PyThaiNLP / pythainlp / 4445565965

4445565965

push

github

GitHub 
Merge pull request #777 from kangkengkhadev/dev

278 of 278 new or added lines in 3 files covered. (100.0%)

5534 of 6126 relevant lines covered (90.34%)

0.9 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

2.26
/pythainlp/khavee/core.py
1 
from pythainlp.tokenize import subword_tokenize
1✔
2 










3



class KhaveeVerifier:




1✔






4



    def check_sara(self,word):




1✔






5



        sara = []




×








6



        countoa = 0




×







7



        # In case การันย์











8



        if '์' in word[-1]:




×








9



            word = word[:-2]




×







10



        # In case สระเดี่ยว











11



        for i in word:




×








12



            if i == 'ะ' or i == 'ั':




×








13



                sara.append('อะ')




×








14



            elif i == 'ิ':




×








15



                sara.append('อิ')




×








16



            elif i == 'ุ':




×








17



                sara.append('อุ')




×








18



            elif i == 'ึ':




×








19



                sara.append('อึ')




×








20



            elif i == 'ี':




×








21



                sara.append('อี')




×








22



            elif i == 'ู':




×








23



                sara.append('อู')




×








24



            elif i == 'ื':




×








25



                sara.append('อือ')




×








26



            elif i == 'เ':




×








27



                sara.append('เอ')




×








28



            elif i == 'แ':




×








29



                sara.append('แอ')




×








30



            elif i == 'า':




×








31



                sara.append('อา') 




×








32



            elif i == 'โ':




×








33



                sara.append('โอ')




×








34



            elif i == 'ำ':




×








35



                sara.append('อำ')




×








36



            elif i == 'อ':




×








37



                countoa += 1




×








38



                sara.append('ออ')




×








39



            elif i == 'ั' and 'ว' in word:




×








40



                sara.append('อัว')




×








41



            elif i == 'ไ' or i == 'ใ':




×








42



                sara.append('ไอ') 




×








43



            elif 'รร' in word:




×








44



                if self.check_marttra(word) == 'กม':




×








45



                    sara.append('อำ')




×







46



                else:











47



                    sara.append('อะ')




×







48



        # Incase ออ











49



        if countoa == 1 and 'อ' in word[-1]:




×








50



            sara.remove('ออ')




×







51



        # In case เอ เอ 











52



        countA = 0




×








53



        for i in sara:




×








54



            if i == 'เอ':




×








55



                countA = countA + 1




×








56



            if countA > 1:




×








57



                sara.remove('เอ')




×








58



                sara.remove('เอ')




×








59



                sara.append('แ')




×







60



        # In case สระประสม











61



        if 'เอ' in sara and 'อะ' in sara:




×








62



            sara.remove('เอ')




×








63



            sara.remove('อะ')




×








64



            sara.append('เอะ')




×








65



        elif 'แอ' in sara and 'อะ' in sara:




×








66



            sara.remove('แอ')




×








67



            sara.remove('อะ')




×








68



            sara.append('แอะ')




×








69



        if 'เอะ' in sara and 'ออ' in sara:




×








70



            sara.remove('เอะ')




×








71



            sara.remove('ออ')




×








72



            sara.append('เออะ')




×








73



        elif 'เอ' in sara and 'อิ' in sara:




×








74



            sara.remove('เอ')




×








75



            sara.remove('อิ')




×








76



            sara.append('เออ')        




×








77



        elif 'เอะ' in sara and 'อา' in sara:




×








78



            sara.remove('เอะ')




×








79



            sara.remove('ออ')




×








80



            sara.append('เอาะ')




×








81



        elif 'เอ' in sara and 'ออ' in sara and 'อ' in word[-1]:




×








82



            sara.remove('เอ')




×








83



            sara.remove('ออ')




×








84



            sara.append('เออ')




×








85



        elif 'โอ' in sara and 'อะ' in sara: 




×








86



            sara.remove('โอ')




×








87



            sara.remove('อะ')




×








88



            sara.append('โอะ')




×








89



        elif 'เอ' in sara and 'อี' in sara: 




×








90



            sara.remove('เอ')




×








91



            sara.remove('อี')




×








92



            sara.append('เอีย')




×








93



        elif 'เอ' in sara and 'อือ' in sara: 




×








94



            sara.remove('เอ')




×








95



            sara.remove('อือ')




×








96



            sara.append('อัว')   




×








97



        elif 'เอ' in sara and 'อา' in sara: 




×








98



            sara.remove('เอ')




×








99



            sara.remove('อา')




×








100



            sara.append('เอา') 




×








101



        if 'อือ' in sara and 'เออ' in sara: 




×








102



            sara.remove('เออ')




×








103



            sara.remove('อือ')




×








104



            sara.append('เอือ')  




×








105



        elif 'ออ' in sara and len(sara) > 1:




×








106



            sara.remove('ออ')        




×








107



        elif 'ว' in word and len(sara) == 0:




×








108



            sara.append('อัว')




×








109



        if 'ั' in word and self.check_marttra(word) == 'กา':




×








110



            sara = []




×








111



            sara.append('ไอ')




×







112



        # In case อ











113



        if word == 'เออะ':




×








114



            sara = []




×








115



            sara.append('เออะ')




×








116



        elif word == 'เออ':




×








117



            sara = []




×








118



            sara.append('เออ')




×








119



        elif word == 'เอ':




×








120



            sara = []




×








121



            sara.append('เอ')




×








122



        elif word == 'เอะ':




×








123



            sara = []




×








124



            sara.append('เอะ')




×








125



        elif word == 'เอา':




×








126



            sara = []




×








127



            sara.append('เอา')




×








128



        if 'ฤา' in word or 'ฦา' in word:




×








129



            sara = []




×








130



            sara.append('อือ') 




×








131



        elif 'ฤ' in word or 'ฦ' in word:




×








132



            sara = []




×








133



            sara.append('อึ') 




×







134



        # In case กน











135



        if sara == [] and len(word) == 2:




×








136



            if word[-1] != 'ร':




×








137



                sara.append('โอะ')




×







138



            else:











139



                sara.append('ออ') 




×








140



        elif sara == [] and len(word) == 3:




×








141



            sara.append('ออ') 




×








142



        if sara == []:




×








143



            return 'Cant find Sara in this word'




×







144



        else:











145



            return sara[0]




×







146












147













148



    def check_marttra(self,word):




1✔






149



        if word[-1] == 'ร' and word[-2] in ['ต','ท'] :




×








150



            word = word[:-1]




×







151



            # print(word)











152



        if '์' in word[-1]:




×








153



            if 'ิ' in word[-2] or 'ุ' in word[-2]:




×








154



                word = word[:-3]




×







155



            else:











156



                word = word[:-2]




×








157



        if 'ำ' in word or ('ํ' in word and 'า' in word) or 'ไ' in word or 'ใ' in word:




×








158



            return 'กา'




×








159



        elif word[-1] in ['า','ะ','ิ','ี','ุ','ู','อ'] or ('ี' in word and 'ย' in word[-1]) or ('ื' in word and 'อ' in word[-1]):




×








160



            return 'กา'




×








161



        elif word[-1] in ['ง']:




×








162



            return 'กง'




×








163



        elif word[-1] in ['ม']:




×








164



            return 'กม'




×








165



        elif word[-1] in ['ย']:




×








166



            if 'ั' in word:




×








167



                return 'กา'




×







168



            else:











169



                return 'เกย'




×








170



        elif word[-1] in ['ว']:




×








171



            return 'เกอว'




×








172



        elif word[-1] in ['ก','ข','ค','ฆ']:




×








173



            return 'กก'




×








174



        elif word[-1] in ['จ','ช','ซ','ฎ','ฏ','ฐ','ฑ','ฒ','ด','ต','ถ','ท','ธ','ศ','ษ','ส'] :




×








175



            return 'กด'




×








176



        elif word[-1] in ['ญ',', ณ' ,'น' ,'ร' ,'ล' ,'ฬ']:




×








177



            return 'กน'




×








178



        elif word[-1] in ['บ', 'ป', 'พ', 'ฟ', 'ภ']:




×








179



            return 'กบ'




×







180



        else:











181



           return 'Cant find Marttra in this word'




×







182













183



    def check_sumpus(self,word1,word2):




1✔






184



        marttra1 = self.check_marttra(word1)




×








185



        marttra2 = self.check_marttra(word2)




×








186



        sara1 = self.check_sara(word1)




×








187



        sara2 = self.check_sara(word2)




×








188



        if sara1 == 'อะ' and marttra1 == 'เกย':




×








189



            sara1 = 'ไอ'




×








190



            marttra1 = 'กา'




×








191



        elif sara2 == 'อะ' and marttra2 == 'เกย':




×








192



            sara2 = 'ไอ'




×








193



            marttra2 = 'กา'




×








194



        if sara1 == 'อำ' and marttra1 == 'กม':




×








195



            sara1 = 'อำ'




×








196



            marttra1 = 'กา'




×








197



        elif sara2 == 'อำ' and marttra2 == 'กม':




×








198



            sara2 = 'อำ'




×








199



            marttra2 = 'กา'




×







200



        # print(marttra1,marttra2)










201



        # print(sara1,sara2)











202



        if marttra1 == marttra2 and sara1 == sara2:




×








203



            return True




×







204



        else:











205



            return False




×







206













207



    def check_klon(self,text,k_type=8):




1✔






208



        if k_type == 8:




×








209



            try:




×








210



                error = []




×








211



                list_sumpus_sent1 = []




×








212



                list_sumpus_sent2h = []




×








213



                list_sumpus_sent2l = []




×








214



                list_sumpus_sent3 = []




×








215



                list_sumpus_sent4 = []




×








216



                for i, sent in enumerate(text.split()):




×








217



                    sub_sent = subword_tokenize(sent, engine='dict')




×







218



                    # print(i)











219



                    if len(sub_sent) > 10:




×








220



                        error.append('In the sentence'+str(i+2)+'there are more than 10 words.'+str(sub_sent))




×








221



                    if (i+1) % 4 == 1:




×








222



                        list_sumpus_sent1.append(sub_sent[-1])




×








223



                    elif (i+1) % 4 == 2:




×








224



                        list_sumpus_sent2h.append([sub_sent[1],sub_sent[2],sub_sent[3],sub_sent[4]])




×








225



                        list_sumpus_sent2l.append(sub_sent[-1])




×








226



                    elif (i+1) % 4 == 3:




×








227



                        list_sumpus_sent3.append(sub_sent[-1])




×








228



                    elif (i+1) % 4 == 0:




×








229



                        list_sumpus_sent4.append(sub_sent[-1])




×








230



                if len(list_sumpus_sent1) != len(list_sumpus_sent2h) or len(list_sumpus_sent2h) != len(list_sumpus_sent2l) or len(list_sumpus_sent2l) != len(list_sumpus_sent3) or len(list_sumpus_sent3) != len(list_sumpus_sent4)  or len(list_sumpus_sent4) != len(list_sumpus_sent1):




×








231



                    return 'The poem does not complete 4 sentences.'




×







232



                else:











233



                    for i in range(len(list_sumpus_sent1)):




×








234



                        countwrong = 0




×








235



                        for j in list_sumpus_sent2h[i]:




×








236



                            if self.check_sumpus(list_sumpus_sent1[i],j) == False:




×








237



                                    countwrong +=1




×








238



                        if  countwrong > 3:




×








239



                            error.append('Cant find rhyme between paragraphs '+str((list_sumpus_sent1[i],list_sumpus_sent2h[i]))+'in paragraph '+str(i+1))




×








240



                        if self.check_sumpus(list_sumpus_sent2l[i],list_sumpus_sent3[i]) == False:




×







241



                            # print(sumpus_sent2l,sumpus_sent3)











242



                            error.append('Cant find rhyme between paragraphs '+str((list_sumpus_sent2l[i],list_sumpus_sent3[i]))+'in paragraph '+str(i+1))




×








243



                        if i > 0:




×








244



                            if self.check_sumpus(list_sumpus_sent2l[i],list_sumpus_sent4[i-1]) == False:




×








245



                                error.append('Cant find rhyme between paragraphs '+str((list_sumpus_sent2l[i],list_sumpus_sent4[i-1]))+'in paragraph '+str(i+1))




×








246



                    if error == []:




×








247



                        return 'The poem is correct according to the principle.'




×







248



                    else:











249



                        return error




×








250



            except:




×








251



                return 'Something went wrong Make sure you enter it in correct form of klon4.'




×








252



        elif k_type == 4:




×








253



            try:




×








254



                error = []




×








255



                list_sumpus_sent1 = []




×








256



                list_sumpus_sent2h = []




×








257



                list_sumpus_sent2l = []




×








258



                list_sumpus_sent3 = []




×








259



                list_sumpus_sent4 = []




×








260



                for i, sent in enumerate(text.split()):




×








261



                    sub_sent = subword_tokenize(sent, engine='dict')




×








262



                    if len(sub_sent) > 5:




×








263



                        error.append('In the sentence'+str(i+2)+'there are more than 4 words.'+str(sub_sent))




×








264



                    if (i+1) % 4 == 1:




×








265



                        list_sumpus_sent1.append(sub_sent[-1])




×








266



                    elif (i+1) % 4 == 2:




×







267



                        # print([sub_sent[1],sub_sent[2]])











268



                        list_sumpus_sent2h.append([sub_sent[1],sub_sent[2]])




×








269



                        list_sumpus_sent2l.append(sub_sent[-1])




×








270



                    elif (i+1) % 4 == 3:




×








271



                        list_sumpus_sent3.append(sub_sent[-1])




×








272



                    elif (i+1) % 4 == 0:




×








273



                        list_sumpus_sent4.append(sub_sent[-1])




×








274



                if len(list_sumpus_sent1) != len(list_sumpus_sent2h) or len(list_sumpus_sent2h) != len(list_sumpus_sent2l) or len(list_sumpus_sent2l) != len(list_sumpus_sent3) or len(list_sumpus_sent3) != len(list_sumpus_sent4)  or len(list_sumpus_sent4) != len(list_sumpus_sent1):




×








275



                    return 'The poem does not complete 4 sentences.'




×







276



                else:











277



                    for i in range(len(list_sumpus_sent1)):




×








278



                        countwrong = 0




×








279



                        for j in list_sumpus_sent2h[i]:




×







280



                            # print(list_sumpus_sent1[i],j)











281



                            if self.check_sumpus(list_sumpus_sent1[i],j) == False:




×








282



                                    countwrong +=1




×








283



                        if  countwrong > 1:




×








284



                            error.append('Cant find rhyme between paragraphs '+str((list_sumpus_sent1[i],list_sumpus_sent2h[i]))+'in paragraph '+str(i+1))




×








285



                        if self.check_sumpus(list_sumpus_sent2l[i],list_sumpus_sent3[i]) == False:




×







286



                            # print(sumpus_sent2l,sumpus_sent3)











287



                            error.append('Cant find rhyme between paragraphs '+str((list_sumpus_sent2l[i],list_sumpus_sent3[i]))+'in paragraph '+str(i+1))




×








288



                        if i > 0:




×








289



                            if self.check_sumpus(list_sumpus_sent2l[i],list_sumpus_sent4[i-1]) == False:




×








290



                                error.append('Cant find rhyme between paragraphs '+str((list_sumpus_sent2l[i],list_sumpus_sent4[i-1]))+'in paragraph '+str(i+1))




×








291



                    if error == []:




×








292



                        return 'The poem is correct according to the principle.'




×







293



                    else:











294



                        return error




×








295



            except:




×








296



                return 'Something went wrong Make sure you enter it in correct form.'




×







297



            










298



        else:











299



            return 'Something went wrong Make sure you enter it in correct form.'



×





















    

  • 

    •