o ja*@sUdZddlmZmZdedefddZdZeed<e eZ e e ed <d d e eDZ ee e fed <d edefddZdede defddZdedefddZdededBfddZdS)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. )DETERMINISTIC_CONFIDENCEDetectionResultdatareturncCsd} |d|}|dkrdS|d|d}|dkrdS||d|}t|dkr>t|ddkr>tdd |Dr>dS|d}q) aCheck 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. rT~{F~}css(|]}d|ko dknVqdS)!~N).0br r ]/var/www/html/fyndo/pharma/fyndo/venv/lib/python3.10/site-packages/chardet/pipeline/escape.py $s&z(_has_valid_hz_regions..)findlenall)rstartbeginendregionr r r_has_valid_hz_regionss   rs@ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/ _B64_CHARS _UTF7_BASE64cCsi|]\}}||qSr r )r icr r r /sr _B64_DECODE b64_bytescCs@t|}|d}|d}|dkr"t|d}d|>d}||@r"dS|d}t|}d}d} d} |D]"} |d>t| B}| d7} | dkrT| d8} || ?d@|| <| d7} q2d} td|dd D]=} || d>|| dB}d |krwd krnn| r~dSd } q_d |krdkrnn| sdSd} q_| rdSd} q_| 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. rrFr iiTii)rr bytearrayrange)rn total_bits padding_bitslast_valmask num_bytesrawbit_buf bit_countout_idxr prev_highr code_unitr r r_is_valid_utf7_b642sH  r3poscCsnttdB}d}|d}|dkr3||}|dvr|d8}q ||vr*|d7}|d8}n |dkS|dks|dkS)aeReturn 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. =rr"> )r frozenset)rr4 b64_with_padcountrrr r r_is_embedded_in_base64os   r<cCsHd} |td|}|dkrdS|d}|t|kr(||tdkr(|d}q|t|krY||tdkrY|t|krV||tdkrV|d7}|t|krV||tdksD|}qt||ra|}q|}|t|kr||tvr|d7}|t|kr||tvso||}|||}|dkr|r|}q|dkrt|rdSt||}q) asCheck 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. rT+rFr"-)rordrr<rislowerr3max)rr shift_posr4rb64_lenb64_datar r r_has_valid_utf7_sequencess>    rFNcCsd|v}d|v}d|v}|s|s|sdS|red|vs"d|vs"d|vr)tdtd d Sd |vr4td td d Sd |vsDd|vsDd|vsDd|vrZd|vrSd|vrStd td d Stdtd d Sd|vretdtdd S|rzd|vrzd|vrzt|rztdtdd S|rt|dkrt|rtdtdd SdS)zDetect ISO-2022, HZ-GB-2312, and UTF-7 from escape/tilde/plus sequences. :param data: The raw byte data to examine. :returns: A :class:`DetectionResult` if an escape encoding is found, or ``None``. ~+Ns$(Os$(Ps$(Qiso2022_jp_2004ja)encoding confidencelanguages(Iiso2022_jp_exts$Bs$@s(Js$(D iso2022_jp_2s$)C iso2022_krkorrhzzhzutf-7)rrrrBrF)rhas_esc has_tildehas_plusr r rdetect_escape_encodingsh  r[)__doc__chardet.pipelinerrbytesboolrr__annotations__r9rint enumeraterdictr3r<rFr[r r r rs  "=<