+ 2ja0t$Rt^RIHt^RIt^RIt^RIHtHt^RI H t H t H t H t ^RIHt^RIHt] 'd^RIHt.R1OtRRlt]!4tR]R &RRR]/tR]R&RR/tR]R &RtR]R&RtR]R&]!^^04tR]R&]!RR04tR ]R!&!R"R ]4t!R#R]4t !R$R ]!4t"R%R&lt#R'R(lt$R)R*lt%!R+R] 4t&!R,R ] 4t'R-R.lt(R/R0lt)R#)2zHTTP transport abstractions for nab-index. Defines minimal protocols for async HTTP GET requests. Implementations can use any async HTTP library (httpx, or urllib3 wrapped in to_thread, etc.). ) annotationsN)PackageNotFoundErrorversion) TYPE_CHECKINGAnyFinalProtocol)IndexAccessError)RETRY_STATUSES)MappingDEFAULT_HEADERSIDENTITY_HEADERS USER_AGENTAsyncHttpTransportContentDecodingError HttpError HttpResponseUnserveableUrlErrorcV^8dQhRR/#returnstr)formats"^/home/user/billing-ledger-validation/.venv/lib/python3.14/site-packages/nab_index/transport.py __annotate__r%s))S)cDR\R4 2# \dR#i;i)zFReturn the ``nab-index/`` name PyPI's API guidelines ask for.z nab-index/z nab-indexznab-index/0.0.0+unknown)rrrrr _user_agentr%s,)GK0122 )()s  z Final[str]zAccept-Encodinggzipz User-AgentzFinal[dict[str, str]]identityir_HTTP_BAD_REQUESTi_HTTP_SERVER_ERRORzFinal[frozenset[int]]_CONTENT_STATUSESzx-gzipzFinal[frozenset[str]] _GZIP_CODINGSc]tRt^@tRtRtR#)rzA request failed, or answered with a status the caller cannot use. Transports raise this from ``get`` and ``raise_for_status`` so callers can handle index failures without importing a specific HTTP backend. rN__name__ __module__ __qualname____firstlineno____doc____static_attributes__rrrrr@src]tRt^HtRtRtR#)ra The index reached a verdict that it will not serve this URL. Raised for a client-error status the retry policy does not treat as a blip, so not a 408 or a 429: a 404 on an advertised PEP 658 sidecar, a 403, a 410. The status is a property of the URL, so asking again gets the same answer, and a caller may treat the artifact as unavailable. A 5xx that outlived the retry budget, and a connection that failed, stay a bare :class:`HttpError`. Those say nothing about the URL, so a caller must not read them as a verdict. rNr'rrrrrHs rc]tRt^VtRtRtR#)rz@A response body did not decode as its Content-Encoding promised.rNr'rrrrrVsJrc V^8dQhRRRR/#)rparamsrrfloatr)rs"rrrZs  S U rcVPR4FHpVPR4wr#pVP4P4R8XgK<\ V4u# R# \ dR#i;i)zzReturn an Accept-Encoding entry's ``q`` value, 1.0 when it carries none. A ``q`` that does not parse reads as a refusal. ;=qgg?)split partitionstriplowerr2 ValueError)r1paramkey_values& r_qualityr@Zsh c",  99;   # % U|# #   s A  A0/A0c V^8dQhRRRR/#)rrequest_headersMapping[str, str]rboolr)rs"rrris"3rc^VP4UUu/uFwrVP4VbK pppVPRR4PR4FOpVP R4wrVpVP 4P4R8XgK<\ V4^8gKNR# R#uuppi)aJWhether ``request_headers`` asked the server for gzip. A static file server derives Content-Encoding from the filename, so it serves a ``.tar.gz`` as its own untouched bytes under ``Content-Encoding: gzip``. Decoding that yields a bare tar, which no published digest covers, so only a coding the request asked for may be undone. zaccept-encoding,r4r TF)itemsr:getr7r8r9r@)rBnamer?foldedentrycodingr>r1s& r accepts_gziprNis6E5J5J5L M5LkddjjlE!5LF M-r288=!OOC06 <<>   !V +0@10D>  NsB)c$V^8dQhRRRRRR/#)rbodybytescontent_encodingz str | Nonerr)rs"rrrzs!11e1z1e1rc$V'dVfV#VP4P4\9dV#\P!V4# \ \ P\P3dpRT 2p\T4ThRp?ii;i)aCReturn ``body`` decoded per ``content_encoding``. Transports fetch bodies undecoded and decode them here: the HTTP libraries' own gzip decoders accept a stream cut before its trailer and hand back a silent prefix under a 200, while :func:`gzip.decompress` checks the trailer (CRC and length) and turns the truncation into :class:`ContentDecodingError`. Only gzip is handled, the one coding the transports advertise; any other coding passes through untouched. An empty body also passes through: a bodiless response (a 304) may still carry the representation's Content-Encoding. Nz,gzip response body is truncated or corrupt: ) r9r:r%r decompressEOFErrorzliberror BadGzipFiler)rPrRexcmsgs&& r decode_bodyr[zs #+ %%'}< 1t$$ djj$"2"2 31rs# <66$!'  )% J&v**& ,=j*I'I5E+4c3Z+@(@(1&(1C'D $D  ) K9K "12) 8) X  2  r