^jEUdZddlmZmZmZddlmZmZee ddZ ee d<dede d e d e fd Zd Zee d <eeZee e d<edZee e d<eeDcic]\}}|| c}}Zee e fe d<dZe e d<ded e fdZdede d e fdZded e fdZde d e fdZdede d e d e fdZded eeefdzfdZded edzfdZycc}}w) aEarly detection of escape-sequence-based encodings (ISO-2022, HZ-GB-2312, UTF-7). These encodings use ESC (0x1B), tilde (~), or plus (+) sequences to switch character sets. They must be detected before binary detection (ESC is a control byte) and before ASCII detection (HZ-GB-2312 and UTF-7 use only printable ASCII bytes plus their respective shift markers). Note: ``from __future__ import annotations`` is intentionally omitted because this module is compiled with mypyc, which does not support PEP 563 string annotations. )EVIDENCE_CAP_BYTES count_deleteddecodes_without_error)DETERMINISTIC_CONFIDENCEDetectionResult! _HZ_GB_BYTESdata max_startmax_endreturncd} |jd|}|dk(s||k\ry|jd|dz|}|dk(ry||dz|}t|dk\r.t|dzdk(rt|tt|k(ry|dz})aFCheck that at least one ~{...~} region contains valid GB2312 byte pairs. In HZ-GB-2312 GB mode, characters are encoded as pairs of bytes in the 0x21-0x7E range. We require at least one region with a non-empty, even- length run of such bytes. *max_start* bounds where a region may open and *max_end* where it may close. Keeping them separate is what lets a region that opens inside the evidence window close beyond it: bounding both at the same offset would cut a straddling region and hide the only escape evidence a document has (ADR-0006). rT~{F~})findlenrr )r r r startbeginendregions Y/var/www/html/metafoliadorv2/venv/lib/python3.12/site-packages/chardet/pipeline/escape.py_has_valid_hz_regionsrs E  %' B;%9,iiuqy'2 "9eai#& K1 F a1$fl3s6{Ba' s@ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/ _B64_CHARS _UTF7_BASE64sABCDEFGHIJKLMNOPQRSTUVWXYZ_B64_UPPERCASE _B64_DECODE_EMBEDDED_B64_RUN b64_bytesct|}|dz}|dz}|dkDrt|d}d|zdz }||zry|dz}t|}d}d} d} |D]1} |dzt| z}| dz } | dk\s| dz} || z dz|| <| dz } 3d} td|dz d D]C} || dz|| dzz}d |cxkrd kr nn| ryd } )d |cxkrdkr nn| syd} >| ryd} E| S)uCheck if base64 bytes decode to valid UTF-16BE with correct padding. A valid UTF-7 shifted sequence must: 1. Contain at least 3 Base64 characters (18 bits, enough for one 16-bit UTF-16 code unit). 2. Have zero-valued trailing padding bits (the unused low bits of the last Base64 sextet after the last complete 16-bit code unit). 3. Decode to valid UTF-16BE — no lone surrogates. This rejects accidental ``+-`` patterns found in URLs, MIME boundaries, hex-encoded hashes (e.g. SHA-1 git refs), and other ASCII data. The caller (``_has_valid_utf7_sequences``) already checks ``b64_len >= 3`` before calling this function, so *b64_bytes* is always at least 3 bytes. rrFriiTii)rr bytearrayrange)r#n total_bits padding_bitslast_valmask num_bytesrawbit_buf bit_countout_idxc prev_highi code_units r_is_valid_utf7_b64r:GsL IAQJ?Lay}-\!Q& d? aI I CGIG a<;q>1Q > NI#y0D8CL qLG I 1i!mQ ' Vq[CAJ. Y (& (I y *F *II =rposcttdz}d}|dz }|dk\r0||}|dvr|dz}||vr|dz }|tk\ry|dz}n y|dk\r0y)uReturn True if the ``+`` at *pos* is embedded in a base64 stream. Walks backward from *pos*, skipping CR/LF, and counts consecutive base64 characters (including ``=`` for padding). If 4 or more are found, the ``+`` is likely part of a PEM certificate, email attachment, or similar base64 blob rather than a real UTF-7 shift character. The walk stops at the fourth character rather than running to the start of the run. Only the ``>= 4`` verdict is used, and newlines are skipped rather than ending the walk, so without the early exit a line-wrapped base64 blob — a PEM file, a MIME attachment, exactly what this guard exists to recognize — makes each ``+`` rescan everything before it, and the caller quadratic in the size of the blob. =rr'> TF)r frozensetr")r r; b64_with_padcountr8bs r_is_embedded_in_base64rDs~$0)D/#AL E aA q& G  FA   QJE)) FA   q& rb64_datac`t|ddzt|ddzzt|dzdz S)z@Decode the first (only) UTF-16 code unit of a single-unit block.r r'r%r)r )rEs r _single_unitrHsK Xa[ !R ' x{ #q ( * hqk " #   runitcd|cxkxrdkncxsjd|cxkxrdkncxsXd|cxkxrdkncxsFd|cxkxrdkncxs4d |cxkxrd kncxs"d |cxkxrd kncxsd |cxkxrdkScS)z>Script ranges where a lone shifted character plausibly occurs.iiii i+i0i0iNiiiii)rIs r_plausible_lone_unitrMs $ &  $ T #V # $ T #V # $ T #V # $ T #V #  $ T #V #  $ T #V # $rc"d}tt||} |jtd|}|dk(s||k\ry|dz}|t|kr||tdk(r|dz}Q|t|krX||tdk(rG|t|kr6||tdk(r%|dz }|t|kr||tdk(r%|}t ||r|}|}||kr!||t vr|dz }||kr ||t vr||k(r|t|kr||t vr|}||z }|||} |dk\rt d | Ds|}5|dk\r1t| r&|d zd zdk(rt| } t| s|}jyt||}x) a&Check that *data* contains at least one valid UTF-7 shifted sequence. A valid shifted sequence is ``+`` terminated by either an explicit ``-`` or any non-Base64 character (per RFC 2152). The base64 portion must decode to valid UTF-16BE with correct zero-padding bits. The sequence ``+-`` is a literal plus sign and is **not** counted. *max_start* bounds where a shift may occur and *max_end* how far its base64 run may reach, so a run that begins inside the evidence window can finish beyond it (ADR-0006). A run that outruns *max_end* is skipped rather than judged on the part that fits: the padding and surrogate checks read a cut run as a different run, and accepting on evidence we did not see is the one direction this must never fail in. rT+rFr'-c3,K|] }|tvyw)N)r).0rCs r z,_has_valid_utf7_sequences..s#JAA$7#Jsr%r&) minrrordrDranyr:rHrMmax) r r r rlimit shift_posr;r8b64_lenrErIs r_has_valid_utf7_sequencesr\s E D 7 #E IIc#h. ?i94!m T?tCyCH4!GE  T?tCyCH4D /d3i3s8&;qD /d3i3s8&;E  "$ 2E  %iDG|3 FA%iDG|3 :%#d)+Q<0GE c'A; a<#J#J JE  a<.x8! "a'#H-+D1EC O rNcfd|vsd|vsd|vryd|vryd|vs d|vsd |vsd |vr d |vrd |vryy d|vryy)zName the ISO-2022 variant whose designations appear in *data*. :returns: ``(encoding, language)`` or ``None`` when no designation is present. s$(Os$(Ps$(Q)iso2022_jp_2004jas(I)iso2022_jp_extr_s$Bs$@s(Js$(D) iso2022_jp_2r_s$)C) iso2022_krkoNrL)r s r_iso2022_variantrfssTZ4/:3E&D% T       d?w$)$T! rcd|v}d|v}d|v}|s|s|sy|r4t|}|'t||drt|dt|dSt t |t }t t |dt z}|r3d |vr/d |vr+t|||rt|d rtd td S|r;|jr+t|||rt|d rtd tdSy)aDetect ISO-2022, HZ-GB-2312, and UTF-7 from escape/tilde/plus sequences. Every answer this stage returns decodes the whole window: a designation is three bytes that ordinary text can contain (``log: ESC $ B is the escape``), and a genuine ISO-2022 or HZ stream can still carry a bad byte past the header. The gate costs one decode per hit only. :param data: The raw byte data to examine. :returns: A :class:`DetectionResult` if an escape encoding is found, or ``None``. ~+Nrr')encoding confidencelanguagerrrhzzhzutf-7) rfrrrrUrrrisasciir\)r has_esc has_tildehas_plusvariantr r s rdetect_escape_encodingru=soG It|H 9X"4(  #8wqz#J" 3   CI12I#d)Q!334G  TM TM !$ 7 ; !$ -/  <  LLN %dIw ? !$ 0/  r) __doc__chardet._utilsrrrchardet.pipelinerrbytesr+r __annotations__intboolrrr@rr enumerater dictr"r:rDrHrMr\tuplestrrfru)r8r6s00rrs~ TSFE$-. e.!!#!!!JX EW(4 in4"++H!I #I1:*0EF1q!tF T#s(^F3:%:D:zSTB 5 S  s t XEXcXCXDXv$5$U38_t%;$N[[?T+A[{Gs( C