o Git$@sddlmZddlmZddlmZddlmZddlmZddlmZddlmZddlm Z ed Z ed d Gd d d ee Z dS)) dataclass)field)Dict)Generic)List)Optional)TypeVar)UnionSymbolF)reprc@seZdZUdZeedZeee fe d< eedZ ee efe d< dZ ee d< dZ e e d< d d Zed ed dfd dZeded dfddZd efddZdefddZd0de deed efddZdeee fd ee effddZd1dd Zd2d!d"Zd#eee fd ee effd$d%Zd#eee fd efd&d'Zd efd(d)Zddd efd*d+Ze d e!efd,d-Z"e d e!e fd.d/Z#dS)3 SymbolTablea:SymbolTable that maps symbol IDs, found on the FSA arcs to actual objects. These objects can be arbitrary Python objects that can serve as keys in a dictionary (i.e. they need to be hashable and immutable). The SymbolTable can only be read to/written from disk if the symbols are strings. )default_factory_id2sym_sym2id_next_available_idepscCs|jD]\}}|j||ksJ|dksJq|jD]\}}|dks(J|j||ks1Jqd|jvrD|j|jd<d|j|j<n|jd|jksNJ|j|jdksXJt|jd|_dS)Nrr)ritemsrrmaxr)selfidxsymrC/home/ubuntu/.local/lib/python3.10/site-packages/k2/symbol_table.py __post_init__9s   zSymbolTable.__post_init__sreturncCst}t}|dD]E}|}t|dkrq t|dks'Jdt||dt|d}}||vs=Jd|||vsHJd||||<|||<q |dd}t|||d S) aBuild a symbol table from a string. The string consists of lines. Every line has two fields separated by space(s), tab(s) or both. The first field is the symbol and the second the integer id of the symbol. Args: s: The input string with the format described above. Returns: An instance of :class:`SymbolTable`.  rz$Expect a line with 2 fields. Given: rzDuplicated symbol zDuplicated id rrrr)dictsplitlenintgetr )rid2symsym2idlinefieldsrrrrrrfrom_strKs     zSymbolTable.from_strfilenamecCsDt|ddd}t|WdS1swYdS)aBuild a symbol table from file. Every line in the symbol table file has two fields separated by space(s), tab(s) or both. The following is an example file: .. code-block:: 0 a 1 b 2 c 3 Args: filename: Name of the symbol table file. Its format is documented above. Returns: An instance of :class:`SymbolTable`. rzutf-8)encodingN)openr r*readstrip)r+frrr from_filels$zSymbolTable.from_filecCs4d}t|jD]\}}||d|d7}q |S)z Returns: Return a string representation of this object. You can pass it to the method ``from_str`` to recreate an identical object.  r)sortedrr)rrrsymbolrrrto_strszSymbolTable.to_strcCsVt|d}t|jD] \}}t|||dq WddS1s$wYdS)aSerialize the SymbolTable to a file. Every line in the symbol table file has two fields separated by space(s), tab(s) or both. The following is an example file: .. code-block:: 0 a 1 b 2 c 3 Args: filename: Name of the symbol table file. Its format is documented above. w)fileN)r.r5rrprint)rr+r1rr6rrrto_files "zSymbolTable.to_fileNr6indexcCsx||jvr |j|S|dur|j}||jvr&td|d|d|j|||j|<||j|<|j|kr:|d|_|S)ahAdd a new symbol to the SymbolTable. Args: symbol: The symbol to be added. index: Optional int id to which the symbol should be assigned. If it is not available, a ValueError will be raised. Returns: The int id to which the symbol has been assigned. NzCannot assign id 'z' to 'z' - already occupied by r)rrr ValueError)rr6r<rrradds       zSymbolTable.addkcCst|tr |j|S|j|S)aPGet a symbol for an id or get an id for a symbol Args: k: If it is an id, it tries to find the symbol corresponding to the id; if it is a symbol, it tries to find the id corresponding to the symbol. Returns: An id or a symbol depending on the given `k`.  isinstancer$rr)rr?rrrr%s  zSymbolTable.getothercCs:||i|j|j}i|j|j}t|||jdS)aCreate a union of two SymbolTables. Raises an AssertionError if the same IDs are occupied by different symbols. Args: other: A symbol table to merge with ``self``. Returns: A new symbol table. r )_check_compatiblerrr r)rrBr&r'rrrmerges zSymbolTable.mergec Cs|j|jksJd|jd|jt|j|j}|D]}||||ks9Jd|d||d||dqt|j|j}|D]}||||ksaJd|d||d||dqEdS) NzMismatched epsilon symbol: z != zID conflict for id: z, self[idx] = "z", other[idx] = ""z, self[sym] = "z", other[sym] = ")rsetr intersectionr)rrB common_idsrcommon_symbolsrrrrrCs(  zSymbolTable._check_compatibleitemcCs ||SN)r%rrJrrr __getitem__ zSymbolTable.__getitem__cCst|tr ||jvS||jvSrKr@rLrrr __contains__s   zSymbolTable.__contains__cCs t|jSrK)r#r)rrrr__len__rNzSymbolTable.__len__cCs:t|t|kr dS|jD] }||||krdSq dS)NFT)r#symbols)rrBrrrr__eq__s zSymbolTable.__eq__cCt|j}||S)zDReturns a list of integer IDs corresponding to the symbols. )listrkeyssortransrrridsszSymbolTable.idscCrS)z\Returns a list of symbols (e.g., strings) corresponding to the integer IDs. )rTrrUrVrWrrrrQszSymbolTable.symbolsrK)rBr rr )rBr rN)$__name__ __module__ __qualname____doc__rr!rrr$r __annotations__rrrr staticmethodstrr*r2r7r;rr>r r%rDrCrMboolrOrPrRpropertyrrYrQrrrrr s:     "  " r N) dataclassesrrtypingrrrrrr r r rrrrs