U H$xe@sddddddddgZdd lmZdd lmZmZdd lmZmZm Z m Z m Z Gd ddeZ e e ed ddZe eeee eefdddZe eee dddZe eeee eefdddZe eeee eefdddZdS)EndSkipAlignerPrefixComparerSuffixComparerhamming_spherehamming_environmentedit_environment edit_distance)IntFlag)IteratorTuple)rrrrrc@s$eZdZdZdZdZdZdZdZdS)rz Flags for the Aligner that indicate which ends of reference or query may be skipped at no cost. Setting all four flags at the same time results in semiglobal alignment. N) __name__ __module__ __qualname____doc__ZREFERENCE_STARTZ QUERY_STARTZ REFERENCE_ENDZ QUERY_STOPZ SEMIGLOBALrr-lib/python3.8/site-packages/cutadapt/align.pyrs )streturnc Cst|}t|}tt|d}td|dD]}|d}|dd7<td|dD]X}t||d||dk}t|d|||d||dd} ||}| ||<qXq.|dS)z Return the edit distance between the strings s and t. The edit distance is the sum of the numbers of insertions, deletions, and mismatches that is minimally necessary to transform one string into the other. r r )lenlistrangeintmin) rrmncostsjprevimatchcrrrr%s    )rkrccs>t|}t|dD]$}t||D]}||||fVq"qdS)a  Find all strings t for which the hamming distance between s and t is at most k, assuming the alphabet is A, C, G, T. Yield tuples (t, e, m), where e is the hamming distance between s and t and m is the number of matches (equal to len(t) - e). r N)rrr)rr)r"errrrr?sccs|V|dkrdSt||dD]}t|}dD]T}t|D]<}|d||}|||dV|||ddVq<||Vq0t|D]"}|d|||ddVqq dS)z Apply all possible edits up to edit distance k to string s and yield the resulting strings. A string may be returned more than once. r Nr ZACGT)naive_edit_environmentrr)rr)r"chr&prefixrrrr+Ms   r+c cs||r|t|nd}t||dt|d}t}t||D]>}||krFq8||||}|dd\}}|||fVq8dS)a Find all strings t for which the edit distance between s and t is at most k, assuming the alphabet is A, C, G, T. Yield tuples (t, e, score), where e is the edit distance between s and t and score is the score of the optimal alignment. r )Zmax_error_rateflagsZ min_overlapN)rrsetr+addZlocate) rr)ZrateZalignerseenrresultZscoreerrorsrrrpy_edit_environmentcs  r5ccst|}d}dtt|ddg|dfg}|r|\}}}t|}|d|krj||d|dfV|||krxq.t||krq.|D]} dg|d} t|d| d<dg|d} td|dD]} || d| krdnd} | dkst|| d| }| | dd}|| d}||krN||krN||| dd| }}n,||krl|| | d}}n||| }}|| | <|| | <q||| | | fqq.dS)a2 Find all strings t for which the edit distance between s and t is at most k, assuming the alphabet is A, C, G, T. Yield tuples (t, e, m), where e is the edit distance between s and t and m is the number of matches in the optimal alignment. This is slow and only used in testing. ZTGCAr r rN)rrrpopr AssertionErrorappend)rr)r"ZalphabetZ work_stackrr#Zmatchesr&r,Z next_costsZ next_matchesr$r'ZdiagleftZupr(r!rrrslow_edit_environmentwsD          r;N)__all__enumr typingr r Zcutadapt._alignrrrrrrstrrrrr+r5r;rrrrs"