Skip to content

周伟奇 / bmw-ocr

Go to a project
Toggle navigation
Toggle navigation pinning
1c6d880f by 周伟奇
Options

add license

1 parent c1c49a8e
Inline Side-by-side
Showing 4 changed files with 319 additions and 179 deletions
src/apps/doc/consts.py
...@@ -95,7 +95,7 @@ HEADERS_MAPPING.update( ...@@ -95,7 +95,7 @@ HEADERS_MAPPING.update(
95 HEADERS_MAPPING.update( 95 HEADERS_MAPPING.update(
96 { 96 {
97 '交易日期': BASE_HEADERS_MAPPING['记账日期'], 97 '交易日期': BASE_HEADERS_MAPPING['记账日期'],
98 '存入': BASE_HEADERS_MAPPING['金额'], 98 # '存入': BASE_HEADERS_MAPPING['金额'],
99 '对方账号': BASE_HEADERS_MAPPING['对方卡号/账号'], 99 '对方账号': BASE_HEADERS_MAPPING['对方卡号/账号'],
100 '对方名称': BASE_HEADERS_MAPPING['对方账户名'], 100 '对方名称': BASE_HEADERS_MAPPING['对方账户名'],
101 '摘要': BASE_HEADERS_MAPPING['附言'], 101 '摘要': BASE_HEADERS_MAPPING['附言'],
...@@ -160,6 +160,12 @@ HEADERS_MAPPING.update( ...@@ -160,6 +160,12 @@ HEADERS_MAPPING.update(
160 '交易地点/对方账号和户名': BASE_HEADERS_MAPPING['对方卡号/账号'], 160 '交易地点/对方账号和户名': BASE_HEADERS_MAPPING['对方卡号/账号'],
161 } 161 }
162 ) 162 )
163 # 农业银行-窄页
164 HEADERS_MAPPING.update(
165 {
166 '交易对手账号': BASE_HEADERS_MAPPING['对方卡号/账号'],
167 }
168 )
163 # 竖版-特殊-农商行 169 # 竖版-特殊-农商行
164 HEADERS_MAPPING.update( 170 HEADERS_MAPPING.update(
165 { 171 {
...@@ -299,17 +305,27 @@ HEADERS_MAPPING.update( ...@@ -299,17 +305,27 @@ HEADERS_MAPPING.update(
299 # ('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） 305 # ('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）
300 # 306 #
301 # # 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 307 # # 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
302 # ('农业银行', (1, None, 3, 5, None, 8, 7, 6, None, None, None, None, None)), # 横版-表格-农业银行-中国农业银行个人账户明细 308 # ('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
309 #
310 # # 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
311 # ('农业银行-10-1', (2, None, 4, 5, None, 3, 7, 6, None, None, None, None, None)),
312 #
313 # # 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
314 # ('农业银行-9', (1, None, 3, 4, None, 2, 8, 7, None, None, None, None, None)),
303 # 315 #
304 # # 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 316 # # 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
305 # ('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 317 # ('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
306 # 318 #
307 # # 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 319 # # 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
308 # ('工商银行', (1, None, 9, 10, None, 7, None, None, None, None, None, None, None)), 320 # ('工商银行', (1, None, 9, 10, None, 7, None, None, None, None, None, None, None)),
309 # 321 #
322 # # 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
323 # ('工商银行-电子账单', (1, None, 9, 10, None, 7, 11, 12, None, None, None, None, None)),
324 #
310 # # 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 325 # # 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
311 # # 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 326 # # 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
312 # ('建设银行', (None, None, None, None, None, 2, None, None, None, None, None, None, None)), 327 # ('建设银行-竖版', (3, None, 4, 5, None, 2, None, 7, None, None, None, None, None)),
328 # ('建设银行-横版', (5, None, 6, 7, None, 2, None, 9, None, None, None, None, None)),
313 # 329 #
314 # # 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 330 # # 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
315 # ('微信', (2, None, 6, None, None, 3, 7, None, None, None, None, None, None)), 331 # ('微信', (2, None, 6, None, None, 3, 7, None, None, None, None, None, None)),
...@@ -320,7 +336,13 @@ HEADERS_MAPPING.update( ...@@ -320,7 +336,13 @@ HEADERS_MAPPING.update(
320 # # -----------------普通打印：部分格线-------------------------------- 336 # # -----------------普通打印：部分格线--------------------------------
321 # 337 #
322 # # 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 338 # # 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
323 # ('农业银行', (1, None, 3, 4, None, 2, None, 5, None, None, None, None, None)), 339 # ('农业银行-5', (1, None, 3, 4, None, 2, None, 5, None, None, None, None, None)),
340 #
341 # # 农业银行：日期 地点 摘要 存入 支出 余额 对方账号 对方户名
342 # ('农业银行-8', (1, None, None, 6, None, 3, 8, 7, None, None, None, 4, 5)),
343
344 # # 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
345 # ('农业银行-窄页', (1, None, 3, 4, None, 2, 7, 6, None, None, None, None, None)),
324 # 346 #
325 # # 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 347 # # 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
326 # ('农商行', (1, None, 2, 3, None, 6, 5, 4, None, None, None, None, None)), 348 # ('农商行', (1, None, 2, 3, None, 6, 5, 4, None, None, None, None, None)),
...@@ -330,6 +352,9 @@ HEADERS_MAPPING.update( ...@@ -330,6 +352,9 @@ HEADERS_MAPPING.update(
330 # 352 #
331 # # 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 353 # # 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
332 # ('平安电子账单', (2, None, 5, 6, None, 4, None, None, None, None, None, None, None)), 354 # ('平安电子账单', (2, None, 5, 6, None, 4, None, None, None, None, None, None, None)),
355
356 # # 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
357 # ('建设银行-电子账单', (5, None, 6, 7, None, 2, None, 9, None, None, None, None, None)),
333 # 358 #
334 # # -----------------普通打印：无格线------------------------------------- 359 # # -----------------普通打印：无格线-------------------------------------
335 # 360 #
...@@ -338,7 +363,8 @@ HEADERS_MAPPING.update( ...@@ -338,7 +363,8 @@ HEADERS_MAPPING.update(
338 # 363 #
339 # # 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 364 # # 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
340 # # 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 365 # # 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
341 # ('邮储银行', (1, None, None, None, None, 2, None, None, None, None, None, None, None)), 366 # ('邮储银行-8', (1, None, 4, 5, None, 2, 6, 7, None, None, None, None, None)),
367 # ('邮储银行-5', (1, None, 3, 4, None, 2, None, None, None, None, None, None, None)),
342 # 368 #
343 # # 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 369 # # 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
344 # ('工商银行电子版', (1, None, 9, 10, None, 7, None, None, None, None, None, None, None)), 370 # ('工商银行电子版', (1, None, 9, 10, None, 7, None, None, None, None, None, None, None)),
...@@ -351,13 +377,15 @@ HEADERS_MAPPING.update( ...@@ -351,13 +377,15 @@ HEADERS_MAPPING.update(
351 # ('民生银行', (None, None, 5, 6, None, None, 7, None, 8, None, None, None, None)), 377 # ('民生银行', (None, None, 5, 6, None, None, 7, None, 8, None, None, None, None)),
352 # 378 #
353 # # 农业银行：交易日期 摘要/附言 交易金额 对方账号和户名 379 # # 农业银行：交易日期 摘要/附言 交易金额 对方账号和户名
354 # ('农业银行', (1, None, 3, None, None, 2, None, 4, None, None, None, None, None)), 380 # ('农业银行-整数', (1, None, 3, None, None, 2, None, 4, None, None, None, None, None)),
355 # 381 #
356 # # 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 382 # # 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
357 # ('农业银行', (1, None, 3, 4, None, 2, None, 5, None, None, None, None, None)), 383 # ('农业银行', (1, None, 3, 4, None, 2, None, 5, None, None, None, None, None)),
358 # 384 #
359 # # 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 385 # # 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
360 # ('农业银行', (1, 2, None, None, None, None, None, None, None, None, None, None, None)), 386 # # 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
387 # ('农业银行', (1, 2, 4, 5, None, 3, None, None, None, None, None, None, None)),
388 # ('农业银行-扩张缩进', (1, 2, 5, 6, None, 4, None, None, None, None, None, None, None)),
361 # 389 #
362 # # 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 390 # # 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
363 # ('交通银行', (1, None, 5, 6, None, 3, None, None, None, None, 4, None, None)), 391 # ('交通银行', (1, None, 5, 6, None, 3, None, None, None, None, 4, None, None)),
...@@ -374,11 +402,10 @@ HEADERS_MAPPING.update( ...@@ -374,11 +402,10 @@ HEADERS_MAPPING.update(
374 # 402 #
375 # # 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 403 # # 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
376 # ('邮储银行', (2, None, 5, 6, None, 4, None, 7, None, None, None, None, None)), 404 # ('邮储银行', (2, None, 5, 6, None, 4, None, 7, None, None, None, None, None)),
377 #
378 # # 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
379 # ('建设银行', (5, None, 6, 7, None, 2, None, 9, None, None, None, None, None)),
380 # ] 405 # ]
381 406
407 OTHER_TUPLE = (None, None, None, None, None, None, None, None, None, None, None, None, None)
408
382 # { 409 # {
383 # "0":"其他", 410 # "0":"其他",
384 # "1":"普通打印-全表格-中国农业银行", 411 # "1":"普通打印-全表格-中国农业银行",
...@@ -408,67 +435,163 @@ HEADERS_MAPPING.update( ...@@ -408,67 +435,163 @@ HEADERS_MAPPING.update(
408 # "22":"针式打印-部分格线-邮储银行一本通绿卡" 435 # "22":"针式打印-部分格线-邮储银行一本通绿卡"
409 # } 436 # }
410 437
438 # CLASSIFY_LIST = [
439 # ('其他', OTHER_TUPLE),
440 # ('农业银行', (1, None, 3, 5, None, 8, 7, 6, None, None, None, None, None)),
441 # ('中国银行', (1, 2, 4, 5, 6, 9, 10, 11, 12, None, None, None, None)),
442 # ('北京银行', (1, None, 4, 5, None, 2, 6, 7, None, None, 3, None, None)),
443 # ('工商银行', (1, None, 9, 10, None, 7, None, None, None, None, None, None, None)),
444 # ('建设银行', (None, None, None, None, None, 2, None, None, None, None, None, None, None)),
445 # ('微信', (2, None, 6, None, None, 3, 7, None, None, None, None, None, None)),
446 # ('支付宝', (2, None, None, 6, None, 3, None, None, None, None, None, 4, 5)),
447 #
448 # ('交通银行', (1, None, 5, 6, None, 3, None, None, None, None, 4, None, None)),
449 # ('农业银行', (1, None, 3, None, None, 2, None, 4, None, None, None, None, None)),
450 # ('农业银行', (1, 2, None, None, None, None, None, None, None, None, None, None, None)),
451 # ('招商银行', (1, None, 3, 4, None, 6, None, None, None, None, None, None, None)),
452 # ('招商银行电子版', (1, None, 3, 4, None, 5, 6, None, None, None, None, None, None)),
453 # ('民生银行', (None, None, 5, 6, None, None, 7, None, 8, None, None, None, None)),
454 #
455 # ('中信银行', (1, None, None, 5, None, 2, 6, 7, 8, None, None, 3, 4)),
456 # ('农业银行', (1, None, 3, 4, None, 2, None, 5, None, None, None, None, None)),
457 # ('农业银行', (1, None, 3, 4, None, 2, None, 5, None, None, None, None, None)),
458 # ('农业银行', (1, None, 3, 4, None, 2, None, 5, None, None, None, None, None)),
459 # ('平安电子账单', (2, None, 5, 6, None, 4, None, None, None, None, None, None, None)),
460 #
461 # ('建设银行', (None, None, None, None, None, None, None, None, None, None, None, None, None)),
462 # ('邮储银行', (2, None, 5, 6, None, 4, None, 7, None, None, None, None, None)),
463 # ('邮储银行', (2, None, 5, 6, None, 4, None, 7, None, None, None, None, None)),
464 # ]
465
466 # "4":"普通打印-全表格-中国银行",
467 # "5":"普通打印-全表格-农业银行-10列",
468 # "6":"普通打印-全表格-农业银行-10列-1",
469 # "7":"普通打印-全表格-农业银行-9列",
470 # "8":"普通打印-全表格-北京银行",
471 # "9":"普通打印-全表格-工商银行",
472 # "10":"普通打印-全表格-工商银行-电子账单",
473 # "11":"普通打印-全表格-建设银行",
474 # "12":"普通打印-全表格-微信账单",
475 # "13":"普通打印-全表格-支付宝账单",
476
477 # "14":"普通打印-无格线-交通银行",
478 # "15":"普通打印-无格线-储蓄银行-5列",
479 # "16":"普通打印-无格线-储蓄银行-8列",
480 # "17":"普通打印-无格线-农业银行-扩张缩进",
481 # "18":"普通打印-无格线-农业银行-整数",
482 # "19":"普通打印-无格线-招商银行",
483 # "20":"普通打印-无格线-招商银行-电子账单",
484 # "21":"普通打印-无格线-民生银行",
485
486 # "22":"普通打印-部分格线-横版-中信银行",
487 # "23":"普通打印-部分格线-竖版-农业银行-5列",
488 # "24":"普通打印-部分格线-竖版-农业银行-8列",
489 # "25":"普通打印-部分格线-竖版-农业银行-窄页",
490 # "26":"普通打印-部分格线-竖版-平安电子账单",
491 # "27":"普通打印-部分格线-竖版-建设银行-电子账单",
492
493 # "34":"针式打印-全格线-建设银行",
494 # "35":"针式打印-部分格线-竖版-邮储银行",
495 # "36":"针式打印-部分格线-竖版-邮储银行-绿卡",
496
411 CLASSIFY_LIST = [ 497 CLASSIFY_LIST = [
412 ('其他', (None, None, None, None, None, None, None, None, None, None, None, None, None)), 498 ('其他', OTHER_TUPLE),
413 ('农业银行', (1, None, 3, 5, None, 8, 7, 6, None, None, None, None, None)), 499 ('其他', OTHER_TUPLE),
414 ('中国银行', (1, 2, 4, 5, 6, 9, 10, 11, 12, None, None, None, None)), 500 ('其他', OTHER_TUPLE),
415 ('北京银行', (1, None, 4, 5, None, 2, 6, 7, None, None, 3, None, None)), 501 ('其他', OTHER_TUPLE),
416 ('工商银行', (1, None, 9, 10, None, 7, None, None, None, None, None, None, None)), 502 ('普通打印-全表格-中国银行', (1, 2, 4, 5, 6, 9, 10, 11, 12, None, None, None, None)),
417 ('建设银行', (None, None, None, None, None, 2, None, None, None, None, None, None, None)), 503 ('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)),
418 ('微信', (2, None, 6, None, None, 3, 7, None, None, None, None, None, None)), 504 ('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)),
419 ('支付宝', (2, None, None, 6, None, 3, None, None, None, None, None, 4, 5)), 505 ('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)),
420 506 ('普通打印-全表格-北京银行', (1, None, 4, 5, None, 2, 6, 7, None, None, 3, None, None)),
421 ('交通银行', (1, None, 5, 6, None, 3, None, None, None, None, 4, None, None)), 507 ('普通打印-全表格-工商银行', (1, None, 9, 10, None, 7, None, None, None, None, None, None, None)),
422 ('农业银行', (1, None, 3, None, None, 2, None, 4, None, None, None, None, None)), 508 ('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)),
423 ('农业银行', (1, 2, None, None, None, None, None, None, None, None, None, None, None)), 509 ('普通打印-全表格-建设银行', (3, None, 4, 5, None, 2, None, 7, None, None, None, None, None)),
424 ('招商银行', (1, None, 3, 4, None, 6, None, None, None, None, None, None, None)), 510 ('普通打印-全表格-微信账单', (2, None, 6, None, None, 3, 7, None, None, None, None, None, None)),
425 ('招商银行电子版', (1, None, 3, 4, None, 5, 6, None, None, None, None, None, None)), 511 ('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)),
426 ('民生银行', (None, None, 5, 6, None, None, 7, None, 8, None, None, None, None)), 512
427 513 ('普通打印-无格线-交通银行', (1, None, 5, 6, None, 3, None, None, None, None, 4, None, None)),
428 ('中信银行', (1, None, None, 5, None, 2, 6, 7, 8, None, None, 3, 4)), 514 ('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)),
429 ('农业银行', (1, None, 3, 4, None, 2, None, 5, None, None, None, None, None)), 515 ('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)),
430 ('农业银行', (1, None, 3, 4, None, 2, None, 5, None, None, None, None, None)), 516 ('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)),
431 ('农业银行', (1, None, 3, 4, None, 2, None, 5, None, None, None, None, None)), 517 ('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)),
432 ('平安电子账单', (2, None, 5, 6, None, 4, None, None, None, None, None, None, None)), 518 ('普通打印-无格线-招商银行', (1, None, 3, 4, None, 6, None, None, None, None, None, None, None)),
433 519 ('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)),
434 ('建设银行', (None, None, None, None, None, None, None, None, None, None, None, None, None)), 520 ('普通打印-无格线-民生银行', (None, None, 5, 6, None, None, 7, None, 8, None, None, None, None)),
435 ('邮储银行', (2, None, 5, 6, None, 4, None, 7, None, None, None, None, None)), 521
436 ('邮储银行', (2, None, 5, 6, None, 4, None, 7, None, None, None, None, None)), 522 ('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)),
523 ('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)),
524 ('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)),
525 ('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)),
526 ('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)),
527 ('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)),
528 ('其他', OTHER_TUPLE),
529 ('其他', OTHER_TUPLE),
530 ('其他', OTHER_TUPLE),
531 ('其他', OTHER_TUPLE),
532 ('其他', OTHER_TUPLE),
533 ('其他', OTHER_TUPLE),
534 ('针式打印-全格线-建设银行', OTHER_TUPLE),
535 ('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)),
536 ('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)),
537 ('其他', OTHER_TUPLE),
437 ] 538 ]
438 539
439 # ----------license相关------------------------------------------------------------------------------------------------ 540 # ----------license相关------------------------------------------------------------------------------------------------
541
542 # "0":"AVT Invioce",
543 # "1":"二手车发票",
544 # "2":"其他",
545 # "3":"护照",
546 # "28":"机动车登记证",
547 # "29":"机动车销售统一发票",
548 # "30":"港澳通行证",
549 # "31":"营业执照",
550 # "32":"行驶证",
551 # "33":"身份证",
552 # "37":"银行卡"
553
554 # 其他
555 OTHER_CLASSIFY = 2
556
557 # 身份证
558 IC_CN_NAME = '身份证'
559 IC_CLASSIFY = 33
560 # 增值税发票
561 VAT_CN_NAME = '增值税发票'
562 VAT_CLASSIFY = 0
563 # 机动车登记证书
564 MVC_CN_NAME = '机动车登记证书'
565 MVC_CLASSIFY = 28
566 # 机动车销售统一发票
567 MVI_CN_NAME = '机动车销售统一发票'
568 MVI_CLASSIFY = 29
569 IC_PID = VAT_PID = MVC_PID = MVI_PID = None
570
440 # 营业执照 571 # 营业执照
441 BL_KEY = 'bl' 572 BL_CN_NAME = '营业执照'
573 BL_CLASSIFY = 31
574 BL_PID = 41
442 # 二手车发票 575 # 二手车发票
443 UCI_KEY = 'uci' 576 UCI_CN_NAME = '二手车发票'
577 UCI_CLASSIFY = 1
578 UCI_PID = 60
444 # 港澳台通行证 579 # 港澳台通行证
445 EEP_KEY = 'eep' 580 EEP_CN_NAME = '港澳台通行证'
581 EEP_CLASSIFY = 30
582 EEP_PID = 1018
446 # 行驶证 583 # 行驶证
447 DL_KEY = 'dl' 584 DL_CN_NAME = '行驶证'
585 DL_CLASSIFY = 32
586 DL_PID = 5
448 # 护照 587 # 护照
449 PP_KEY = 'pp' 588 PP_CN_NAME = '护照'
589 PP_CLASSIFY = 3
590 PP_PID = 8
450 # 银行卡 591 # 银行卡
451 BC_KEY = 'bc' 592 BC_CN_NAME = '银行卡'
452 # 身份证 593 BC_CLASSIFY = 37
453 IC_KEY = 'ic' 594 BC_PID = 4
454 # 机动车登记证书
455 MVC_KEY = 'mvc'
456 # 机动车销售统一发票
457 MVI_KEY = 'mvi'
458 # 增值税发票
459 VAT_KEY = 'vat'
460
461 LICENSE_ORDER = ((MVI_KEY, '机动车销售统一发票'),
462 (IC_KEY, '身份证'),
463 (BC_KEY, '银行卡'),
464 (BL_KEY, '营业执照'),
465 (UCI_KEY, '二手车发票'),
466 (EEP_KEY, '港澳台通行证'),
467 (DL_KEY, '行驶证'),
468 (PP_KEY, '护照'),
469 (MVC_KEY, '机动车登记证书'),
470 (VAT_KEY, '增值税发票'))
471
472 BC_FIELD = (('CardNum', '银行卡号'), 595 BC_FIELD = (('CardNum', '银行卡号'),
473 ('BankName', '发卡行名称'), 596 ('BankName', '发卡行名称'),
474 ('CardName', '银行卡名称'), 597 ('CardName', '银行卡名称'),
...@@ -478,14 +601,19 @@ BC_FIELD = (('CardNum', '银行卡号'), ...@@ -478,14 +601,19 @@ BC_FIELD = (('CardNum', '银行卡号'),
478 601
479 SUCCESS_CODE_SET = {'0', 0} 602 SUCCESS_CODE_SET = {'0', 0}
480 603
481 BC_PID = 4 604 LICENSE_ORDER = ((MVI_CLASSIFY, (MVI_PID, MVI_CN_NAME)),
482 605 (IC_CLASSIFY, (IC_PID, IC_CN_NAME)),
483 OTHER_SET = {0, 1, 2} 606 (BC_CLASSIFY, (BC_PID, BC_CN_NAME)),
484 BS_SET = {10, 11, 12} 607 (BL_CLASSIFY, (BL_PID, BL_CN_NAME)),
485 LICENSE_SET_1 = {110, 111, 112} 608 (UCI_CLASSIFY, (UCI_PID, UCI_CN_NAME)),
486 LICENSE_SET_2 = {1110, 1111, 1112} 609 (EEP_CLASSIFY, (EEP_PID, EEP_CN_NAME)),
487 610 (DL_CLASSIFY, (DL_PID, DL_CN_NAME)),
488 CLASSIFY_PID_DICT = { 611 (PP_CLASSIFY, (PP_PID, PP_CN_NAME)),
489 0: (4, BC_KEY) # 银行卡 612 (MVC_CLASSIFY, (MVC_PID, MVC_CN_NAME)),
490 } 613 (VAT_CLASSIFY, (VAT_PID, VAT_CN_NAME)))
491 614
615 LICENSE_CLASSIFY_MAPPING = dict(LICENSE_ORDER)
616
617 OTHER_CLASSIFY_SET = {OTHER_CLASSIFY}
618 LICENSE_CLASSIFY_SET_1 = {IC_CLASSIFY, VAT_CLASSIFY, MVC_CLASSIFY, MVI_CLASSIFY}
619 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
...@@ -81,9 +81,14 @@ class Command(BaseCommand, LoggerMixin): ...@@ -81,9 +81,14 @@ class Command(BaseCommand, LoggerMixin):
81 return doc_data_path, excel_path, src_excel_path, pdf_path 81 return doc_data_path, excel_path, src_excel_path, pdf_path
82 82
83 @staticmethod 83 @staticmethod
84 def bs_process(wb, sheets, bs_summary, unknown_summary, pno, img_idx, classify, confidence): 84 def bs_process(wb, ocr_data, bs_summary, unknown_summary, img_path, classify):
85 sheets = ocr_data.get('data', [])
86 if not sheets:
87 return
88 confidence = ocr_data.get('confidence', 1)
89 img_name, _ = os.path.splitext(os.path.basename(img_path))
85 for i, sheet in enumerate(sheets): 90 for i, sheet in enumerate(sheets):
86 sheet_name = 'page_{0}_img_{1}_{2}'.format(pno, img_idx, i) 91 sheet_name = '{0}_{1}'.format(img_name, i)
87 # ['户名', '卡号', '页码', '回单验证码', '打印时间', '起始时间', '终止时间'] 92 # ['户名', '卡号', '页码', '回单验证码', '打印时间', '起始时间', '终止时间']
88 summary = sheet.get('summary') 93 summary = sheet.get('summary')
89 card = summary[1] 94 card = summary[1]
...@@ -139,38 +144,52 @@ class Command(BaseCommand, LoggerMixin): ...@@ -139,38 +144,52 @@ class Command(BaseCommand, LoggerMixin):
139 words = cell.get('words') 144 words = cell.get('words')
140 ws.cell(row=r1+1, column=c1+1, value=words) 145 ws.cell(row=r1+1, column=c1+1, value=words)
141 146
142 def license2_process(self, img_path, license_summary, pid, license_key): 147 @staticmethod
143 with open(img_path, 'rb') as f: 148 def license1_process(ocr_data, license_summary, classify):
144 base64_data = base64.b64encode(f.read()) 149 license_data = ocr_data.get('data', [])
145 # 获取解码后的base64值 150 if not license_data:
146 filedata = base64_data.decode() 151 return
147 # pid 产品的pid, key, secret 登录之后能够查看到 152 _, license_key = consts.CLASSIFY_PID_DICT.get(classify)
148 datas = { 153 for license_dict in license_data:
149 "pid": str(pid), 154 res_list = []
150 "key": conf.OCR_KEY, 155 for field, value in license_dict.items():
151 "secret": conf.OCR_SECRET, 156 res_list.append((field, value))
152 "file": filedata 157 license_summary.setdefault(license_key, []).append(res_list)
153 } 158
154 r = requests.post(self.ocr_url_2, data=datas) 159 @staticmethod
155 if r.status_code == 200: 160 def license2_process(ocr_res_2, license_summary, pid, classify):
156 # 识别结果 161 if ocr_res_2.get('ErrorCode') in consts.SUCCESS_CODE_SET:
157 response = r.json() 162 if pid == consts.BC_PID:
158 if response.get('ErrorCode') in consts.SUCCESS_CODE_SET: 163 # 银行卡
159 if pid == consts.BC_PID: 164 res_list = []
160 # 银行卡 165 for en_key, chn_key in consts.BC_FIELD:
161 res_list = [] 166 res_list.append((chn_key, ocr_res_2.get(en_key, '')))
162 for en_key, chn_key in consts.BC_FIELD: 167 license_summary.setdefault(classify, []).append(res_list)
163 res_list.append((chn_key, response.get(en_key, ''))) 168 else:
164 license_summary.setdefault(license_key, []).append(res_list) 169 # 营业执照、行驶证等
165 else: 170 for result_dict in ocr_res_2.get('ResultList', []):
166 # 营业执照、行驶证等 171 res_list = []
167 for result_dict in response.get('ResultList', []): 172 for field_dict in result_dict.get('FieldList', []):
168 res_list = [] 173 res_list.append(
169 for field_dict in result_dict.get('FieldList', []): 174 (field_dict.get('chn_key', ''), field_dict.get('value', '')))
170 res_list.append((field_dict.get('chn_key', ''), field_dict.get('value', ''))) 175 license_summary.setdefault(classify, []).append(res_list)
171 license_summary.setdefault(license_key, []).append(res_list) 176
172 177 # async def fetch_ocr_result(self, img_path):
173 def ocr_2_wb(self, res, wb, pno, img_idx, bs_summary, unknown_summary, license_summary): 178 # async with aiohttp.ClientSession(
179 # headers=self.ocr_header, connector=aiohttp.TCPConnector(ssl=False)
180 # ) as session:
181 # json_data = self.get_ocr_json(img_path)
182 # async with session.post(self.ocr_url, json=json_data) as response:
183 # return await response.json()
184 #
185 # async def img_2_ocr_2_wb(self, wb, img_path, summary):
186 # res = await self.fetch_ocr_result(img_path)
187 # self.cronjob_log.info('{0} [fetch ocr result success] [img={1}] [res={2}]'.format(self.log_base, img_path, res))
188 # sheets_list = res.get('result').get('res')
189 # img_name = os.path.basename(img_path)
190 # self.append_sheet(wb, sheets_list, img_name, summary)
191
192 def img_2_ocr_2_wb(self, wb, img_path, bs_summary, unknown_summary, license_summary):
174 # # 流水 193 # # 流水
175 # res = { 194 # res = {
176 # 'code': 1, 195 # 'code': 1,
...@@ -178,7 +197,7 @@ class Command(BaseCommand, LoggerMixin): ...@@ -178,7 +197,7 @@ class Command(BaseCommand, LoggerMixin):
178 # 'data': { 197 # 'data': {
179 # 'classify': 0, 198 # 'classify': 0,
180 # 'confidence': 0.999, 199 # 'confidence': 0.999,
181 # 'sheets': [ 200 # 'data': [
182 # { 201 # {
183 # 'summary': ['户名', '卡号', '页码', '回单验证码', '打印时间', '起始时间', '终止时间'], 202 # 'summary': ['户名', '卡号', '页码', '回单验证码', '打印时间', '起始时间', '终止时间'],
184 # 'cells': [] 203 # 'cells': []
...@@ -220,55 +239,52 @@ class Command(BaseCommand, LoggerMixin): ...@@ -220,55 +239,52 @@ class Command(BaseCommand, LoggerMixin):
220 # 'confidence': 0.999, 239 # 'confidence': 0.999,
221 # } 240 # }
222 # } 241 # }
242 with open(img_path, 'rb') as f:
243 base64_data = base64.b64encode(f.read())
244 # 获取解码后的base64值
245 file_data = base64_data.decode()
246 json_data_1 = {
247 "file": file_data
248 }
249 response_1 = requests.post(self.ocr_url_1, data=json_data_1)
250 if response_1.status_code == 200:
251 ocr_res_1 = response_1.json()
252 self.cronjob_log.info('{0} [ocr_1 result] [img={1}] [res={2}]'.format(
253 self.log_base, img_path, ocr_res_1))
254
255 if ocr_res_1.get('code') == 1:
256 ocr_data = ocr_res_1.get('data', {})
257 classify = ocr_data.get('classify')
258 if classify is None:
259 return
260 elif classify in consts.OTHER_CLASSIFY_SET: # 其他类
261 return
262 elif classify in consts.LICENSE_CLASSIFY_SET_1: # 证件1
263 self.license1_process(ocr_data, license_summary, classify)
264 elif classify in consts.LICENSE_CLASSIFY_SET_2: # 证件2
265 pid, _ = consts.LICENSE_CLASSIFY_MAPPING.get(classify)
266 json_data_2 = {
267 "pid": str(pid),
268 "key": conf.OCR_KEY,
269 "secret": conf.OCR_SECRET,
270 "file": file_data
271 }
272 response_2 = requests.post(self.ocr_url_2, data=json_data_2)
273 if response_2.status_code == 200:
274 # 识别结果
275 ocr_res_2 = response_2.json()
276 self.cronjob_log.info('{0} [ocr_2 result] [img={1}] [res={2}]'.format(
277 self.log_base, img_path, ocr_res_2))
278 self.license2_process(ocr_res_2, license_summary, pid, classify)
279 else:
280 raise Exception('ocr 2 error, img_path={0}'.format(img_path))
281 else: # 流水处理
282 self.bs_process(wb, ocr_data, bs_summary, unknown_summary, img_path, classify)
283 else:
284 pass
285 else:
286 raise Exception('ocr 1 error, img_path={0}'.format(img_path))
223 287
224 data = res.get('data', {})
225 classify = data.get('classify')
226 if classify is None:
227 return
228 elif classify in consts.OTHER_SET: # 其他类
229 return
230 elif classify in consts.BS_SET: # 流水处理
231 sheets = data.get('sheets', [])
232 if not sheets:
233 return
234 confidence = data.get('confidence', 1)
235 self.bs_process(wb, sheets, bs_summary, unknown_summary, pno, img_idx, classify, confidence)
236 elif classify in consts.LICENSE_SET_1: # 证件1
237 # self.license1_process() # TODO license1
238 pass
239 elif classify in consts.LICENSE_SET_2: # 证件2
240 pid, license_key = consts.CLASSIFY_PID_DICT.get(classify)
241 self.license2_process(license_summary, pid, license_key) # TODO reuse img data?
242
243 # async def fetch_ocr_result(self, img_path):
244 # async with aiohttp.ClientSession(
245 # headers=self.ocr_header, connector=aiohttp.TCPConnector(ssl=False)
246 # ) as session:
247 # json_data = self.get_ocr_json(img_path)
248 # async with session.post(self.ocr_url, json=json_data) as response:
249 # return await response.json()
250 #
251 # async def img_2_ocr_2_wb(self, wb, img_path, summary):
252 # res = await self.fetch_ocr_result(img_path)
253 # self.cronjob_log.info('{0} [fetch ocr result success] [img={1}] [res={2}]'.format(self.log_base, img_path, res))
254 # sheets_list = res.get('result').get('res')
255 # img_name = os.path.basename(img_path)
256 # self.append_sheet(wb, sheets_list, img_name, summary)
257
258 def fetch_ocr_result(self, img_path):
259 files = [
260 ('img', open(img_path, 'rb'))
261 ]
262 response = requests.request("POST", self.ocr_url_1, files=files)
263 if response.status_code == 200:
264 return response.json()
265
266 def img_2_ocr_2_wb(self, wb, img_info, bs_summary, unknown_summary, license_summary):
267 res = self.fetch_ocr_result(img_info[0])
268 self.cronjob_log.info('{0} [fetch ocr result success] [img={1}] [res={2}]'.format(
269 self.log_base, img_info[0], res))
270 if res.get('code') == 1:
271 self.ocr_2_wb(res, wb, img_info[1], img_info[2], bs_summary, unknown_summary, license_summary)
272 288
273 @staticmethod 289 @staticmethod
274 def get_most(value_list): 290 def get_most(value_list):
...@@ -414,7 +430,6 @@ class Command(BaseCommand, LoggerMixin): ...@@ -414,7 +430,6 @@ class Command(BaseCommand, LoggerMixin):
414 # EDMS异常：下载异常-->回队列-->邮件；上传异常-->重新上传队列-->邮件 430 # EDMS异常：下载异常-->回队列-->邮件；上传异常-->重新上传队列-->邮件
415 # 算法异常：第一道异常-->识别失败-->邮件；第二道异常-->识别失败-->邮件 431 # 算法异常：第一道异常-->识别失败-->邮件；第二道异常-->识别失败-->邮件
416 # TODO 数据库断联问题 432 # TODO 数据库断联问题
417 # TODO 非流水证件处理
418 # TODO EDMS API GATEWAY 433 # TODO EDMS API GATEWAY
419 def handle(self, *args, **kwargs): 434 def handle(self, *args, **kwargs):
420 sleep_second = int(conf.SLEEP_SECOND) 435 sleep_second = int(conf.SLEEP_SECOND)
...@@ -466,8 +481,8 @@ class Command(BaseCommand, LoggerMixin): ...@@ -466,8 +481,8 @@ class Command(BaseCommand, LoggerMixin):
466 # loop.run_until_complete(asyncio.wait(tasks)) 481 # loop.run_until_complete(asyncio.wait(tasks))
467 # loop.close() 482 # loop.close()
468 483
469 for img_info in pdf_handler.img_info_list: 484 for img_path in pdf_handler.img_path_list:
470 self.img_2_ocr_2_wb(wb, img_info, bs_summary, unknown_summary, license_summary) 485 self.img_2_ocr_2_wb(wb, img_path, bs_summary, unknown_summary, license_summary)
471 486
472 self.cronjob_log.info('{0} [bs_summary={1}] [unknown_summary={2}] [license_summary={3}]'.format( 487 self.cronjob_log.info('{0} [bs_summary={1}] [unknown_summary={2}] [license_summary={3}]'.format(
473 self.log_base, bs_summary, unknown_summary, license_summary)) 488 self.log_base, bs_summary, unknown_summary, license_summary))
......
src/apps/doc/ocr/wb.py
...@@ -168,7 +168,7 @@ class BSWorkbook(Workbook): ...@@ -168,7 +168,7 @@ class BSWorkbook(Workbook):
168 month_mapping.setdefault(month_list[-1], []).insert( 168 month_mapping.setdefault(month_list[-1], []).insert(
169 0, (ws.title, idx_list[-1] + min_row, ws.max_row, 0)) 169 0, (ws.title, idx_list[-1] + min_row, ws.max_row, 0))
170 170
171 def build_metadata_rows(self, classify, confidence, role, code, print_time, start_date, end_date): 171 def build_metadata_rows(self, confidence, code, print_time, start_date, end_date):
172 if start_date is None or end_date is None: 172 if start_date is None or end_date is None:
173 timedelta = None 173 timedelta = None
174 else: 174 else:
...@@ -176,10 +176,6 @@ class BSWorkbook(Workbook): ...@@ -176,10 +176,6 @@ class BSWorkbook(Workbook):
176 metadata_rows = [ 176 metadata_rows = [
177 ('流水识别置信度', confidence), 177 ('流水识别置信度', confidence),
178 self.blank_row, 178 self.blank_row,
179 ('分类结果', classify),
180 self.blank_row,
181 ('户名', role),
182 self.blank_row,
183 self.code_header, 179 self.code_header,
184 ] 180 ]
185 metadata_rows.extend(code) 181 metadata_rows.extend(code)
...@@ -200,19 +196,19 @@ class BSWorkbook(Workbook): ...@@ -200,19 +196,19 @@ class BSWorkbook(Workbook):
200 ms = self.create_sheet('{0}({1})'.format(self.meta_sheet_title, card)) 196 ms = self.create_sheet('{0}({1})'.format(self.meta_sheet_title, card))
201 return ms 197 return ms
202 198
203 def build_meta_sheet(self, card, classify, confidence, role, code, print_time, start_date, end_date): 199 def build_meta_sheet(self, card, confidence, code, print_time, start_date, end_date):
204 metadata_rows = self.build_metadata_rows(classify, confidence, role, code, print_time, start_date, end_date) 200 metadata_rows = self.build_metadata_rows(confidence, code, print_time, start_date, end_date)
205 ms = self.create_meta_sheet(card) 201 ms = self.create_meta_sheet(card)
206 for row in metadata_rows: 202 for row in metadata_rows:
207 ms.append(row) 203 ms.append(row)
208 return ms 204 return ms
209 205
210 def build_month_sheet(self, role, month_mapping, ms, is_reverse): 206 def build_month_sheet(self, card, month_mapping, ms, is_reverse):
211 tmp_ws = self.create_sheet('tmp_ws') 207 tmp_ws = self.create_sheet('tmp_ws')
212 for month in sorted(month_mapping.keys()): 208 for month in sorted(month_mapping.keys()):
213 # 3.1.拷贝数据 209 # 3.1.拷贝数据
214 parts = month_mapping.get(month) 210 parts = month_mapping.get(month)
215 new_ws = self.create_sheet('{0}({1})'.format(month, role)) 211 new_ws = self.create_sheet('{0}({1})'.format(month, card[-6:]))
216 new_ws.append(consts.FIXED_HEADERS) 212 new_ws.append(consts.FIXED_HEADERS)
217 for part in parts: 213 for part in parts:
218 ws = self.get_sheet_by_name(part[0]) 214 ws = self.get_sheet_by_name(part[0])
...@@ -338,9 +334,7 @@ class BSWorkbook(Workbook): ...@@ -338,9 +334,7 @@ class BSWorkbook(Workbook):
338 334
339 # 2.元信息提取表 335 # 2.元信息提取表
340 ms = self.build_meta_sheet(card, 336 ms = self.build_meta_sheet(card,
341 summary['classify'],
342 summary['confidence'], 337 summary['confidence'],
343 summary['role'],
344 summary['code'], 338 summary['code'],
345 summary['print_time'], 339 summary['print_time'],
346 start_date, 340 start_date,
...@@ -359,13 +353,16 @@ class BSWorkbook(Workbook): ...@@ -359,13 +353,16 @@ class BSWorkbook(Workbook):
359 self.remove(self.get_sheet_by_name(sheet)) 353 self.remove(self.get_sheet_by_name(sheet))
360 354
361 def license_rebuild(self, license_summary): 355 def license_rebuild(self, license_summary):
362 for en_key, cn_key in consts.LICENSE_ORDER: 356 for classify, (_, name) in consts.LICENSE_ORDER:
363 ws = self.create_sheet(cn_key) 357 res = license_summary.get(classify)
364 for bl in license_summary.get(en_key, []): 358 if res is None:
359 continue
360 ws = self.create_sheet(name)
361 for bl in res:
365 for bl_field in bl: 362 for bl_field in bl:
366 ws.append(bl_field) 363 ws.append(bl_field)
367 ws.append((None, )) 364 ws.append((None, ))
368 365
369 def rebuild(self, bs_summary, license_summary): 366 def rebuild(self, bs_summary, license_summary):
370 self.bs_rebuild(bs_summary) 367 self.bs_rebuild(bs_summary)
371 # self.license_rebuild(license_summary) 368 self.license_rebuild(license_summary)
......
src/common/tools/pdf_to_img.py
...@@ -25,7 +25,7 @@ class PDFHandler: ...@@ -25,7 +25,7 @@ class PDFHandler:
25 def __init__(self, path, img_dir_path): 25 def __init__(self, path, img_dir_path):
26 self.path = path 26 self.path = path
27 self.img_dir_path = img_dir_path 27 self.img_dir_path = img_dir_path
28 self.img_info_list = [] 28 self.img_path_list = []
29 self.xref_set = set() 29 self.xref_set = set()
30 30
31 def get_img_save_path(self, pno, img_index=0, ext='png'): 31 def get_img_save_path(self, pno, img_index=0, ext='png'):
...@@ -38,7 +38,7 @@ class PDFHandler: ...@@ -38,7 +38,7 @@ class PDFHandler:
38 pm = page.getPixmap(matrix=trans_2, alpha=False) 38 pm = page.getPixmap(matrix=trans_2, alpha=False)
39 img_save_path = self.get_img_save_path(page.number) 39 img_save_path = self.get_img_save_path(page.number)
40 pm.writePNG(img_save_path) 40 pm.writePNG(img_save_path)
41 self.img_info_list.append((img_save_path, page.number, 0)) 41 self.img_path_list.append(img_save_path)
42 42
43 @staticmethod 43 @staticmethod
44 def getimage(pix): 44 def getimage(pix):
...@@ -88,7 +88,7 @@ class PDFHandler: ...@@ -88,7 +88,7 @@ class PDFHandler:
88 with open(img_save_path, "wb") as f: 88 with open(img_save_path, "wb") as f:
89 f.write(img_data) 89 f.write(img_data)
90 self.xref_set.add(xref) 90 self.xref_set.add(xref)
91 self.img_info_list.append((img_save_path, pno, img_index)) 91 self.img_path_list.append(img_save_path)
92 92
93 @staticmethod 93 @staticmethod
94 def split_il(il): 94 def split_il(il):
...@@ -179,7 +179,7 @@ class PDFHandler: ...@@ -179,7 +179,7 @@ class PDFHandler:
179 img_save_path = self.get_img_save_path(pno, img_index, im_list[0][2]) 179 img_save_path = self.get_img_save_path(pno, img_index, im_list[0][2])
180 new_img.save(img_save_path) 180 new_img.save(img_save_path)
181 page_to_png = False 181 page_to_png = False
182 self.img_info_list.append((img_save_path, pno, img_index)) 182 self.img_path_list.append(img_save_path)
183 183
184 # 3.3 碎图分组大于2、全过滤、含特殊filter，特殊处理：整个页面保存为png图片 184 # 3.3 碎图分组大于2、全过滤、含特殊filter，特殊处理：整个页面保存为png图片
185 if page_to_png: 185 if page_to_png:
......
Styling with Markdown is supported
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!