mdhtml2docx 0.1.0__tar.gz → 0.1.2__tar.gz

{mdhtml2docx-0.1.0 → mdhtml2docx-0.1.2}/CHANGELOG.md

@@ -1,5 +1,21 @@
 <!-- do not remove -->
 
+## 0.1.2
+
+### New Features
+
+- Support cross-references to text targets (spans, definition terms) via shared mdhtml Resolver; bookmark span and dt elements ([#10](https://github.com/AnswerDotAI/mdhtml2docx/issues/10))
+
+
+## 0.1.1
+
+### New Features
+
+- Move template marker rendering into converter as literal «body» runs; tmpl callable now takes token node dict ([#8](https://github.com/AnswerDotAI/mdhtml2docx/issues/8))
+- Move `jinja_literal` to mdhtml.jinja and import `mustache_kind` from mdhtml.mustache ([#7](https://github.com/AnswerDotAI/mdhtml2docx/issues/7))
+- Add details-div degradation to bold label and switch dash handling to explicit replacements callback ([#6](https://github.com/AnswerDotAI/mdhtml2docx/issues/6))
+
+
 ## 0.1.0
 
 ### New Features
@@ -8,4 +24,3 @@
 - Rename to mdhtml2docx and switch input from xhtmlmd XHTML to mdhtml JustHTML DOM with support for raw-data encodings, reference tokens/groups, HTML5 repair, inert templates, and figcaption alt text ([#2](https://github.com/AnswerDotAI/mdhtml2docx/issues/2))
 - Add xhtml2docx conversion engine, WML helpers, style management, validation, Word scripting, and syntax highlighting modules ([#1](https://github.com/AnswerDotAI/mdhtml2docx/issues/1))
 - Support MDHTML raw-data encodings, reference tokens and groups, HTML5 body repair, inert templates, and figure accessibility text from `figcaption`.
-

{mdhtml2docx-0.1.0/mdhtml2docx.egg-info → mdhtml2docx-0.1.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mdhtml2docx
-Version: 0.1.0
+Version: 0.1.2
 Summary: Convert MDHTML to Word docx files
 Author-email: Jeremy Howard <j@fast.ai>
 License-Expression: Apache-2.0
@@ -11,7 +11,7 @@ Requires-Python: >=3.10
 Description-Content-Type: text/markdown
 License-File: LICENSE
 Requires-Dist: lxml
-Requires-Dist: mdhtml>=0.1.15
+Requires-Dist: mdhtml>=0.1.32
 Requires-Dist: fast5ever>=0.1.1
 Provides-Extra: dev
 Requires-Dist: fastship; extra == "dev"
@@ -49,7 +49,7 @@ warnings = convert(document, 'out.docx')
 
 Body-position text and phrasing content become implicit Word paragraphs. Whitespace between blocks remains inert. HTML templates remain inert and are not rendered.
 
-Supported: headings, paragraphs, bold, italic, strikethrough, underline (the `u` element), highlight, superscript and subscript, inline code, links (external and internal, via bookmarks), images (local files are embedded with correct dimensions; remote URLs degrade to links), block quotes, fenced and indented code blocks, bullet and numbered lists (including nesting, `start`, and task-list checkboxes), pipe and grid tables (including row and column spans and header rows), definition lists, footnotes, horizontal rules, and abbreviations. Math spans and blocks become native Word math zones (`m:oMath`) holding the source text as-is, whatever the dialect (TeX, UnicodeMath, AsciiMath). Build-up/rendering is left downstream, and pandoc reads the zones back as math.
+Supported: headings, paragraphs, bold, italic, strikethrough, underline (the `u` element), highlight, superscript and subscript, inline code, links (external and internal, via bookmarks), images (local files are embedded with correct dimensions; remote URLs degrade to links), block quotes, fenced and indented code blocks, bullet and numbered lists (including nesting, `start`, and task-list checkboxes), pipe and grid tables (including row and column spans and header rows), definition lists, footnotes, horizontal rules, and abbreviations. A `details` div (the dialect's collapsible block) degrades to its label as a bold line above the body, never a numbered heading — Word has no folding. Math spans and blocks become native Word math zones (`m:oMath`) holding the source text as-is, whatever the dialect (TeX, UnicodeMath, AsciiMath). Build-up/rendering is left downstream, and pandoc reads the zones back as math.
 
 Code blocks with a language are syntax colored when [fastpylight](https://github.com/AnswerDotAI/fastpylight) is installed: tokens get `Hl*` character styles (e.g. `Hl Keyword`), whose colors come from the theme entry in the reference list below, so a finished document can be restyled in Word like everything else.
 

{mdhtml2docx-0.1.0 → mdhtml2docx-0.1.2}/README.md

@@ -28,7 +28,7 @@ warnings = convert(document, 'out.docx')
 
 Body-position text and phrasing content become implicit Word paragraphs. Whitespace between blocks remains inert. HTML templates remain inert and are not rendered.
 
-Supported: headings, paragraphs, bold, italic, strikethrough, underline (the `u` element), highlight, superscript and subscript, inline code, links (external and internal, via bookmarks), images (local files are embedded with correct dimensions; remote URLs degrade to links), block quotes, fenced and indented code blocks, bullet and numbered lists (including nesting, `start`, and task-list checkboxes), pipe and grid tables (including row and column spans and header rows), definition lists, footnotes, horizontal rules, and abbreviations. Math spans and blocks become native Word math zones (`m:oMath`) holding the source text as-is, whatever the dialect (TeX, UnicodeMath, AsciiMath). Build-up/rendering is left downstream, and pandoc reads the zones back as math.
+Supported: headings, paragraphs, bold, italic, strikethrough, underline (the `u` element), highlight, superscript and subscript, inline code, links (external and internal, via bookmarks), images (local files are embedded with correct dimensions; remote URLs degrade to links), block quotes, fenced and indented code blocks, bullet and numbered lists (including nesting, `start`, and task-list checkboxes), pipe and grid tables (including row and column spans and header rows), definition lists, footnotes, horizontal rules, and abbreviations. A `details` div (the dialect's collapsible block) degrades to its label as a bold line above the body, never a numbered heading — Word has no folding. Math spans and blocks become native Word math zones (`m:oMath`) holding the source text as-is, whatever the dialect (TeX, UnicodeMath, AsciiMath). Build-up/rendering is left downstream, and pandoc reads the zones back as math.
 
 Code blocks with a language are syntax colored when [fastpylight](https://github.com/AnswerDotAI/fastpylight) is installed: tokens get `Hl*` character styles (e.g. `Hl Keyword`), whose colors come from the theme entry in the reference list below, so a finished document can be restyled in Word like everything else.
 

{mdhtml2docx-0.1.0 → mdhtml2docx-0.1.2}/mdhtml2docx/__init__.py

@@ -1,3 +1,5 @@
-__version__ = "0.1.0"
+__version__ = "0.1.2"
+
+
 
 from .convert import convert

{mdhtml2docx-0.1.0 → mdhtml2docx-0.1.2}/mdhtml2docx/convert.py

@@ -10,14 +10,14 @@ from pathlib import Path
 from fast5ever import Comment, Element, Node, Text
 from lxml import etree
 from mdhtml import parse_mdhtml
-from mdhtml.export import REFTYPES, SCHEMES, decode_raw, group_plan, mustache_kind, ref_tokens, ref_variant
+from mdhtml.export import REFTYPES, SCHEMES, decode_raw, tmpl_node, group_plan, ref_tokens, ref_variant, target_kind, Resolver
 from .styles import STYLE_MAP, style_id, theme_styles
 from .styles import ref_path as _refpath
 from .wml import *
 from .wml import qn
 from .hilite import segments, tokenize
 
-__all__ = ['convert', 'mustache_fields', 'jinja_literal']
+__all__ = ['convert', 'mustache_fields']
 
 def _sid(key): return style_id(STYLE_MAP[key])
 
@@ -195,34 +195,30 @@ class Converter:
         return self._bknames[id]
 
     def ref_prefix(self, el, fmt, plural=False):
-        "Literal runs before a reference field: override text, the type prefix word, or nothing for bare and caption refs"
-        pre = el.to_text().strip()
-        if not pre:
-            tgt = (_get(el, 'href') or '#')[1:]
-            if 'bare' in ref_tokens(_get(el, 'data-ref')) or self.reftarget.get(tgt) == 'caption': return []
-            t = tgt.split('-')[0]
-            if t not in self.reftypes: raise ValueError(f'unknown reference type {t!r}; pass reftypes= to define its prefix')
-            pre = self.reftypes[t][plural]
-        return self.text_runs(pre + ' ', fmt)
+        "Literal runs before a reference field: the shared `Resolver.prefix` (override text, the type word, or nothing for bare, caption, and text refs)"
+        tgt = (_get(el, 'href') or '#')[1:]
+        pre = self.res.prefix(el.to_text().strip(), tgt, ref_tokens(_get(el, 'data-ref')), plural)
+        return self.text_runs(pre, fmt) if pre else []
 
     def ref_fld(self, el, fmt):
         """REF/PAGEREF field for a cross-reference `a`, with a cached placeholder Word replaces on update.
         Heading/paragraph targets number via `\\w`; caption targets return their bookmarked 'Label N' text
-        (or the number-only `_n` bookmark for bare/leaf/rel refs), so `\\w` never applies to them."""
+        (or the number-only `_n` bookmark for bare/leaf/rel refs), and text targets (spans, definition
+        terms) their bookmark text, so `\\w` never applies to either."""
         tgt = (_get(el, 'href') or '#')[1:]
         tokens = ref_tokens(_get(el, 'data-ref'))
-        if tgt not in self.reftarget:
-            raise ValueError(f'cross-reference target #{tgt} not found (targets are headings, paragraphs, figures, and tables with ids)')
-        kind = ref_variant(tokens)
+        self.res.check(tgt)
+        variant = ref_variant(tokens)
         nm, self.has_fields = self.bkname(tgt), True
-        if kind == 'page': instr, cached = rf' PAGEREF {nm} \h ', '#'
+        if variant == 'page': instr, cached = rf' PAGEREF {nm} \h ', '#'
         elif self.reftarget[tgt] == 'caption':
-            bare = 'bare' in tokens or kind in ('leaf', 'rel')
+            bare = 'bare' in tokens or variant in ('leaf', 'rel')
             instr, cached = (rf' REF {nm}_n \h ' if bare else rf' REF {nm} \h '), '#'
+        elif self.reftarget[tgt] == 'text': instr, cached = rf' REF {nm} \h ', self.res.core(tgt, tokens)
         else:
-            sw = self.REFSWITCH[kind]
+            sw = self.REFSWITCH[variant]
             instr = rf' REF {nm} {sw} \h ' if sw else rf' REF {nm} \h '
-            cached = self.idtext.get(tgt, '#') if kind == 'text' else '#'
+            cached = self.idtext.get(tgt, '#') if variant == 'text' else '#'
         return [E('w:fldSimple', {'w:instr': instr}, E('w:r', self.rpr(fmt), E('w:t', cached)))]
 
     def ref_group(self, el, fmt):
@@ -247,11 +243,11 @@ class Converter:
         return next((self.refstyles[c.lower()] for c in _classes(el) if c.lower() in self.refstyles), None)
 
     def span(self, el, fmt):
-        "Inline span: math -> inline m:oMath zone (linear source, dialect-agnostic), custom style -> rStyle, else transparent"
+        "Inline span: math -> inline m:oMath zone (linear source, dialect-agnostic), custom style -> rStyle, else transparent; an id bookmarks the runs"
         if _get(el, 'data-refs') is not None: return self.ref_group(el, fmt)
-        if 'math' in _classes(el): return [self.omath(el)]
-        if sid := self.custom_style(el, 'character'): return self.runs(el, fmt | {'rstyle': sid})
-        return self.runs(el, fmt)
+        if 'math' in _classes(el): return self.bookmark(el, [self.omath(el)])
+        if sid := self.custom_style(el, 'character'): return self.bookmark(el, self.runs(el, fmt | {'rstyle': sid}))
+        return self.bookmark(el, self.runs(el, fmt))
 
     def omath(self, el):
         "An m:oMath zone holding `el`'s text as linear-format math runs"
@@ -469,15 +465,20 @@ class Converter:
     def table(self, el):
         "w:tbl (+ caption paragraph before, spacer paragraph after)"
         cap = None
-        rows, nhead = [], 0
+        rows, nhead, markers = [], 0, {}
+        def _add(c):
+            "Rows in order; template markers recorded at the row index they precede"
+            if _tag(c) == 'template': markers.setdefault(len(rows), []).append(c)
+            else: rows.append(c)
         for sec in _els(el):
             t = _tag(sec)
             if t == 'caption': cap = sec
             elif t == 'thead':
-                rows += _els(sec)
+                for c in _els(sec): _add(c)
                 nhead = len(rows)
-            elif t in ('tbody', 'tfoot'): rows += _els(sec)
-            elif t == 'tr': rows.append(sec)
+            elif t in ('tbody', 'tfoot'):
+                for c in _els(sec): _add(c)
+            elif t in ('tr', 'template'): _add(sec)
         placed, ncols = self.table_grid(rows)
         dxa, all_fr = self.col_widths(el, ncols)
         def _tcw(ci, cs):
@@ -494,8 +495,13 @@ class Converter:
                 'w:firstColumn': 0, 'w:lastColumn': 0, 'w:noHBand': 0, 'w:noVBand': 1}))
         gw = dxa or [self.content_w // ncols] * ncols   # pandoc's docx reader drops tables whose gridCols lack w:w
         grid = E('w:tblGrid', *[E('w:gridCol', {'w:w': gw[i]}) for i in range(ncols)])
+        def _marker_tr(mel):
+            "A range marker between rows: one full-width literal cell, so forms keep their markers visible"
+            tcpr = E('w:tcPr', _tcw(0, ncols), E('w:gridSpan', {'w:val': ncols}) if ncols > 1 else None)
+            return E('w:tr', E('w:tc', tcpr, self.para(self.tmpl_runs(mel, {}, 'row'))))
         trs = []
         for ri, rowcells in enumerate(placed):
+            for mel in markers.get(ri, []): trs.append(_marker_tr(mel))
             tcs = []
             for item in rowcells:
                 if item[0] == 'cont':
@@ -512,6 +518,7 @@ class Converter:
                     if not len(body) or etree.QName(body[-1]).localname != 'p': body.append(E('w:p'))
                     tcs.append(E('w:tc', tcpr, *body))
             trs.append(E('w:tr', E('w:trPr', E('w:tblHeader')) if ri < nhead else None, *tcs))
+        for mel in markers.get(len(rows), []): trs.append(_marker_tr(mel))
         out = self.caption_para(el, 'tbl', cap)
         return out + [E('w:tbl', tblpr, grid, *trs), E('w:p')]
 
@@ -582,6 +589,9 @@ class Converter:
         if tag == 'div':
             cls = _classes(el)
             if 'math' in cls and 'display' in cls: return [E('w:p', E('m:oMathPara', self.omath(el)))]
+            if 'details' in cls and (kids := _els(el)) and _tag(kids[0]) in ('h1', 'h2', 'h3', 'h4', 'h5', 'h6'):
+                label = self.para(self.runs(kids[0], {'b': True}), style)   # the dialect's collapsible block: label as a bold line, never a numbered heading
+                return [label, *self.block_nodes(kids[1:], style, sid)]
             return self.blocks(el, style, self.custom_style(el, 'paragraph') or sid)
         if tag in BLOCK_TAGS and any(_tag(c) in BLOCK_TAGS for c in _els(el)):
             return self.blocks(el, style, sid)   # unknown container: recurse
@@ -594,16 +604,21 @@ class Converter:
     BIND_ID = '{8E2C9A44-7D31-4E5B-9C0D-1A6F2B3C4D5E}'   # fixed datastore id, so builds are reproducible
 
     def tmpl_runs(self, el, fmt, form):
-        """Template-token runs via the `tmpl` callable: str is a literal text run, ('field', instr) a live
-        field, ('control', name) an interactive plain-text content control, None dropped"""
+        """Template-token runs. Range markers and unknown tokens are converter policy: literal
+        `«body»` runs, so an unfilled form shows its markers. Var tokens go through the `tmpl`
+        callable with the token node dict (see `mdhtml.export.tmpl_node`): str is a literal text
+        run, ('field', instr) a live field, ('control', name) an interactive plain-text content
+        control, ('bound', name) a data-bound one, None dropped"""
+        node = tmpl_node(el, form)
+        if node["kind"] != "var": return self.text_runs(f'«{node["body"].strip()}»', fmt)
         if self.tmpl is None: return []
-        res = self.tmpl(el.to_text(), _get(el, 'data-template', ''), form)
+        res = self.tmpl(node)
         if res is None: return []
         if isinstance(res, str): return self.text_runs(res, fmt)
         kind, val = res
         if kind == 'field':
             self.has_fields = True
-            return [E('w:fldSimple', {'w:instr': f' {val.strip()} '}, E('w:r', self.rpr(fmt), E('w:t', f'«{el.to_text().strip()}»')))]
+            return [E('w:fldSimple', {'w:instr': f' {val.strip()} '}, E('w:r', self.rpr(fmt), E('w:t', f'«{node["name"]}»')))]
         if kind in ('control', 'bound'):
             self.has_controls = True
             sdtpr = E('w:sdtPr', E('w:alias', {'w:val': val}), E('w:tag', {'w:val': val}), E('w:showingPlcHdr'))
@@ -634,7 +649,7 @@ class Converter:
         out = []
         for c in _els(el):
             t = _tag(c)
-            if t == 'dt': out.append(self.para(self.runs(c, {}), 'dt'))
+            if t == 'dt': out.append(self.para(self.bookmark(c, self.runs(c, {})), 'dt'))
             elif t == 'dd':
                 blocky = any(_tag(k) in BLOCK_TAGS for k in _els(c))
                 out += self.blocks(c, 'dd') if blocky else [self.para(self.runs(c, {}), 'dd')]
@@ -832,18 +847,16 @@ class Converter:
         self.fndefs.update({_get(li, 'id'): li for sec in fn for li in _walk(sec) if _tag(li) == 'li' and _get(li, 'id')})
         return body
 
-    BOOKMARKABLE = {'p', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6'}
-
     def to_docx(self, mdhtml, dest):
         root = parse_frag(mdhtml)
         nodes = self.harvest_footnotes(root.children)
-        self.idtext, self.reftarget = {}, {}
+        self.idtext, self.reftarget, self.res = {}, {}, Resolver(self.reftypes)
         for e in (e for node in nodes if isinstance(node, Element) for e in _walk(node)):
             if not (i := _get(e, 'id')): continue
             self.idtext[i] = ' '.join(e.to_text().split())
-            t = _tag(e)
-            if t in ('figure', 'table'): self.reftarget[i] = 'caption'
-            elif t in self.BOOKMARKABLE: self.reftarget[i] = 'block'
+            k = target_kind(_tag(e))
+            if k: self.reftarget[i] = k
+            self.res.register(i, k, self.idtext[i])
         self.ids = set(self.idtext)
         blocks = self.block_nodes(nodes)
         docxml = self.document(blocks)
@@ -888,16 +901,10 @@ class Converter:
         root.insert(pos, E('w:updateFields', {'w:val': 'true'}))
         return etree.tostring(root, xml_declaration=True, encoding='UTF-8', standalone=True)
 
-def mustache_fields(body, syntax, form):
-    "Mustache variables as live Word `MERGEFIELD`s; section markers (`#`/`/`/`^` sigils) stay literal text"
-    if mustache_kind(body) == 'section': return '{{' + body + '}}'
-    return 'field', f'MERGEFIELD {body.strip()}'
-
+def mustache_fields(node):
+    "Template variables as live Word `MERGEFIELD`s (markers never reach `tmpl`: the converter shows them literally)"
+    return 'field', f'MERGEFIELD {node["name"]}'
 
-def jinja_literal(body, syntax, form):
-    "Jinja tokens re-spelled canonically (`{{ x }}`/`{% x %}`) as text, for docxtpl-style downstream pipelines"
-    o, c = ('{%', '%}') if syntax == 'jinja-stmt' else ('{{', '}}')
-    return f'{o} {body.strip()} {c}'
 
 
 def convert(mdhtml, dest, reference=None, base=None, reftypes=None, number_headings=None, tmpl=None):
@@ -911,9 +918,10 @@ def convert(mdhtml, dest, reference=None, base=None, reftypes=None, number_headi
     live REF fields; `reftypes` maps type tokens to (singular, plural) prefix words beyond the built-in
     `sec`, and `number_headings` (a styles.SCHEMES name such as 'legal', or a {lvlText: numFmt} dict, one entry per heading level)
     numbers the headings via a multilevel list so `\\w` fields resolve. Template tokens are dropped
-    unless `tmpl` is given: a callable `(body, syntax, form) -> str` for a literal text run,
+    unless `tmpl` is given: a callable taking the token node dict (`mdhtml.export.tmpl_node`: `body`,
+    `syntax`, `form`, `kind`, `name`, `inverted`) and returning a str for a literal text run,
     `('field', instr)` for a live field, `('control', name)` for an interactive plain-text content
     control, `('bound', name)` for a content control data-bound to a shared per-variable XML node
-    (same-name controls stay in sync as one is filled), or None to drop - `mustache_fields` and
-    `jinja_literal` are ready-made recipes."""
+    (same-name controls stay in sync as one is filled), or None to drop; range markers never reach `tmpl` and render as literal «body» runs - `mustache_fields` here
+    is the ready-made recipe."""
     return Converter(reference, base, reftypes, number_headings, tmpl).to_docx(mdhtml, dest)

{mdhtml2docx-0.1.0 → mdhtml2docx-0.1.2/mdhtml2docx.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mdhtml2docx
-Version: 0.1.0
+Version: 0.1.2
 Summary: Convert MDHTML to Word docx files
 Author-email: Jeremy Howard <j@fast.ai>
 License-Expression: Apache-2.0
@@ -11,7 +11,7 @@ Requires-Python: >=3.10
 Description-Content-Type: text/markdown
 License-File: LICENSE
 Requires-Dist: lxml
-Requires-Dist: mdhtml>=0.1.15
+Requires-Dist: mdhtml>=0.1.32
 Requires-Dist: fast5ever>=0.1.1
 Provides-Extra: dev
 Requires-Dist: fastship; extra == "dev"
@@ -49,7 +49,7 @@ warnings = convert(document, 'out.docx')
 
 Body-position text and phrasing content become implicit Word paragraphs. Whitespace between blocks remains inert. HTML templates remain inert and are not rendered.
 
-Supported: headings, paragraphs, bold, italic, strikethrough, underline (the `u` element), highlight, superscript and subscript, inline code, links (external and internal, via bookmarks), images (local files are embedded with correct dimensions; remote URLs degrade to links), block quotes, fenced and indented code blocks, bullet and numbered lists (including nesting, `start`, and task-list checkboxes), pipe and grid tables (including row and column spans and header rows), definition lists, footnotes, horizontal rules, and abbreviations. Math spans and blocks become native Word math zones (`m:oMath`) holding the source text as-is, whatever the dialect (TeX, UnicodeMath, AsciiMath). Build-up/rendering is left downstream, and pandoc reads the zones back as math.
+Supported: headings, paragraphs, bold, italic, strikethrough, underline (the `u` element), highlight, superscript and subscript, inline code, links (external and internal, via bookmarks), images (local files are embedded with correct dimensions; remote URLs degrade to links), block quotes, fenced and indented code blocks, bullet and numbered lists (including nesting, `start`, and task-list checkboxes), pipe and grid tables (including row and column spans and header rows), definition lists, footnotes, horizontal rules, and abbreviations. A `details` div (the dialect's collapsible block) degrades to its label as a bold line above the body, never a numbered heading — Word has no folding. Math spans and blocks become native Word math zones (`m:oMath`) holding the source text as-is, whatever the dialect (TeX, UnicodeMath, AsciiMath). Build-up/rendering is left downstream, and pandoc reads the zones back as math.
 
 Code blocks with a language are syntax colored when [fastpylight](https://github.com/AnswerDotAI/fastpylight) is installed: tokens get `Hl*` character styles (e.g. `Hl Keyword`), whose colors come from the theme entry in the reference list below, so a finished document can be restyled in Word like everything else.
 

{mdhtml2docx-0.1.0 → mdhtml2docx-0.1.2}/mdhtml2docx.egg-info/requires.txt

@@ -1,5 +1,5 @@
 lxml
-mdhtml>=0.1.15
+mdhtml>=0.1.32
 fast5ever>=0.1.1
 
 [dev]

{mdhtml2docx-0.1.0 → mdhtml2docx-0.1.2}/pyproject.toml

@@ -15,7 +15,7 @@ classifiers = [
   "Programming Language :: Python :: 3 :: Only",
 ]
 
-dependencies = ["lxml", "mdhtml>=0.1.15", "fast5ever>=0.1.1"]
+dependencies = ["lxml", "mdhtml>=0.1.32", "fast5ever>=0.1.1"]
 
 [project.optional-dependencies]
 dev = [
@@ -40,5 +40,6 @@ include = ["mdhtml2docx"]
 mdhtml2docx = ["schemas/*", "templates/*", "word.sdef"]
 
 [tool.pytest.ini_options]
-markers = ["slow: needs live GUI apps or long runtimes; deselected by default"]
-addopts = "-m 'not slow'"
+markers = ["slow: needs live GUI apps or long runtimes; deselected by default",
+    "checkout: needs the mdhtml source tree (examples/sample.md), not just the wheel; deselected by default"]
+addopts = "-m 'not slow and not checkout'"

{mdhtml2docx-0.1.0 → mdhtml2docx-0.1.2}/tests/test_convert.py

@@ -1,11 +1,13 @@
-import base64, subprocess, zipfile
+import base64, pytest, subprocess, zipfile
 from pathlib import Path
 
 from fastcore.test import test_eq as teq, test as tt, test_fail as tfail
 from fastcore.utils import in_
 
-from mdhtml import JINJA, MUSTACHE, mustache_kind, parse_mdhtml, sample_md, to_mdhtml
-from mdhtml2docx.convert import convert, jinja_literal, mustache_fields
+from mdhtml import DASHES, replacements, parse_mdhtml, to_mdhtml
+from mdhtml.mustache import MUSTACHE
+from mdhtml.tools import SAMPLE_MD, sample_md
+from mdhtml2docx.convert import convert, mustache_fields
 from mdhtml2docx.validate import fast_checks
 
 
@@ -29,6 +31,13 @@ def test_skeleton(tmp_path):
     assert doc.index('w:val="FirstParagraph"') < doc.index('w:val="BodyText"')
 
 
+def test_dash_replacements(tmp_path):
+    out = tmp_path/'t.docx'
+    convert(to_mdhtml('scores 3--5 -- nice... `a--b`', callbacks={'text': replacements(*DASHES)}), out)
+    doc = zipfile.ZipFile(out).read('word/document.xml').decode()
+    for s in ('3–5', ' – nice… ', 'a--b'): tt(s, doc, in_)
+
+
 def test_html5_fragment_and_mutable_dom_input(tmp_path):
     out = tmp_path/'t.docx'
     dom = parse_mdhtml('<p>Before <div>middle</div> after.</p><span zoop:33="x">tail</span>'
@@ -165,14 +174,14 @@ def test_math(tmp_path):
     tt('E = mc^2', doc, in_)
 
 
+@pytest.mark.checkout
 def test_sample(tmp_path):
     "The loop-closer: mdhtml's packaged sample -> docx (smart, legally numbered); every emitted style resolves"
     from lxml import etree
     from mdhtml2docx.styles import STYLE_MAP, style_id
     out = tmp_path/'sample.docx'
-    warns = convert(to_mdhtml(sample_md(), smart=True, auto_ids=True, implicit_figures=True), out, number_headings='legal')
-    teq(warns, ['remote image not embedded: https://dummyimage.com/96x48/eeeeee/333333.png&text=demo',
-        'remote image not embedded: https://dummyimage.com/96x48/eeeeee/333333.png&text=fig'])
+    warns = convert(to_mdhtml(sample_md(), callbacks={'text': replacements(*DASHES)}, implicit_figures=True), out, base=SAMPLE_MD.parent, number_headings='legal')
+    teq(warns, [])
     teq(fast_checks(out), 'valid')
     W = 'http://schemas.openxmlformats.org/wordprocessingml/2006/main'
     z = zipfile.ZipFile(out)
@@ -188,7 +197,7 @@ def test_sample(tmp_path):
     doc = z.read('word/document.xml').decode()
     for s in (r'REF sec_late \w \h', r'REF sec_payment \w \h', r'PAGEREF sec_late \h',
         r'REF fig_diagram \h', r'REF tbl_stages_n \h', ' SEQ Figure ', ' SEQ Table ',
-        '<w:br w:type="page"/>', 'Delivery stages', '“'): tt(s, doc, in_)
+        '<w:br w:type="page"/>', 'Delivery stages'): tt(s, doc, in_)
     tt('updateFields', z.read('word/settings.xml').decode(), in_)
     md = pandoc(out)
     for s in ('# mdhtml feature sample', '###### Week one', 'Temperature 1961-1990', '-89.2',
@@ -299,6 +308,21 @@ def test_xrefs(tmp_path):
     tfail(lambda: convert('<p><a href="#sec-pay" data-ref="page text"></a></p>', out), contains='conflicting data-ref')
 
 
+def test_text_targets(tmp_path):
+    out = tmp_path/'t.docx'
+    warns = convert(
+        '<h1 id="sec-d">Defs</h1>\n'
+        '<dl><dt id="def-term"><strong>"Term"</strong></dt><dt id="def-ap">Agreement Period</dt>'
+        '<dd>the deal period</dd></dl>\n'
+        '<p>The <span id="def-eff">Effective Date</span> is set.</p>\n'
+        '<p>The <a href="#def-term" data-ref=""></a> starts on the <a href="#def-eff" data-ref=""></a>; '
+        'see <a href="#def-ap" data-ref="page"></a>.</p>', out)
+    teq(warns, [])
+    teq(fast_checks(out), 'valid')
+    doc = zipfile.ZipFile(out).read('word/document.xml').decode()
+    for s in (r'REF def_term \h', r'REF def_eff \h', r'PAGEREF def_ap \h',
+        'w:name="def_term"', 'w:name="def_eff"', '"Term"', 'Effective Date'): tt(s, doc, in_)
+
 def test_xrefs_numbered_reference_doc(tmp_path):
     "A numbered docx (made by us) as reference: numbering merges with list numbering, no reinjection"
     ref = tmp_path/'ref.docx'
@@ -365,19 +389,15 @@ def test_caption_refs(tmp_path):
 def test_template_tokens(tmp_path):
     out = tmp_path/'t.docx'
     src = to_mdhtml('Pay {{sal}} to {{name}}.\n\n{{#opt}}\n\nGranted.\n\n{{/opt}}\n', templates=MUSTACHE)
-    convert(src, out)                                    # default: tokens dropped, as before
+    convert(src, out)                                    # default: vars dropped, markers literal
     doc = zipfile.ZipFile(out).read('word/document.xml').decode()
-    assert 'sal' not in doc and '{{' not in doc
+    assert 'sal' not in doc and '{{' not in doc and '«#opt»' in doc   # vars dropped; markers stay visible
     warns = convert(src, out, tmpl=mustache_fields)
     doc = zipfile.ZipFile(out).read('word/document.xml').decode()
     assert 'MERGEFIELD sal' in doc and 'MERGEFIELD name' in doc
-    assert '{{#opt}}' in doc and '{{/opt}}' in doc       # section markers stay literal, own paragraphs
+    assert '«#opt»' in doc and '«/opt»' in doc          # markers: converter-rendered literals, own paragraphs
     teq(warns, [])
     teq(fast_checks(out), 'valid')
-    convert(to_mdhtml('V {{ v }}.\n\n{% if x %}\n', templates=JINJA), out, tmpl=jinja_literal)
-    doc = zipfile.ZipFile(out).read('word/document.xml').decode()
-    assert '{{ v }}' in doc and '{% if x %}' in doc
-    teq(fast_checks(out), 'valid')
 
 def test_table_custom_style(tmp_path):
     out = tmp_path/'t.docx'
@@ -391,9 +411,7 @@ def test_table_custom_style(tmp_path):
 
 
 def test_template_controls(tmp_path):
-    def controls(body, syntax, form):
-        if mustache_kind(body) == 'section': return None
-        return 'control', body
+    def controls(node): return 'control', node['name']
     out = tmp_path/'t.docx'
     convert(to_mdhtml('Pay {{sal}} to {{name}}.\n\n{{#opt}}\n', templates=MUSTACHE), out, tmpl=controls)
     doc = zipfile.ZipFile(out).read('word/document.xml').decode()
@@ -401,14 +419,12 @@ def test_template_controls(tmp_path):
     assert doc.count('<w:showingPlcHdr/>') == 2 and 'w:val="PlaceholderText"' in doc
     styles = zipfile.ZipFile(out).read('word/styles.xml').decode()
     assert 'w:styleId="PlaceholderText"' in styles and 'w:val="808080"' in styles
-    assert '{{' not in doc                                   # section marker dropped by this callable
+    assert '«#opt»' in doc                                # marker: converter-rendered literal
     teq(fast_checks(out), 'valid')
 
 
 def test_template_bound(tmp_path):
-    def bound(body, syntax, form):
-        if mustache_kind(body) == 'section': return None
-        return 'bound', body
+    def bound(node): return 'bound', node['name']
     out = tmp_path/'t.docx'
     convert(to_mdhtml('Pay {{sal}} to {{sal}} and {{name}}.\n', templates=MUSTACHE), out, tmpl=bound)
     z = zipfile.ZipFile(out)
@@ -419,3 +435,13 @@ def test_template_bound(tmp_path):
     assert 'customXmlProperties' in z.read('[Content_Types].xml').decode()
     assert 'customXml' in z.read('word/_rels/document.xml.rels').decode()
     teq(fast_checks(out), 'valid')
+
+
+def test_details_degrades_to_bold_label(tmp_path):
+    out = tmp_path/'t.docx'
+    warns = convert(to_mdhtml('::: {.details .tool-usage-details}\n## the label\n\nbody text\n:::\n'), out)
+    assert not warns
+    md = pandoc(out)
+    tt('**the label**', md, in_)
+    tt('body text', md, in_)
+    assert '# the label' not in md   # label is a bold line, not a heading

mdhtml2docx-0.1.2/tests/test_word.py

@@ -0,0 +1,17 @@
+import pytest
+from pathlib import Path
+
+from mdhtml import DASHES, replacements, to_mdhtml
+from mdhtml.tools import SAMPLE_MD, sample_md
+from mdhtml2docx import convert
+
+DATA = Path(__file__).parent.parent/'_data'
+
+pytestmark = [pytest.mark.slow, pytest.mark.checkout]
+
+
+def test_sample():
+    docx = DATA/'sample.docx'
+    warns = convert(to_mdhtml(sample_md(), callbacks={'text': replacements(*DASHES)}, implicit_figures=True), docx, base=SAMPLE_MD.parent, number_headings='legal')
+    assert not warns
+    assert docx.stat().st_size

mdhtml2docx-0.1.0/tests/test_word.py

@@ -1,16 +0,0 @@
-import pytest
-from pathlib import Path
-
-from mdhtml import sample_md, to_mdhtml
-from mdhtml2docx import convert
-
-DATA = Path(__file__).parent.parent/'_data'
-
-pytestmark = pytest.mark.slow
-
-
-def test_sample():
-    docx = DATA/'sample.docx'
-    warns = convert(to_mdhtml(sample_md(), smart=True, auto_ids=True, implicit_figures=True), docx, number_headings='legal')
-    assert len(warns) == 2 and all('remote image not embedded' in w for w in warns)
-    assert docx.stat().st_size