mdhtml2docx 0.1.0__py3-none-any.whl

mdhtml2docx/hilite.py

@@ -0,0 +1,27 @@
+"""Optional syntax scopes for code blocks, via fastpylight (Jeremy's tree-sitter highlighter).
+When fastpylight is absent or the language unknown, callers fall back to plain runs, so the
+package keeps its lxml-only hard dependency. Colors live in the reference doc's Hl* character
+styles (see styles.theme_styles); this module only tokenizes."""
+
+# theme_colors is probed too: a pre-theme_colors fastpylight is treated as absent, not half-working
+try: from fastpylight import tokenize, theme_colors  # chkstyle: ignore
+except ImportError: tokenize = None
+
+__all__ = ['segments']
+
+def segments(code, lang):
+    """Split `code` into [(text, scope)] using fastpylight's byte-offset tokens, or None when
+    fastpylight or `lang` is unavailable (caller then emits plain runs)"""
+    if tokenize is None or not lang: return None
+    try: toks = tokenize(code, lang)
+    except ValueError: return None
+    b = code.encode()
+    out, pos = [], 0
+    def emit(s, e, scope=None):
+        if e > s: out.append((b[s:e].decode(), scope))
+    for s, e, scope in toks:
+        emit(pos, s)
+        emit(s, e, scope)
+        pos = e
+    emit(pos, len(b))
+    return out

mdhtml2docx/schemas/SOURCES.md

@@ -0,0 +1,7 @@
+# Schema provenance
+
+The `*.xsd` files except `xml.xsd` are the wordprocessing import closure (12 of 26 files) of the ECMA-376 Part 4 5th edition transitional schema set, from `OfficeOpenXML-XMLSchema-Transitional.zip` inside https://ecma-international.org/wp-content/uploads/ECMA-376-4_5th_edition_december_2016.zip (sml/pml and their closures omitted: spreadsheets and presentations are out of scope).
+
+`xml.xsd` is the W3C schema for the XML namespace, from https://www.w3.org/2001/xml.xsd.
+
+Local modification, needed because lxml cannot resolve a namespace-only import: in `wml.xsd` and `shared-math.xsd`, the `<xsd:import namespace="http://www.w3.org/XML/1998/namespace"/>` element gained `schemaLocation="xml.xsd"`. No other edits.