_AMOUNT + 1): 94 # BORROW_KEY: 8,
64 if header_col in header_col_set or header_col == consts.RESULT_HEADER_COL: 95 # 'header': ('日期', '金额')
65 continue 96 # }
66 fix_col = consts.CLASSIFY_LIST[classify][1][header_col - 1] 97 statistics_header_info = {}
67 if fix_col is None: 98 sheet_order_list = sorted(sheet_header_info, reverse=True,
99 key=lambda x: sheet_header_info[x][consts.FIND_COUNT_KEY])
100 best_sheet_info = sheet_header_info.get(sheet_order_list[0])
101 if best_sheet_info.get(consts.FIND_COUNT_KEY, 0) == 0:
102 for key, value in consts.CLASSIFY_MAP.items():
103 col = consts.CLASSIFY_LIST[classify][1][value]
104 statistics_header_info[key] = col - 1 if isinstance(col, int) else None
105 statistics_header_info[consts.HEADER_KEY] = consts.CLASSIFY_HEADER_LIST[classify]
106 else:
107 find_col_set = best_sheet_info.get(consts.FIND_COL_KEY, set())
108 # SUMMARY_KEY DATE_KEY OVER_KEY BORROW_KEY
109 for key in consts.KEY_LIST:
110 col = best_sheet_info.get(key)
111 if col is None:
112 col_dict = header_info.get(key, {})
113 for idx in sorted(col_dict, key=lambda x: col_dict[x], reverse=True):
114 if idx in find_col_set:
68 continue 115 continue
69 fix_col = fix_col + consts.FIXED_COL_AMOUNT 116 col = idx
70 if fix_col in moved_col_set: 117 find_col_set.add(col)
71 break 118 break
72 letter = get_column_letter(fix_col) 119 else:
73 ws.move_range("{0}1:{0}{1}".format(letter, ws.max_row), cols=header_col - fix_col) 120 fixed_col = consts.CLASSIFY_LIST[classify][1][consts.CLASSIFY_MAP[key]]
121 if fixed_col not in find_col_set and isinstance(fixed_col, int):
122 col = fixed_col - 1
123 find_col_set.add(col)
124 statistics_header_info[key] = col
125 statistics_header_info[consts.HEADER_KEY] = best_sheet_info.get(consts.HEADER_KEY)
126 return statistics_header_info
74 127
75 ws.delete_cols(consts.FIXED_COL_AMOUNT + 1, amount=ws.max_column) 128 @staticmethod
76 min_row = 1 if len(moved_col_set) == 0 else 2 129 def get_data_col_min_row(sheet, sheet_header_info, header_info, classify):
77 return min_row 130 date_col = sheet_header_info.get(sheet, {}).get(consts.DATE_KEY)
131 if date_col is None:
132 date_col_dict = header_info.get(consts.DATE_KEY, {})
133 find_col_set = sheet_header_info.get(sheet, {}).get(consts.FIND_COL_KEY, set())
134 for idx in sorted(date_col_dict, key=lambda x: date_col_dict[x], reverse=True):
135 if idx in find_col_set:
136 continue
137 date_col = idx
138 break
139 else:
140 fixed_col = consts.CLASSIFY_LIST[classify][1][consts.CLASSIFY_MAP[consts.DATE_KEY]]
141 if fixed_col not in find_col_set and isinstance(fixed_col, int):
142 date_col = fixed_col - 1
143 min_row = sheet_header_info.get(sheet, {}).get(consts.MIN_ROW_KEY, 2)
144 return date_col, min_row
78 145
79 @staticmethod 146 @staticmethod
80 def month_split(dti, date_list, date_statistics): 147 def month_split(dti, date_list, date_statistics):
...@@ -122,8 +189,14 @@ class BSWorkbook(Workbook): ...@@ -122,8 +189,14 @@ class BSWorkbook(Workbook):
122 reverse_trend = -1 189 reverse_trend = -1
123 return reverse_trend 190 return reverse_trend
124 191
125 def sheet_split(self, ws, month_mapping, reverse_trend_list, min_row, date_list, date_statistics): 192 def sheet_split(self, ws, date_col, min_row, month_mapping, reverse_trend_list, date_list, date_statistics):
126 for date_tuple_src in ws.iter_cols(min_col=1, max_col=1, min_row=min_row, values_only=True): 193 if date_col is None:
194 # month_info process
195 month_info = month_mapping.setdefault('xxxx-xx', [])
196 month_info.append((ws.title, min_row, ws.max_row, 0))
197 return
198 date_col = date_col + 1
199 for date_tuple_src in ws.iter_cols(min_col=date_col, max_col=date_col, min_row=min_row, values_only=True):
127 date_tuple = [date[:10] if isinstance(date, str) else date for date in date_tuple_src] 200 date_tuple = [date[:10] if isinstance(date, str) else date for date in date_tuple_src]
128 dt_array, tz_parsed = tslib.array_to_datetime( 201 dt_array, tz_parsed = tslib.array_to_datetime(
129 np.array(date_tuple, copy=False, dtype=np.object_), 202 np.array(date_tuple, copy=False, dtype=np.object_),
...@@ -199,11 +272,12 @@ class BSWorkbook(Workbook): ...@@ -199,11 +272,12 @@ class BSWorkbook(Workbook):
199 return amount_str 272 return amount_str
200 # 1.替换 273 # 1.替换
201 res_str = amount_str.translate(consts.TRANS) 274 res_str = amount_str.translate(consts.TRANS)
202 # 2.删除多余的- 275 # 2.首字符处理
203 res_str = res_str[0] + res_str[1:].replace('-', '') 276 first_char = res_str[0]
204 # 3.首字符处理 277 if first_char in consts.ERROR_CHARS:
205 if res_str[0] in consts.ERROR_CHARS: 278 first_char = '-'
206 res_str = '-{0}'.format(res_str[1:]) 279 # 3.删除多余的-
280 res_str = first_char + res_str[1:].replace('-', '')
207 # 4.逗号与句号处理 281 # 4.逗号与句号处理
208 if len(res_str) >= 4: 282 if len(res_str) >= 4:
209 period_idx = len(res_str) - 3 283 period_idx = len(res_str) - 3
...@@ -213,90 +287,132 @@ class BSWorkbook(Workbook): ...@@ -213,90 +287,132 @@ class BSWorkbook(Workbook):
213 res_str = '{0}.{1}'.format(res_str[:period_idx], res_str[period_idx + 1:]) 287 res_str = '{0}.{1}'.format(res_str[:period_idx], res_str[period_idx + 1:])
214 return res_str 288 return res_str
215 289
216 def build_month_sheet(self, card, month_mapping, ms, is_reverse): 290 def build_month_sheet(self, ms, card, month_mapping, is_reverse, statistics_header_info, max_column):
291 summary_cell_idx = statistics_header_info.get(consts.SUMMARY_KEY)
292 date_cell_idx = statistics_header_info.get(consts.DATE_KEY)
293 amount_cell_idx = statistics_header_info.get(consts.AMOUNT_KEY) # None or src or append
294 over_cell_idx = statistics_header_info.get(consts.OVER_KEY)
295 income_cell_idx = statistics_header_info.get(consts.IMCOME_KEY)
296 outlay_cell_idx = statistics_header_info.get(consts.OUTLAY_KEY)
297 borrow_cell_idx = statistics_header_info.get(consts.BORROW_KEY)
298 header = list(statistics_header_info.get(consts.HEADER_KEY))
299 src_header_len = len(header)
300 if max_column > src_header_len:
301 for i in range(max_column - src_header_len):
302 header.append(None)
303
304 add_col = ['核对结果']
305 if amount_cell_idx is None:
306 if income_cell_idx is not None or outlay_cell_idx is not None:
307 add_col = ['金额', '核对结果']
308 amount_cell_idx = len(header)
309 header.extend(add_col)
310 result_idx = len(header) - 1
311
217 tmp_ws = self.create_sheet('tmp_ws') 312 tmp_ws = self.create_sheet('tmp_ws')
218 for month in sorted(month_mapping.keys()): 313 for month in sorted(month_mapping.keys()):
219 # 3.1.拷贝数据 314 # 3.1.拷贝数据
220 parts = month_mapping.get(month) 315 parts = month_mapping.get(month)
221 new_ws = self.create_sheet('{0}({1})'.format(month, card[-6:])) 316 new_ws = self.create_sheet('{0}({1})'.format(month, card[-6:]))
222 new_ws.append(consts.FIXED_HEADERS) 317 new_ws.append(header)
223 for part in parts: 318 for part in parts:
224 ws = self.get_sheet_by_name(part[0]) 319 ws = self.get_sheet_by_name(part[0])
225 for row_value in ws.iter_rows(min_row=part[1], max_row=part[2], values_only=True): 320 for row_value in ws.iter_rows(min_row=part[1], max_row=part[2], values_only=True):
321 if any(row_value):
226 new_ws.append(row_value) 322 new_ws.append(row_value)
227 # 3.2.提取信息、高亮 323 # 3.2.提取信息、高亮
228 amount_mapping = {} 324 amount_mapping = {}
229 amount_fill_row = set() 325 amount_fill_row = set()
326
230 for rows in new_ws.iter_rows(min_row=2): 327 for rows in new_ws.iter_rows(min_row=2):
231 summary_cell = rows[consts.SUMMARY_IDX] 328 # TODO 删除空行
232 date_cell = rows[consts.DATE_IDX] 329 summary_cell = None if summary_cell_idx is None else rows[summary_cell_idx]
233 amount_cell = rows[consts.AMOUNT_IDX] 330 date_cell = None if date_cell_idx is None else rows[date_cell_idx]
234 row = summary_cell.row 331 amount_cell = None if amount_cell_idx is None else rows[amount_cell_idx]
332 over_cell = None if over_cell_idx is None else rows[over_cell_idx]
333 income_cell = None if income_cell_idx is None else rows[income_cell_idx]
334 outlay_cell = None if outlay_cell_idx is None else rows[outlay_cell_idx]
335 borrow_cell = None if borrow_cell_idx is None else rows[borrow_cell_idx]
336
337 summary_cell_value = None if summary_cell is None else summary_cell.value
338 date_cell_value = None if date_cell is None else date_cell.value
339 amount_cell_value = None if amount_cell is None else amount_cell.value
340 over_cell_value = None if over_cell is None else over_cell.value
341 income_cell_value = None if income_cell is None else income_cell.value
342 outlay_cell_value = None if outlay_cell is None else outlay_cell.value
343 borrow_cell_value = None if borrow_cell is None else borrow_cell.value
344
345 # row = summary_cell.row
346 if summary_cell is not None:
235 # 关键词1提取 347 # 关键词1提取
236 if summary_cell.value in self.interest_keyword: 348 if summary_cell_value in self.interest_keyword:
237 ms.append((summary_cell.value, date_cell.value, amount_cell.value)) 349 ms.append((summary_cell_value, date_cell_value, amount_cell_value))
238 # 关键词2提取至临时表 350 # 关键词2提取至临时表
239 elif summary_cell.value in self.salary_keyword: 351 elif summary_cell_value in self.salary_keyword:
240 tmp_ws.append((summary_cell.value, date_cell.value, amount_cell.value)) 352 tmp_ws.append((summary_cell_value, date_cell_value, amount_cell_value))
241 # 贷款关键词高亮 353 # 贷款关键词高亮
242 elif summary_cell.value in self.loan_keyword: 354 elif summary_cell_value in self.loan_keyword:
243 summary_cell.fill = self.loan_fill 355 summary_cell.fill = self.loan_fill
244 356
245 amount_error = False
246 # 3.3.余额转数值 357 # 3.3.余额转数值
247 over_cell = rows[consts.OVER_IDX] 358 over_success = False
359 if over_cell is not None:
248 try: 360 try:
249 over_cell.value = locale.atof(self.amount_format(over_cell.value)) 361 over_cell.value = locale.atof(self.amount_format(over_cell_value))
250 except Exception as e: 362 except Exception as e:
251 amount_error = True 363 pass
252 else: 364 else:
365 over_success = True
253 over_cell.number_format = numbers.FORMAT_NUMBER_00 366 over_cell.number_format = numbers.FORMAT_NUMBER_00
254 367
255 # 3.4.金额转数值 368 # 3.4.金额转数值
369 amount_success = False
370 if amount_cell is not None:
256 try: 371 try:
257 try: 372 try:
258 amount_cell.value = locale.atof(self.amount_format(amount_cell.value)) 373 amount_cell.value = locale.atof(self.amount_format(amount_cell_value))
259 except Exception as e: 374 except Exception as e:
260 try: 375 try:
261 amount_cell.value = locale.atof(self.amount_format(rows[consts.INCOME_IDX].value)) 376 amount_cell.value = locale.atof(self.amount_format(income_cell_value))
262 if amount_cell.value == 0: 377 if amount_cell.value == 0:
263 raise 378 raise
264 elif amount_cell.value < 0: 379 elif amount_cell.value < 0:
265 amount_cell.value = -amount_cell.value 380 amount_cell.value = -amount_cell.value
266 except Exception as e: 381 except Exception as e:
267 amount_cell.value = locale.atof(self.amount_format(rows[consts.OUTLAY_IDX].value)) 382 amount_cell.value = locale.atof(self.amount_format(outlay_cell_value))
268 if amount_cell.value > 0: 383 if amount_cell.value > 0:
269 amount_cell.value = -amount_cell.value 384 amount_cell.value = -amount_cell.value
270 except Exception as e: 385 except Exception as e:
271 amount_error = True 386 pass
272 else: 387 else:
273 if rows[consts.BORROW_IDX].value in consts.BORROW_OUTLAY_SET: 388 amount_success = True
389 if borrow_cell_value in consts.BORROW_OUTLAY_SET:
274 amount_cell.value = -amount_cell.value 390 amount_cell.value = -amount_cell.value
275 amount_cell.number_format = numbers.FORMAT_NUMBER_00 391 amount_cell.number_format = numbers.FORMAT_NUMBER_00
392 if date_cell is not None:
276 same_amount_mapping = amount_mapping.get(date_cell.value, {}) 393 same_amount_mapping = amount_mapping.get(date_cell.value, {})
277 fill_rows = same_amount_mapping.get(-amount_cell.value) 394 fill_rows = same_amount_mapping.get(-amount_cell.value)
278 if fill_rows: 395 if fill_rows:
279 amount_fill_row.add(row) 396 amount_fill_row.add(amount_cell.row)
280 amount_fill_row.update(fill_rows) 397 amount_fill_row.update(fill_rows)
281 amount_mapping.setdefault(date_cell.value, {}).setdefault( 398 amount_mapping.setdefault(date_cell.value, {}).setdefault(
282 amount_cell.value, []).append(row) 399 amount_cell.value, []).append(amount_cell.row)
283 400
284 # 3.5.核对结果 401 # 3.5.核对结果
285 if row > 2 and not amount_error: 402 if amount_success and over_success and amount_cell.row > 2:
403 amount_col_letter = get_column_letter(amount_cell_idx + 1)
404 over_col_letter = get_column_letter(over_cell_idx + 1)
286 if is_reverse: 405 if is_reverse:
287 rows[consts.RESULT_IDX].value = '=IF(D{0}=ROUND(SUM(D{1},C{0}),2), "{2}", "{3}")'.format( 406 rows[result_idx].value = '=IF({2}{0}=ROUND(SUM({2}{1},{3}{0}),4), "{4}", "{5}")'.format(
288 row - 1, row, *self.proof_res) 407 amount_cell.row - 1, amount_cell.row, over_col_letter, amount_col_letter, *self.proof_res)
289 else: 408 else:
290 rows[consts.RESULT_IDX].value = '=IF(D{0}=ROUND(SUM(D{1},C{0}),2), "{2}", "{3}")'.format( 409 rows[result_idx].value = '=IF({2}{0}=ROUND(SUM({2}{1},{3}{0}),4), "{4}", "{5}")'.format(
291 row, row - 1, *self.proof_res) 410 amount_cell.row, amount_cell.row - 1, over_col_letter, amount_col_letter, *self.proof_res)
292
293 # 删除金额辅助列
294 new_ws.delete_cols(consts.BORROW_HEADER_COL, amount=new_ws.max_column)
295 411
296 # 3.6.同一天相同进出账高亮 412 # 3.6.同一天相同进出账高亮
297 del amount_mapping 413 del amount_mapping
298 for row in amount_fill_row: 414 for row in amount_fill_row:
299 new_ws[row][consts.AMOUNT_IDX].fill = self.amount_fill 415 new_ws[row][amount_cell_idx].fill = self.amount_fill
300 416
301 # 关键词2信息提取 417 # 关键词2信息提取
302 ms.append(self.blank_row) 418 ms.append(self.blank_row)
...@@ -319,21 +435,29 @@ class BSWorkbook(Workbook): ...@@ -319,21 +435,29 @@ class BSWorkbook(Workbook):
319 # } 435 # }
320 # } 436 # }
321 for card, summary in bs_summary.items(): 437 for card, summary in bs_summary.items():
322 # 1.原表修剪、排列、按照月份分割 438 # 1.原表表头收集、按照月份分割
439 # 1.1 总结首行信息
440 classify = summary.get('classify', 0)
441 sheet_header_info = {}
442 header_info = {}
443 max_column_list = []
444 for sheet in summary.get('sheet', []):
445 ws = self.get_sheet_by_name(sheet)
446 self.header_collect(ws, sheet_header_info, header_info, max_column_list, classify)
447 statistics_header_info = self.header_statistics(sheet_header_info, header_info, classify)
448 max_column = max(max_column_list)
449
450 # 1.2.按月份分割 min_row 正文第一行 date_col 日期行
323 start_date = summary.get('start_date') 451 start_date = summary.get('start_date')
324 end_date = summary.get('end_date') 452 end_date = summary.get('end_date')
325 date_statistics = False 453 date_statistics = True if start_date is None or end_date is None else False # 用于判断是否需要收集各表中日期
326 if start_date is None or end_date is None: 454 date_list = [] # 用于收集各表中日期
327 date_statistics = True 455 month_mapping = {} # 用于创建月份表
328 date_list = [] 456 reverse_trend_list = [] # 用于判断倒序与正序
329 month_mapping = {}
330 reverse_trend_list = []
331 for sheet in summary.get('sheet', []): 457 for sheet in summary.get('sheet', []):
332 ws = self.get_sheet_by_name(sheet) 458 ws = self.get_sheet_by_name(sheet)
333 # 1.1.删除多余列、排列 459 date_col, min_row = self.get_data_col_min_row(sheet, sheet_header_info, header_info, classify)
334 min_row = self.sheet_prune(ws, summary.get('classify', 0)) 460 self.sheet_split(ws, date_col, min_row, month_mapping, reverse_trend_list, date_list, date_statistics)
335 # 1.2.按月份分割
336 self.sheet_split(ws, month_mapping, reverse_trend_list, min_row, date_list, date_statistics)
337 461
338 if date_statistics is True and len(date_list) > 1: 462 if date_statistics is True and len(date_list) > 1:
339 start_date = min(date_list) if start_date is None else start_date 463 start_date = min(date_list) if start_date is None else start_date
...@@ -353,7 +477,7 @@ class BSWorkbook(Workbook): ...@@ -353,7 +477,7 @@ class BSWorkbook(Workbook):
353 for month_list in month_mapping.values(): 477 for month_list in month_mapping.values():
354 month_list.sort(key=lambda x: x[-1], reverse=is_reverse) 478 month_list.sort(key=lambda x: x[-1], reverse=is_reverse)
355 479
356 self.build_month_sheet(card, month_mapping, ms, is_reverse) 480 self.build_month_sheet(ms, card, month_mapping, is_reverse, statistics_header_info, max_column)
357 481
358 # 4.删除原表 482 # 4.删除原表
359 for sheet in summary.get('sheet'): 483 for sheet in summary.get('sheet'):
......
src/apps/doc/ocr/wb_bak.py 0 → 100644
1 import locale
2 import numpy as np
3 from pandas._libs import tslib
4 from pandas._libs.tslibs.nattype import NaTType
5 from pandas.core.indexes.datetimes import DatetimeIndex
6 from openpyxl import Workbook
7 from openpyxl.styles import Border, Side, PatternFill, numbers
8 from openpyxl.utils import get_column_letter
9 from apps.doc import consts
10
11
12 class BSWorkbook(Workbook):
13
14 def __init__(self, interest_keyword, salary_keyword, loan_keyword, *args, **kwargs):
15 super().__init__(*args, **kwargs)
16 locale.setlocale(locale.LC_NUMERIC, 'en_US.UTF-8')
17 self.meta_sheet_title = '关键信息提取和展示'
18 self.blank_row = (None,)
19 self.code_header = ('页数', '电子回单验证码')
20 self.date_header = ('打印时间', '起始日期', '终止日期', '流水区间结果')
21 self.keyword_header = ('关键词', '记账日期', '金额')
22 self.interest_keyword = interest_keyword
23 self.salary_keyword = salary_keyword
24 self.loan_keyword = loan_keyword
25 self.proof_res = ('对', '错')
26 self.loan_fill = PatternFill("solid", fgColor="00FFCC00")
27 self.amount_fill = PatternFill("solid", fgColor="00FFFF00")
28 # self.bd = Side(style='thin', color="000000")
29 # self.border = Border(left=self.bd, top=self.bd, right=self.bd, bottom=self.bd)
30 self.MAX_MEAN = 31
31
32 @staticmethod
33 def sheet_prune(ws, classify):
34 ws.insert_cols(1, amount=consts.FIXED_COL_AMOUNT)
35 moved_col_set = set()
36 header_col_set = set()
37 # 根据第一行关键词排列
38 for col in range(consts.FIXED_COL_AMOUNT + 1, ws.max_column + 1):
39 header_value = ws.cell(1, col).value
40 header_col = consts.HEADERS_MAPPING.get(header_value)
41 if header_col is not None and header_col not in header_col_set:
42 letter = get_column_letter(col)
43 ws.move_range("{0}1:{0}{1}".format(letter, ws.max_row), cols=header_col - col)
44 moved_col_set.add(col)
45 header_col_set.add(header_col)
46 elif header_value in consts.BORROW_HEADERS_SET:
47 letter = get_column_letter(col)
48 ws.move_range("{0}1:{0}{1}".format(letter, ws.max_row), cols=consts.BORROW_HEADER_COL - col)
49 moved_col_set.add(col)
50 header_col_set.add(consts.BORROW_HEADER_COL)
51 elif header_value in consts.INCOME_HEADERS_SET:
52 letter = get_column_letter(col)
53 ws.move_range("{0}1:{0}{1}".format(letter, ws.max_row), cols=consts.INCOME_HEADER_COL - col)
54 moved_col_set.add(col)
55 header_col_set.add(consts.INCOME_HEADER_COL)
56 elif header_value in consts.OUTLAY_HEADERS_SET:
57 letter = get_column_letter(col)
58 ws.move_range("{0}1:{0}{1}".format(letter, ws.max_row), cols=consts.OUTLAY_HEADER_COL - col)
59 moved_col_set.add(col)
60 header_col_set.add(consts.OUTLAY_HEADER_COL)
61
62 # 缺失表头再次查找
63 for header_col in range(1, consts.FIXED_COL_AMOUNT + 1):
64 if header_col in header_col_set or header_col == consts.RESULT_HEADER_COL:
65 continue
66 fix_col = consts.CLASSIFY_LIST[classify][1][header_col - 1]
67 if fix_col is None:
68 continue
69 fix_col = fix_col + consts.FIXED_COL_AMOUNT
70 if fix_col in moved_col_set:
71 break
72 letter = get_column_letter(fix_col)
73 ws.move_range("{0}1:{0}{1}".format(letter, ws.max_row), cols=header_col - fix_col)
74
75 ws.delete_cols(consts.FIXED_COL_AMOUNT + 1, amount=ws.max_column)
76 min_row = 1 if len(moved_col_set) == 0 else 2
77 return min_row
78
79 @staticmethod
80 def month_split(dti, date_list, date_statistics):
81 month_list = []
82 idx_list = []
83 month_pre = None
84 for idx, month_str in enumerate(dti.strftime('%Y-%m')):
85 if isinstance(month_str, float):
86 continue
87 if month_str != month_pre:
88 month_list.append(month_str)
89 if month_pre is None:
90 if date_statistics:
91 date_list.append(dti[idx].date())
92 idx = 0
93 idx_list.append(idx)
94 month_pre = month_str
95 if date_statistics:
96 for idx in range(len(dti) - 1, -1, -1):
97 if isinstance(dti[idx], NaTType):
98 continue
99 date_list.append(dti[idx].date())
100 break
101 return month_list, idx_list
102
103 @staticmethod
104 def get_reverse_trend(day_idx, idx_list):
105 reverse_trend = 0
106 pre_day = None
107 for idx, day in enumerate(day_idx):
108 if np.isnan(day):
109 continue
110 if idx in idx_list or pre_day is None:
111 pre_day = day
112 continue
113 if day < pre_day:
114 reverse_trend += 1
115 pre_day = day
116 elif day > pre_day:
117 reverse_trend -= 1
118 pre_day = day
119 if reverse_trend > 0:
120 reverse_trend = 1
121 elif reverse_trend < 0:
122 reverse_trend = -1
123 return reverse_trend
124
125 def sheet_split(self, ws, month_mapping, reverse_trend_list, min_row, date_list, date_statistics):
126 for date_tuple_src in ws.iter_cols(min_col=1, max_col=1, min_row=min_row, values_only=True):
127 date_tuple = [date[:10] if isinstance(date, str) else date for date in date_tuple_src]
128 dt_array, tz_parsed = tslib.array_to_datetime(
129 np.array(date_tuple, copy=False, dtype=np.object_),
130 errors="coerce",
131 utc=False,
132 dayfirst=False,
133 yearfirst=False,
134 require_iso8601=True,
135 )
136 dti = DatetimeIndex(dt_array, tz=None, name=None)
137
138 month_list, idx_list = self.month_split(dti, date_list, date_statistics)
139
140 if len(month_list) == 0:
141 # month_info process
142 month_info = month_mapping.setdefault('xxxx-xx', [])
143 month_info.append((ws.title, min_row, ws.max_row, 0))
144 else:
145 # reverse_trend_list process
146 reverse_trend = self.get_reverse_trend(dti.day, idx_list)
147 reverse_trend_list.append(reverse_trend)
148 # month_info process
149 day_idx = dti.day
150 idx_list_max_idx = len(idx_list) - 1
151 for i, item in enumerate(month_list):
152 if i == idx_list_max_idx:
153 day_mean = np.mean(day_idx[idx_list[i]:].dropna())
154 month_mapping.setdefault(item, []).append(
155 (ws.title, idx_list[i] + min_row, ws.max_row, day_mean))
156 else:
157 day_mean = np.mean(day_idx[idx_list[i]: idx_list[i + 1]].dropna())
158 month_mapping.setdefault(item, []).append(
159 (ws.title, idx_list[i] + min_row, idx_list[i + 1] + min_row - 1, day_mean))
160
161 def build_metadata_rows(self, confidence, code, print_time, start_date, end_date):
162 if start_date is None or end_date is None:
163 timedelta = None
164 else:
165 timedelta = (end_date - start_date).days
166 metadata_rows = [
167 ('流水识别置信度', confidence),
168 self.blank_row,
169 self.code_header,
170 ]
171 metadata_rows.extend(code)
172 metadata_rows.extend(
173 [self.blank_row,
174 self.date_header,
175 (print_time, start_date, end_date, timedelta),
176 self.blank_row,
177 self.keyword_header]
178 )
179 return metadata_rows
180
181 def create_meta_sheet(self, card):
182 if self.worksheets[0].title == 'Sheet':
183 ms = self.worksheets[0]
184 ms.title = '{0}({1})'.format(self.meta_sheet_title, card[-6:])
185 else:
186 ms = self.create_sheet('{0}({1})'.format(self.meta_sheet_title, card[-6:]))
187 return ms
188
189 def build_meta_sheet(self, card, confidence, code, print_time, start_date, end_date):
190 metadata_rows = self.build_metadata_rows(confidence, code, print_time, start_date, end_date)
191 ms = self.create_meta_sheet(card)
192 for row in metadata_rows:
193 ms.append(row)
194 return ms
195
196 @staticmethod
197 def amount_format(amount_str):
198 if not isinstance(amount_str, str) or amount_str == '':
199 return amount_str
200 # 1.替换
201 res_str = amount_str.translate(consts.TRANS)
202 # 2.首字符处理
203 first_char = res_str[0]
204 if first_char in consts.ERROR_CHARS:
205 first_char = '-'
206 # 3.删除多余的-
207 res_str = first_char + res_str[1:].replace('-', '')
208 # 4.逗号与句号处理
209 if len(res_str) >= 4:
210 period_idx = len(res_str) - 3
211 if res_str[period_idx] == '.' and res_str[period_idx - 1] == ',':
212 res_str = '{0}{1}'.format(res_str[:period_idx - 1], res_str[period_idx:])
213 elif res_str[period_idx] == ',':
214 res_str = '{0}.{1}'.format(res_str[:period_idx], res_str[period_idx + 1:])
215 return res_str
216
217 def build_month_sheet(self, card, month_mapping, ms, is_reverse):
218 tmp_ws = self.create_sheet('tmp_ws')
219 for month in sorted(month_mapping.keys()):
220 # 3.1.拷贝数据
221 parts = month_mapping.get(month)
222 new_ws = self.create_sheet('{0}({1})'.format(month, card[-6:]))
223 new_ws.append(consts.FIXED_HEADERS)
224 for part in parts:
225 ws = self.get_sheet_by_name(part[0])
226 for row_value in ws.iter_rows(min_row=part[1], max_row=part[2], values_only=True):
227 new_ws.append(row_value)
228 # 3.2.提取信息、高亮
229 amount_mapping = {}
230 amount_fill_row = set()
231 for rows in new_ws.iter_rows(min_row=2):
232 summary_cell = rows[consts.SUMMARY_IDX]
233 date_cell = rows[consts.DATE_IDX]
234 amount_cell = rows[consts.AMOUNT_IDX]
235 row = summary_cell.row
236 # 关键词1提取
237 if summary_cell.value in self.interest_keyword:
238 ms.append((summary_cell.value, date_cell.value, amount_cell.value))
239 # 关键词2提取至临时表
240 elif summary_cell.value in self.salary_keyword:
241 tmp_ws.append((summary_cell.value, date_cell.value, amount_cell.value))
242 # 贷款关键词高亮
243 elif summary_cell.value in self.loan_keyword:
244 summary_cell.fill = self.loan_fill
245
246 amount_error = False
247 # 3.3.余额转数值
248 over_cell = rows[consts.OVER_IDX]
249 try:
250 over_cell.value = locale.atof(self.amount_format(over_cell.value))
251 except Exception as e:
252 amount_error = True
253 else:
254 over_cell.number_format = numbers.FORMAT_NUMBER_00
255
256 # 3.4.金额转数值
257 try:
258 try:
259 amount_cell.value = locale.atof(self.amount_format(amount_cell.value))
260 except Exception as e:
261 try:
262 amount_cell.value = locale.atof(self.amount_format(rows[consts.INCOME_IDX].value))
263 if amount_cell.value == 0:
264 raise
265 elif amount_cell.value < 0:
266 amount_cell.value = -amount_cell.value
267 except Exception as e:
268 amount_cell.value = locale.atof(self.amount_format(rows[consts.OUTLAY_IDX].value))
269 if amount_cell.value > 0:
270 amount_cell.value = -amount_cell.value
271 except Exception as e:
272 amount_error = True
273 else:
274 if rows[consts.BORROW_IDX].value in consts.BORROW_OUTLAY_SET:
275 amount_cell.value = -amount_cell.value
276 amount_cell.number_format = numbers.FORMAT_NUMBER_00
277 same_amount_mapping = amount_mapping.get(date_cell.value, {})
278 fill_rows = same_amount_mapping.get(-amount_cell.value)
279 if fill_rows:
280 amount_fill_row.add(row)
281 amount_fill_row.update(fill_rows)
282 amount_mapping.setdefault(date_cell.value, {}).setdefault(
283 amount_cell.value, []).append(row)
284
285 # 3.5.核对结果
286 if row > 2 and not amount_error:
287 if is_reverse:
288 rows[consts.RESULT_IDX].value = '=IF(D{0}=ROUND(SUM(D{1},C{0}),2), "{2}", "{3}")'.format(
289 row - 1, row, *self.proof_res)
290 else:
291 rows[consts.RESULT_IDX].value = '=IF(D{0}=ROUND(SUM(D{1},C{0}),2), "{2}", "{3}")'.format(
292 row, row - 1, *self.proof_res)
293
294 # 删除金额辅助列
295 new_ws.delete_cols(consts.BORROW_HEADER_COL, amount=new_ws.max_column)
296
297 # 3.6.同一天相同进出账高亮
298 del amount_mapping
299 for row in amount_fill_row:
300 new_ws[row][consts.AMOUNT_IDX].fill = self.amount_fill
301
302 # 关键词2信息提取
303 ms.append(self.blank_row)
304 ms.append(self.keyword_header)
305 for row in tmp_ws.iter_rows(values_only=True):
306 ms.append(row)
307 self.remove(tmp_ws)
308
309 def bs_rebuild(self, bs_summary):
310 # bs_summary = {
311 # '卡号': {
312 # 'classify': 0,
313 # 'confidence': 0.9,
314 # 'role': '柳雪',
315 # 'code': [('page', 'code')],
316 # 'print_time': 'datetime',
317 # 'start_date': 'datetime',
318 # 'end_date': 'datetime',
319 # 'sheet': ['sheet_name']
320 # }
321 # }
322 for card, summary in bs_summary.items():
323 # 1.原表修剪、排列、按照月份分割
324 start_date = summary.get('start_date')
325 end_date = summary.get('end_date')
326 date_statistics = False
327 if start_date is None or end_date is None:
328 date_statistics = True
329 date_list = []
330 month_mapping = {}
331 reverse_trend_list = []
332 for sheet in summary.get('sheet', []):
333 ws = self.get_sheet_by_name(sheet)
334 # 1.1.删除多余列、排列
335 min_row = self.sheet_prune(ws, summary.get('classify', 0))
336 # 1.2.按月份分割
337 self.sheet_split(ws, month_mapping, reverse_trend_list, min_row, date_list, date_statistics)
338
339 if date_statistics is True and len(date_list) > 1:
340 start_date = min(date_list) if start_date is None else start_date
341 end_date = max(date_list) if end_date is None else end_date
342
343 # 2.元信息提取表
344 ms = self.build_meta_sheet(card,
345 summary.get('confidence', 1),
346 summary.get('code'),
347 summary.get('print_time'),
348 start_date,
349 end_date)
350
351 # 3.创建月份表、提取/高亮关键行
352 # 倒序处理
353 is_reverse = True if sum(reverse_trend_list) > 0 else False
354 for month_list in month_mapping.values():
355 month_list.sort(key=lambda x: x[-1], reverse=is_reverse)
356
357 self.build_month_sheet(card, month_mapping, ms, is_reverse)
358
359 # 4.删除原表
360 for sheet in summary.get('sheet'):
361 self.remove(self.get_sheet_by_name(sheet))
362
363 def license_rebuild(self, license_summary, document_scheme):
364 for classify, (_, name, field_order, side_diff, scheme_diff) in consts.LICENSE_ORDER:
365 license_list = license_summary.get(classify)
366 if not license_list:
367 continue
368 ws = self.create_sheet(name)
369 if scheme_diff and document_scheme == consts.DOC_SCHEME_LIST[1]:
370 classify = consts.MVC_CLASSIFY_SE
371 for license_dict in license_list:
372 if classify == consts.IC_CLASSIFY and license_dict.get('类别') == '1':
373 license_summary.setdefault(consts.RP_CLASSIFY, []).append(license_dict)
374 continue
375 if side_diff:
376 key, field_order_yes, field_order_no = consts.FIELD_ORDER_MAP.get(classify)
377 field_order = field_order_yes if key in license_dict else field_order_no
378 for search_field, write_field in field_order:
379 ws.append((write_field, license_dict.get(search_field, '')))
380 ws.append((None, ))
381
382 def skip_img_sheet(self, skip_img):
383 if skip_img:
384 ws = self.create_sheet(consts.SKIP_IMG_SHEET_NAME)
385 ws.append(consts.SKIP_IMG_SHEET_HEADER)
386 for img_tuple in skip_img:
387 ws.append(img_tuple)
388
389 def rebuild(self, bs_summary, license_summary, skip_img, document_scheme):
390 self.bs_rebuild(bs_summary)
391 self.license_rebuild(license_summary, document_scheme)
392 self.skip_img_sheet(skip_img)
Styling with Markdown is supported
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!