asciimath2 0.1.0__tar.gz

asciimath2-0.1.0/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Alex Lee
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

asciimath2-0.1.0/PKG-INFO

@@ -0,0 +1,151 @@
+Metadata-Version: 2.4
+Name: asciimath2
+Version: 0.1.0
+Summary: LaTeX -> AsciiMath2 conversion (AsciiMath2 = asciimath.org + the asciimath-parser extensions, see SPEC.md)
+Keywords: latex,asciimath,asciimath2,math,converter,markup
+Author: Alex Lee
+Author-email: Alex Lee <iasandcb@gmail.com>
+License-Expression: MIT
+License-File: LICENSE
+Classifier: Development Status :: 4 - Beta
+Classifier: Intended Audience :: Developers
+Classifier: Intended Audience :: Science/Research
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Programming Language :: Python :: 3.13
+Classifier: Programming Language :: Python :: 3.14
+Classifier: Topic :: Scientific/Engineering :: Mathematics
+Classifier: Topic :: Text Processing :: Markup :: LaTeX
+Classifier: Typing :: Typed
+Requires-Dist: pylatexenc>=2.11
+Requires-Python: >=3.12
+Project-URL: Homepage, https://github.com/iasandcb/asciimath2
+Project-URL: Repository, https://github.com/iasandcb/asciimath2
+Project-URL: Issues, https://github.com/iasandcb/asciimath2/issues
+Project-URL: Specification, https://github.com/iasandcb/asciimath2/blob/main/SPEC.md
+Description-Content-Type: text/markdown
+
+# asciimath2
+
+**AsciiMath2** is asciimath.org's grammar extended with everything
+[`asciimath-parser`](https://github.com/widcardw/asciimath-parser) already
+implements (matrices, aligned multi-line equations, colors, fonts, accents,
+...), plus two more contributions on top of that:
+[render-time configurability](https://github.com/widcardw/asciimath-parser/pull/22)
+and a [symbol-table isolation contract](https://github.com/widcardw/asciimath-parser/pull/21).
+The full grammar is written down in **[SPEC.md](https://github.com/iasandcb/asciimath2/blob/main/SPEC.md)** — read that first.
+
+This directory is the standard plus one from-scratch implementation of one
+direction through it:
+
+```
+LaTeX  --[this package]-->  AsciiMath2  --[asciimath-parser]-->  LaTeX
+```
+
+`asciimath-parser` already goes AsciiMath2 → LaTeX. This package goes the
+other way, which nothing implemented before. Together they round-trip.
+
+## Install & use
+
+```bash
+uv sync            # or: pip install -e .
+uv run asciimath2 '\frac{1}{2} + \sqrt{x^2+1}'
+# 1/2 + sqrt(x^2 + 1)
+```
+
+```python
+from asciimath2 import convert, Config, UnsupportedLatexError
+
+convert(r"\sum_{i=1}^n i^2")
+# 'sum_(i = 1)^n i^2'
+
+convert(r"\begin{pmatrix} a & b \\ c & d \end{pmatrix}")
+# '(a, b; c, d)'
+
+try:
+    convert(r"\begin{tikzpicture}...\end{tikzpicture}")
+except UnsupportedLatexError:
+    ...  # fall back to keeping the original LaTeX -- see "On failure" below
+```
+
+`Config` mirrors the three render options from PR #22
+(`multiline_env`, `single_newline_break`, `bar_as_mid` — see SPEC.md §2.2);
+only `single_newline_break` changes what this package emits.
+
+## How it works
+
+`src/asciimath2/`:
+
+- `latex_context.py` — extends `pylatexenc`'s macro database with argument
+  specs for macros it doesn't know the arity of (`\binom`, `\overset`,
+  `\operatorname`, ...), so pylatexenc attaches their `{...}` groups as real
+  arguments instead of leaving them as unrelated sibling nodes.
+- `symbols.py` — the LaTeX-command → AsciiMath2-token tables. Ground truth
+  is `asciimath-parser`'s own `SYMBOLMAP`
+  (`packages/core/src/symbols.ts`), read directly from source, not
+  paraphrased from docs.
+- `render.py` — flattens a pylatexenc node tree into a linear token stream,
+  then does a single left-to-right pass building AsciiMath2 text. `^`/`_`
+  bind to exactly the token before/after them; any compound expression used
+  as a required single argument gets wrapped in `(...)`, which AsciiMath2's
+  own grammar strips invisibly in exactly those positions — so this
+  converter never has to reimplement LaTeX operator precedence itself.
+- `cli.py` — `asciimath2 '<latex>'` / reads stdin.
+
+## Scope, honestly
+
+This is a broad, tested common subset of LaTeX math — not a general LaTeX
+interpreter (that's not a well-defined target; LaTeX can define arbitrary
+macros). What's covered: arithmetic, fractions/roots, sub/superscripts,
+`\left`/`\right` delimiters (all standard pairs), big operators with limits,
+trig/log/etc. functions, Greek letters, relations/logic/arrows, fonts and
+accents, `\text`, `\binom`/`\operatorname`/`\pmod`/`\color`, and matrix-like
+and aligned-like environments (`pmatrix`/`bmatrix`/`vmatrix`/`Vmatrix`/
+`matrix`/`array`/`cases`, `aligned`/`align`/`gather`/`split`/...).
+
+Two deliberate choices about what happens outside that:
+
+- **`tex(...)` escape hatch** — an unrecognized *bare* macro with no
+  arguments (e.g. `\digamma`) degrades to `tex(\digamma)`, AsciiMath2's own
+  raw-LaTeX passthrough (SPEC.md §2.4), rather than failing the whole
+  expression.
+- **On failure, raise, don't guess** — anything else unsupported (an unknown
+  macro that's ambiguous about whether it takes arguments, an unrecognized
+  environment, a `\left` with no matching `\right`, ...) raises
+  `UnsupportedLatexError` for the *whole* expression rather than silently
+  emitting something plausible-looking but wrong. This is the same contract
+  mark-vector's current `backend/app/latex_to_asciimath.py` already expects
+  from `py-asciimath` (`None`/exception on failure → keep the original LaTeX
+  visible via `\(...\)`/`\[...\]`), so this package is a drop-in candidate
+  for that integration point if/when it's wired in — see "Not yet done"
+  below.
+
+## Integration status
+
+Both follow-ups from the initial version of this package are now done:
+
+- **Wired into mark-vector's backend.** `backend/app/latex_to_asciimath.py`
+  uses this package (added as a local editable dependency in
+  `backend/pyproject.toml`) instead of `py-asciimath` — which is the whole
+  reason this package exists (`py-asciimath` couldn't do matrices, `\\`,
+  and other constructs this package handles). Multi-row output (aligned
+  systems, etc.) is preserved as blank-line-separated rows inside the
+  `$$...$$` block it writes back to the document, rather than being
+  collapsed to one line, so the row structure survives to the frontend.
+- **Frontend renders AsciiMath2's extensions.** `frontend/static/js/markdown.js`
+  imports `asciimath-parser` directly as an ES module from jsDelivr
+  (`https://cdn.jsdelivr.net/npm/asciimath-parser@<version>/dist/index.js` —
+  it ships no browser-global/UMD build, unlike the `asciimath2tex` package it
+  replaced) instead of `asciimath2tex`, which only implemented the base
+  asciimath.org grammar. For the AsciiMath dialect, a full `$$...$$` block is
+  now handed to the parser in one call rather than being split line-by-line,
+  so `asciimath-parser`'s own row/`&`-alignment handling (SPEC.md §2.2) does
+  the work instead of the frontend approximating it with a `gathered`
+  environment. The LaTeX dialect's per-line `gathered` stacking is
+  unchanged, since LaTeX has no equivalent convention.
+
+Verified end to end (`asciimath2.convert()` output fed through the actual
+published `asciimath-parser@0.6.11` and, in a real browser via Playwright,
+through KaTeX) for derivatives/limits, matrix multiplication, aligned
+systems of equations, piecewise `cases`, and norms — no parse errors,
+correct LaTeX, correct rendering.

asciimath2-0.1.0/README.md

@@ -0,0 +1,124 @@
+# asciimath2
+
+**AsciiMath2** is asciimath.org's grammar extended with everything
+[`asciimath-parser`](https://github.com/widcardw/asciimath-parser) already
+implements (matrices, aligned multi-line equations, colors, fonts, accents,
+...), plus two more contributions on top of that:
+[render-time configurability](https://github.com/widcardw/asciimath-parser/pull/22)
+and a [symbol-table isolation contract](https://github.com/widcardw/asciimath-parser/pull/21).
+The full grammar is written down in **[SPEC.md](https://github.com/iasandcb/asciimath2/blob/main/SPEC.md)** — read that first.
+
+This directory is the standard plus one from-scratch implementation of one
+direction through it:
+
+```
+LaTeX  --[this package]-->  AsciiMath2  --[asciimath-parser]-->  LaTeX
+```
+
+`asciimath-parser` already goes AsciiMath2 → LaTeX. This package goes the
+other way, which nothing implemented before. Together they round-trip.
+
+## Install & use
+
+```bash
+uv sync            # or: pip install -e .
+uv run asciimath2 '\frac{1}{2} + \sqrt{x^2+1}'
+# 1/2 + sqrt(x^2 + 1)
+```
+
+```python
+from asciimath2 import convert, Config, UnsupportedLatexError
+
+convert(r"\sum_{i=1}^n i^2")
+# 'sum_(i = 1)^n i^2'
+
+convert(r"\begin{pmatrix} a & b \\ c & d \end{pmatrix}")
+# '(a, b; c, d)'
+
+try:
+    convert(r"\begin{tikzpicture}...\end{tikzpicture}")
+except UnsupportedLatexError:
+    ...  # fall back to keeping the original LaTeX -- see "On failure" below
+```
+
+`Config` mirrors the three render options from PR #22
+(`multiline_env`, `single_newline_break`, `bar_as_mid` — see SPEC.md §2.2);
+only `single_newline_break` changes what this package emits.
+
+## How it works
+
+`src/asciimath2/`:
+
+- `latex_context.py` — extends `pylatexenc`'s macro database with argument
+  specs for macros it doesn't know the arity of (`\binom`, `\overset`,
+  `\operatorname`, ...), so pylatexenc attaches their `{...}` groups as real
+  arguments instead of leaving them as unrelated sibling nodes.
+- `symbols.py` — the LaTeX-command → AsciiMath2-token tables. Ground truth
+  is `asciimath-parser`'s own `SYMBOLMAP`
+  (`packages/core/src/symbols.ts`), read directly from source, not
+  paraphrased from docs.
+- `render.py` — flattens a pylatexenc node tree into a linear token stream,
+  then does a single left-to-right pass building AsciiMath2 text. `^`/`_`
+  bind to exactly the token before/after them; any compound expression used
+  as a required single argument gets wrapped in `(...)`, which AsciiMath2's
+  own grammar strips invisibly in exactly those positions — so this
+  converter never has to reimplement LaTeX operator precedence itself.
+- `cli.py` — `asciimath2 '<latex>'` / reads stdin.
+
+## Scope, honestly
+
+This is a broad, tested common subset of LaTeX math — not a general LaTeX
+interpreter (that's not a well-defined target; LaTeX can define arbitrary
+macros). What's covered: arithmetic, fractions/roots, sub/superscripts,
+`\left`/`\right` delimiters (all standard pairs), big operators with limits,
+trig/log/etc. functions, Greek letters, relations/logic/arrows, fonts and
+accents, `\text`, `\binom`/`\operatorname`/`\pmod`/`\color`, and matrix-like
+and aligned-like environments (`pmatrix`/`bmatrix`/`vmatrix`/`Vmatrix`/
+`matrix`/`array`/`cases`, `aligned`/`align`/`gather`/`split`/...).
+
+Two deliberate choices about what happens outside that:
+
+- **`tex(...)` escape hatch** — an unrecognized *bare* macro with no
+  arguments (e.g. `\digamma`) degrades to `tex(\digamma)`, AsciiMath2's own
+  raw-LaTeX passthrough (SPEC.md §2.4), rather than failing the whole
+  expression.
+- **On failure, raise, don't guess** — anything else unsupported (an unknown
+  macro that's ambiguous about whether it takes arguments, an unrecognized
+  environment, a `\left` with no matching `\right`, ...) raises
+  `UnsupportedLatexError` for the *whole* expression rather than silently
+  emitting something plausible-looking but wrong. This is the same contract
+  mark-vector's current `backend/app/latex_to_asciimath.py` already expects
+  from `py-asciimath` (`None`/exception on failure → keep the original LaTeX
+  visible via `\(...\)`/`\[...\]`), so this package is a drop-in candidate
+  for that integration point if/when it's wired in — see "Not yet done"
+  below.
+
+## Integration status
+
+Both follow-ups from the initial version of this package are now done:
+
+- **Wired into mark-vector's backend.** `backend/app/latex_to_asciimath.py`
+  uses this package (added as a local editable dependency in
+  `backend/pyproject.toml`) instead of `py-asciimath` — which is the whole
+  reason this package exists (`py-asciimath` couldn't do matrices, `\\`,
+  and other constructs this package handles). Multi-row output (aligned
+  systems, etc.) is preserved as blank-line-separated rows inside the
+  `$$...$$` block it writes back to the document, rather than being
+  collapsed to one line, so the row structure survives to the frontend.
+- **Frontend renders AsciiMath2's extensions.** `frontend/static/js/markdown.js`
+  imports `asciimath-parser` directly as an ES module from jsDelivr
+  (`https://cdn.jsdelivr.net/npm/asciimath-parser@<version>/dist/index.js` —
+  it ships no browser-global/UMD build, unlike the `asciimath2tex` package it
+  replaced) instead of `asciimath2tex`, which only implemented the base
+  asciimath.org grammar. For the AsciiMath dialect, a full `$$...$$` block is
+  now handed to the parser in one call rather than being split line-by-line,
+  so `asciimath-parser`'s own row/`&`-alignment handling (SPEC.md §2.2) does
+  the work instead of the frontend approximating it with a `gathered`
+  environment. The LaTeX dialect's per-line `gathered` stacking is
+  unchanged, since LaTeX has no equivalent convention.
+
+Verified end to end (`asciimath2.convert()` output fed through the actual
+published `asciimath-parser@0.6.11` and, in a real browser via Playwright,
+through KaTeX) for derivatives/limits, matrix multiplication, aligned
+systems of equations, piecewise `cases`, and norms — no parse errors,
+correct LaTeX, correct rendering.

asciimath2-0.1.0/pyproject.toml

@@ -0,0 +1,49 @@
+[project]
+name = "asciimath2"
+version = "0.1.0"
+description = "LaTeX -> AsciiMath2 conversion (AsciiMath2 = asciimath.org + the asciimath-parser extensions, see SPEC.md)"
+readme = "README.md"
+requires-python = ">=3.12"
+license = "MIT"
+license-files = ["LICENSE"]
+keywords = [
+    "latex",
+    "asciimath",
+    "asciimath2",
+    "math",
+    "converter",
+    "markup",
+]
+classifiers = [
+    "Development Status :: 4 - Beta",
+    "Intended Audience :: Developers",
+    "Intended Audience :: Science/Research",
+    "Programming Language :: Python :: 3",
+    "Programming Language :: Python :: 3.12",
+    "Programming Language :: Python :: 3.13",
+    "Programming Language :: Python :: 3.14",
+    "Topic :: Scientific/Engineering :: Mathematics",
+    "Topic :: Text Processing :: Markup :: LaTeX",
+    "Typing :: Typed",
+]
+dependencies = ["pylatexenc>=2.11"]
+
+[[project.authors]]
+name = "Alex Lee"
+email = "iasandcb@gmail.com"
+
+[project.urls]
+Homepage = "https://github.com/iasandcb/asciimath2"
+Repository = "https://github.com/iasandcb/asciimath2"
+Issues = "https://github.com/iasandcb/asciimath2/issues"
+Specification = "https://github.com/iasandcb/asciimath2/blob/main/SPEC.md"
+
+[project.scripts]
+asciimath2 = "asciimath2.cli:main"
+
+[build-system]
+requires = ["uv_build>=0.12.4,<0.13.0"]
+build-backend = "uv_build"
+
+[dependency-groups]
+dev = ["pytest>=9.1.1"]

asciimath2-0.1.0/pyproject.toml.orig

@@ -0,0 +1,45 @@
+[project]
+name = "asciimath2"
+version = "0.1.0"
+description = "LaTeX -> AsciiMath2 conversion (AsciiMath2 = asciimath.org + the asciimath-parser extensions, see SPEC.md)"
+readme = "README.md"
+authors = [
+    { name = "Alex Lee", email = "iasandcb@gmail.com" }
+]
+requires-python = ">=3.12"
+license = "MIT"
+license-files = ["LICENSE"]
+keywords = ["latex", "asciimath", "asciimath2", "math", "converter", "markup"]
+classifiers = [
+    "Development Status :: 4 - Beta",
+    "Intended Audience :: Developers",
+    "Intended Audience :: Science/Research",
+    "Programming Language :: Python :: 3",
+    "Programming Language :: Python :: 3.12",
+    "Programming Language :: Python :: 3.13",
+    "Programming Language :: Python :: 3.14",
+    "Topic :: Scientific/Engineering :: Mathematics",
+    "Topic :: Text Processing :: Markup :: LaTeX",
+    "Typing :: Typed",
+]
+dependencies = [
+    "pylatexenc>=2.11",
+]
+
+[project.urls]
+Homepage = "https://github.com/iasandcb/asciimath2"
+Repository = "https://github.com/iasandcb/asciimath2"
+Issues = "https://github.com/iasandcb/asciimath2/issues"
+Specification = "https://github.com/iasandcb/asciimath2/blob/main/SPEC.md"
+
+[project.scripts]
+asciimath2 = "asciimath2.cli:main"
+
+[build-system]
+requires = ["uv_build>=0.12.4,<0.13.0"]
+build-backend = "uv_build"
+
+[dependency-groups]
+dev = [
+    "pytest>=9.1.1",
+]

asciimath2-0.1.0/src/asciimath2/__init__.py

@@ -0,0 +1,10 @@
+"""AsciiMath2: LaTeX -> AsciiMath2 conversion.
+
+See SPEC.md (repo root) for the AsciiMath2 standard this implements.
+"""
+
+from .config import Config
+from .errors import UnsupportedLatexError
+from .render import convert
+
+__all__ = ["convert", "Config", "UnsupportedLatexError"]

asciimath2-0.1.0/src/asciimath2/cli.py

@@ -0,0 +1,28 @@
+import argparse
+import sys
+
+from .errors import UnsupportedLatexError
+from .render import convert
+
+
+def main(argv: list[str] | None = None) -> int:
+    parser = argparse.ArgumentParser(
+        prog="asciimath2",
+        description="Convert a LaTeX math expression to AsciiMath2 (see SPEC.md).",
+    )
+    parser.add_argument(
+        "latex", nargs="?", help="LaTeX expression. Reads stdin if omitted."
+    )
+    args = parser.parse_args(argv)
+
+    latex = args.latex if args.latex is not None else sys.stdin.read()
+    try:
+        print(convert(latex))
+    except UnsupportedLatexError as e:
+        print(f"error: {e}", file=sys.stderr)
+        return 1
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

asciimath2-0.1.0/src/asciimath2/config.py

@@ -0,0 +1,46 @@
+from dataclasses import dataclass, field
+
+
+@dataclass(frozen=True)
+class Config:
+    """Mirrors the three render-time knobs from
+    widcardw/asciimath-parser#22 (see SPEC.md section 2.2), plus two
+    extension points for macros this package doesn't know about out of the
+    box.
+
+    Only `single_newline_break` changes what this library *emits* (it
+    controls how many newlines separate rows of a multi-line expression).
+    `multiline_env` and `bar_as_mid` are downstream-rendering choices made
+    when the AsciiMath2 text is turned back into LaTeX; they're recorded
+    here purely so a caller can pass them through to whatever renders the
+    AsciiMath2 output, without needing a second place to look them up.
+
+    `custom_macros` and `custom_text` let a caller extend the fixed
+    `symbols.py` tables per-conversion without forking this package.
+    Neither ever *overrides* a macro this package already gives structural
+    meaning to (an ignored sizing command, an accent, a font, ...) - both
+    are consulted only where the built-in tables have nothing to say, so an
+    empty (default) value is always a strict no-op:
+
+    - `custom_macros`: bare, no-argument LaTeX macro name (no backslash) ->
+      AsciiMath2 token, e.g. `{"sqcup": "suu"}` for `\\sqcup`. Consulted only
+      for a macro name none of the built-in tables recognize.
+    - `custom_text`: exact `\\text{...}`/`\\operatorname{...}`/
+      `\\operatorname*{...}` argument content -> AsciiMath2 token, e.g.
+      `{"pmf": "pmf"}` to make `\\text{pmf}` emit the bare word `pmf` instead
+      of the default quoted `"pmf"` (still correct AsciiMath2, just not
+      round-trippable through a renderer that doesn't know `pmf` is a
+      symbol of its own). Checked before the default text/operatorname
+      handling, since that default has no structural meaning of its own to
+      protect.
+    """
+
+    multiline_env: str = "aligned"
+    single_newline_break: bool = False
+    bar_as_mid: bool = True
+    custom_macros: dict[str, str] = field(default_factory=dict)
+    custom_text: dict[str, str] = field(default_factory=dict)
+
+    @property
+    def row_break(self) -> str:
+        return "\n" if self.single_newline_break else "\n\n"

asciimath2-0.1.0/src/asciimath2/errors.py

@@ -0,0 +1,15 @@
+class UnsupportedLatexError(Exception):
+    """Raised when a LaTeX construct has no known AsciiMath2 rendering.
+
+    Carries the offending macro/environment name so callers can decide
+    whether to degrade the whole expression to raw LaTeX (the safe default)
+    or report it.
+    """
+
+    def __init__(self, what: str, detail: str = ""):
+        self.what = what
+        self.detail = detail
+        message = f"unsupported LaTeX construct: {what}"
+        if detail:
+            message += f" ({detail})"
+        super().__init__(message)

asciimath2-0.1.0/src/asciimath2/latex_context.py

@@ -0,0 +1,50 @@
+"""Extends pylatexenc's default macro database with argument specs for
+macros it doesn't already know the arity of, so their `{...}` groups are
+attached to the macro node (`nodeargd.argnlist`) instead of appearing as
+unrelated sibling groups in the node stream.
+"""
+
+from pylatexenc import macrospec
+from pylatexenc.latexwalker import LatexWalker, get_default_latex_context_db
+
+_EXTRA_MACROS = [
+    macrospec.MacroSpec("binom", "{{"),
+    macrospec.MacroSpec("dbinom", "{{"),
+    macrospec.MacroSpec("tbinom", "{{"),
+    macrospec.MacroSpec("overset", "{{"),
+    macrospec.MacroSpec("underset", "{{"),
+    macrospec.MacroSpec("stackrel", "{{"),
+    macrospec.MacroSpec("boldsymbol", "{"),
+    # `\operatorname*` is not a distinct macro name to pylatexenc's lexer -
+    # macro names never include punctuation, so the lexer always reads just
+    # "operatorname" and leaves any "*" as a separate following token. A
+    # macro that takes an optional star flag must say so in its own argspec
+    # (`*` here) instead of trying to register a second "operatorname*"
+    # macro name, which pylatexenc would never actually match (verified: it
+    # silently split `\operatorname*{X}` into a zero-arg `\operatorname`
+    # plus a stray literal "*" plus an unrelated bare `{X}` group before this
+    # fix). The star itself carries no AsciiMath2-visible distinction (same
+    # "display-style detail this package doesn't model" trade-off as
+    # \dfrac/\tfrac/\cfrac all collapsing to plain `/`), so render.py's
+    # `_render_operatorname` just ignores argnlist[0] and reads the mandatory
+    # group from argnlist[1].
+    macrospec.MacroSpec("operatorname", "*{"),
+    macrospec.MacroSpec("pmod", "{"),
+    macrospec.MacroSpec("bmod", ""),
+    macrospec.MacroSpec("cfrac", "{{"),
+    macrospec.MacroSpec("dfrac", "{{"),
+    macrospec.MacroSpec("tfrac", "{{"),
+    macrospec.MacroSpec("color", "{{"),
+    macrospec.MacroSpec("textcolor", "{{"),
+    macrospec.MacroSpec("boxed", "{"),
+]
+
+
+def build_context_db() -> macrospec.LatexContextDb:
+    db = get_default_latex_context_db()
+    db.add_context_category("asciimath2-extra", prepend=True, macros=_EXTRA_MACROS)
+    return db
+
+
+def make_walker(latex: str) -> LatexWalker:
+    return LatexWalker(latex, latex_context=build_context_db())