o Rŀg}@sdZddlmZddlmZddlmZddlmZddlm Z ddl m Z d d Z d d Z d dZddZddZGddde ZdS)zBio.SeqIO support for the "gck" file format. The GCK binary format is generated by the Gene Construction Kit software from Textco BioSoftware, Inc. )unpack)Seq) SeqFeature)SimpleLocation) SeqRecord)SequenceIteratorcCs*||}t||krtd|d|S)z9Read the specified number of bytes from the given handle.z Cannot read z bytes from handle)readlen ValueErrorhandlelengthdatarC/var/www/html/myenv/lib/python3.10/site-packages/Bio/SeqIO/GckIO.py_reads  rcCs*t|d}td|d}t||}||fS)aLRead a length-prefixed packet. Parts of a GCK file are made of "packets" comprising of 4 bytes giving the packet's size, followed by the packet's data. There is no type tag. The type of a packet, and thus the type of data it contains, is solely indicated by the position of the packet within the GCK file. >Ir)rrr rrr _read_packets  rcC,t|d}td|d}t||d}|S)zeRead a Pascal string. A Pascal string is one byte for length followed by the actual string. rz>BrASCIIrrdecoder rrr _read_pstring/ rcCr)zcRead a 32-bit Pascal string. Similar to a Pascal string but length is encoded on 4 bytes. rrrrrr rrr_read_p4string:rrccs&|d}|s tdt|dkrtdt|\}}td|ddd}||dkr1td|ddd}tt|}t|t|\}}td |dd \}}|t|kr_td|d |d krktd t|D]c}d |d } || | d } td | \} } } }}}}|dkrd}nd}t | | |d}| dkrd} nd} i}|dkrt |}|g|d<|dkrt |}|g|d<|dkrqot || |d}|j |qot|\}}td |dd \}}|d |dkrtdt|D]'}d |d} || | d}td|\} } }}|rt ||rt |qt|t|d}td|d}t||d}t|D]#}|d} || | d}td|ddd}|dkrYt |q7t|dt |}|dd|_|_||_t|d}td |d}|dkrd!|jd"<nd#|jd"<|VdS)$Nz Empty file.z1Improper header, cannot read 24 bytes from handlerrrz"Conflicting sequence length valuesrz>IH\z9Features packet size inconsistent with number of featuresz>II6xH14xB17xII35xBr)strandCDS misc_featurelabelnote)type qualifiersXz3Sites packet size inconsistent with number of sitesz >II24xII48xz>Hii z>16xBcirculartopologylinear)r r r rrrrrrangerrrrfeaturesappendrsplitnameid description annotations)r rpacketr seq_lengthsequencerecord num_featuresioffset feature_datastartendr&r!has_name has_commentversionlocationr'r4commentfeature num_sites site_data num_versionsversions version_dataflags circularityrrr_parseEs                       rOcs(eZdZdZfddZddZZS) GckIteratorzParser for GCK files.cstj|ddddS)z+Break up a GCK file into SeqRecord objects.bGCK)modefmtN)super__init__)selfsource __class__rrrVszGckIterator.__init__cCs t|}|S)zStart parsing the file, and return a SeqRecord generator. Note that a GCK file can only contain one sequence, so this iterator will always return a single record. )rO)rWr recordsrrrparseszGckIterator.parse)__name__ __module__ __qualname____doc__rVr\ __classcell__rrrYrrPs rPN)r`structrBio.SeqrBio.SeqFeaturerr Bio.SeqRecordr InterfacesrrrrrrOrPrrrrs