o ޥia0@srddlmZmZmZmZddlZddlZddlZddlZdZ dZ GdddZ Gddde Z Gd d d e ZdS) )unicode_literalsdivisionabsolute_importprint_functionNconfig dictionaryc@sJeZdZdddZddZgfddZdd Zd d Zd d ZddZ dS)OpenCCNcCsPd|_||_d|_t|_t|_t|_t d|_ |jdur&| dSdS)a init OpenCC :param conversion: the conversion of usage, options are 'hk2s', 's2hk', 's2t', 's2tw', 's2twp', 't2hk', 't2s', 't2tw', 'tw2s', 'tw2sp', etc check the json file names in config directory :return: None Fu(\s+|-|,|\.|\?|!|\*|　|，|。|、|；|：|？|！|…|“|”|‘|’|『|』|「|」|﹁|﹂|—|－|（|）|《|》|〈|〉|～|．|／|＼|︒|︑|︔|︓|︿|﹀|︹|︺|︙|︐|［|﹇|］|﹈|︕|︖|︰|︳|︴|︽|︾|︵|︶|｛|︷|｝|︸|﹃|﹄|【|︻|】|︼)N) conversion_name conversion_dict_init_donelist _dict_chain_dict_chain_datadict dict_cacherecompilesplit_chars_re _init_dictselfr rA/home/ubuntu/.local/lib/python3.10/site-packages/opencc/opencc.py__init__$s  zOpenCC.__init__cCsv|js |d|_g}|j|}tdt|D]}|ddkr.|||||jq|||qd |S)z] Convert string from Simplified Chinese to Traditional Chinese or vice versa Trr ) r rrsplitrangelenappend_convertrjoin)rstringresultsplit_string_listirrrconvert9s   zOpenCC.convertcCs<t|}|D]}||td|}qd|S)av Convert string from Simplified Chinese to Traditional Chinese or vice versa If a dictionary is part of a group of dictionaries, stop conversion on a word after the first match is found. :param string: the input string :param dictionary: list of dictionaries to be applied against the string :return: converted string r ) StringTreecreate_parse_treer!inorder)rr"rtreec_dictrrrr Ps   zOpenCC._convertcCs|jdur tdg|_|jd}tjtjtt|}t | }t |}Wdn1s1wY| d|_ | dD] }||j| dqAg|_||j|jt|jD]\}}t|trm|g|j|<q^d|_dS)zR initialize the dict with chosen conversion :return: None Nzconversion is not setz.jsonnameconversion_chainrT)r ValueErrorrospathr!dirname__file__ CONFIG_DIRopenjsonloadgetr _add_dict_chainr_add_dictionaries enumerate isinstancetupler )rr config_filef setting_jsonchainindexr+rrrr_s$        zOpenCC._init_dictc Cs|D]w}t|trg}|||||q||jvrqi}d}d}tj|ddd.}|D]#} | d\} } | || <t | |krFt | }t | |krPt | }q-Wdn1s[wY||||f|||f|j|<q||j|qdS)Nirzutf-8)encoding ) r;r r9rrior4striprr) r chain_list chain_dataitemr@map_dictmax_lenmin_lenr>linekeyvaluerrrr9zs0      zOpenCC._add_dictionariescCs~|ddkrg}|dD]}|||q||dS|ddkr=|d}tjtjtt|}||dSdS)z add dict chain :param dict_chain: the dict chain to add to :param dict_dict: the dict to be added in :return: None typegroupdictstxtfileN) r7r8rr/r0r!r1r2DICT_DIR)r dict_chain dict_dictr@ dict_itemfilename dict_filerrrr8s zOpenCC._add_dict_chaincCs|j|krdSd|_||_dS)a  set conversion :param conversion: the conversion of usage, options are 'hk2s', 's2hk', 's2t', 's2tw', 's2twp', 't2hk', 't2s', 't2tw', 'tw2s', and 'tw2sp' check the json file names in config directory :return: None NF)r r rrrrset_conversions  zOpenCC.set_conversionN) __name__ __module__ __qualname__rr&r rr9r8r\rrrrr#s   rc@s>eZdZdZdZdddZddZdd Zd d Zd d Z dS)TreeNoderrBNcCs ddg|_||_d|_||_dS)NF)branchrPmatched length_hint)rrPhintrrrrs  zTreeNode.__init__cC ||_dSr])rc)rrcrrr set_matched zTreeNode.set_matchedcCrfr])rP)rrPrrr set_valuerhzTreeNode.set_valuecCs||j|<dSr])rb)rrbnoderrr set_branchzTreeNode.set_branchcCrfr])rd)rrerrrset_hintrhzTreeNode.set_hintr]) r^r_r`LEFTRIGHTrrgrirkrmrrrrras  rac@s.eZdZddZddZddZd dd ZdS) r'cCst||_dSr])raroot)rr"rrrrrlzStringTree.__init__c Cs|jg}g}|D]]}|r_|}||j||j\}}}} |rU|||d|d|rAt|| } | | | tj | |rTt|| } | | | tj | n| |d|_|s |} |}| }qdS)a Compare smaller and smaller sub-strings going from left to rightin root node value against a test_dict_list entry. If match is found, create tree nodes for remaining left and right string portions and place these nodes on a stack for processing. :param test_dict_list: a list of tuples of the max key length and dict currently being applied against the string NT) rppop_StringTree__findMatchrPrdrirmrgrarrkrnro) rtest_dict_list working_stackunmatched_stack test_dictcurrrPlstringrstringtest_lenrjtemprrrr(s4         zStringTree.create_parse_treecCs\g}g}|j} |r|||jtj}|s |r*|}||j|jtj}n |Sq)zd Do a non-recursive inorder traversal of the tree. :return: list of strings )rprrbrarnrqrPro)r return_valstackrwrrrr)s   zStringTree.inorderNc Cst|}d}d}t||d}|rt||}|d}||krztd||dD]J} || | ||dvrq| dkr?|d| }| ||krM|| |d}|d|| | |} t| ddkri| dd} | |||fSq'|d8}||ksdS)a8 Compare smaller and smaller sub-strings going from left to right against test_dict. If an entry is found, return it as well as the remaining string(s) and the test length. :param cstring: the string to find a match :param test_dict: a tuple of the max key length and dict currently being applied against the string :return: the new matched value, old string to left of the match, old string to right of the match (may be all None if no match found), last test length NrrBr )NNNN)rminrr) rr"rvre string_lenrxryrzrMr%rPrrr __findMatch s,    zStringTree.__findMatchr])r^r_r`rr(r)rrrrrrr's 'r') __future__rrrrrFr/r5rr3rVrobjectrar'rrrrs