1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513151415151516151715181519152015211522152315241525152615271528152915301531153215331534153515361537153815391540154115421543154415451546154715481549155015511552155315541555155615571558155915601561156215631564156515661567156815691570157115721573157415751576157715781579158015811582158315841585158615871588158915901591159215931594159515961597159815991600160116021603160416051606160716081609161016111612161316141615161616171618161916201621162216231624162516261627162816291630163116321633163416351636163716381639164016411642164316441645164616471648164916501651165216531654165516561657165816591660166116621663166416651666166716681669167016711672167316741675167616771678167916801681168216831684168516861687168816891690169116921693169416951696169716981699170017011702170317041705170617071708170917101711171217131714171517161717171817191720172117221723172417251726172717281729173017311732173317341735173617371738173917401741174217431744174517461747174817491750175117521753175417551756175717581759176017611762176317641765176617671768176917701771177217731774177517761777177817791780178117821783178417851786178717881789179017911792179317941795179617971798179918001801180218031804180518061807180818091810181118121813181418151816181718181819182018211822182318241825182618271828182918301831183218331834183518361837183818391840184118421843184418451846184718481849185018511852185318541855185618571858185918601861186218631864186518661867186818691870187118721873187418751876187718781879188018811882188318841885188618871888188918901891189218931894189518961897189818991900190119021903190419051906190719081909191019111912191319141915191619171918191919201921192219231924192519261927192819291930193119321933193419351936193719381939194019411942194319441945194619471948194919501951195219531954195519561957195819591960196119621963196419651966196719681969197019711972197319741975197619771978197919801981198219831984198519861987198819891990199119921993199419951996199719981999200020012002200320042005200620072008200920102011201220132014201520162017201820192020202120222023202420252026202720282029203020312032203320342035203620372038203920402041204220432044204520462047204820492050205120522053205420552056205720582059206020612062206320642065206620672068206920702071207220732074207520762077207820792080208120822083208420852086208720882089209020912092209320942095209620972098209921002101210221032104210521062107210821092110211121122113211421152116211721182119212021212122212321242125212621272128212921302131213221332134213521362137213821392140214121422143214421452146214721482149215021512152215321542155215621572158215921602161216221632164216521662167216821692170217121722173217421752176217721782179218021812182218321842185218621872188218921902191219221932194219521962197219821992200220122022203220422052206220722082209221022112212221322142215221622172218221922202221222222232224222522262227222822292230223122322233223422352236223722382239224022412242224322442245224622472248224922502251225222532254225522562257225822592260226122622263226422652266226722682269227022712272227322742275227622772278227922802281228222832284228522862287228822892290229122922293229422952296229722982299230023012302230323042305230623072308230923102311231223132314231523162317231823192320232123222323232423252326232723282329233023312332233323342335233623372338233923402341234223432344234523462347234823492350235123522353235423552356235723582359236023612362236323642365236623672368236923702371237223732374237523762377237823792380238123822383238423852386238723882389239023912392239323942395239623972398239924002401240224032404240524062407240824092410241124122413241424152416241724182419242024212422242324242425242624272428242924302431243224332434243524362437243824392440244124422443244424452446244724482449245024512452245324542455245624572458245924602461246224632464246524662467246824692470247124722473247424752476247724782479248024812482248324842485248624872488248924902491249224932494249524962497249824992500250125022503250425052506250725082509251025112512251325142515251625172518251925202521252225232524252525262527252825292530253125322533253425352536253725382539254025412542254325442545254625472548254925502551255225532554255525562557255825592560256125622563256425652566256725682569257025712572257325742575257625772578257925802581258225832584258525862587258825892590259125922593259425952596259725982599260026012602260326042605260626072608260926102611261226132614261526162617261826192620262126222623262426252626262726282629263026312632263326342635263626372638263926402641264226432644264526462647264826492650265126522653265426552656265726582659266026612662266326642665266626672668266926702671267226732674267526762677267826792680268126822683268426852686268726882689269026912692269326942695269626972698269927002701270227032704270527062707270827092710271127122713271427152716271727182719272027212722272327242725272627272728272927302731273227332734273527362737273827392740274127422743274427452746274727482749275027512752275327542755275627572758275927602761276227632764276527662767276827692770277127722773277427752776277727782779278027812782278327842785278627872788278927902791279227932794279527962797279827992800280128022803280428052806280728082809281028112812281328142815281628172818281928202821282228232824282528262827282828292830283128322833283428352836283728382839284028412842284328442845284628472848284928502851285228532854285528562857285828592860286128622863286428652866286728682869287028712872287328742875287628772878287928802881288228832884288528862887288828892890289128922893289428952896289728982899290029012902290329042905290629072908290929102911291229132914291529162917291829192920292129222923292429252926292729282929293029312932 |
- import re
- def find_repeat(source, elmt): # 去重后重新定位数字索引
- elmt_index = []
- s_index = 0
- e_index = len(source)
- while (s_index < e_index):
- try:
- temp = source.index(elmt, s_index, e_index)
- elmt_index.append(temp)
- s_index = temp + 1
- except ValueError:
- break
- return elmt_index
- def ocr_key_words(rect,type_score_dict): # 将ocr识别得到的文字与模型得到的type_score对应
- '''
- :param rect: OCR识别结果数组,格式:res = {'chars': [},'coordinates': [(),()},'words': []}
- :param type_score_dict: 模型得到的type_score(与模型得到的边框相对应)
- :return: 字典中添加word
- '''
- len_ocr = len(rect['chars'])
- xmin = type_score_dict['type_box'][0]
- ymin = type_score_dict['type_box'][1]
- xmax = type_score_dict['type_box'][2]
- ymax = type_score_dict['type_box'][3]
- words=[]
- for j in range(len_ocr):
- if rect['coordinates'][j][0] - xmin > -30 and rect['coordinates'][j][1] - ymin > -30 and rect['coordinates'][j][2] - xmax < 30 and rect['coordinates'][j][3] - ymax < 30:
- word = rect['chars'][j]
- words.append(word)
- type_score_dict['words']= words
- type_score_dict_ocr = type_score_dict
- return type_score_dict_ocr
- def key_words(type_score_dict_ocr): # 根据OCR结果结合关键字解析
- total_score = 0
- volume_score = 0
- volume_structure_item = 0
- volume_structure = []
- Score_structure_item = 0
- Score_structure = []
- all_structure = {}
- keyword_volume = ['第卷', '第部']
- keyword_type = ['选择', '非选择题', '综合题', '问答题', '主观题', '客观题', '解答题','计算题']
- len_keyword_type = len(keyword_type)
- keyword_item1 = ['共分', '合计分', '总共分', '总计分', '小题满分', '本小题', '满分', '共计', '共.分', '合计.分', '总共.分', '总计.分', '小题满分.','本小题.', '满分.', '共计.']
- len_keyword_item1 = len(keyword_item1)
- keyword_item2 = ['每题分', '每小题分', '空分', '每小题.分', '每题.分', '空.分'] # '分/题'暂未考虑
- len_keyword_item2 = len(keyword_item2)
- keyword_item3 = ['共题', '共小题', '分小题', '本题小题', '共个小题', '分为小题', '分个小题','本大题共小题']
- len_keyword_item3 = len(keyword_item3)
- keyword_item4 = ['分']
- len_keyword_item4 = len(keyword_item4)
- keyword_item5 = ['分/题']
- len_keyword_item5 = len(keyword_item5)
- keyword_item5 = ['题', '.', '、']
- ocr_1 = type_score_dict_ocr['words']
- s = ''.join((str(x) for x in ocr_1)) # 合并为一个字符串
- if s.find('IV') != -1 or s.find('Ⅳ') != -1:
- s = s.replace('Ⅳ', '4')
- s = s.replace('IV', '4')
- elif s.find('III') != -1 or s.find('Ⅲ') != -1:
- s = s.replace('Ⅲ', '3')
- s = s.replace('III', '3')
- elif s.find('II') != -1 or s.find('Ⅱ') != -1:
- s = s.replace('Ⅱ', '2')
- s = s.replace('II', '2')
- elif s.find('VI') != -1 or s.find('Ⅵ') != -1:
- s = s.replace('Ⅵ', '6')
- s = s.replace('VI', '6')
- elif s.find('VII') != -1 or s.find('Ⅶ') != -1:
- s = s.replace('Ⅶ', '7')
- s = s.replace('VII', '7')
- elif s.find('VIII') != -1 or s.find('Ⅷ') != -1:
- s = s.replace('Ⅷ', '8')
- s = s.replace('VIII', '8')
- elif s.find('IX') != -1 or s.find('Ⅸ') != -1:
- s = s.replace('Ⅸ', '9')
- s = s.replace('IX', '9')
- elif s.find('X') != -1 or s.find('Ⅹ') != -1:
- s = s.replace('Ⅹ', '10')
- s = s.replace('X', '10')
- elif s.find('I') != -1 or s.find('Ⅰ') != -1:
- s = s.replace('Ⅰ', '1')
- s = s.replace('I', '1')
- elif s.find('V') != -1 or s.find('Ⅴ') != -1:
- s = s.replace('Ⅴ', '5')
- s = s.replace('V', '5')
- C_s = re.sub("[A-Za-z0-9\!\%\[\]\,\。]", "", s) # 提取汉字
- E_s = ''.join(re.findall(r'[A-Za-z]', s)) # 提取英文字符
- N_s = re.findall('\d+', s) # 提取阿拉伯数字
- if len(N_s) == 1 and len(N_s[0]) < 6 and len(E_s) == 0 and (C_s == keyword_item5[0] or C_s == keyword_item5[1] or len(C_s) == 0):
- type_score_dict_ocr['item_N'] = int(N_s[0])
- type_score_dict_ocr['item_total_score'] = -1
- type_score_dict_ocr['item_count'] = -1
- type_score_dict_ocr['item_score'] = -1
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- elif N_s != []:
- for iiii in range(len(keyword_volume)):
- Score_structure_item = {}
- if C_s.find(keyword_volume[iiii]) != -1:
- '''
- 对应试卷中存在分卷信息的情况,根据包含数字的个数分为5类,暂定包含信息的有效数字个数小于5,并处理小题分数和总分可能包含小数点的情况
- 暂定小题个数不包含小数
- 暂定总分数中不存在有意义的小数位
- '''
- if len(N_s) == 1:
- num_index = s.index(N_s[0])
- num_infer = s[num_index - len(N_s[0])]
- num_back = s[num_index + len(N_s[0])]
- if num_back == '分': # 第卷/部*分
- volume_score = int(N_s[0])
- type_score_dict_ocr['volume_N'] = -1
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = -1
- type_score_dict_ocr['volume_score'] = -1
- elif num_back == '卷' or num_back == '部': # 第*卷
- volume_N = int(N_s[0])
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = -1
- type_score_dict_ocr['volume_count'] = -1
- type_score_dict_ocr['volume_score'] = -1
- elif len(N_s) == 2:
- num_index1 = s.index(N_s[0])
- num_infer1 = s[num_index1 - len(N_s[0])]
- num_back1 = s[num_index1 + len(N_s[0])]
- all_1 = find_repeat(s, N_s[1])
- temp1 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[1]:
- temp1 = temp1 + 1
- num_index2 = all_1[temp1]
- num_infer2 = s[num_index2 - len(N_s[1])]
- num_back2 = s[num_index2 + len(N_s[1])]
- if isinstance(N_s[0], str):
- N_s[0] = int(N_s[0])
- if isinstance(N_s[1], str):
- N_s[1] = int(N_s[1])
- for k in range(len_keyword_item1):
- if C_s.find(keyword_item1[k]) != -1:
- if (num_back1 == '卷' or num_back1 == '部') and num_back2 == '分': # 第*卷*分
- volume_N = N_s[0]
- volume_score = N_s[1]
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = -1
- type_score_dict_ocr['volume_score'] = -1
- break
- elif num_back1 == '.' and num_infer2 == '.' and num_back2 == '分': # 第卷,共*.*分
- volume_N = -1
- volume_score = N_s[0]
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = -1
- type_score_dict_ocr['volume_score'] = -1
- break
- else:
- for l in range(len_keyword_item2):
- if C_s.find(keyword_item2[l]) != -1:
- if (num_infer1 == '题' or num_infer1 == '空') and num_back2 == '分': # 第卷,每小题*分,共*分
- volume_score = N_s[1]
- item_score = N_s[0]
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = -1
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_infer2 == '题' or num_infer2 == '空') and num_back1 == '分': # 第卷,共*分 ,每小题*分
- volume_score = N_s[0]
- item_score = N_s[1]
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = -1
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif l == len(keyword_item2) - 1:
- for m in range(len_keyword_item3):
- if C_s.find(keyword_item3[l]) != -1:
- if num_back2 == '分': # 第卷,共*小题,共*分
- volume_score = N_s[1]
- item_count = N_s[0]
- item_score = volume_score / item_count
- type_score_dict_ocr['volume_N'] = -1
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_back1 == '分': # 第卷,共*分 ,共*小题
- volume_score = N_s[0]
- item_count = N_s[1]
- item_score = volume_score / item_count
- type_score_dict_ocr['volume_N'] = -1
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- break
- break
- elif k == len_keyword_item1 - 1:
- for l in range(len_keyword_item2):
- if C_s.find(keyword_item2[l]) != -1:
- for m in range(len_keyword_item3):
- if C_s.find(keyword_item3[l]) != -1:
- if num_back2 == '分': # 第卷,共*小题,每小题*分
- item_count = N_s[0]
- item_score = N_s[1]
- volume_score = item_score * item_count
- type_score_dict_ocr['volume_N'] = -1
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_back1 == '分': # 第卷,每小题*分 ,共*小题
- item_count = N_s[1]
- item_score = N_s[0]
- volume_score = item_count * item_score
- type_score_dict_ocr['volume_N'] = -1
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif m == len_keyword_item3 - 1:
- if num_back2 == '分': # 第卷,每小题*.*分
- volume_score = -1
- item_count = -1
- item_score = float(N_s[0] + '.' + N_s[1])
- type_score_dict_ocr['volume_N'] = -1
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- break
- break
- elif len(N_s) == 3:
- num_index1 = s.index(N_s[0])
- num_infer1 = s[num_index1 - len(N_s[0])]
- num_back1 = s[num_index1 + len(N_s[0])]
- all_1 = find_repeat(s, N_s[1])
- temp1 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[1]:
- temp1 = temp1 + 1
- num_index2 = all_1[temp1]
- num_infer2 = s[num_index2 - len(N_s[1])]
- num_back2 = s[num_index2 + len(N_s[1])]
- all_2 = find_repeat(s, N_s[2])
- temp2 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[2]:
- temp2 = temp2 + 1
- for jj in range(len(N_s[1])):
- if N_s[1][jj] == N_s[2]:
- temp2 = temp2 + 1
- num_index3 = all_2[temp2]
- num_infer3 = s[num_index3 - len(N_s[2])]
- num_back3 = s[num_index3 + len(N_s[2])]
- if isinstance(N_s[0], str):
- N_s[0] = int(N_s[0])
- if isinstance(N_s[1], str):
- N_s[1] = int(N_s[1])
- if isinstance(N_s[2], str):
- N_s[2] = int(N_s[2])
- for l in range(len_keyword_item3):
- if C_s.find(keyword_item3[l]) != -1:
- for m in range(len_keyword_item2):
- if C_s.find(keyword_item2[m]) != -1:
- if (num_back1 == '卷' or num_back1 == '部') and num_back3 == '分': # 第*卷,共*题,每题*分
- volume_N = N_s[0]
- item_count = N_s[1]
- item_score = N_s[2]
- volume_score = item_count * item_score
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_back1 == '卷' or num_back1 == '部') and num_back2 == '分': # 第*卷,每题*分,共*题
- volume_N = N_s[0]
- item_count = N_s[2]
- item_score = N_s[1]
- volume_score = item_count * item_score
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_back1 == '卷' or num_back1 == '部') and num_back2 == '分': # 第卷,每题*.*分,共*题
- volume_N = -1
- item_score = float(N_s[0] + '.' + N_s[1])
- item_count = N_s[2]
- volume_score = item_score * item_count
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_back1 == '卷' or num_back1 == '部') and num_back3 == '分': # 第卷,共*题,每题*.*分
- volume_N = -1
- item_score = float(N_s[1] + '.' + N_s[2])
- item_count = N_s[0]
- volume_score = item_score * item_count
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_back1 == '分' and (num_infer2 == '题' or num_infer2 == '空') and num_back2 == '分': # 第卷,共*分,每题*分,共*题
- volume_score = N_s[0]
- item_count = N_s[2]
- item_score = N_s[1]
- type_score_dict_ocr['volume_N'] = -1
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_back1 == '分' and (num_infer3 == '题' or num_infer3 == '空') and num_back3 == '分': # 第卷,共*分,共*题,每题*分
- volume_score = N_s[0]
- item_count = N_s[1]
- item_score = N_s[2]
- type_score_dict_ocr['volume_N'] = -1
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_back2 == '分' and (num_infer3 == '题' or num_infer3 == '空') and num_back3 == '分': # 第卷,共*题,共*分,每题*分
- volume_score = N_s[1]
- item_count = N_s[0]
- item_score = N_s[2]
- type_score_dict_ocr['volume_N'] = -1
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_infer3 == '分' and (num_infer1 == '题' or num_infer1 == '空') and num_back1 == '分': # 第卷,每题*分,共*题,共*分
- volume_score = N_s[2]
- item_count = N_s[1]
- item_score = N_s[0]
- type_score_dict_ocr['volume_N'] = -1
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_back3 == '分' and (num_infer2 == '题' or num_infer2 == '空') and num_back2 == '分': # 第卷,共*题,每题*分,共*分
- volume_score = N_s[2]
- item_count = N_s[0]
- item_score = N_s[1]
- type_score_dict_ocr['volume_N'] = -1
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_back3 == '分' and (num_infer1 == '题' or num_infer1 == '空') and num_back1 == '分': # 第卷,每题*分,共*题,共*分
- volume_score = N_s[2]
- item_count = N_s[1]
- item_score = N_s[0]
- type_score_dict_ocr['volume_N'] = -1
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif m == len_keyword_item2 - 1:
- for n in range(len_keyword_item1):
- if C_s.find(keyword_item1[n]) != -1:
- if (num_back1 == '卷' or num_back1 == '部') and (num_back2 == '题' or num_back2 == '小') and num_back3 == '分': # 第*卷,共*题,共*分
- volume_N = N_s[0]
- volume_score = N_s[2]
- item_count = N_s[1]
- item_score = volume_score / item_count
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_back1 == '卷' or num_back1 == '部') and (num_back3 == '题' or num_back3 == '小') and num_back2 == '分': # 第*卷,共*分,共*题
- volume_N = N_s[0]
- volume_score = N_s[1]
- item_count = N_s[2]
- item_score = volume_score / item_count
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_back1 == '.' and num_infer2 == '.' and (num_back3 == '题' or num_back3 == '小') and num_back2 == '分': # 第卷,共*.*分,共*题
- volume_N = -1
- volume_score = N_s[0]
- item_count = N_s[2]
- item_score = volume_score / item_count
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_back2 == '.' and num_infer3 == '.' and (num_back1 == '题' or num_back1 == '小') and num_back3 == '分': # 第卷,共*题,共*.*分
- volume_N = -1
- volume_score = N_s[1]
- item_count = N_s[0]
- item_score = volume_score / item_count
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- break
- break
- elif l == len_keyword_item3 - 1:
- for p in range(len_keyword_item1):
- if C_s.find(keyword_item1[p]) != -1:
- for q in range(len_keyword_item2):
- if C_s.find(keyword_item2[q]) != -1:
- if (num_back1 == '卷' or num_back1 == '部') and num_back2 == '分' and num_back3 == '分': # 第*卷,共*分,每题*分 / 第*卷,每题*分,共*分
- volume_N = int(N_s[0])
- if N_s[1] > N_s[2]:
- volume_score = N_s[1]
- item_score = N_s[2]
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- else:
- volume_score = N_s[2]
- item_score = N_s[1]
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_back1 == '.' and num_infer2 == '.' and num_back2 == '分' and num_back3 == '分': # 第卷,共*.*分,每题*分 / 第卷,每题*.*分,共*分
- volume_N = -1
- if N_s[0] > N_s[2]:
- volume_score = N_s[0]
- item_score = N_s[2]
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- else:
- volume_score = N_s[2]
- item_score = float(N_s[0] + '.' + N_s[1])
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_back2 == '.' and num_infer3 == '.' and num_back1 == '分' and num_back3 == '分': # 第卷,共*分,每题*.*分 / 第卷,每题*分,共*.*分
- volume_N = -1
- if N_s[0] > N_s[1]:
- volume_score = N_s[0]
- item_score = float(N_s[1] + '.' + N_s[2])
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- else:
- volume_score = N_s[1]
- item_score = N_s[0]
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- break
- break
- elif len(N_s) == 4:
- num_index1 = s.index(N_s[0])
- num_infer1 = s[num_index1 - len(N_s[0])]
- num_back1 = s[num_index1 + len(N_s[0])]
- all_1 = find_repeat(s, N_s[1])
- temp1 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[1]:
- temp1 = temp1 + 1
- num_index2 = all_1[temp1]
- num_infer2 = s[num_index2 - len(N_s[1])]
- num_back2 = s[num_index2 + len(N_s[1])]
- all_2 = find_repeat(s, N_s[2])
- temp2 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[2]:
- temp2 = temp2 + 1
- for jj in range(len(N_s[1])):
- if N_s[1][jj] == N_s[2]:
- temp2 = temp2 + 1
- num_index3 = all_2[temp2]
- num_infer3 = s[num_index3 - len(N_s[2])]
- num_back3 = s[num_index3 + len(N_s[2])]
- all_3 = find_repeat(s, N_s[3])
- temp3 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[3]:
- temp3 = temp3 + 1
- for jj in range(len(N_s[1])):
- if N_s[1][jj] == N_s[3]:
- temp3 = temp3 + 1
- for kk in range(len(N_s[2])):
- if N_s[2][kk] == N_s[3]:
- temp3 = temp3 + 1
- num_index4 = all_3[temp3]
- num_infer4 = s[num_index4 - len(N_s[3])]
- num_back4 = s[num_index4 + len(N_s[3])]
- if isinstance(N_s[0], str):
- N_s[0] = int(N_s[0])
- if isinstance(N_s[1], str):
- N_s[1] = int(N_s[1])
- if isinstance(N_s[2], str):
- N_s[2] = int(N_s[2])
- if isinstance(N_s[3], str):
- N_s[3] = int(N_s[3])
- for l in range(len_keyword_item1):
- if C_s.find(keyword_item1[l]) != -1:
- for m in range(len_keyword_item2):
- if C_s.find(keyword_item2[m]) != -1:
- for n in range(len_keyword_item3):
- if C_s.find(keyword_item3[n]) != -1:
- if (num_back1 == '卷' or num_back1 == '部') and (num_infer2 == '题' or num_infer2 == '空') and num_back2 == '分' and num_back4 == '分': # 第*卷,每题*分,共*题,共*分
- volume_N = N_s[0]
- volume_score = N_s[3]
- item_count = N_s[2]
- item_score = N_s[1]
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_back1 == '卷' or num_back1 == '部') and (num_infer3 == '题' or num_infer3 == '空') and num_back3 == '分' and num_back4 == '分': # 第*卷,共*题,每题*分,共*分
- volume_N = N_s[0]
- volume_score = N_s[3]
- item_count = N_s[1]
- item_score = N_s[2]
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_back1 == '卷' or num_back1 == '部') and (num_infer4 == '题' or num_infer4 == '空') and num_back4 == '分' and num_back2 == '分': # 第*卷,共*分,共*题,每题*分
- volume_N = N_s[0]
- volume_score = N_s[1]
- item_count = N_s[2]
- item_score = N_s[3]
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_back1 == '卷' or num_back1 == '部') and (num_infer4 == '题' or num_infer4 == '空') and num_back4 == '分' and num_back3 == '分': # 第*卷,共*题,共*分,每题*分
- volume_N = N_s[0]
- volume_score = N_s[2]
- item_count = N_s[1]
- item_score = N_s[3]
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_back1 == '卷' or num_back1 == '部') and (num_infer3 == '题' or num_infer3 == '空') and num_back3 == '分' and num_back2 == '分': # 第*卷,共*分,每题*分,共*题
- volume_N = N_s[0]
- volume_score = N_s[1]
- item_count = N_s[3]
- item_score = N_s[2]
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_back1 == '卷' or num_back1 == '部') and (num_infer2 == '题' or num_infer2 == '空') and num_back2 == '分' and num_back3 == '分': # 第*卷,每题*分,共*分,共*题
- volume_N = N_s[0]
- volume_score = N_s[2]
- item_count = N_s[3]
- item_score = N_s[1]
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_back1 == '.' and num_infer2 == '.') and num_back2 == '分' and num_back3 == '分': # 第卷,每题*.*分,共*分,共*题/第卷,共*.*分,每题*分,共*题
- volume_N = -1
- if int(N_s[0]) > int(N_s[2]):
- volume_score = N_s[0]
- item_score = N_s[2]
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- else:
- volume_score = N_s[2]
- item_score = float(N_s[0] + '.' + N_s[1])
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_back2 == '.' and num_infer3 == '.') and num_back1 == '分' and num_back3 == '分': # 第卷,每题*分,共*.*分,共*题/第卷,共*分,每题*.*分,共*题
- volume_N = -1
- if int(N_s[0]) > int(N_s[1]):
- volume_score = N_s[0]
- item_score = float(N_s[1] + '.' + N_s[2])
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- else:
- volume_score = N_s[1]
- item_score = N_s[0]
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_back2 == '.' and num_infer3 == '.') and num_back3 == '分' and num_back4 == '分': # 第卷,共*题,共*.*分,每题*分/第卷,共*题,每题*.*分,共*分
- volume_N = -1
- if N_s[1] > N_s[3]:
- volume_score = N_s[1]
- item_score = N_s[3]
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- else:
- volume_score = N_s[4]
- item_score = float(N_s[1] + '.' + N_s[2])
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_back3 == '.' and num_infer4 == '.') and num_back2 == '分' and num_back4 == '分': # 第卷,共*题,共*分,每题*.*分/第卷,共*题,每题*分,共*.*分
- volume_N = -1
- if int(N_s[1]) > int(N_s[2]):
- volume_score = N_s[1]
- item_score = float(N_s[2] + '.' + N_s[3])
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- else:
- volume_score = N_s[2]
- item_score = N_s[1]
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_back1 == '.' and num_infer2 == '.') and num_back2 == '分' and num_back4 == '分': # 第卷,共*.*分,共*题,每题*分/第卷,每题*.*分,共*题,共*分
- volume_N = -1
- if int(N_s[0]) > int(N_s[3]):
- volume_score = N_s[0]
- item_score = N_s[3]
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- else:
- volume_score = N_s[3]
- item_score = float(N_s[0] + '.' + N_s[1])
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_back3 == '.' and num_infer4 == '.') and num_back1 == '分' and num_back4 == '分': # 第卷,共*分,共*题,每题*.*分/第卷,每题*分,共*题,共*.*分
- volume_N = -1
- if int(N_s[0]) > int(N_s[2]):
- volume_score = N_s[0]
- item_score = N_s[2] + '.' + N_s[3]
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- else:
- volume_score = N_s[2]
- item_score = N_s[0]
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif n == len_keyword_item3 - 1:
- if (num_back1 == '卷' or num_back1 == '部') and num_back2 == '.' and num_infer3 == '.' and num_back3 == '分' and num_back4 == '分': # 第*卷,每题*.*分,共*分/第*卷,共*.*分,每题*分
- volume_N = int(N_s[0])
- if N_s[1] > N_s[3]:
- volume_score = N_s[1]
- item_score = N_s[3]
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- else:
- volume_score = N_s[3]
- item_score = float(N_s[1] + '.' + N_s[2])
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_back1 == '卷' or num_back1 == '部') and num_back3 == '.' and num_infer4 == '.' and num_back2 == '分' and num_back4 == '分': # 第*卷,每题*分,共*.*分/第*卷,共*分,每题*.*分
- volume_N = int(N_s[0])
- if int(N_s[1]) > int(N_s[2]):
- volume_score = N_s[1]
- item_score = float(N_s[2] + '.' + N_s[3])
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- else:
- volume_score = N_s[2]
- item_score = N_s[1]
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_back1 == '.' and num_infer2 == '.' and num_back2 == '分' and num_back3 == '.' and num_infer4 == '.' and num_back4 == '分': # 第卷,每题*.*分,共*.*分/第卷,共*.*分,每题*.*分
- volume_N = -1
- if N_s[0] > N_s[2]:
- volume_score = float(N_s[2] + '.' + N_s[3])
- item_score = N_s[3]
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- else:
- volume_score = N_s[3]
- item_score = float(N_s[1] + '.' + N_s[2])
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_back1 == '卷' or num_back1 == '部') and (num_infer2 == '题' or num_infer2 == '空') and num_back2 == '分' and num_back3 == '分': # 第*卷,每题*分,共*分
- volume_N = N_s[0]
- volume_score = N_s[2]
- item_count = -1
- item_score = N_s[1]
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_back1 == '卷' or num_back1 == '部') and (num_infer3 == '题' or num_infer3 == '空') and num_back2 == '分' and num_back3 == '分': # 第*卷,共*分,每题*分
- volume_N = N_s[0]
- volume_score = N_s[1]
- item_count = -1
- item_score = N_s[2]
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- break
- break
- elif len(N_s) == 5:
- num_index1 = s.index(N_s[0])
- num_infer1 = s[num_index1 - len(N_s[0])]
- num_back1 = s[num_index1 + len(N_s[0])]
- all_1 = find_repeat(s, N_s[1])
- temp1 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[1]:
- temp1 = temp1 + 1
- num_index2 = all_1[temp1]
- num_infer2 = s[num_index2 - len(N_s[1])]
- num_back2 = s[num_index2 + len(N_s[1])]
- all_2 = find_repeat(s, N_s[2])
- temp2 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[2]:
- temp2 = temp2 + 1
- for jj in range(len(N_s[1])):
- if N_s[1][jj] == N_s[2]:
- temp2 = temp2 + 1
- num_index3 = all_2[temp2]
- num_infer3 = s[num_index3 - len(N_s[2])]
- num_back3 = s[num_index3 + len(N_s[2])]
- all_3 = find_repeat(s, N_s[3])
- temp3 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[3]:
- temp3 = temp3 + 1
- for jj in range(len(N_s[1])):
- if N_s[1][jj] == N_s[3]:
- temp3 = temp3 + 1
- for kk in range(len(N_s[2])):
- if N_s[2][kk] == N_s[3]:
- temp3 = temp3 + 1
- num_index4 = all_3[temp3]
- num_infer4 = s[num_index4 - len(N_s[3])]
- num_back4 = s[num_index4 + len(N_s[3])]
- all_4 = find_repeat(s, N_s[4])
- temp4 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[4]:
- temp4 = temp4 + 1
- for jj in range(len(N_s[1])):
- if N_s[1][jj] == N_s[4]:
- temp4 = temp4 + 1
- for kk in range(len(N_s[2])):
- if N_s[2][kk] == N_s[4]:
- temp4 = temp4 + 1
- for ll in range(len(N_s[3])):
- if N_s[3][ll] == N_s[4]:
- temp4 = temp4 + 1
- num_index5 = all_4[temp4]
- num_infer5 = s[num_index5 - len(N_s[4])]
- num_back5 = s[num_index5 + len(N_s[4])]
- if isinstance(N_s[0], str):
- N_s[0] = int(N_s[0])
- if isinstance(N_s[1], str):
- N_s[1] = int(N_s[1])
- if isinstance(N_s[2], str):
- N_s[2] = int(N_s[2])
- if isinstance(N_s[3], str):
- N_s[3] = int(N_s[3])
- if isinstance(N_s[4], str):
- N_s[4] = int(N_s[4])
- for l in range(len_keyword_item1):
- if C_s.find(keyword_item1[l]) != -1:
- for m in range(len_keyword_item2):
- if C_s.find(keyword_item2[m]) != -1:
- for n in range(len_keyword_item3):
- if C_s.find(keyword_item3[n]) != -1:
- if (num_back1 == '卷' or num_back1 == '部') and (num_back2 == '.' and num_infer3 == '.') and num_back3 == '分' and num_back4 == '分': # 第*卷,每题*.*分,共*分,共*题/第卷,共*.*分,每题*分,共*题
- volume_N = N_s[0]
- if N_s[1] > N_s[3]:
- volume_score = N_s[1]
- item_score = N_s[3]
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- else:
- volume_score = N_s[3]
- item_score = float(N_s[1] + '.' + N_s[2])
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_back1 == '卷' or num_back1 == '部') and (num_back3 == '.' and num_infer4 == '.') and num_back2 == '分' and num_back4 == '分': # 第*卷,每题*分,共*.*分,共*题/第卷,共*分,每题*.*分,共*题
- volume_N = N_s[0]
- if N_s[1] > N_s[2]:
- volume_score = N_s[1]
- item_score = float(N_s[2] + '.' + N_s[3])
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- else:
- volume_score = N_s[2]
- item_score = N_s[1]
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_back1 == '卷' or num_back1 == '部') and (num_back3 == '.' and num_infer4 == '.') and num_back4 == '分' and num_back5 == '分': # 第卷,共*题,共*.*分,每题*分/第卷,共*题,每题*.*分,共*分
- volume_N = N_s[0]
- if N_s[2] > N_s[4]:
- volume_score = N_s[2]
- item_score = N_s[4]
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- else:
- volume_score = N_s[4]
- item_score = float(N_s[2] + '.' + N_s[3])
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_back1 == '卷' or num_back1 == '部') and (num_back4 == '.' and num_infer5 == '.') and num_back3 == '分' and num_back5 == '分': # 第*卷,共*题,共*分,每题*.*分/第卷,共*题,每题*分,共*.*分
- volume_N = N_s[0]
- if N_s[2] > N_s[3]:
- volume_score = N_s[2]
- item_score = float(N_s[3] + '.' + N_s[4])
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- else:
- volume_score = N_s[3]
- item_score = N_s[2]
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_back1 == '卷' or num_back1 == '部') and (num_back2 == '.' and num_infer3 == '.') and num_back3 == '分' and num_back5 == '分': # 第*卷,共*.*分,共*题,每题*分/第*卷,每题*.*分,共*题,共*分
- volume_N = N_s[0]
- if N_s[1] > N_s[4]:
- volume_score = N_s[1]
- item_score = N_s[4]
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- else:
- volume_score = N_s[4]
- item_score = float(N_s[1] + '.' + N_s[2])
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_back1 == '卷' or num_back1 == '部') and (num_back4 == '.' and num_infer5 == '.') and num_back2 == '分' and num_back5 == '分': # 第*卷,共*分,共*题,每题*.*分/第卷,每题*分,共*题,共*.*分
- volume_N = N_s[0]
- if N_s[1] > N_s[3]:
- volume_score = N_s[0]
- item_score = float(N_s[4] + '.' + N_s[4])
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- else:
- volume_score = N_s[3]
- item_score = N_s[1]
- item_count = int(volume_score / item_score)
- type_score_dict_ocr['volume_N'] = volume_N
- type_score_dict_ocr['volume_total_score'] = volume_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- break
- break
- if 'volume_N' not in type_score_dict_ocr.keys():
- all_structure = {'volume_structure': -1,
- 'Score_structure': -1}
- break
- else:
- for i in range(len_keyword_type):
- if C_s.find(keyword_type[i]) != -1 and C_s.find('非') != -1:
- type_score_dict_ocr['keyword_type'] = keyword_type[1]
- break
- elif C_s.find(keyword_type[0]) != -1:
- type_score_dict_ocr['keyword_type'] = keyword_type[0]
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- break
- elif C_s.find(keyword_type[i]) != -1:
- type_score_dict_ocr['keyword_type'] = keyword_type[i]
- break
- elif i == len_keyword_type - 1:
- type_score_dict_ocr['keyword_type'] = keyword_type[0]
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- break
- volume_structure_item = type_score_dict_ocr
- volume_structure.append(volume_structure_item)
- if Score_structure == []:
- all_structure = {'volume_structure': volume_structure,
- 'Score_structure': -1}
- else:
- all_structure = {'volume_structure': volume_structure,
- 'Score_structure': Score_structure}
- break
- elif iiii == len(keyword_volume) - 1:
- '''
- 对应试卷中不存在分卷信息的情况,根据包含数字的个数分为4类,暂定包含信息的有效数字个数小于4,并处理小题分数和总分可能包含小数点的情况
- 暂定小题个数不包含小数
- 暂定总分数中不存在有意义的小数位
- '''
- for xxx in range(len_keyword_type):
- if C_s.find(keyword_type[xxx]) != -1:
- for x in range(len_keyword_item1):
- if C_s.find(keyword_item1[x]) != -1:
- if len(N_s) == 1:
- num_index = s.index(N_s[0])
- num_infer = s[num_index - len(N_s[0])]
- num_back = s[num_index + len(N_s[0])]
- if isinstance(N_s[0], str):
- N_s[0] = int(N_s[0])
- if num_back == '分': # 选择题/主观题,共*分
- item_total_score = N_s[0]
- type_score_dict_ocr['volume_N'] = -1
- type_score_dict_ocr['volume_total_score'] = int(item_total_score)
- type_score_dict_ocr['volume_count'] = -1
- type_score_dict_ocr['volume_score'] = -1
- break
- elif len(N_s) == 2:
- num_index1 = s.index(N_s[0])
- num_infer1 = s[num_index1 - len(N_s[0])]
- num_back1 = s[num_index1 + len(N_s[0])]
- all_1 = find_repeat(s, N_s[1])
- temp1 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[1]:
- temp1 = temp1 + 1
- num_index2 = all_1[temp1]
- num_infer2 = s[num_index2 - len(N_s[1])]
- num_back2 = s[num_index2 + len(N_s[1])]
- if isinstance(N_s[0], str):
- N_s[0] = int(N_s[0])
- if isinstance(N_s[1], str):
- N_s[1] = int(N_s[1])
- for y in range(len_keyword_item2):
- if C_s.find(keyword_item2[y]) != -1:
- if num_back1 == '分' and (num_infer2 == '题' or num_infer2 == '空'): # 选择题/主观题/客观题,共*分,每题*分
- item_total_score = int(N_s[0])
- item_count = int(N_s[0] / N_s[1])
- item_score = N_s[1]
- type_score_dict_ocr['volume_N'] = -1
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif ( num_infer1 == '题' or num_infer1 == '空') and num_back2 == '分': # 选择题/主观题,每题*分,共*分
- item_total_score = N_s[1]
- item_count = int(N_s[1] / N_s[0])
- item_score = N_s[0]
- type_score_dict_ocr['volume_N'] = -1
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif y == len_keyword_item2 - 1:
- for u in range(len_keyword_item3):
- if C_s.find(keyword_item3[u]) != -1:
- if num_back1 == '分': # 选择题/主观题,共*分,共*题
- item_total_score = N_s[0]
- item_count = N_s[1]
- item_score = N_s[0] / N_s[1]
- type_score_dict_ocr['volume_N'] = -1
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_back2 == '分': # 选择题/主观题,共*题,共*分
- item_total_score = N_s[1]
- item_count = N_s[0]
- item_score = N_s[1] / N_s[0]
- type_score_dict_ocr['volume_N'] = -1
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif u == len_keyword_item3 - 1:
- if num_back1 == '.' and num_infer2 == '.' and num_back2 == '分': # *.*分
- item_N = -1
- item_total_score = N_s[0]
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = -1
- type_score_dict_ocr['volume_score'] = -1
- break
- elif num_back2 == '分': # *,*分
- item_N = N_s[0]
- item_total_score = int(N_s[1])
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = -1
- type_score_dict_ocr['volume_score'] = -1
- break
- break
- break
- elif len(N_s) == 3:
- num_index1 = s.index(N_s[0])
- num_infer1 = s[num_index1 - len(N_s[0])]
- num_back1 = s[num_index1 + len(N_s[0])]
- all_1 = find_repeat(s, N_s[1])
- temp1 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[1]:
- temp1 = temp1 + 1
- num_index2 = all_1[temp1]
- num_infer2 = s[num_index2 - len(N_s[1])]
- num_back2 = s[num_index2 + len(N_s[1])]
- all_2 = find_repeat(s, N_s[2])
- temp2 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[2]:
- temp2 = temp2 + 1
- for jj in range(len(N_s[1])):
- if N_s[1][jj] == N_s[2]:
- temp2 = temp2 + 1
- num_index3 = all_2[temp2]
- num_infer3 = s[num_index3 - len(N_s[2])]
- num_back3 = s[num_index3 + len(N_s[2])]
- if isinstance(N_s[0], str):
- N_s[0] = int(N_s[0])
- if isinstance(N_s[1], str):
- N_s[1] = int(N_s[1])
- if isinstance(N_s[2], str):
- N_s[2] = int(N_s[2])
- for v in range(len_keyword_item2):
- if C_s.find(keyword_item2[v]) != -1:
- for w in range(len_keyword_item3):
- if C_s.find(keyword_item3[w]) != -1:
- if (num_infer1 == '题' or num_infer1 == '空') and num_back1 == '分' and num_back3 == '分': # 每题*分,共*题,共*分
- item_total_score = N_s[2]
- item_count = N_s[1]
- item_score = N_s[0]
- if item_total_score < item_count * item_score:
- item_total_score = item_count * item_score
- type_score_dict_ocr['volume_N'] = -1
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_infer1 == '题' or num_infer1 == '空') and num_back1 == '分' and num_back2 == '分': # 每题*分,共*分,共*题
- item_total_score = N_s[1]
- item_count = N_s[2]
- item_score = N_s[0]
- if item_total_score < item_count * item_score:
- item_total_score = item_count * item_score
- type_score_dict_ocr['volume_N'] = -1
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_infer2 == '题' or num_infer2 == '空') and num_back2 == '分' and num_back3 == '分': # 共*题,每题*分,共*分
- item_total_score = N_s[2]
- item_count = N_s[0]
- item_score = N_s[1]
- if item_total_score < item_count * item_score:
- item_total_score = item_count * item_score
- type_score_dict_ocr['volume_N'] = -1
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_infer2 == '题' or num_infer2 == '空') and num_back2 == '分' and num_back1 == '分': # 共*分,每题*分,共*题
- item_total_score = N_s[0]
- item_count = N_s[2]
- item_score = N_s[1]
- if item_total_score < item_count * item_score:
- item_total_score = item_count * item_score
- type_score_dict_ocr['volume_N'] = -1
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_infer3 == '题' or num_infer3 == '空') and num_back3 == '分' and num_back2 == '分': # 共*题,共*分,每题*分
- item_total_score = N_s[1]
- item_count = N_s[0]
- item_score = N_s[2]
- if item_total_score < item_count * item_score:
- item_total_score = item_count * item_score
- type_score_dict_ocr['volume_N'] = -1
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_infer3 == '题' or num_infer3 == '空') and num_back3 == '分' and num_back1 == '分': # 共*分,共*题, 每题*分
- item_total_score = N_s[0]
- item_count = N_s[1]
- item_score = N_s[2]
- if item_total_score < item_count * item_score:
- item_total_score = item_count * item_score
- type_score_dict_ocr['volume_N'] = -1
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif w == len_keyword_item3 - 1:
- if num_back1 != '.' and num_back2 == '分' and (num_infer3 == '题' or num_infer3 == '空') and num_back3 == '分': # *,共*分,每题*分
- item_N = N_s[0]
- item_total_score = N_s[1]
- item_count = int(N_s[1] / N_s[2])
- item_score = N_s[2]
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_back1 != '.' and num_back3 == '分' and (num_infer2 == '题' or num_infer2 == '空') and num_back2 == '分': # *,每题*分,共*分,
- item_N = N_s[0]
- item_total_score = N_s[2]
- item_count = int(N_s[2] / N_s[1])
- item_score = N_s[1]
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_back1 == '.' and num_infer2 == '.' and num_back3 == '分' and num_back2 == '分': # 每题*.*分,共*分/共*.*分,每题*分
- item_N = -1
- if int(N_s[0]) > int(N_s[2]):
- item_total_score = N_s[0]
- item_score = N_s[2]
- item_count = int(item_total_score/item_score)
- else:
- item_total_score = N_s[2]
- item_score = float(N_s[0]+'.'+N_s[1])
- item_count = int(item_total_score / item_score)
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_back2 == '.' and num_infer3 == '.' and num_back1 == '分' and num_back3 == '分': # 每题*分,共*.*分/共*分,每题*.*分
- item_N = -1
- if int(N_s[0]) > int(N_s[1]):
- item_total_score = N_s[0]
- item_score = float(N_s[1]+'.'+N_s[2])
- item_count = int(item_total_score/item_score)
- else:
- item_total_score = N_s[1]
- item_score = N_s[0]
- item_count = int(item_total_score / item_score)
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- break
- elif v == len_keyword_item2 - 1:
- for w in range(len_keyword_item3):
- if C_s.find(keyword_item3[w]) != -1:
- if num_back3 == '分' and num_infer3 =='.' and num_back2 =='.': # *小题,共*.*分,
- item_N = -1
- item_total_score = N_s[1]
- item_count = N_s[0]
- item_score = N_s[1]/N_s[0]
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_back2 == '分' and num_infer2 =='.'and num_back1 =='.': # 共*.*分,*小题
- item_N = -1
- item_total_score = N_s[0]
- item_count = N_s[2]
- item_score = N_s[0]/N_s[2]
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_back3 == '分' and num_infer3 !='.': # *,*小题,共*分,
- item_N = N_s[0]
- item_total_score = N_s[2]
- item_count = N_s[1]
- item_score = N_s[2]/N_s[1]
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_back2 == '分' and num_infer2 !='.': # *,共*分,共*小题
- item_N = N_s[0]
- item_total_score = N_s[1]
- item_count = N_s[2]
- item_score = N_s[1] / N_s[2]
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif w == len_keyword_item3-1:
- if num_back3 == '分' and num_infer3 =='.' and num_back2 =='.': # *,共*.*分,
- item_N = N_s[0]
- item_total_score = N_s[1]
- item_count = -1
- item_score = -1
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_back3 == '分':
- item_total_score = N_s[1]
- item_N = -1
- item_count = -1
- item_score = -1
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- break
- break
- elif len(N_s) == 4:
- num_index1 = s.index(N_s[0])
- num_infer1 = s[num_index1 - len(N_s[0])]
- num_back1 = s[num_index1 + len(N_s[0])]
- all_1 = find_repeat(s, N_s[1])
- temp1 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[1]:
- temp1 = temp1 + 1
- num_index2 = all_1[temp1]
- num_infer2 = s[num_index2 - len(N_s[1])]
- num_back2 = s[num_index2 + len(N_s[1])]
- all_2 = find_repeat(s, N_s[2])
- temp2 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[2]:
- temp2 = temp2 + 1
- for jj in range(len(N_s[1])):
- if N_s[1][jj] == N_s[2]:
- temp2 = temp2 + 1
- num_index3 = all_2[temp2]
- num_infer3 = s[num_index3 - len(N_s[2])]
- num_back3 = s[num_index3 + len(N_s[2])]
- all_3 = find_repeat(s, N_s[3])
- temp3 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[3]:
- temp3 = temp3 + 1
- for jj in range(len(N_s[1])):
- if N_s[1][jj] == N_s[3]:
- temp3 = temp3 + 1
- for kk in range(len(N_s[2])):
- if N_s[2][kk] == N_s[3]:
- temp3 = temp3 + 1
- num_index4 = all_3[temp3]
- num_infer4 = s[num_index4 - len(N_s[3])]
- num_back4 = s[num_index4 + len(N_s[3])]
- if isinstance(N_s[0], str):
- N_s[0] = int(N_s[0])
- if isinstance(N_s[1], str):
- N_s[1] = int(N_s[1])
- if isinstance(N_s[2], str):
- N_s[2] = int(N_s[2])
- if isinstance(N_s[3], str):
- N_s[3] = int(N_s[3])
- for y in range(len_keyword_item1):
- if C_s.find(keyword_item1[y]) != -1:
- for z in range(len_keyword_item2):
- if C_s.find(keyword_item2[z]) != -1:
- for u in range(len_keyword_item3):
- if C_s.find(keyword_item3[u]) != -1:
- if (num_infer2 == '题' or num_infer2 == '空') and num_back2 == '分' and num_back4 == '分': # *,每题*分,共*题,共*分
- item_N = N_s[0]
- item_total_score = N_s[3]
- item_count = N_s[2]
- item_score = N_s[1]
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr[
- 'volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_infer2 == '题' or num_infer2 == '空') and num_back2 == '分' and num_back3 == '分': # *,每题*分,共*分,共*题
- item_N = N_s[0]
- item_total_score = N_s[2]
- item_count = N_s[3]
- item_score = N_s[1]
- type_score_dict_ocr['volume_N'] = item_N
- if item_total_score < item_count * item_score:
- item_total_score = item_count * item_score
- type_score_dict_ocr[
- 'volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_infer3 == '题' or num_infer3 == '空') and num_back3 == '分' and num_back4 == '分': # *,共*题,每题*分,共*分
- item_N = N_s[0]
- item_total_score = N_s[3]
- item_count = N_s[1]
- item_score = N_s[2]
- if item_total_score < item_count * item_score:
- item_total_score = item_count * item_score
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr[
- 'volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_infer3 == '题' or num_infer3 == '空') and num_back3 == '分' and num_back2 == '分': # *,共*分,每题*分,共*题
- item_N = N_s[0]
- item_total_score = N_s[1]
- item_count = N_s[3]
- item_score = N_s[2]
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr[
- 'volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_infer4 == '题' or num_infer4 == '空') and num_back4 == '分' and num_back3 == '分': # *,共*题,共*分,每题*分
- item_N = N_s[0]
- item_total_score = N_s[2]
- item_count = N_s[1]
- item_score = N_s[3]
- type_score_dict_ocr['volume_N'] = item_N
- if item_total_score < item_count * item_score:
- item_total_score = item_count * item_score
- type_score_dict_ocr[
- 'volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif (num_infer4 == '题' or num_infer4 == '空') and num_back4 == '分' and num_back2 == '分': # *,共*分,共*题, 每题*分
- item_N = N_s[0]
- item_total_score = N_s[1]
- item_count = N_s[2]
- item_score = N_s[3]
- type_score_dict_ocr['item_N'] = item_N
- if item_total_score < item_count * item_score:
- item_total_score = item_count * item_score
- type_score_dict_ocr[
- 'item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- break
- elif num_back1== '.' and num_infer2== '.' and num_back2 == '分'and num_back4 == '分' : # 共*.*分,共*题, 每题*分/每题*.*分,共*题,共*分
- item_N = -1
- if N_s[0] > N_s[3]:
- item_total_score = N_s[0]
- item_score = N_s[3]
- item_count = int(item_total_score / item_score)
- else:
- item_total_score = N_s[3]
- item_score = float(N_s[0] + '.' + N_s[1])
- item_count = int(item_total_score / item_score)
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr[ 'item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- break
- elif num_back3== '.' and num_infer4== '.' and num_back1 == '分'and num_back4 == '分' : # 共*分,共*题, 每题*.*分/每题*分,共*题,共*.*分
- item_N = -1
- if N_s[0] > N_s[2]:
- item_total_score = N_s[0]
- item_score = float(N_s[2] + '.' + N_s[3])
- item_count = int(item_total_score / item_score)
- else:
- item_total_score = N_s[2]
- item_score = N_s[0]
- item_count = int(item_total_score / item_score)
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr[ 'item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- break
- elif num_back2== '.' and num_infer3== '.' and num_back3 == '分'and num_back4 == '分' : # 共*题,共*.*分,每题*分/共*题,每题*.*分,共*分
- item_N = -1
- if N_s[1] > N_s[3]:
- item_total_score = N_s[1]
- item_score = N_s[3]
- item_count = int(item_total_score / item_score)
- else:
- item_total_score = N_s[3]
- item_score = float(N_s[1] + '.' + N_s[2])
- item_count = int(item_total_score / item_score)
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_back3== '.' and num_infer4== '.' and num_back4 == '分'and num_back2 == '分' : # 共*题,共*分,每题*.*分/共*题,每题*分,共*.*分
- item_N = -1
- if N_s[1] > N_s[2]:
- item_total_score = N_s[1]
- item_score = float(N_s[2] + '.' + N_s[3])
- item_count = int(item_total_score / item_score)
- else:
- item_total_score = N_s[2]
- item_score = N_s[1]
- item_count = int(item_total_score / item_score)
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_back1== '.' and num_infer2== '.' and num_back2 == '分'and num_back3 == '分' : # 每题*.*分,共*分,共*题/共*.*分,每题*分,共*题
- item_N = -1
- if N_s[0] > N_s[2]:
- item_total_score = N_s[0]
- item_score = N_s[2]
- item_count = int(item_total_score / item_score)
- else:
- item_total_score = N_s[2]
- item_score = float(N_s[0] + '.' + N_s[1])
- item_count = int(item_total_score / item_score)
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_back2== '.' and num_infer3== '.' and num_back3 == '分'and num_back1 == '分' : # 每题*分,共*.*分,共*题/共*分,每题*.*分,共*题
- item_N = -1
- if N_s[0] > N_s[1]:
- item_total_score = N_s[0]
- item_score = float(N_s[1] + '.' + N_s[2])
- item_count = int(item_total_score / item_score)
- else:
- item_total_score = N_s[1]
- item_score = N_s[0]
- item_count = int(item_total_score / item_score)
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif u == len_keyword_item3-1:
- if num_back2== '.' and num_infer3== '.' and num_back3 == '分'and num_back4 == '分' : # *,共*.*分, 每题*分/每题*.*分,共*分
- item_N = N_s[0]
- if N_s[1] > N_s[3]:
- item_total_score = N_s[1]
- item_score = N_s[3]
- item_count = int(item_total_score / item_score)
- else:
- item_total_score = N_s[3]
- item_score = float(N_s[1] + '.' + N_s[2])
- item_count = int(item_total_score / item_score)
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr[ 'item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- break
- elif num_back3== '.' and num_infer4== '.' and num_back2 == '分'and num_back4 == '分' : # *,共*分, 每题*.*分/*,每题*分,共*.*分
- item_N = int(N_s[0])
- if N_s[1] > N_s[2]:
- item_total_score = N_s[1]
- item_score = float(N_s[2] + '.' + N_s[3])
- item_count = int(item_total_score / item_score)
- else:
- item_total_score = N_s[2]
- item_score = N_s[1]
- item_count = int(item_total_score / item_score)
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr[ 'item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- break
- elif num_back2== '.' and num_infer3== '.' and num_back3 == '分'and num_back4 == '分' : # *,共*.*分,每题*分/*,每题*.*分,共*分
- item_N = N_s[0]
- if N_s[1] > N_s[3]:
- item_total_score = N_s[1]
- item_score = N_s[3]
- item_count = int(item_total_score / item_score)
- else:
- item_total_score = N_s[3]
- item_score = float(N_s[1] + '.' + N_s[2])
- item_count = int(item_total_score / item_score)
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_back3== '.' and num_infer4== '.' and num_back4 == '分'and num_back2 == '分' : # *,共*分,每题*.*分/*,每题*分,共*.*分
- item_N = N_s[0]
- if N_s[1] > N_s[2]:
- item_total_score = N_s[1]
- item_score = float(N_s[2] + '.' + N_s[3])
- item_count = int(item_total_score / item_score)
- else:
- item_total_score = N_s[2]
- item_score = N_s[1]
- item_count = int(item_total_score / item_score)
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_back1== '.' and num_infer2== '.' and num_back2 == '分'and num_back3 == '分' : # *,每题*.*分,共*分/*,共*.*分,每题*分
- item_N = N_s[0]
- if N_s[1] > N_s[3]:
- item_total_score = N_s[1]
- item_score = N_s[3]
- item_count = int(item_total_score / item_score)
- else:
- item_total_score = N_s[3]
- item_score = float(N_s[1] + '.' + N_s[2])
- item_count = int(item_total_score / item_score)
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_back2== '.' and num_infer3== '.' and num_back3 == '分'and num_back1 == '分' : # *,每题*分,共*.*分/*,共*分,每题*.*分
- item_N = N_s[0]
- if N_s[1] > N_s[2]:
- item_total_score = N_s[1]
- item_score = float(N_s[2] + '.' + N_s[3])
- item_count = int(item_total_score / item_score)
- else:
- item_total_score = N_s[2]
- item_score = N_s[1]
- item_count = int(item_total_score / item_score)
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- break
- break
- elif y == len_keyword_item1 - 1 and num_back4 == '分':
- item_total_score = N_s[3]
- item_N = -1
- item_score = -1
- item_count = -1
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- break
- break
- elif x == len_keyword_item1 - 1:
- for y in range(len_keyword_item2):
- if C_s.find(keyword_item2[y]) != -1:
- if len(N_s) == 1:
- num_index1 = s.index(N_s[0])
- num_infer1 = s[num_index1 - len(N_s[0])]
- num_back1 = s[num_index1 + len(N_s[0])]
- if isinstance(N_s[0], str):
- N_s[0] = int(N_s[0])
- if num_back1 == '分': # 每题*分
- item_score = N_s[0]
- type_score_dict_ocr['volume_N'] = -1
- type_score_dict_ocr['volume_total_score'] = -1
- type_score_dict_ocr['volume_count'] = -1
- type_score_dict_ocr['volume_score'] = item_score
- break
- if len(N_s) == 2:
- num_index1 = s.index(N_s[0])
- num_infer1 = s[num_index1 - len(N_s[0])]
- num_back1 = s[num_index1 + len(N_s[0])]
- all_1 = find_repeat(s, N_s[1])
- temp1 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[1]:
- temp1 = temp1 + 1
- num_index2 = all_1[temp1]
- num_infer2 = s[num_index2 - len(N_s[1])]
- num_back2 = s[num_index2 + len(N_s[1])]
- if isinstance(N_s[0], str):
- N_s[0] = int(N_s[0])
- if isinstance(N_s[1], str):
- N_s[1] = int(N_s[1])
- for z in range(len(keyword_item3)):
- if C_s.find(keyword_item3[z]) != -1:
- if num_back2 == '分': # 共*题,每题*分
- item_total_score = N_s[0] * N_s[1]
- item_count = N_s[0]
- item_score = N_s[1]
- type_score_dict_ocr['volume_N'] = -1
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_back1 == '分': # 每题*分,共*题
- item_total_score = int(N_s[0]) * int(N_s[1])
- item_count = int(N_s[1])
- item_score = int(N_s[0])
- type_score_dict_ocr['volume_N'] = -1
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif z == len(keyword_item3) - 1:
- if num_back2 == '分' and num_back1 == '.' and num_infer2 == '.': # *.*分
- item_N = -1
- item_score = float(N_s[0]+'.'+N_s[1])
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = -1
- type_score_dict_ocr['volume_count'] = -1
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_back2 == '分': # *,*分
- item_N = int(N_s[0])
- item_score = int(N_s[1])
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = -1
- type_score_dict_ocr['volume_count'] = -1
- type_score_dict_ocr['volume_score'] = item_score
- break
- break
- if len(N_s) == 3:
- num_index1 = s.index(N_s[0])
- num_infer1 = s[num_index1 - len(N_s[0])]
- num_back1 = s[num_index1 + len(N_s[0])]
- all_1 = find_repeat(s, N_s[1])
- temp1 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[1]:
- temp1 = temp1 + 1
- num_index2 = all_1[temp1]
- num_infer2 = s[num_index2 - len(N_s[1])]
- num_back2 = s[num_index2 + len(N_s[1])]
- all_2 = find_repeat(s, N_s[2])
- temp2 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[2]:
- temp2 = temp2 + 1
- for jj in range(len(N_s[1])):
- if N_s[1][jj] == N_s[2]:
- temp2 = temp2 + 1
- num_index3 = all_2[temp2]
- num_infer3 = s[num_index3 - len(N_s[2])]
- num_back3 = s[num_index3 + len(N_s[2])]
- if isinstance(N_s[0], str):
- N_s[0] = int(N_s[0])
- if isinstance(N_s[1], str):
- N_s[1] = int(N_s[1])
- if isinstance(N_s[2], str):
- N_s[2] = int(N_s[2])
- if num_back3 == '分' and (num_back2 == '题' or num_back2 == '小' or num_back2 == '空') and num_back1 != '分': # *,共*题,每题*分
- item_N = int(N_s[0])
- item_total_score = int(N_s[1]) * int(N_s[2])
- item_count = int(N_s[1])
- item_score = int(N_s[2])
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_back2 == '分' and (num_back3 == '题' or num_back3 == '小' or num_back3 == '空') and num_back1 != '分': # *,每题*分,共*题
- item_N = int(N_s[0])
- item_total_score = int(N_s[1]) * int(N_s[2])
- item_count = int(N_s[2])
- item_score = int(N_s[1])
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_infer3 == '.' and num_back3 == '分' and num_back2 == '.': # 共*题,每题*.*分
- item_N = -1
- item_count = int(N_s[0])
- item_score = float(N_s[1]+'.'+N_s[2])
- item_total_score = int(item_count * item_score)
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_infer2 == '.' and num_back2 == '分' and num_back1 == '.' : # 每题*.*分,共*题
- item_N = -1
- item_count = int(N_s[2])
- item_score = float(N_s[0] + '.' + N_s[1])
- item_total_score = int(item_count * item_score)
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif num_back3 == '分': # * * ,每题*分
- item_N = -1
- item_count = -1
- item_score = -1
- item_total_score = int(N_s[2])
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- break
- if len(N_s) == 4:
- num_index1 = s.index(N_s[0])
- num_infer1 = s[num_index1 - len(N_s[0])]
- num_back1 = s[num_index1 + len(N_s[0])]
- all_1 = find_repeat(s, N_s[1])
- temp1 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[1]:
- temp1 = temp1 + 1
- num_index2 = all_1[temp1]
- num_infer2 = s[num_index2 - len(N_s[1])]
- num_back2 = s[num_index2 + len(N_s[1])]
- all_2 = find_repeat(s, N_s[2])
- temp2 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[2]:
- temp2 = temp2 + 1
- for jj in range(len(N_s[1])):
- if N_s[1][jj] == N_s[2]:
- temp2 = temp2 + 1
- num_index3 = all_2[temp2]
- num_infer3 = s[num_index3 - len(N_s[2])]
- num_back3 = s[num_index3 + len(N_s[2])]
- all_3 = find_repeat(s, N_s[3])
- temp3 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[3]:
- temp3 = temp3 + 1
- for jj in range(len(N_s[1])):
- if N_s[1][jj] == N_s[3]:
- temp3 = temp3 + 1
- num_index4 = all_3[temp3]
- num_infer4 = s[num_index4 - len(N_s[3])]
- num_back4 = s[num_index4 + len(N_s[3])]
- if isinstance(N_s[0], str):
- N_s[0] = int(N_s[0])
- if isinstance(N_s[1], str):
- N_s[1] = int(N_s[1])
- if isinstance(N_s[2], str):
- N_s[2] = int(N_s[2])
- if isinstance(N_s[3], str):
- N_s[3] = int(N_s[3])
- if num_back3 == '.' and num_infer4 == '.' and num_back4 == '分' and (num_back2 == '题' or num_back2 == '小' or num_back2 == '空') and num_back1 != '分': # *,共*题,每题*.*分
- item_N = int(N_s[0])
- item_count = int(N_s[1])
- item_score = float(N_s[2] + '.' + N_s[3])
- item_total_score = int(item_count * item_score)
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- elif num_back2 == '.' and num_infer3 == '.' and num_back3 == '分' and (num_back4 == '题' or num_back4 == '小' or num_back4 == '空') and num_back1 != '分': # *,每题*.*分,共*题
- item_N = int(N_s[0])
- item_count = int(N_s[3])
- item_score = float(N_s[1] + '.' + N_s[2])
- item_total_score = int(item_count * item_score)
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = item_count
- type_score_dict_ocr['volume_score'] = item_score
- break
- elif y == len_keyword_item2 - 1:
- if C_s.find(keyword_item4[0]) != -1:
- if len(N_s) == 2: # *,*分
- num_index1 = s.index(N_s[0])
- num_infer1 = s[num_index1 - len(N_s[0])]
- num_back1 = s[num_index1 + len(N_s[0])]
- all_1 = find_repeat(s, N_s[1])
- temp1 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[1]:
- temp1 = temp1 + 1
- num_index2 = all_1[temp1]
- num_infer2 = s[num_index2 - len(N_s[1])]
- num_back2 = s[num_index2 + len(N_s[1])]
- if isinstance(N_s[0], str):
- N_s[0] = int(N_s[0])
- if isinstance(N_s[1], str):
- N_s[1] = int(N_s[1])
- if num_back1 == '分':
- item_N = -1
- item_total_score = int(N_s[0])
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = -1
- type_score_dict_ocr['volume_score'] = -1
- elif num_back2 == '分':
- item_N = int(N_s[0])
- item_total_score = int(N_s[1])
- type_score_dict_ocr['volume_N'] = item_N
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = -1
- type_score_dict_ocr['volume_score'] = -1
- break
- elif len(N_s) == 1:
- num_index1 = s.index(N_s[0])
- num_infer1 = s[num_index1 - len(N_s[0])]
- num_back1 = s[num_index1 + len(N_s[0])]
- item_total_score = int(N_s[0])
- type_score_dict_ocr['volume_N'] = -1
- type_score_dict_ocr['volume_total_score'] = item_total_score
- type_score_dict_ocr['volume_count'] = -1
- type_score_dict_ocr['volume_score'] = -1
- # if num_back1 == '分': # *分
- # test_9_23=0
- break
- break
- else:
- type_score_dict_ocr['volume_N'] = -1
- type_score_dict_ocr['volume_total_score'] = -1
- type_score_dict_ocr['volume_count'] = -1
- type_score_dict_ocr['volume_score'] = -1
- if 'volume_N' not in type_score_dict_ocr.keys():
- all_structure = {'volume_structure': -1,
- 'Score_structure': -1}
- break
- else:
- if C_s.find(keyword_type[1]) != -1:
- type_score_dict_ocr['keyword_type'] = keyword_type[1]
- elif C_s.find(keyword_type[0]) != -1:
- type_score_dict_ocr['keyword_type'] = keyword_type[0]
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- elif C_s.find(keyword_type[xxx]) != -1:
- type_score_dict_ocr['keyword_type'] = keyword_type[xxx]
- elif xxx == len_keyword_type - 1:
- type_score_dict_ocr['keyword_type'] = -2
- type_score_dict_ocr['item_N'] = type_score_dict_ocr.pop(
- 'volume_N')
- type_score_dict_ocr[
- 'item_total_score'] = type_score_dict_ocr.pop(
- 'volume_total_score')
- type_score_dict_ocr['item_count'] = type_score_dict_ocr.pop(
- 'volume_count')
- type_score_dict_ocr['item_score'] = type_score_dict_ocr.pop(
- 'volume_score')
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- volume_structure_item = type_score_dict_ocr
- volume_structure.append(volume_structure_item)
- if Score_structure == []:
- all_structure = {'volume_structure': volume_structure,
- 'Score_structure': -1}
- elif Score_structure[0]['keyword_type'] != -2:
- all_structure = {'volume_structure': volume_structure,
- 'Score_structure': Score_structure}
- else:
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- elif xxx == len_keyword_type - 1:
- for x in range(len_keyword_item1):
- if C_s.find(keyword_item1[x]) != -1:
- if len(N_s) == 1:
- num_index = s.index(N_s[0])
- num_infer = s[num_index - len(N_s[0])]
- num_back = s[num_index + len(N_s[0])]
- if isinstance(N_s[0], str):
- N_s[0] = int(N_s[0])
- if num_back == '分': # 共*分
- item_total_score = N_s[0]
- type_score_dict_ocr['item_N'] = -1
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = -1
- type_score_dict_ocr['item_score'] = -1
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- break
- elif len(N_s) == 2:
- num_index1 = s.index(N_s[0])
- num_infer1 = s[num_index1 - len(N_s[0])]
- num_back1 = s[num_index1 + len(N_s[0])]
- all_1 = find_repeat(s, N_s[1])
- temp1 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[1]:
- temp1 = temp1 + 1
- num_index2 = all_1[temp1]
- num_infer2 = s[num_index2 - len(N_s[1])]
- num_back2 = s[num_index2 + len(N_s[1])]
- if isinstance(N_s[0], str):
- N_s[0] = int(N_s[0])
- if isinstance(N_s[1], str):
- N_s[1] = int(N_s[1])
- for y in range(len_keyword_item2):
- if C_s.find(keyword_item2[y]) != -1:
- if num_back1 == '分' and (num_infer2 == '题' or num_infer2 == '空'): # 共*分,每题*分
- item_total_score = int(N_s[0])
- item_count = int(N_s[0]) / int(N_s[1])
- item_score = int(N_s[1])
- type_score_dict_ocr['item_N'] = -1
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- elif (num_infer1 == '题' or num_infer1 == '空') and num_back2 == '分': # 每题*分,共*分
- item_total_score = int(N_s[1])
- item_count = int(N_s[1]) / int(N_s[0])
- item_score = int(N_s[0])
- type_score_dict_ocr['item_N'] = -1
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- break
- elif y == len_keyword_item2 - 1:
- for u in range(len_keyword_item3):
- if C_s.find(keyword_item3[u]) != -1:
- if num_back1 == '分': # 共*分,共*题
- item_total_score = int(N_s[0])
- item_count = int(N_s[1])
- item_score = int(N_s[0]) / int(N_s[1])
- type_score_dict_ocr['item_N'] = -1
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- elif num_back2 == '分': # 共*题,共*分
- item_total_score = int(N_s[1])
- item_count = int(N_s[0])
- item_score = int(N_s[1]) / int(N_s[0])
- type_score_dict_ocr['item_N'] = -1
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- elif u == len_keyword_item3 - 1:
- if num_back2 == '分' and num_infer2 == '.'and num_back1 == '.': # *.*分
- item_N = -1
- item_total_score = int(N_s[0])
- type_score_dict_ocr['item_N'] = item_N
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = -1
- type_score_dict_ocr['item_score'] = -1
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- elif num_back2 == '分':
- item_N = int(N_s[0])
- item_total_score = int(N_s[1])
- type_score_dict_ocr['item_N'] = item_N
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = -1
- type_score_dict_ocr['item_score'] = -1
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- break
- elif len(N_s) == 3:
- num_index1 = s.index(N_s[0])
- num_infer1 = s[num_index1 - len(N_s[0])]
- num_back1 = s[num_index1 + len(N_s[0])]
- all_1 = find_repeat(s, N_s[1])
- temp1 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[1]:
- temp1 = temp1 + 1
- num_index2 = all_1[temp1]
- num_infer2 = s[num_index2 - len(N_s[1])]
- num_back2 = s[num_index2 + len(N_s[1])]
- all_2 = find_repeat(s, N_s[2])
- temp2 = 0
- if len(N_s[0]) == len(N_s[2]):
- if N_s[0] == N_s[2]:
- temp2 = temp2 + 1
- else:
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[2]:
- temp2 = temp2 + 1
- if len(N_s[1]) == len(N_s[2]):
- if N_s[1] == N_s[2]:
- temp2 = temp2 + 1
- else:
- for jj in range(len(N_s[1])):
- if N_s[1][jj] == N_s[2]:
- temp2 = temp2 + 1
- num_index3 = all_2[temp2]
- num_infer3 = s[num_index3 - len(N_s[2])]
- if num_index3 + len(N_s[2]) < len(s):
- num_back3 = s[num_index3 + len(N_s[2])]
- else:
- num_back3 = []
- if isinstance(N_s[0], str):
- N_s[0] = int(N_s[0])
- if isinstance(N_s[1], str):
- N_s[1] = int(N_s[1])
- if isinstance(N_s[2], str):
- N_s[2] = int(N_s[2])
- for v in range(len_keyword_item2):
- if C_s.find(keyword_item2[v]) != -1:
- for w in range(len_keyword_item3):
- if C_s.find(keyword_item3[w]) != -1:
- if (num_infer1 == '题' or num_infer1 == '空') and num_back1 == '分' and num_back3 == '分': # 每题*分,共*题,共*分
- item_total_score = int(N_s[2])
- item_count = int(N_s[1])
- item_score = int(N_s[0])
- if item_total_score < item_count * item_score:
- item_total_score = item_count * item_score
- type_score_dict_ocr['item_N'] = -1
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- elif (num_infer1 == '题' or num_infer1 == '空') and num_back1 == '分' and num_back2 == '分': # 每题*分,共*分,共*题
- item_total_score = int(N_s[1])
- item_count = int(N_s[2])
- item_score = int(N_s[0])
- if item_total_score < item_count * item_score:
- item_total_score = item_count * item_score
- type_score_dict_ocr['item_N'] = -1
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- elif (num_infer2 == '题' or num_infer2 == '空') and num_back2 == '分' and num_back3 == '分': # 共*题,每题*分,共*分
- item_total_score = int(N_s[2])
- item_count = int(N_s[0])
- item_score = int(N_s[1])
- if item_total_score < item_count * item_score:
- item_total_score = item_count * item_score
- type_score_dict_ocr['item_N'] = -1
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- elif (num_infer2 == '题' or num_infer2 == '空') and num_back2 == '分' and num_back1 == '分': # 共*分,每题*分,共*题
- item_total_score = int(N_s[0])
- item_count = int(N_s[2])
- item_score = int(N_s[1])
- if item_total_score < item_count * item_score:
- item_total_score = item_count * item_score
- type_score_dict_ocr['item_N'] = -1
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- elif (num_infer3 == '题' or num_infer3 == '空') and num_back3 == '分' and num_back2 == '分': # 共*题,共*分,每题*分
- item_total_score = int(N_s[1])
- item_count = int(N_s[0])
- item_score = int(N_s[2])
- if item_total_score < item_count * item_score:
- item_total_score = item_count * item_score
- type_score_dict_ocr['item_N'] = -1
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- elif (num_infer3 == '题' or num_infer3 == '空') and num_back3 == '分' and num_back1 == '分': # 共*分,共*题, 每题*分
- item_total_score = int(N_s[0])
- item_count = int(N_s[1])
- item_score = int(N_s[2])
- if item_total_score < item_count * item_score:
- item_total_score = item_count * item_score
- type_score_dict_ocr['item_N'] = -1
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- elif w == len_keyword_item3 - 1:
- if num_back2 == '分' and (num_infer3 == '题' or num_infer3 == '空') and num_back3 == '分': # *,共*分,每题*分
- item_N = int(N_s[0])
- item_total_score = int(N_s[1])
- item_count = int(N_s[1]) / int(N_s[2])
- item_score = int(N_s[2])
- type_score_dict_ocr['item_N'] = item_N
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- elif num_back3 == '分' and (num_infer2 == '题' or num_infer2 == '空') and num_back2 == '分': # *,每题*分,共*分,
- item_N = int(N_s[0])
- item_total_score = int(N_s[2])
- item_count = int(N_s[2]) / int(N_s[1])
- item_score = int(N_s[1])
- type_score_dict_ocr['item_N'] = item_N
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- elif num_back3 == '分' and num_infer2 == '.' and num_back1 == '.' and num_back2 == '分': # 每题*.*分,共*分/共*.*分,每题*分
- item_N = -1
- if int(N_s[0]) > int(N_s[2]):
- item_total_score = int(N_s[0])
- item_score = int(N_s[2])
- item_count = int(item_total_score / item_score)
- else:
- item_total_score = int(N_s[2])
- item_score = float(N_s[0] + '.' + N_s[1])
- item_count = int(item_total_score / item_score)
- type_score_dict_ocr['item_N'] = item_N
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- elif num_back3 == '分' and num_infer3 == '.' and num_back2 == '.' and num_back1 == '分': # 每题*分,共*.*分/共*分,每题*.*分
- item_N = -1
- if int(N_s[0]) > int(N_s[2]):
- item_total_score = int(N_s[0])
- item_score = float(N_s[1] + '.' + N_s[2])
- item_count = int(item_total_score / item_score)
- else:
- item_total_score = int(N_s[1])
- item_score = int(N_s[0])
- item_count = int(item_total_score / item_score)
- type_score_dict_ocr['item_N'] = item_N
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- break
- elif v == len_keyword_item2 - 1:
- for w in range(len_keyword_item3):
- if C_s.find(keyword_item3[w]) != -1:
- if num_back3 == '分' and num_infer3 == '.'and num_back2 == '.': # *小题,共*.*分,
- item_N = -1
- item_total_score = int(N_s[1])
- item_count = int(N_s[0])
- item_score = int(N_s[1]) / int(N_s[0])
- type_score_dict_ocr['item_N'] = item_N
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- elif num_back2 == '分' and num_infer2 == '.'and num_back1 == '.': # 共*.*分,*小题
- item_N = -1
- item_total_score = int(N_s[0])
- item_count = int(N_s[1])
- item_score = int(N_s[0]) / int(N_s[2])
- type_score_dict_ocr['item_N'] = item_N
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- elif num_back2 == '分': # *,*小题,共*分,
- item_N = int(N_s[0])
- item_total_score = int(N_s[1])
- item_count = int(N_s[2])
- item_score = int(N_s[1]) / int(N_s[2])
- type_score_dict_ocr['item_N'] = item_N
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- elif num_back3 == '分': # *,共*分,共*小题
- item_N = int(N_s[0])
- item_total_score = int(N_s[2])
- item_count = int(N_s[1])
- item_score = int(N_s[2]) / int(N_s[1])
- type_score_dict_ocr['item_N'] = item_N
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- elif w == len_keyword_item3 - 1:
- if num_back3 == '分' and num_infer3 == '.'and num_back2 == '.': # *,共*.*分,
- item_N = int(N_s[0])
- item_total_score = int(N_s[1])
- item_count = -1
- item_score = -1
- type_score_dict_ocr['item_N'] = item_N
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- break
- break
- elif len(N_s) == 4:
- num_index1 = s.index(N_s[0])
- num_infer1 = s[num_index1 - len(N_s[0])]
- num_back1 = s[num_index1 + len(N_s[0])]
- all_1 = find_repeat(s, N_s[1])
- temp1 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[1]:
- temp1 = temp1 + 1
- num_index2 = all_1[temp1]
- num_infer2 = s[num_index2 - len(N_s[1])]
- num_back2 = s[num_index2 + len(N_s[1])]
- all_2 = find_repeat(s, N_s[2])
- temp2 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[2]:
- temp2 = temp2 + 1
- for jj in range(len(N_s[1])):
- if N_s[1][jj] == N_s[2]:
- temp2 = temp2 + 1
- num_index3 = all_2[temp2]
- num_infer3 = s[num_index3 - len(N_s[2])]
- num_back3 = s[num_index3 + len(N_s[2])]
- all_3 = find_repeat(s, N_s[3])
- temp3 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[3]:
- temp3 = temp3 + 1
- for jj in range(len(N_s[1])):
- if N_s[1][jj] == N_s[3]:
- temp3 = temp3 + 1
- for kk in range(len(N_s[2])):
- if N_s[2][kk] == N_s[3]:
- temp3 = temp3 + 1
- num_index4 = all_3[temp3]
- num_infer4 = s[num_index4 - len(N_s[3])]
- num_back4 = s[num_index4 + len(N_s[3])]
- if isinstance(N_s[0], str):
- N_s[0] = int(N_s[0])
- if isinstance(N_s[1], str):
- N_s[1] = int(N_s[1])
- if isinstance(N_s[2], str):
- N_s[2] = int(N_s[2])
- if isinstance(N_s[3], str):
- N_s[3] = int(N_s[3])
- for y in range(len_keyword_item1):
- if C_s.find(keyword_item1[y]) != -1:
- for z in range(len_keyword_item2):
- if C_s.find(keyword_item2[z]) != -1:
- for u in range(len_keyword_item3):
- if C_s.find(keyword_item3[u]) != -1:
- if (num_infer2 == '题' or num_infer2 == '空') and num_back2 == '分' and num_back4 == '分': # *,每题*分,共*题,共*分
- item_N = int(N_s[0])
- item_total_score = int(N_s[3])
- item_count = int(N_s[2])
- item_score = int(N_s[1])
- elif (num_infer2 == '题' or num_infer2 == '空') and num_back2 == '分' and num_back3 == '分': # *,每题*分,共*分,共*题
- item_N = int(N_s[0])
- item_total_score = int(N_s[2])
- item_count = int(N_s[3])
- item_score = int(N_s[1])
- elif (num_infer3 == '题' or num_infer3 == '空') and num_back3 == '分' and num_back4 == '分': # *,共*题,每题*分,共*分
- item_N = int(N_s[0])
- item_total_score = int(N_s[3])
- item_count = int(N_s[1])
- item_score = int(N_s[2])
- elif (num_infer3 == '题' or num_infer3 == '空') and num_back3 == '分' and num_back2 == '分': # *,共*分,每题*分,共*题
- item_N = int(N_s[0])
- item_total_score = int(N_s[1])
- item_count = int(N_s[3])
- item_score = int(N_s[2])
- elif (num_infer4 == '题' or num_infer4 == '空') and num_back4 == '分' and num_back3 == '分': # *,共*题,共*分,每题*分
- item_N = int(N_s[0])
- item_total_score = int(N_s[2])
- item_count = int(N_s[1])
- item_score = int(N_s[3])
- type_score_dict_ocr['item_N'] = item_N
- elif (num_infer4 == '题' or num_infer4 == '空') and num_back4 == '分' and num_back2 == '分': # *,共*分,共*题, 每题*分
- item_N = int(N_s[0])
- item_total_score = int(N_s[1])
- item_count = int(N_s[2])
- item_score = int(N_s[3])
- elif num_back1== '.' and num_infer2== '.' and num_back2 == '分'and num_back4 == '分' : # 共*.*分,共*题, 每题*分/每题*.*分,共*题,共*分
- item_N = -1
- if int(N_s[0]) > int(N_s[3]):
- item_total_score = int(N_s[0])
- item_score = int(N_s[3])
- item_count = int(item_total_score / item_score)
- else:
- item_total_score = int(N_s[3])
- item_score = float(N_s[0] + '.' + N_s[1])
- item_count = int(item_total_score / item_score)
- elif num_back3== '.' and num_infer4== '.' and num_back1 == '分'and num_back4 == '分' : # 共*分,共*题, 每题*.*分/每题*分,共*题,共*.*分
- item_N = -1
- if int(N_s[0]) > int(N_s[2]):
- item_total_score = int(N_s[0])
- item_score = float(N_s[2] + '.' + N_s[3])
- item_count = int(item_total_score / item_score)
- else:
- item_total_score = int(N_s[2])
- item_score = int(N_s[0])
- item_count = int(item_total_score / item_score)
- elif num_back2== '.' and num_infer3== '.' and num_back3 == '分'and num_back4 == '分' : # 共*题,共*.*分,每题*分/共*题,每题*.*分,共*分
- item_N = -1
- if int(N_s[1]) > int(N_s[3]):
- item_total_score = int(N_s[1])
- item_score = int(N_s[3])
- item_count = int(item_total_score / item_score)
- else:
- item_total_score = int(N_s[3])
- item_score = float(N_s[1] + '.' + N_s[2])
- item_count = int(item_total_score / item_score)
- elif num_back3== '.' and num_infer4== '.' and num_back4 == '分'and num_back2 == '分' : # 共*题,共*分,每题*.*分/共*题,每题*分,共*.*分
- item_N = -1
- if int(N_s[1]) > int(N_s[2]):
- item_total_score = int(N_s[1])
- item_score = float(N_s[2] + '.' + N_s[3])
- item_count = int(item_total_score / item_score)
- else:
- item_total_score = int(N_s[2])
- item_score = int(N_s[1])
- item_count = int(item_total_score / item_score)
- elif num_back1== '.' and num_infer2== '.' and num_back2 == '分'and num_back3 == '分' : # 每题*.*分,共*分,共*题/共*.*分,每题*分,共*题
- item_N = -1
- if int(N_s[0]) > int(N_s[2]):
- item_total_score = int(N_s[0])
- item_score = int(N_s[2])
- item_count = int(item_total_score / item_score)
- else:
- item_total_score = int(N_s[2])
- item_score = float(N_s[0] + '.' + N_s[1])
- item_count = int(item_total_score / item_score)
- elif num_back2== '.' and num_infer3== '.' and num_back3 == '分'and num_back1 == '分' : # 每题*分,共*.*分,共*题/共*分,每题*.*分,共*题
- item_N = -1
- if int(N_s[0]) > int(N_s[1]):
- item_total_score = int(N_s[0])
- item_score = float(N_s[1] + '.' + N_s[2])
- item_count = int(item_total_score / item_score)
- else:
- item_total_score = int(N_s[1])
- item_score = int(N_s[0])
- item_count = int(item_total_score / item_score)
- else:
- break
- if item_total_score < item_count * item_score:
- item_total_score = item_count * item_score
- type_score_dict_ocr['item_N'] = item_N
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- break
- break
- break
- elif x == len_keyword_item1 - 1:
- for y in range(len_keyword_item2):
- if C_s.find(keyword_item2[y]) != -1:
- if len(N_s) == 1:
- num_index1 = s.index(N_s[0])
- num_infer1 = s[num_index1 - len(N_s[0])]
- num_back1 = s[num_index1 + len(N_s[0])]
- if isinstance(N_s[0], str):
- N_s[0] = int(N_s[0])
- if num_back1 == '分': # 每题*分
- item_score = int(N_s[0])
- type_score_dict_ocr['item_N'] = -1
- type_score_dict_ocr['item_total_score'] = -1
- type_score_dict_ocr['item_count'] = -1
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- elif len(N_s) == 2:
- num_index1 = s.index(N_s[0])
- num_infer1 = s[num_index1 - len(N_s[0])]
- num_back1 = s[num_index1 + len(N_s[0])]
- all_1 = find_repeat(s, N_s[1])
- temp1 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[1]:
- temp1 = temp1 + 1
- num_index2 = all_1[temp1]
- num_infer2 = s[num_index2 - len(N_s[1])]
- num_back2 = s[num_index2 + len(N_s[1])]
- if isinstance(N_s[0], str):
- N_s[0] = int(N_s[0])
- if isinstance(N_s[1], str):
- N_s[1] = int(N_s[1])
- for z in range(len(keyword_item3)):
- if C_s.find(keyword_item3[z]) != -1:
- if num_back2 == '分': # 共*题,每题*分
- item_total_score = int(N_s[0]) * int(N_s[1])
- item_count = int(N_s[0])
- item_score = int(N_s[1])
- type_score_dict_ocr['item_N'] = -1
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- elif num_back1 == '分': # 每题*分,共*题
- item_total_score = int(N_s[0]) * int(N_s[1])
- item_count = int(N_s[1])
- item_score = int(N_s[0])
- type_score_dict_ocr['item_N'] = -1
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- elif z == len(keyword_item3) - 1:
- if num_back2 == '分' and num_infer2 == '.' and num_back1 == '.': # *.*分
- item_N = -1
- item_score = float(N_s[0] + '.'+N_s[1])
- type_score_dict_ocr['item_N'] = item_N
- type_score_dict_ocr['item_total_score'] = -1
- type_score_dict_ocr['item_count'] = -1
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- elif num_back2 == '分': # *,*分
- item_N = int(N_s[0])
- item_score = int(N_s[1])
- type_score_dict_ocr['item_N'] = item_N
- type_score_dict_ocr['item_total_score'] = -1
- type_score_dict_ocr['item_count'] = -1
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- break
- elif len(N_s) == 3: # 9月16号修改
- num_index1 = s.index(N_s[0])
- num_infer1 = s[num_index1 - len(N_s[0])]
- num_back1 = s[num_index1 + len(N_s[0])]
- all_1 = find_repeat(s, N_s[1])
- temp1 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[1]:
- temp1 = temp1 + 1
- num_index2 = all_1[temp1]
- num_infer2 = s[num_index2 - len(N_s[1])]
- num_back2 = s[num_index2 + len(N_s[1])]
- all_2 = find_repeat(s, N_s[2])
- temp2 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[2]:
- temp2 = temp2 + 1
- for jj in range(len(N_s[1])):
- if N_s[1][jj] == N_s[2]:
- temp2 = temp2 + 1
- num_index3 = all_2[temp2]
- num_infer3 = s[num_index3 - len(N_s[2])]
- num_back3 = s[num_index3 + len(N_s[2])]
- if isinstance(N_s[0], str):
- N_s[0] = int(N_s[0])
- if isinstance(N_s[1], str):
- N_s[1] = int(N_s[1])
- if isinstance(N_s[2], str):
- N_s[2] = int(N_s[2])
- if num_back3 == '分' and num_back2 != '分' and num_infer3 != '.': # *,共*题,每题*分
- item_N = int(N_s[0])
- item_total_score = int(N_s[1]) * int(N_s[2])
- item_count = int(N_s[1])
- item_score = int(N_s[2])
- elif num_back2 == '分' and num_back3 != '分' and num_infer2 != '.': # *,每题*分,共*题
- item_N = int(N_s[0])
- item_total_score = int(N_s[1]) * int(N_s[2])
- item_count = int(N_s[2])
- item_score = int(N_s[1])
- type_score_dict_ocr['item_N'] = item_N
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- elif num_back2 == '分' and num_back3 == '分'and num_infer3 != '.' and num_infer2 != '.': # *,*分,每题*分
- item_N = int(N_s[0])
- item_total_score = int(N_s[1])
- item_count = int(N_s[1]) / int(N_s[2])
- item_score = int(N_s[2])
- type_score_dict_ocr['item_N'] = item_N
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- elif num_back3 == '分' and num_back2 == '.' and num_infer3 == '.' and (num_back1 =='题' or num_back1 =='小'or num_back1 =='空'): # 共*题,每题*.*分
- item_N = -1
- item_count = int(N_s[0])
- item_score = float(N_s[1]+'.'+N_s[2])
- item_total_score = int(item_score * item_count)
- type_score_dict_ocr['item_N'] = item_N
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- elif num_back2 == '分' and num_back1 == '.' and num_infer2 == '.'and (num_back3 =='题' or num_back3 =='小'or num_back3 =='空'): # 每题*.*分,共*题
- item_N = -1
- item_count = int(N_s[2])
- item_score = float(N_s[0] + '.' + N_s[1])
- item_total_score = int(item_score * item_count)
- type_score_dict_ocr['item_N'] = item_N
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- elif num_back3 == '分' and num_back2 == '.' and num_infer3 == '.': # *,每题*.*分
- item_N = N_s[0]
- item_count = -1
- item_score = float(N_s[1] + '.' + N_s[2])
- item_total_score = -1
- type_score_dict_ocr['item_N'] = item_N
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- elif len(N_s) == 4: # 9月16号修改
- num_index1 = s.index(N_s[0])
- num_infer1 = s[num_index1 - len(N_s[0])]
- num_back1 = s[num_index1 + len(N_s[0])]
- all_1 = find_repeat(s, N_s[1])
- temp1 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[1]:
- temp1 = temp1 + 1
- num_index2 = all_1[temp1]
- num_infer2 = s[num_index2 - len(N_s[1])]
- num_back2 = s[num_index2 + len(N_s[1])]
- all_2 = find_repeat(s, N_s[2])
- temp2 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[2]:
- temp2 = temp2 + 1
- for jj in range(len(N_s[1])):
- if N_s[1][jj] == N_s[2]:
- temp2 = temp2 + 1
- num_index3 = all_2[temp2]
- num_infer3 = s[num_index3 - len(N_s[2])]
- num_back3 = s[num_index3 + len(N_s[2])]
- all_3 = find_repeat(s, N_s[3])
- temp3 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[3]:
- temp3 = temp3 + 1
- for jj in range(len(N_s[1])):
- if N_s[1][jj] == N_s[2]:
- temp3 = temp3 + 1
- num_index4 = all_3[temp3]
- num_infer4 = s[num_index4 - len(N_s[3])]
- num_back4 = s[num_index4 + len(N_s[3])]
- if isinstance(N_s[0], str):
- N_s[0] = int(N_s[0])
- if isinstance(N_s[1], str):
- N_s[1] = int(N_s[1])
- if isinstance(N_s[2], str):
- N_s[2] = int(N_s[2])
- if isinstance(N_s[3], str):
- N_s[3] = int(N_s[3])
- if num_back4 == '分' and num_back2 != '分' and num_infer4 == '.'and num_back3 == '.': # *,共*题,每题*.*分
- item_N = int(N_s[0])
- item_count = int(N_s[1])
- item_score = float(N_s[2]+'.'+N_s[3])
- item_total_score = int(item_score * item_count)
- type_score_dict_ocr['item_N'] = item_N
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- elif num_back3 == '分' and num_back4 != '分' and num_infer3 == '.'and num_back2 == '.': # *,每题*.*分,共*题
- item_N = int(N_s[0])
- item_count = int(N_s[3])
- item_score = float(N_s[1] + '.' + N_s[2])
- item_total_score = int(item_score * item_count)
- type_score_dict_ocr['item_N'] = item_N
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = item_count
- type_score_dict_ocr['item_score'] = item_score
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- elif y == len_keyword_item2 - 1:
- if C_s.find(keyword_item4[0]) != -1:
- if len(N_s) == 2: # *,*分
- num_index1 = s.index(N_s[0])
- num_infer1 = s[num_index1 - len(N_s[0])]
- num_back1 = s[num_index1 + len(N_s[0])]
- if num_infer1 == '( ' or num_back1 == ')' or num_infer1 == '( ' or num_back1 == ')':
- break
- else:
- all_1 = find_repeat(s, N_s[1])
- temp1 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[1]:
- temp1 = temp1 + 1
- num_index2 = all_1[temp1]
- num_infer2 = s[num_index2 - len(N_s[1])]
- num_back2 = s[num_index2 + len(N_s[1])]
- if isinstance(N_s[0], str):
- N_s[0] = int(N_s[0])
- if isinstance(N_s[1], str):
- N_s[1] = int(N_s[1])
- if int(N_s[0]) > 1000:
- item_N =0
- item_N1 = int(N_s[0][-4] + N_s[0][-3])
- item_N2 = int(N_s[0][-2] + N_s[0][-1])
- if item_N2 - item_N1 == 1:
- item_N = [0, 0]
- item_N = [item_N1, item_N2]
- elif item_N2 - item_N1 == 2:
- item_N = [0, 0, 0]
- item_N = [item_N1, item_N1 + 1, item_N2]
- elif item_N2 - item_N1 == 3:
- item_N = [0, 0, 0, 0]
- item_N = [item_N1, item_N1 + 1, item_N1 + 2, item_N2]
- type_score_dict_ocr['item_N'] = item_N
- else:
- item_N = int(N_s[0])
- item_total_score = int(N_s[1])
- type_score_dict_ocr['item_N'] = item_N
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = -1
- type_score_dict_ocr['item_score'] = -1
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- elif len(N_s) == 3: # *,*分
- num_index1 = s.index(N_s[0])
- num_infer1 = s[num_index1 - len(N_s[0])]
- num_back1 = s[num_index1 + len(N_s[0])]
- all_1 = find_repeat(s, N_s[1])
- temp1 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[1]:
- temp1 = temp1 + 1
- num_index2 = all_1[temp1]
- num_infer2 = s[num_index2 - len(N_s[1])]
- num_back2 = s[num_index2 + len(N_s[1])]
- all_2 = find_repeat(s, N_s[2])
- temp2 = 0
- for ii in range(len(N_s[0])):
- if N_s[0][ii] == N_s[2]:
- temp2 = temp2 + 1
- for jj in range(len(N_s[1])):
- if N_s[1][jj] == N_s[2]:
- temp2 = temp2 + 1
- num_index3 = all_2[temp2]
- num_infer3 = s[num_index3 - len(N_s[2])]
- if num_index3 + len(N_s[2]) < len(s):
- num_back3 = s[num_index3 + len(N_s[2])]
- else:
- num_back3 = []
- if isinstance(N_s[0], str):
- N_s[0] = int(N_s[0])
- if isinstance(N_s[1], str):
- N_s[1] = int(N_s[1])
- if isinstance(N_s[2], str):
- N_s[2] = int(N_s[2])
- if num_back2 == '分':
- item_N = N_s[0]
- item_total_score = int(N_s[1])
- type_score_dict_ocr['item_total_score'] = item_total_score
- item_total_score = int(N_s[2])
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_N'] = item_N
- type_score_dict_ocr['item_count'] = -1
- type_score_dict_ocr['item_score'] = -1
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- elif num_back3 == '分':
- if int(N_s[1]) - int(N_s[0]) == 1:
- item_N = [0, 0]
- item_N = [int(N_s[0]), int(N_s[1])]
- elif int(N_s[1]) - int(N_s[0]) == 2:
- item_N = [0, 0, 0]
- item_N = [int(N_s[0]), int(N_s[0]) + 1, int(N_s[1])]
- elif int(N_s[1]) - int(N_s[0]) == 3:
- item_N = [0, 0, 0, 0]
- item_N = [int(N_s[0]), int(N_s[0]) + 1, int(N_s[0]) + 2,
- int(N_s[1])]
- else:
- break
- item_total_score = int(N_s[2])
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_N'] = item_N
- type_score_dict_ocr['item_count'] = -1
- type_score_dict_ocr['item_score'] = -1
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- elif len(N_s) == 1:
- num_index1 = s.index(N_s[0])
- num_infer1 = s[num_index1 - len(N_s[0])]
- if num_index1 + len(N_s[0]) < len(s):
- num_back1 = s[num_index1 + len(N_s[0])]
- item_total_score = int(N_s[0])
- type_score_dict_ocr['item_N'] = -1
- type_score_dict_ocr['item_total_score'] = item_total_score
- type_score_dict_ocr['item_count'] = -1
- type_score_dict_ocr['item_score'] = -1
- if num_back1 == '分': # *分
- Score_structure_item = type_score_dict_ocr
- Score_structure.append(Score_structure_item)
- all_structure = {'volume_structure': -1,
- 'Score_structure': Score_structure}
- break
- break
- return all_structure
|