o UݢgO@s UddlmZddlZddlmZmZddlmZddlZddl Z ddl m Z ddl mZddlmZe e jjZe e jjZeefZd8dd Zd8d d Zd d ZddZddZddZd9ddZddZd:ddZ d;d!d"Z!ej"d#d$eZ#d%e$d&<d8dd6d7Z'dS)?) annotationsN)CallableIterable)unescape) ensure_str)tablesFcCs&t|}|s|rtd|d|S)zA wrapper around h5py.is_hdf5, optionally can throw an exception. Args: filename: The name of the file to test throw_exception: Should we raise an error if not? Returns: bool as to whether the file is valid zFile: z is not a valid HDF5 file.)h5pyis_hdf5OSError)filenameZthrow_exceptionvalidr ^/oak/stanford/groups/akundaje/marinovg/programs/cellranger-9.0.1/lib/python/cellranger/hdf5.pyr s r cCsXd}|rd}t||}|D]\}}|||<qWddS1s%wYdS)Nwa)rFileitems)r dataappendfilemodefkeyvaluer r rwrite_h5(s "rc Csrtj|dd(}z t|dtj}Wn tyd}Yn wWd|SWd|S1s2wY|S)Nr)mode/)r open_filerZ get_node_attr h5_constantsH5_FILETYPE_KEYAttributeError)r rfiletyper r rget_h5_filetype1s   r"cCsDt|d}|j|j||dWddS1swYdS)z(Save an array to the root of an h5 file.r)objN)rrZ create_carrayroot)r namearrrr r r save_array_h5:s"r'cCs@t|d}t|j|WdS1swYdS)z*Load an array from the root of an h5 file.rN)rrgetattrr$read)r r%rr r r load_array_h5@s$r*cKs|dust|dr t|dkr ttjdf}|j||ddSt|tjr,|jjt ks\s$z-create_hdf5_string_dataset..cSsg|] }|r t|ndqS)encode_ascii_xmlr3r r r `sz.create_hdf5_string_dataset..cs|]}t|VqdSr/lenr3r r rr6b)rr.) hasattrr=npr.bytes_create_datasetr0ndarraycharSTR_DTYPE_CHARlistallmax)groupr%rkwargsr. fixed_lenr r rcreate_hdf5_string_datasetFsrLr5 str | bytesreturnr2cCs6t|tr|St|trt|Stdt|)z4Decode a string from 7-bit ASCII + XML into unicode.Expected string type, got type )r0r2r1rdecode ValueErrortyper5r r rdecode_ascii_xmlls   rTcCsZt|tjr|jjtkr|Sdd|D}tdd|D}ttj|f}tj||dS)zYDecode an array-like container of strings from 7-bit ASCII + XML. into unicode. cSsg|]}t|qSr )rTr3r r rr:~z*decode_ascii_xml_array..csr;r/r<r4sr r rr6r>z)decode_ascii_xml_array..r-) r0r@rCr.rDUNICODE_DTYPE_CHARrHstr_array)rZ unicode_datarKr.r r rdecode_ascii_xml_arrayvs r[cCs6t|tr |ddSt|tr|Stdt|)zEncode a string as fixed-length 7-bit ASCII with XML-encoding. for characters outside of 7-bit ASCII. Respect python2 and python3, either unicode or binary. asciixmlcharrefreplacerO)r0r2encoder1rQrRrSr r rr9s   r9r"np.ndarray | Iterable[str | bytes]0np.ndarray[tuple[int, int], np.dtype[np.bytes_]]cs|t|tjr|jjtkr|jjdkr|Sddfdd|D}tdd|D}td|}ttj|f}tj ||d S) zEncode an array-like container of strings as fixed-length 7-bit ASCII. with XML-encoding for characters outside of 7-bit ASCII. rcSs|durt|SdS)Nr7r8)rWr r rconvertsz'encode_ascii_xml_array..convertcg|]}|qSr r r3rar rr:rUz*encode_ascii_xml_array..csr;r/r<rVr r rr6r>z)encode_ascii_xml_array..r,r-) r0r@rCr.rDrEitemsizerHrArZ)rZ ascii_datarKr.r rcrencode_ascii_xml_arrays    re)maxsizezCallable[[bytes | str], str]_cached_decodedataset h5py.Datasetmemoizebool list[str]cs8|jdurgS|dd}t|rtfdd|DS)aRead a dataset of strings from HDF5 (h5py). Args: dataset (h5py.Dataset): Data to read. memoize (bool): Whether to use memoization to make string conversion more efficient. Returns: list[unicode]: Strings in the dataset. Ncrbr r r3rPr rr:rUz,read_hdf5_string_dataset..)shaperTrh)rirkrr rnrread_hdf5_string_datasets  rpr%r#str | bytes | Iterable[str | bytes]cCsxt|}t|ttBrt|}n%t|tjr!|jjtvr!t |}nt|t t Br5t|dttBr5t |}||j |<dS)a)Set an attribute of an HDF5 dataset/group. Strings are stored as fixed-length 7-bit ASCII with XML-encoding for characters outside of 7-bit ASCII. This is inspired by the choice made for the Loom spec: https://github.com/linnarsson-lab/loompy/blob/master/doc/format/index.rst rN) r9r0r2r1r@rCr.rDSTRING_DTYPE_CHARSrerFtupleattrs)rir%rr r r set_hdf5_attrs   ruoutfilefiles_to_combineIterable[str | bytes]c Cst|tsJtt|d0}|D]$}tt|d}|j|j|jddWdn1s1wYqWddS1sBwYdS)NrrT) recursive)r0rFrrrZ copy_childrenr$)rvrwoutfnamerr r rcombine_h5s_into_ones"r|)F)r5rMrNr2)r5rM)rr_rNr`)rirjrkrlrNrm)r%rMrrq)rvrMrwrx)( __future__r functoolscollections.abcrrZhtmlrrnumpyr@sixrcellranger.h5_constantsrZcellranger.wrapped_tablesrr.rArDrErYrXrrr rr"r'r*rLrTr[r9re lru_cacherh__annotations__rprur|r r r rs4         &