o DfC@sdZddlZddlZddlmZddlmZmZmZm Z m Z m Z m Z ddl m Z ddlmZddlmZmZmZmZmZmZmZmZmZmZeeZGd d d ZGd d d ZdS) z Parse adapter specifications N)Path)TypeOptionalListTupleIteratorAnyDict)xopen) FastaReader) Adapter FrontAdapterNonInternalFrontAdapter BackAdapterNonInternalBackAdapterAnywhereAdapter PrefixAdapter SuffixAdapter LinkedAdapterInvalidCharacterc @seZdZdZdedeededefddZededefd d Zd d Z d dZ e dedefddZ e dede eeeffddZdddddddddd ZedefddZeddZe ddZdd ZdS)!AdapterSpecificationa@ # noqa: E501 Description of a single non-linked adapter. These are the attributes: - name (None or str) - restriction (None, 'anchored', or 'noninternal') - sequence (nucleotide sequence as string) - parameters (dict with extra parameters such as 'max_errors', 'min_overlap') - cmdline_type ('front' for -a, 'back' for -g and 'anywhere' for -b) >>> AdapterSpecification.parse('a_name=ACGT;anywhere', 'back') AdapterSpecification(name='a_name', restriction=None, sequence='ACGT', parameters={'anywhere': True}, cmdline_type='back') name restrictionsequence cmdline_typecCs:|dvsJ|dvs J||_||_||_||_||_dS)N)Nanchored noninternalfrontbackanywhererrr parametersr)selfrrrr#rr%X/var/www/html/software/conda/envs/catlas/lib/python3.10/site-packages/cutadapt/parser.py__init__"s   zAdapterSpecification.__init__speccCs$|||\}}}}||||||S)zd ksFntd |q t|trc|dkrTtd |dd |d |}d}q |dkrktd d}q t|tsw|dkr{td|S)z Replace all occurrences of ``x{n}`` (where x is any character) with n occurrences of x. Raise ValueError if the expression cannot be parsed. >>> AdapterSpecification.expand_braces('TGA{5}CT') 'TGAAAAACT' Nz([{}]){z""{" must be used after a character}z"}" cannot be used hereri'zValue {} invalidz "}" expectedz Expected "{"zUnterminated expression)resplit ValueErrorintr, isinstance)rresultprevsr%r%r& expand_bracesEs8   z"AdapterSpecification.expand_bracescCs<|dd}d}t|dkr|\}}|}|}||fS)z_ Parse an adapter specification given as 'name=adapt' into 'name' and 'adapt'. =r N)r:lenstrip)r(fieldsrr%r%r& _extract_namems  z"AdapterSpecification._extract_namemax_error_rate max_errors min_overlapN) eZ error_raterGorHrIr!requiredoptionalc Cs$|d}t}|D]n}|}|sq |d\}}}|dkr'|dkr'td|}||jvr7td||j|durJ|j|}|j|dus>|}|dkrUd}nzt|}Wn tyht |}Ynw||vrttd||||<q d |vrd |vrtd d |vrd |d <|d =|S) z/Parse key=value;key=value;key=value into a dict;rBr5zNo value givenzUnknown parameter {}NTzKey {} specified twicerMrLz>'optional' and 'required' cannot be specified at the same timeF) r:dictrD partitionr;allowed_parametersKeyErrorr,r<float)r*r(rEr>fieldkeyequalsvaluer%r%r&_parse_parameterss@       z&AdapterSpecification._parse_parametersc Cs|dvrtdtd}|d\}}}||\}}|}||}||}t|ddkr8|d|ifSd}|drGd }|d d}|dr[|durT|d }| d }d} | d rjd } |dd}| dr~| durw|d } | d }t t |t t | } | d kr||dkr| rtd|dkr|rtd|dus| dusJ|dur|} n| } |dkr| durtd|| ||fS)a1 Parse an adapter specification for a non-linked adapter (without '...') Allow: 'back' and ADAPTER 'back' and ADAPTERX 'back' and ADAPTER$ 'front' and ADAPTER 'front' and XADAPTER 'front' and ^ADAPTER 'anywhere' and ADAPTER rz,cmdline_type must be front, back or anywherezYou cannot use multiple placement restrictions for an adapter at the same time. Choose one of ^ADAPTER, ADAPTER$, XADAPTER or ADAPTERXrNXrN^rr rxX$r8rzIAllowed placement restrictions for a 5' adapter are XADAPTER and ^ADAPTERr zIAllowed placement restrictions for a 3' adapter are ADAPTERX and ADAPTER$r!zPPlacement restrictions (with X, ^, $) not supported for 'anywhere' (-b) adapters)r;rPrFrDrXrArC startswithupperlstripendswithrstripr<bool) r*r(rerrormiddleZparameters_specrr#Zfront_restrictionZback_restrictionZn_placement_restrictionsrr%r%r&r)sd            zAdapterSpecification._parsecCs|dkr|dur tS|dkrtS|dkrtStd||dkr:|dur'tS|dkr-tS|dkr3tStd||dks@J|durFtStd ) zA restriction: None, "anchored", or "noninternal" rNrrz,Value {} for a front restriction not allowedr z+Value {} for a back restriction not allowedr!z5No placement may be specified for "anywhere" adapters) rrrr;r,rrrr)rrr%r%r&_restriction_to_classs0 z*AdapterSpecification._restriction_to_classcCs||j|jSr1)rerrr/r%r%r& adapter_classsz"AdapterSpecification.adapter_class)r. __module__ __qualname____doc__strrr' classmethodr+r0r3 staticmethodrArrFrQrXr)rerfr%r%r%r&rsH  ' $ H  rc @seZdZdZddZddededeed efd d Ze d ed ed e eeffddZ dedeeded efddZ d ed edeeded e f ddZddeded eefddZdee eefd eefddZdS) AdapterParserz Factory for Adapter classes that all use the same default parameters (error rate, indels etc.). The given **kwargs will be passed to the Adapter constructors. cKs ||_dSr1)default_parameters)r$kwargsr%r%r&r' s zAdapterParser.__init__r Nr(rrr4cCst|dvr td||d\}}}|dkr#|r#|r#|||||S|dkr1||||\}}n|}||||S)a Parse an adapter specification not using ``file:`` notation and return an object of an appropriate Adapter class. name -- Adapter name if not included as part of the spec. (If spec is 'name=ADAPTER', name will be 'name'.) cmdline_type -- describes which commandline parameter was used (``-a`` is 'back', ``-b`` is 'anywhere', and ``-g`` is 'front'). rzcmdline_type cannot be {!r}z...)r;r,rP _parse_linked_normalize_ellipsis_parse_not_linked)r$r(rrspec1rdspec2r%r%r&r)$s zAdapterParser._parsersrtcCs`|dkrtd|s|dkr|}||fStd|s,|dkr&d}|}||fS|}||fStd)Nr!z2No ellipsis ("...") allowed in "anywhere" adaptersr zInvalid adapter specificationrzExpected either spec1 or spec2)r;)rsrtrr(r%r%r&rq;sz!AdapterParser._normalize_ellipsiscCst||}|}|jddr|ttfvrd|jd<d|jvr%td|j }| |j|d|j |dur;|j n|d|S) Nr!FTZforce_anywhererLzA'required' and 'optional' can only be used within linked adapters)rrr%) rr+rfr#poprrr;rncopyupdaterr)r$r(rrZaspecrfr#r%r%r&rrRs     zAdapterParser._parse_not_linkedcCs|dkrtdt|d}t|d}|dur|j}|jdu}|jdu}|j} | |j|j} | |j|dkrDd} d} n|} |} | d| } | d| } | |j fddi| } | |j fddi| }t | || | |d S) z6Return a linked adapter from two specification stringsr!z*'anywhere' (-b) adapters may not be linkedrr NTrLr) front_adapter back_adapterfront_required back_requiredr) r;rr+rrrnrvrwr#rurfrr)r$rsrtrrZ front_specZ back_specZfront_anchoredZ back_anchoredZfront_parametersZback_parametersrzr{rxryr%r%r&rp^s@          zAdapterParser._parse_linkedc cs|drFt|ddddd+}t|}|D]}|jdd}|r'|dnd}|j|j||dVqWddS1s?wYdSz |j||ddVWdSty|}zt| rwd |d d |}t|j dd |d}~ww) aQ Parse an adapter specification and yield appropriate Adapter classes. This works like the _parse_no_file() function above, but also supports the ``file:`` notation for reading adapters from an external FASTA file. Since a file can contain multiple adapters, this function is a generator. zfile:Nrbr)modethreadsr )rzA file exists named '{}'. zFTo use the sequences in that file as adapter sequences, write 'file:' z!before the path, as in 'file:{}'. ) r]r r rr:r)rrrexistsr,args) r$r(rfZfastarecordrrJZ extra_messager%r%r&r+s. " zAdapterParser.parsetype_spec_pairscCs8g}|D]\}}|dvrtd||||q|S)ap Parse all three types of commandline options that can be used to specify adapters. adapters must be a list of (str, str) pairs, where the first is the adapter type (either 'front', 'back' or 'anywhere') and the second is the adapter specification given on the commandline Return a list of appropriate Adapter classes. >r r!rz,adapter type must be front, back or anywhere)r;extendr+)r$radaptersrr(r%r%r& parse_multis  zAdapterParser.parse_multi)r N)r )r.rgrhrir'rjrr r)rlrrqrrrrprr+rrr%r%r%r&rms  " ,&rm) rir9loggingpathlibrtypingrrrrrrr r Z dnaio.readersr rr rrrrrrrrr getLoggerr.loggerrrmr%r%r%r&s $  0