regen.mde 0.2.2 → 0.7.0

package/src/build_corpus/equations.py

@@ -0,0 +1,80 @@
+"""LaTeX -> OMML (Office Math) conversion for Markdown -> Word export.
+
+Word renders equations from OMML (`<m:oMath>`), not from raw text. The previous
+exporter only set the font to "Cambria Math" and emitted the LaTeX source as
+literal characters, so anything with real commands (\\sum, \\Delta, \\rightarrow,
+\\leq) showed up as raw text. This module performs a genuine conversion:
+
+    LaTeX --(latex2mathml)--> presentation MathML --(mathml2omml)--> OMML
+
+and returns a parsed `<m:oMath>` / `<m:oMathPara>` element ready to splice into a
+python-docx paragraph. On any failure it returns ``None`` so callers can fall
+back to plain text rather than crash.
+"""
+
+from __future__ import annotations
+
+from docx.oxml import parse_xml
+from docx.oxml.ns import nsdecls
+
+_MATH_NS = "http://schemas.openxmlformats.org/officeDocument/2006/math"
+
+
+def _latex_to_omml_string(latex: str) -> str | None:
+    """Convert a LaTeX fragment to an OMML `<m:oMath>` XML string, or None."""
+    latex = latex.strip()
+    if not latex:
+        return None
+    try:
+        # Imported lazily so a missing optional dependency degrades to plain
+        # text instead of breaking the whole export.
+        import latex2mathml.converter as _l2m
+        from mathml2omml import convert as _mml2omml
+    except Exception:
+        return None
+    try:
+        mathml = _l2m.convert(latex)
+        omml = _mml2omml(mathml)
+    except Exception:
+        return None
+    if not omml or "oMath" not in omml:
+        return None
+    return omml
+
+
+def _parse_with_namespaces(omml: str):
+    """Parse an OMML fragment, injecting the math/word namespace decls."""
+    decls = nsdecls("m", "w")
+    for tag in ("<m:oMath>", "<m:oMathPara>"):
+        if omml.startswith(tag):
+            opener = tag[:-1] + f" {decls}>"
+            omml = opener + omml[len(tag):]
+            break
+    else:
+        return None
+    try:
+        return parse_xml(omml)
+    except Exception:
+        return None
+
+
+def latex_to_omath(latex: str):
+    """Return an inline `<m:oMath>` element for ``latex``, or None on failure."""
+    omml = _latex_to_omml_string(latex)
+    if omml is None:
+        return None
+    return _parse_with_namespaces(omml)
+
+
+def latex_to_omath_para(latex: str):
+    """Return a display `<m:oMathPara>` element for ``latex``, or None.
+
+    Wraps the inline `<m:oMath>` in `<m:oMathPara>` so Word treats it as a
+    centered display equation on its own line.
+    """
+    omml = _latex_to_omml_string(latex)
+    if omml is None:
+        return None
+    if omml.startswith("<m:oMath>"):
+        omml = "<m:oMathPara>" + omml + "</m:oMathPara>"
+    return _parse_with_namespaces(omml)