mdhtml2term 0.1.0__tar.gz → 0.1.1__tar.gz

{mdhtml2term-0.1.0 → mdhtml2term-0.1.1}/CHANGELOG.md

@@ -1,5 +1,10 @@
 <!-- do not remove -->
 
+## 0.1.1
+
+- API update
+
+
 ## 0.1.0
 
 ### New Features

{mdhtml2term-0.1.0/mdhtml2term.egg-info → mdhtml2term-0.1.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mdhtml2term
-Version: 0.1.0
+Version: 0.1.1
 Summary: Render MDHTML in the terminal with Rich
 Author-email: Jeremy Howard <github@jhoward.fastmail.fm>
 License-Expression: Apache-2.0
@@ -10,7 +10,7 @@ Classifier: Programming Language :: Python :: 3 :: Only
 Requires-Python: >=3.10
 Description-Content-Type: text/markdown
 License-File: LICENSE
-Requires-Dist: mdhtml>=0.1.39
+Requires-Dist: mdhtml>=0.1.43
 Requires-Dist: fastcore
 Requires-Dist: rich
 Provides-Extra: dev
@@ -42,7 +42,7 @@ console = Console()
 for block in md_blocks(markdown_text): console.print(block)
 ```
 
-`md_blocks` takes a Markdown string and returns a list of Rich renderables. `theme` names the Pygments theme for code blocks, and defaults to `ansi_dark`. Other keyword arguments go to `mdhtml.md2dom`.
+`md_blocks` takes a Markdown string and returns a list of Rich renderables. `theme` names the Pygments theme for code blocks and defaults to `ansi_dark`. Other keyword arguments go to `mdhtml.parse_md`. The shared Rust document engine resolves references and assigns numbers. Unresolved references show their target ids.
 
 ## Supported content
 
@@ -53,4 +53,4 @@ for block in md_blocks(markdown_text): console.print(block)
 - Task list checkboxes, as `[ ]` and `[x]`.
 - A numbered list that resumes after other content keeps its numbering.
 
-Other inline elements show their text. Other block containers render their children.
+Other inline elements show their text. Other block containers render their children. Raw payloads and unresolved template instructions are omitted.

{mdhtml2term-0.1.0 → mdhtml2term-0.1.1}/README.md

@@ -20,7 +20,7 @@ console = Console()
 for block in md_blocks(markdown_text): console.print(block)
 ```
 
-`md_blocks` takes a Markdown string and returns a list of Rich renderables. `theme` names the Pygments theme for code blocks, and defaults to `ansi_dark`. Other keyword arguments go to `mdhtml.md2dom`.
+`md_blocks` takes a Markdown string and returns a list of Rich renderables. `theme` names the Pygments theme for code blocks and defaults to `ansi_dark`. Other keyword arguments go to `mdhtml.parse_md`. The shared Rust document engine resolves references and assigns numbers. Unresolved references show their target ids.
 
 ## Supported content
 
@@ -31,4 +31,4 @@ for block in md_blocks(markdown_text): console.print(block)
 - Task list checkboxes, as `[ ]` and `[x]`.
 - A numbered list that resumes after other content keeps its numbering.
 
-Other inline elements show their text. Other block containers render their children.
+Other inline elements show their text. Other block containers render their children. Raw payloads and unresolved template instructions are omitted.

{mdhtml2term-0.1.0 → mdhtml2term-0.1.1}/mdhtml2term/__init__.py

@@ -1,3 +1,4 @@
-__version__ = "0.1.0"
+__version__ = "0.1.1"
+
 
 from .core import md_blocks

mdhtml2term-0.1.1/mdhtml2term/core.py

@@ -0,0 +1,111 @@
+"Render MDHTML Markdown as Rich renderables for the terminal, one per top-level block."
+import mdhtml
+from fastcore.meta import delegates
+from rich import box
+from rich.cells import cell_len
+from rich.console import Group
+from rich.segment import Segment
+from rich.style import Style
+from rich.syntax import Syntax
+from rich.table import Table
+from rich.text import Text
+
+__all__ = ['md_blocks']
+
+INLINE = dict(strong='bold', emph='italic', code='cyan', strike='strike', link='underline bright_blue', highlight='reverse')
+
+class Prefixed:
+    "Renders `rends` one after another, with `first` before the first line and `rest` before each later line"
+    def __init__(self, rends, first, rest=None, style=''):
+        self.rends,self.first,self.rest = rends,first,first if rest is None else rest
+        self.style = Style.parse(style) if style else None
+
+    def __rich_console__(self, console, options):
+        opts = options.update_width(max(1, options.max_width - cell_len(self.first)))
+        lines = [l for r in self.rends for l in console.render_lines(r, opts, pad=False)] or [[]]
+        for i,line in enumerate(lines):
+            yield Segment(self.first if i == 0 else self.rest, self.style)
+            yield from line
+            yield Segment.line()
+
+def _inline(nodes, text=None, style=''):
+    if text is None: text = Text()
+    for node in nodes:
+        kind = node['type']
+        sty = f"{style} {INLINE.get(kind, '')}".strip()
+        if kind in ('text', 'code'): text.append(node['text'], style=sty or None)
+        elif kind in ('soft_break', 'hard_break'): text.append('\n')
+        elif kind == 'image': text.append(f"[{node['alt'] or 'image'}]", style=f'{style} dim'.strip())
+        elif kind == 'math_inline': text.append(node['tex'], style=sty or None)
+        elif kind == 'footnote_ref': text.append(f"[{node['label']}]", style=sty or None)
+        else:
+            href = node['url'] if kind == 'link' else None
+            if href: sty = f'{sty} link {href}'
+            start = len(text)
+            _inline(node.get('children', []), text, sty)
+            if href and text.plain[start:] != href: text.append(f' ({href})', style='dim')
+    return text
+
+def _list(node, theme):
+    items = []
+    for i,item in enumerate(node['items'], node['start']):
+        bullet = f'{i}. ' if node['ordered'] else '• '
+        check = '' if item['checked'] is None else '[x] ' if item['checked'] else '[ ] '
+        items.append(Prefixed(_blocks(item['blocks'], theme), bullet + check, ' ' * cell_len(bullet), style='bold'))
+    return Group(*items)
+
+def _caption(node):
+    text = _inline(node['caption'])
+    if number := node['number']: text = Text(number['display'] + (': ' if text.plain else '')) + text
+    return text
+
+def _table(node, theme):
+    head = next((r for r in node['head'] if r['type'] == 'row'), None)
+    rows = [r for r in [*node['rows'], *node['foot']] if r['type'] == 'row']
+    first = head if head is not None else rows[0] if rows else None
+    table = Table(box=box.SIMPLE_HEAD, show_header=head is not None, show_edge=False, pad_edge=False)
+    if first is not None:
+        for cell in first['cells']:
+            table.add_column(Group(*_blocks(cell['content'], theme)) if head is not None else '', justify=cell['align'] or 'left')
+        for row in rows: table.add_row(*(Group(*_blocks(c['content'], theme)) for c in row['cells']))
+    return Group(_caption(node), table) if node['caption'] or node['number'] else table
+
+def _block(node, theme):
+    kind = node['type']
+    if kind == 'heading':
+        prefix = '#' * node['level'] + ' ' + (node['number']['display'] + ' ' if node['number'] else '')
+        return _inline(node['children'], Text(prefix, style='bold magenta'), 'bold magenta')
+    if kind == 'paragraph': return _inline(node['children'])
+    if kind == 'code_block': return Syntax(node['text'].rstrip('\n'), node['lang'] or 'text', theme=theme)
+    if kind == 'list': return _list(node, theme)
+    if kind == 'block_quote': return Prefixed(_blocks(node['children'], theme), '│ ', style='dim')
+    if kind == 'thematic_break': return Text('─' * 40, style='dim')
+    if kind == 'table': return _table(node, theme)
+    if kind == 'math_block':
+        text = node['tex'] + (f" ({node['number']['display']})" if node['number'] else '')
+        return Text(text)
+    if kind == 'figure': return Group(*_blocks(node['body'], theme), _caption(node))
+    if kind == 'panel':
+        title = _inline(node['title']['children'], style='bold') if node['title'] else Text(node['label'], style='bold')
+        return Group(title, *_blocks(node['body'], theme))
+    if kind == 'definition_list':
+        out = []
+        for item in node['items']:
+            out.extend(_inline(term['children'], style='bold') for term in item['terms'])
+            for definition in item['definitions']: out.extend(_blocks(definition, theme))
+        return Group(*out)
+    if kind in ('raw_block', 'script_block', 'operation', 'html_block'): return None
+    return Group(*_blocks(node['children'], theme))
+
+def _blocks(nodes, theme): return [r for node in nodes if (r := _block(node, theme)) is not None]
+
+@delegates(mdhtml.parse_md)
+def md_blocks(
+    md:str, # Markdown source
+    theme:str='ansi_dark', # Pygments theme for code blocks
+    **kwargs
+):
+    "One Rich renderable per top-level semantic block of `md`."
+    doc = mdhtml.parse_md(md, **kwargs).bake(refs='lenient', links=False)
+    notes = [Prefixed(_blocks(body, theme), f'[{label}] ', ' ' * (cell_len(label) + 3)) for label,body in doc.footnotes.items()]
+    return _blocks(doc.blocks, theme) + notes

{mdhtml2term-0.1.0 → mdhtml2term-0.1.1/mdhtml2term.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mdhtml2term
-Version: 0.1.0
+Version: 0.1.1
 Summary: Render MDHTML in the terminal with Rich
 Author-email: Jeremy Howard <github@jhoward.fastmail.fm>
 License-Expression: Apache-2.0
@@ -10,7 +10,7 @@ Classifier: Programming Language :: Python :: 3 :: Only
 Requires-Python: >=3.10
 Description-Content-Type: text/markdown
 License-File: LICENSE
-Requires-Dist: mdhtml>=0.1.39
+Requires-Dist: mdhtml>=0.1.43
 Requires-Dist: fastcore
 Requires-Dist: rich
 Provides-Extra: dev
@@ -42,7 +42,7 @@ console = Console()
 for block in md_blocks(markdown_text): console.print(block)
 ```
 
-`md_blocks` takes a Markdown string and returns a list of Rich renderables. `theme` names the Pygments theme for code blocks, and defaults to `ansi_dark`. Other keyword arguments go to `mdhtml.md2dom`.
+`md_blocks` takes a Markdown string and returns a list of Rich renderables. `theme` names the Pygments theme for code blocks and defaults to `ansi_dark`. Other keyword arguments go to `mdhtml.parse_md`. The shared Rust document engine resolves references and assigns numbers. Unresolved references show their target ids.
 
 ## Supported content
 
@@ -53,4 +53,4 @@ for block in md_blocks(markdown_text): console.print(block)
 - Task list checkboxes, as `[ ]` and `[x]`.
 - A numbered list that resumes after other content keeps its numbering.
 
-Other inline elements show their text. Other block containers render their children.
+Other inline elements show their text. Other block containers render their children. Raw payloads and unresolved template instructions are omitted.

{mdhtml2term-0.1.0 → mdhtml2term-0.1.1}/mdhtml2term.egg-info/requires.txt

@@ -1,4 +1,4 @@
-mdhtml>=0.1.39
+mdhtml>=0.1.43
 fastcore
 rich
 

{mdhtml2term-0.1.0 → mdhtml2term-0.1.1}/pyproject.toml

@@ -15,7 +15,7 @@ classifiers = [
   "Programming Language :: Python :: 3 :: Only",
 ]
 
-dependencies = ["mdhtml>=0.1.39", "fastcore", "rich"]
+dependencies = ["mdhtml>=0.1.43", "fastcore", "rich"]
 
 [project.optional-dependencies]
 dev = [

mdhtml2term-0.1.0/mdhtml2term/core.py

@@ -1,127 +0,0 @@
-"Render MDHTML Markdown as Rich renderables for the terminal, one per top-level block."
-import mdhtml
-from fastcore.meta import delegates
-from rich import box
-from rich.cells import cell_len
-from rich.console import Group
-from rich.segment import Segment
-from rich.style import Style
-from rich.syntax import Syntax
-from rich.table import Table
-from rich.text import Text
-
-__all__ = ['md_blocks']
-
-INLINE = {'strong': 'bold', 'em': 'italic', 'code': 'cyan', 'del': 'strike', 'a': 'underline bright_blue', 'mark': 'reverse'}
-HEADINGS = {'h1', 'h2', 'h3', 'h4', 'h5', 'h6'}
-BLOCKS = HEADINGS | {'p', 'pre', 'ul', 'ol', 'blockquote', 'hr', 'table', 'div', 'section', 'aside', 'figure', 'figcaption', 'details', 'summary', 'dl', 'dt', 'dd'}
-
-class Prefixed:
-    "Renders `rends` one after another, with `first` before the first line and `rest` before each later line"
-    def __init__(self, rends, first, rest=None, style=''):
-        self.rends,self.first,self.rest = rends,first,first if rest is None else rest
-        self.style = Style.parse(style) if style else None
-
-    def __rich_console__(self, console, options):
-        opts = options.update_width(max(1, options.max_width - cell_len(self.first)))
-        lines = [l for r in self.rends for l in console.render_lines(r, opts, pad=False)] or [[]]
-        for i,line in enumerate(lines):
-            yield Segment(self.first if i == 0 else self.rest, self.style)
-            yield from line
-            yield Segment.line()
-
-def _elem(c, t, style=''):
-    nm = getattr(c, 'name', None)
-    if nm == '#text': return t.append(c.text, style=style or None)
-    if nm == 'br': return t.append('\n')
-    a = c.attrs or {}
-    if nm == 'img': return t.append(f"[{a.get('alt') or 'image'}]", style=f'{style} dim'.strip())
-    if nm == 'input' and a.get('type') == 'checkbox': return t.append('[x]' if 'checked' in a else '[ ]', style=style or None)
-    sty = f"{style} {INLINE.get(nm, '')}".strip()
-    href = nm == 'a' and a.get('href')
-    if href: sty = f'{sty} link {href}'
-    n = len(t)
-    _inline(c, t, sty)
-    if href and t.plain[n:] != href: t.append(f' ({href})', style='dim')
-
-def _inline(node, t, style=''):
-    for c in node.children: _elem(c, t, style)
-    return t
-
-def _find(el, name):
-    for c in getattr(el, 'children', []):
-        if getattr(c, 'name', None) == name: yield c
-        else: yield from _find(c, name)
-
-def _code_block(pre, theme):
-    code = next(_find(pre, 'code'), None)
-    if code is None: return _inline(pre, Text())
-    src = ''.join(c.text for c in code.children if getattr(c, 'name', None) == '#text')
-    cls = (code.attrs or {}).get('class') or ''
-    lang = next((w.removeprefix('language-') for w in cls.split() if w.startswith('language-')), 'text')
-    return Syntax(src.rstrip('\n'), lang, theme=theme)
-
-def _list(el, theme):
-    ordered = el.name == 'ol'
-    n = int((el.attrs or {}).get('start', 1))
-    items = []
-    for li in el.children:
-        if getattr(li, 'name', None) != 'li': continue
-        bullet = f'{n}. ' if ordered else '• '
-        items.append(Prefixed(_blocks(li, theme), bullet, ' ' * cell_len(bullet), style='bold'))
-        n += 1
-    return Group(*items)
-
-def _cells(tr): return [c for c in tr.children if getattr(c, 'name', None) in ('td', 'th')]
-
-def _table(el):
-    head, rows = None, []
-    for sec in el.children:
-        nm = getattr(sec, 'name', None)
-        if nm == 'thead': head = next(_find(sec, 'tr'), None)
-        elif nm in ('tbody', 'tfoot'): rows += list(_find(sec, 'tr'))
-        elif nm == 'tr': rows.append(sec)
-    first = head if head is not None else rows[0] if rows else None
-    t = Table(box=box.SIMPLE_HEAD, show_header=head is not None, show_edge=False, pad_edge=False)
-    if first is None: return t
-    for c in _cells(first): t.add_column(_inline(c, Text()) if head is not None else '', justify=(c.attrs or {}).get('align') or 'left')
-    for tr in rows: t.add_row(*[_inline(c, Text()) for c in _cells(tr)])
-    return t
-
-def _block(el, theme):
-    nm = el.name
-    if nm in HEADINGS: return _inline(el, Text('#' * int(nm[1]) + ' ', style='bold magenta'), 'bold magenta')
-    if nm == 'p': return _inline(el, Text())
-    if nm == 'pre': return _code_block(el, theme)
-    if nm in ('ul', 'ol'): return _list(el, theme)
-    if nm == 'blockquote': return Prefixed(_blocks(el, theme), '│ ', style='dim')
-    if nm == 'hr': return Text('─' * 40, style='dim')
-    if nm == 'table': return _table(el)
-    return Group(*_blocks(el, theme))
-
-def _blocks(el, theme):
-    "One renderable per block child of `el`, plus one `Text` per run of inline content between them"
-    out, t = [], Text()
-    def flush():
-        nonlocal t
-        t.rstrip()
-        if t.plain: out.append(t)
-        t = Text()
-    for c in getattr(el, 'children', []):
-        nm = getattr(c, 'name', None)
-        if nm in BLOCKS:
-            flush()
-            out.append(_block(c, theme))
-        elif nm == '#text' and not t.plain: t.append(c.text.lstrip())
-        else: _elem(c, t)
-    flush()
-    return out
-
-@delegates(mdhtml.md2dom)
-def md_blocks(
-    md:str, # Markdown source
-    theme:str='ansi_dark', # Pygments theme for code blocks
-    **kwargs
-):
-    "One Rich renderable per top-level block of `md`"
-    return _blocks(mdhtml.md2dom(md, **kwargs), theme)