Pluck9527 · Pluck9527 · Aug 11, 2026 · Aug 11, 2026 · Aug 11, 2026 · Aug 11, 2026
diff --git a/docs/ctf-codecs.md b/docs/ctf-codecs.md
@@ -0,0 +1,99 @@
+# CTF-oriented codecs
+
+This fork adds codecs required by the Chinese CTF Misc workflow while preserving CodExt's registry and chaining model.
+
+| Codec | Example name | Notes |
+|---|---|---|
+| Base92 | `base92` | Compatible with thenoviceoof/base92; `flag` encodes to `F#S<I`. |
+| Base1024 / Ecoji | `base1024`, `ecoji` | Apache-licensed Ecoji Base1024 repertoire; `flag` encodes to `👪😑🌟🙋`. |
+| Base2048 | `base2048` | Uses qntm's Base2048 repertoire. |
+| Base65536 | `base65536`, `base65535` | Uses qntm's repertoire; `base65535` is a guide-compatible alias. |
+| Base64 padding steganography | `base64-stego`, `hide_base64_padding(...)` | Round-trip generated carriers plus caller-supplied Base64 carrier lines; decoded carrier bytes remain unchanged. |
+| Custom-alphabet Base64 | `add_custom_base64(alphabet, name)` | Registers a named codec from exactly 64 unique non-padding, non-whitespace characters. |
+| MD5 CTF helpers | `md5-16[-upper]`, `md5-upper`, `crack-md5` | 16/32-character case variants plus built-in or caller-supplied dictionary lookup through `crack_md5(...)`. |
+| Zero-width steganography | `zero-width` | U+200B/U+200C bits separated by U+200D. |
+| Hill | `hill-3,3,2,5`, `hill-GYBNQKURP` | Any invertible square matrix from 2x2 through 6x6, supplied as integers or letters. |
+| Gronsfeld | `gronsfeld-31415` | Numeric repeating key. |
+| Music symbols | `music-symbol` | Offline CTF music-symbol mapping with its `§=` terminator. |
+| Vigenere arbitrary key | `vigenere-key-h687474703a2f2f...`, `add_vigenere_codec(...)` | Hex parameter/runtime registration accepts URL keys; punctuation is ignored and ASCII letters form the key stream. |
+| Cloud shadow | `cloud-shadow` | Also registered as `yunying`. |
+| XXencode | `xxencode` | Binary-to-text encoding with 45-byte lines. |
+| PGP word list | `pgp-word-list` | Enforces alternating even/odd word lists. |
+| Core values | `core-values` | Chinese socialist core-values hexadecimal mapping. |
+| Chinese Telegraph Code | `chinese-telegraph[-mainland|-taiwan]` | Fixed `zhtelecode==0.1.0` codebooks. |
+| Enigma I | `enigma-iv.ii.i-c-lfp-hrq-hr.qp.fz.sw.eu` | Rotors, reflector, positions, rings and plugboard are encoded in the name. |
+| Text encoding brute force | `text-encoding-brute-force` | Recursive breadth-first repair (depth 3); returns ranked JSON with the full conversion path. |
+| Mojibake repair | `mojibake-gbk` | Reverses common UTF-8-as-GB18030 mojibake, including the `锟斤拷` replacement symptom. |
+| Quoted-Printable | `quoted-printable`, `qp` | UTF-8 payloads through Python's MIME-compatible Quoted-Printable implementation. |
+| Unicode escapes | `unicode-escape` | Supports `\\u`, `\\U`, `\\x`, `U+`, `+U`, decimal and hexadecimal HTML entities. |
+| Brainfuck / Ook! | `brainfuck`, `ook`, `short-ook` | Bounded interpreter, bracket validation and standard/short Ook token pairs. |
+| AAEncode | `aaencode` | Emits executable AAEncoded JavaScript and decodes it without evaluating JavaScript. |
+| Decabit | `decabit` | Complete 0-126 ten-pulse table; `DECA` matches the guide example. |
+| DNA | `dna1` through `dna8`, `dna-triplet` | Eight complementary-pairing maps plus the guide's three-nucleotide 63-character table. |
+| SMS PDU | `sms-pdu[-DESTINATION]`, `sms-pdu-gsm7-DESTINATION`, `sms-pdu-8bit-DESTINATION`, `sms-pdu-info` | SUBMIT/DELIVER/status-report, GSM-7/8-bit/UCS-2, 8/16-bit concatenation, application ports, national-shift UDH metadata and multipart reassembly. |
+| Differential Manchester | `differential-manchester[-inverted]` | Start-of-bit rule plus mandatory mid-bit transition validation. |
+| SNOW whitespace steganography | `snow`, `snow-compressed`, `snow-compressed-p-h70617373` | stegsnow-compatible bit order, `-C` Huffman and `-p` ICE-CFB password mode. |
+| Whitespace language | `whitespace-lang`, `whitespace-lang-input-h41` | Bounded interpreter with stack, arithmetic, heap, flow-control and byte/number input. |
+| AES / DES / 3DES | `aes-cbc-hKEY-hIV-b64-pkcs7` | ECB/CBC, hex/Base64, PKCS#7/zero/no padding; keys and IVs use hex parameters. |
+| RC4 | `rc4-hKEY-b64` | Keyed RC4 with hex or Base64 ciphertext. |
+| SM2 / SM3 / SM4 / SM9 | `sm3`, `sm4-ecb-hKEY-hex-none`, `add_sm2_codec(...)` | SM3, SM4, SM2 encryption plus SM2 and identity-based SM9 key generation, signing and verification helpers/codecs. |
+| Rabbit | `rabbit-hKEY-hIV-hex`, `rabbit-pbe-hPASSWORD` | Raw Rabbit plus CryptoJS/OpenSSL `Salted__` passphrase envelopes. |
+| Emoji-AES | `emoji-aes-hPASSWORD[-ROTATION]` | Compatible with the original CryptoJS Emoji-AES alphabet and optional rotation. |
+| ADFGX / ADFGVX | `adfgx-german`, `adfgvx-german` | Guide default squares; optional custom square as `h<UTF-8-hex>`. |
+| Substitution analysis | `frequency-analysis`, `quipqiup`, `substitution-hALPHABET` | English quadgram simulated annealing, ranked local candidates and explicit monoalphabetic keys; no service call. |
+| Key recovery | `crack-vigenere`, `hill-crack`, `hill-recover-3`, `enigma-crack[-CRIB]` | Vigenere frequency recovery, exhaustive 2x2 Hill, known-plaintext NxN Hill and bounded Enigma rotor-position search. |
+| Keyboard encodings | `keyboard-coordinates`, `keyboard-qwe`, `keyboard-symbol-shift`, `phone-26`, `phone-t9-coordinate` | Coordinates, QWE remapping, repeated-symbol vertical movement, 26-key conversion and both T9 token conventions. |
+| Numeric tap code | `tap-numeric` | Accepts and emits guide-style `5,2 3,1 ...` row/column pairs. |
+| Chinese ASCII | `chinese-ascii`, `unicode-decimal` | Decimal Unicode code points and decimal/hex HTML entities. |
+| Chinese niche encodings | `pinyin-tone`, `pawnshop`, `chinese-strokes` | Tone-derived ASCII, pawnshop digits and the guide's 1-12 stroke table. |
+| VBE | `vbe` | Microsoft Script Encoder-compatible encoding and multi-block decoding without script execution. |
+| Twitter Secret Messages | `twitter-secret` | Compatible Unicode homoglyph steganography; helpers accept separate cover and secret strings. |
+| Spammimic | `spammimic`, `spammimic-online` | Deterministic UTF-8/CRC-protected offline spam cover; the separately named online adapter remains for the official proprietary grammar. |
+
+```python
+import codext
+
+alphabet = "/+9876543210zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA"
+codext.add_custom_base64(alphabet, "base64-ctf")
+plaintext = codext.decode("i5qMi/==", "base64-ctf")
+```
+
+All codecs work through `codext.encode`, `codext.decode`, the CLI registry and codec chains. Chinese Telegraph Code uses Mainland/Taiwan codebooks derived from Unicode Unihan mappings.
+
+Examples for the new CTF batch:
+
+```python
+import json
+import codext
+
+candidates = json.loads(codext.decode("涓枃", "text-encoding-brute-force"))
+assert any(candidate["text"] == "中文" for candidate in candidates)
+
+pdu = codext.encode("你好flag", "sms-pdu-13800138000")
+assert codext.decode(pdu, "sms-pdu") == "你好flag"
+
+carrier = codext.encode("flag{SNOW}", "snow-compressed")
+assert codext.decode(carrier, "snow-compressed") == "flag{SNOW}"
+
+aes = "aes-cbc-h30313233343536373839616263646566-h66656463626139383736353433323130-b64-pkcs7"
+assert codext.decode(codext.encode("flag{AES}", aes), aes) == "flag{AES}"
+
+hidden = codext.hide_twitter_secret("This is a sufficiently long ASCII cover. " * 4, "flag-test")
+assert codext.reveal_twitter_secret(hidden) == "flag-test"
+
+assert codext.decode("5,2 3,1 3,1 3,2", "tap-numeric") == "wllm"
+assert codext.decode("ooo yyy ii", "phone-26") == "you"
+assert codext.decode("20013 25991", "chinese-ascii") == "中文"
+
+private_key, public_key = codext.generate_sm2_signing_keypair()
+signature = codext.sm2_sign("flag{SM2}", private_key)
+assert codext.sm2_verify("flag{SM2}", signature, public_key)
+
+master_secret, master_public, user_key = codext.generate_sm9_signing_keys("alice@example.com")
+signature = codext.sm9_sign("flag{SM9}", "alice@example.com", master_public, user_key)
+assert codext.sm9_verify("flag{SM9}", signature, "alice@example.com", master_public)
+```
+
+`snow-compressed` corresponds to SNOW's `-C` mode. Password-bearing names append `-p-h<password bytes in hex>`. `whitespace-lang` executes at most 1,000,000 instructions; the plain codec supplies an empty input channel and the `-input-h...` form supplies exact input bytes. Dynamic crypto parameters deliberately use `h<hex>` so URLs, punctuation and binary keys remain unambiguous.
+
+Sources: thenoviceoof/base92 (MIT), Ecoji (Apache-2.0), qntm/base2048 and qntm/base65536 (MIT), the PGPfone word list, zhtelecode 0.1.0 (MIT), gmalg (MIT), 3GPP TS 23.038/23.040, the original AAEncode implementation, Ascetics/ctfkit's reverse-engineered CTF music map, dCode's Decabit table, stegsnow's Apache-2.0 whitespace/Huffman/ICE implementation, CryptoJS Rabbit/AES formats, the Emoji-AES project, Didier Stevens' public-domain VBE decoder and Twitter Secret Messages' published homoglyph table. The local English quadgram model is derived from Project Gutenberg eBook 11, which is public domain in the United States.
diff --git a/pyproject.toml b/pyproject.toml
@@ -26,8 +26,14 @@ classifiers = [
   "Topic :: Software Development :: Libraries :: Python Modules",
 ]
 dependencies = [
+  "ecoji==0.1.1",
+  "gmalg>=1.1.1",
+  "gmssl>=3.2.2",
   "legacycrypt; python_version >= '3.13'",
   "markdown2>=2.5.4",
+  "pycryptodome>=3.23",
+  "pypinyin>=0.55",
+  "zhtelecode==0.1.0",
 ]
 dynamic = ["version"]
 
@@ -76,3 +82,6 @@ base100 = "codext.base.base100:main100"
 base122 = "codext.base.base122:main122"
 codext = "codext.__init__:main"
 unbase = "codext.base.__init__:main"
+
+[tool.setuptools.package-data]
+codext = ["VERSION.txt", "macros.json", "others/pgp_word_list.json", "crypto/english_quadgrams.txt"]
diff --git a/requirements.txt b/requirements.txt
@@ -1,2 +1,7 @@
 legacycrypt; python_version >= '3.13'
 markdown2>=2.5.4
+gmssl>=3.2.2
+pycryptodome>=3.23
+pypinyin>=0.55
+gmalg>=1.1.1
+ecoji==0.1.1
diff --git a/src/codext/__init__.py b/src/codext/__init__.py
@@ -21,6 +21,33 @@
 
 reset()
 
+from .base.baseN import add_custom_base64
+from .hashing.md import crack_md5
+__all__.append("add_custom_base64")
+__all__.append("crack_md5")
+
+from .crypto.modern import add_symmetric_codec
+from .crypto.keyed_vigenere import add_vigenere_codec
+from .crypto.sm import (add_sm2_codec, add_sm2_signature_codec, add_sm9_signature_codec, generate_sm2_keypair,
+                        generate_sm2_signing_keypair, generate_sm9_signing_keys, sm2_sign, sm2_verify, sm9_sign,
+                        sm9_verify)
+from .crypto.substitution import frequency_report, solve_substitution, solve_substitution_candidates
+from .crypto.analysis import crack_enigma, crack_hill, crack_vigenere, recover_hill_key, recover_vigenere_key
+from .stegano.twitter_secret import hide_twitter_secret, reveal_twitter_secret
+from .stegano.base64_stego import hide_base64_padding
+from .stegano.spammimic import spammimic_decode, spammimic_encode
+from .stegano.whitespace_lang import run_whitespace
+from .others.sms_pdu import parse_sms_pdu, sms_pdu_encode
+from .others.vbe import decode_vbe_payload, encode_vbe_payload
+__all__.extend(["add_symmetric_codec", "add_vigenere_codec", "add_sm2_codec", "add_sm2_signature_codec",
+                "add_sm9_signature_codec", "generate_sm2_keypair", "generate_sm2_signing_keypair",
+                "generate_sm9_signing_keys", "sm2_sign", "sm2_verify", "sm9_sign", "sm9_verify",
+                "frequency_report", "solve_substitution", "solve_substitution_candidates", "crack_enigma",
+                "crack_hill", "crack_vigenere", "recover_hill_key", "recover_vigenere_key", "hide_twitter_secret",
+                "reveal_twitter_secret", "hide_base64_padding", "spammimic_encode", "spammimic_decode",
+                "run_whitespace", "parse_sms_pdu", "sms_pdu_encode", "decode_vbe_payload",
+                "encode_vbe_payload"])
+
 
 # populate codext with attributes from codecs that were not modified
 for attr in codecs.__all__:

diff --git a/src/codext/base/__init__.py b/src/codext/base/__init__.py
@@ -5,8 +5,10 @@
 from .base45 import *
 from .base85 import *
 from .base91 import *
+from .base92 import *
 from .base100 import *
 from .base122 import *
+from .unicode_bases import *
 from .baseN import *
 from ..__common__ import *
 from ..__info__ import __version__

diff --git a/src/codext/base/base92.py b/src/codext/base/base92.py
@@ -0,0 +1,66 @@
+# -*- coding: UTF-8 -*-
+"""Base92 codec compatible with thenoviceoof/base92."""
+from ..__common__ import *
+
+
+ALPHABET = "!" + "".join(chr(value) for value in range(ord("#"), ord("_") + 1)) + \
+           "".join(chr(value) for value in range(ord("a"), ord("}") + 1))
+
+
+def base92_encode(text, errors="strict"):
+    data, buffer, bits, result = b(text), 0, 0, []
+    if not data:
+        return "~", 1
+    for byte in data:
+        buffer = (buffer << 8) | byte
+        bits += 8
+        while bits >= 13:
+            value = buffer >> (bits - 13)
+            buffer &= (1 << (bits - 13)) - 1
+            bits -= 13
+            result.extend((ALPHABET[value // 91], ALPHABET[value % 91]))
+    if bits:
+        width = 6 if bits < 7 else 13
+        value = buffer << (width - bits)
+        result.extend((ALPHABET[value],) if width == 6 else (ALPHABET[value // 91], ALPHABET[value % 91]))
+    return (encoded := "".join(result)), len(encoded)
+
+
+def base92_decode(text, errors="strict"):
+    text = ensure_str(text)
+    if text in ("", "~"):
+        return b"", len(text)
+    if len(text) == 1:
+        raise Base92DecodeError("one character is not a valid Base92 encoding")
+    error = handle_error("base92", errors, decode=True)
+    buffer, bits, result, pairs = 0, 0, bytearray(), len(text) // 2 * 2
+    for index in range(0, pairs, 2):
+        try:
+            value = ALPHABET.index(text[index]) * 91 + ALPHABET.index(text[index + 1])
+        except ValueError:
+            replacement = error(text[index:index + 2], index)
+            if replacement:
+                result.extend(b(replacement))
+            continue
+        if value >= 8192:
+            raise Base92DecodeError("invalid Base92 pair at position %d" % index)
+        buffer = (buffer << 13) | value
+        bits += 13
+        while bits >= 8:
+            result.append(buffer >> (bits - 8))
+            buffer &= (1 << (bits - 8)) - 1
+            bits -= 8
+    if pairs < len(text):
+        try:
+            buffer = (buffer << 6) | ALPHABET.index(text[-1])
+            bits += 6
+        except ValueError:
+            error(text[-1], len(text) - 1)
+        while bits >= 8:
+            result.append(buffer >> (bits - 8))
+            buffer &= (1 << (bits - 8)) - 1
+            bits -= 8
+    return bytes(result), len(text)
+
+
+add("base92", base92_encode, base92_decode, r"^base[-_]?92$", expansion_factor=1.25)
diff --git a/src/codext/base/baseN.py b/src/codext/base/baseN.py
@@ -114,6 +114,20 @@
 main64url = main(64, "RFC 4648 / Base64URL", "url", False)
 
 
+def add_custom_base64(alphabet, name="base64-custom"):
+    """Register a Base64 codec backed by a caller-supplied 64-character alphabet."""
+    if not isinstance(alphabet, str) or len(alphabet) != 64:
+        raise ValueError("Base64 alphabet must contain exactly 64 characters")
+    if len(set(alphabet)) != 64:
+        raise ValueError("Base64 alphabet characters must be unique")
+    if "=" in alphabet or any(char.isspace() for char in alphabet):
+        raise ValueError("Base64 alphabet cannot contain padding or whitespace characters")
+    if not isinstance(name, str) or not re.match(r"^[a-zA-Z0-9][a-zA-Z0-9_-]*$", name):
+        raise ValueError("Bad custom Base64 codec name")
+    base2n(alphabet, r"^(%s)$" % re.escape(name), name=name, padding_char="=", guess=[])
+    return name
+
+
 B67 = {
     r'':                 upper + lower + digits + "-_.!~",
     r'[-_]inv(erted)?$': lower + upper + digits + "-_.!~",