artlbv · June 16, 2026 08:12
diff --git a/pptx_to_twiki.py b/pptx_to_twiki.py
 #!/usr/bin/env python3
 """
 pptx_to_twiki.py
 ================
 Convert a CMS DP-note PowerPoint file into a TWiki source file and a folder
 of individual plot PNG images, following the style of CMS public DP note pages
 (e.g. https://twiki.cern.ch/twiki/bin/view/CMSPublic/DP2026001).

 Usage
 -----
    python pptx_to_twiki.py input.pptx [options]

 Options
 -------
    --out-dir DIR       Directory to write outputs (default: <pptx_stem>_twiki/)
    --dp-number STR     DP note number, e.g. 2026/042  (default: 2026/XXX)
    --cds-record STR    CDS record number or full URL  (default: XXXXXXX)
    --title STR         Override document title (default: taken from slide 1)
    --figures-section N Place the Figures section as section N (default: auto,
                        = last text section + 1)
    --no-latex          Skip automatic physics-notation LaTeX substitutions

 Requirements
 ------------
    pip install python-pptx Pillow
 """

 import argparse
 import os
 import re
 import shutil
 import sys
 import textwrap
 from pathlib import Path

 try:
    from pptx import Presentation
    from pptx.enum.shapes import MSO_SHAPE_TYPE
 except ImportError:
    sys.exit("ERROR: python-pptx is required.  Run: pip install python-pptx")


 # ---------------------------------------------------------------------------
 # Physics-notation LaTeX substitutions
 # Applied to all text before writing to the TWiki source.
 # Order matters: longer patterns before shorter ones.
 # ---------------------------------------------------------------------------
 LATEX_SUBS = [
    # Arrows and Greek in decay chains
    (r'→',            r'\\rightarrow '),
    (r'->',           r'\\rightarrow '),
    (r'τh',           r'\\tau_h'),
    (r'τµ',           r'\\tau_\\mu'),
    (r'τ',            r'\\tau'),
    (r'µ',            r'\\mu'),
    (r'η',            r'\\eta'),
    # Superscripts / subscripts written as plain text
    (r'fb-1',         r'fb%$^{-1}$%'),
    (r'fb−1',         r'fb%$^{-1}$%'),
    # Kinematic variables
    (r'\bpT\b',       r'%$p_T$%'),
    (r'\bp_T\b',      r'%$p_T$%'),
    (r'\bHT\b',       r'%$H_T$%'),
    (r'\bH_T\b',      r'%$H_T$%'),
    (r'\bmHH\b',      r'%$m_{HH}$%'),
    (r'\bm_HH\b',     r'%$m_{HH}$%'),
    (r'\bm_{HH}\b',   r'%$m_{HH}$%'),
    (r'\bsqrt\(s\)',  r'%$\\sqrt{s}$%'),
    (r'√s',           r'%$\\sqrt{s}$%'),
    # tt-bar
    (r'tt̄',           r'%$t\\bar{t}$%'),
    (r'tt-bar',       r'%$t\\bar{t}$%'),
    # Delta
    (r'\bΔR\b',       r'%$\\Delta R$%'),
    (r'\bΔη\b',       r'%$\\Delta\\eta$%'),
    # Wrap bare numbers with units that follow GeV
    (r'(\d+)\s*GeV',  r'%$\1$% GeV'),
    (r'(\d+)\s*TeV',  r'%$\1$% TeV'),
    # Subscript formatting for L1HT
    (r'\bL1HT\b',     r'L1%$H_T$%'),
 ]

 # Slide titles that trigger special handling instead of going into the
 # generic Figures section or text sections.
 REFERENCES_TITLES = {'references', 'bibliography'}
 GLOSSARY_TITLES   = {'glossary', 'abbreviations'}
 SUMMARY_TITLES    = {'summary', 'conclusions', 'conclusion'}
 SKIP_TITLES       = {'title', ''}   # slide 1 (title slide) is skipped


 # ---------------------------------------------------------------------------
 # Helpers
 # ---------------------------------------------------------------------------

 def slugify(text: str, max_len: int = 50) -> str:
    """Turn arbitrary text into a filesystem-safe slug."""
    text = text.strip()
    text = re.sub(r'[^\w\s\-]', '', text)
    text = re.sub(r'\s+', '_', text)
    text = re.sub(r'_+', '_', text)
    return text[:max_len].strip('_')


 def apply_latex(text: str, do_latex: bool) -> str:
    """Apply physics-notation LaTeX substitutions."""
    if not do_latex:
        return text
    for pattern, replacement in LATEX_SUBS:
        try:
            text = re.sub(pattern, replacement, text)
        except re.error:
            text = text.replace(pattern, replacement)
    return text


 def wrap_math(text: str) -> str:
    """Wrap bare \\frac{} etc that aren't already in %$...$% markers."""
    # Ensure any remaining bare \command{} blocks are wrapped
    text = re.sub(r'(?<!%\$)(\\[a-zA-Z]+\{[^}]*\})(?!\$%)', r'%$\1$%', text)
    return text


 def get_slide_title(slide) -> str:
    """Return the title text of a slide (first placeholder of type TITLE or
    the first text box whose text is short enough to be a title)."""
    for shape in slide.placeholders:
        if shape.placeholder_format.idx == 0:   # idx 0 = title
            return shape.text.strip()
    # Fallback: first short text shape
    for shape in slide.shapes:
        if hasattr(shape, 'text') and shape.text.strip():
            t = shape.text.strip().split('\n')[0]
            if len(t) < 120:
                return t
    return ''


 def get_slide_body_texts(slide) -> list[str]:
    """Return all non-title, non-page-number text blocks from a slide."""
    texts = []
    title = get_slide_title(slide)
    for shape in slide.shapes:
        if not hasattr(shape, 'text'):
            continue
        t = shape.text.strip()
        if not t or t == '‹#›' or t == title:
            continue
        # Skip tiny label-only shapes (likely legend labels etc.)
        if len(t) < 4:
            continue
        texts.append(t)
    return texts


 def get_visible_images(slide, slide_w: float, slide_h: float) -> list[dict]:
    """Return list of dicts for each image fully or partly on the slide canvas."""
    images = []
    for shape in slide.shapes:
        if shape.shape_type != 13:   # MSO_SHAPE_TYPE.PICTURE = 13
            continue
        try:
            rId  = shape._element.blipFill.blip.rEmbed
            rel  = slide.part.rels[rId]
            fname = os.path.basename(rel.target_ref)
            img   = shape.image
        except Exception:
            continue
        w = shape.width.inches
        h = shape.height.inches
        l = shape.left.inches
        t = shape.top.inches
        # Skip if entirely outside the slide canvas
        if l >= slide_w or t >= slide_h or l + w <= 0 or t + h <= 0:
            continue
        # Skip tiny images (logos, watermarks, labels) — heuristic: < 0.5" in
        # either dimension
        if w < 0.5 or h < 0.5:
            continue
        images.append({
            'fname': fname,
            'left':  l,
            'top':   t,
            'width': w,
            'height': h,
            'blob':  img.blob,
            'ext':   img.ext,   # 'png', 'jpeg', …
        })
    # Sort left-to-right
    images.sort(key=lambda x: x['left'])
    return images


 def figure_filename(slide_num: int, sub_idx: int, title_slug: str,
                    n_images: int) -> str:
    """Build a figure PNG filename."""
    prefix = f'Figure_{slide_num:03d}'
    if n_images > 1:
        suffix = chr(ord('a') + sub_idx)   # a, b, c, …
        return f'{prefix}-{suffix}_{title_slug}.png'
    return f'{prefix}_{title_slug}.png'


 def twiki_figure_block(png_name: str, caption: str) -> str:
    """Render the standard CMS TWiki figure table block."""
    return (
        '| Figure | Caption |||\n'
        f'| [[%ATTACHURLPATH%/{png_name}]'
        f'[<img alt="" src="%ATTACHURL%/{png_name}" height="300/" />]] '
        f'| <p> {caption} </br> '
        f'[[[%ATTACHURL%/{png_name}][Get png version]]] <br> </p> |||\n'
    )


 def format_references(text: str) -> str:
    """Convert a slide reference list into TWiki numbered list with links."""
    lines = []
    for line in text.split('\n'):
        line = line.strip()
        if not line:
            continue
        # Try to extract a URL and linkify it
        url_match = re.search(r'(https?://\S+)', line)
        if url_match:
            url = url_match.group(1).rstrip('.,;)')
            line = line.replace(url_match.group(1), f'[[{url}][{url}]]')
        lines.append(f'   1. {line}')
    return '\n'.join(lines)


 def format_glossary(text: str) -> str:
    """Convert a bullet-list glossary slide into TWiki bullet list."""
    lines = []
    for line in text.split('\n'):
        line = line.strip().lstrip('•*–-').strip()
        if not line:
            continue
        lines.append(f'   * {line}')
    return '\n'.join(lines)


 # ---------------------------------------------------------------------------
 # Main converter
 # ---------------------------------------------------------------------------

 def convert(pptx_path: str, out_dir: str, dp_number: str, cds_record: str,
            title_override: str | None, figures_section: int | None,
            do_latex: bool) -> None:

    prs = Presentation(pptx_path)
    slide_w = prs.slide_width.inches
    slide_h = prs.slide_height.inches

    out_path    = Path(out_dir)
    figs_path   = out_path / 'figures'
    out_path.mkdir(parents=True, exist_ok=True)
    figs_path.mkdir(parents=True, exist_ok=True)

    # ---- Pass 1: classify every slide ----
    classified = []   # list of dicts
    for i, slide in enumerate(prs.slides):
        slide_num = i + 1
        title     = get_slide_title(slide)
        tl        = title.lower().strip()
        bodies    = get_slide_body_texts(slide)
        images    = get_visible_images(slide, slide_w, slide_h)

        if slide_num == 1:
            kind = 'title'
        elif tl in REFERENCES_TITLES:
            kind = 'references'
        elif tl in GLOSSARY_TITLES:
            kind = 'glossary'
        elif tl in SUMMARY_TITLES:
            kind = 'summary'
        else:
            kind = 'figures' if images else 'text'

        classified.append({
            'num':    slide_num,
            'title':  title,
            'bodies': bodies,
            'images': images,
            'kind':   kind,
        })

    # ---- Determine document title ----
    doc_title = title_override or classified[0]['title'] if classified else 'CMS DP Note'

    # ---- Determine auto section numbering ----
    text_slides   = [s for s in classified if s['kind'] in ('text', 'summary')]
    figure_slides = [s for s in classified if s['kind'] == 'figures']

    n_text_sections = len(text_slides)
    if figures_section is None:
        fig_sec_num = n_text_sections + 1
    else:
        fig_sec_num = figures_section
    ref_sec_num  = fig_sec_num + 1
    glos_sec_num = fig_sec_num + 2

    # ---- Pass 2: extract PNG files ----
    # Map (slide_num, sub_idx) -> png filename
    fig_map = {}
    for s in figure_slides:
        n_imgs    = len(s['images'])
        title_slug = slugify(s['title'])
        for j, img in enumerate(s['images']):
            png_name = figure_filename(s['num'], j, title_slug, n_imgs)
            dest     = figs_path / png_name
            with open(dest, 'wb') as fh:
                fh.write(img['blob'])
            fig_map[(s['num'], j)] = png_name

    # ---- Pass 3: build TWiki source ----
    lines = []

    # Header
    lines += [
        '<noautolink />',
        '<div style="text-align: justify;">',
        f'---+  {doc_title}   (CMS-DP-{dp_number})',
        f'*Status:* Results are published in the DPS note '
        f'[[https://cds.cern.ch/record/{cds_record}][DP-{dp_number}]]',
    ]

    # Text sections
    sec_num = 0
    for s in classified:
        if s['kind'] not in ('text', 'summary'):
            continue
        sec_num += 1
        title_tw = apply_latex(s['title'], do_latex)
        lines += ['', f'---++ {sec_num}. {title_tw}']
        for body in s['bodies']:
            body_tw = apply_latex(body, do_latex)
            # Split long body into paragraphs at double newlines
            paras = re.split(r'\n{2,}', body_tw)
            for para in paras:
                para = para.strip()
                if para:
                    lines += [para, '']

    # Figures section
    lines += [
        '',
        f'---++ {fig_sec_num}. Figures',
        '<span style="font-size: small;">'
        '<span style="-webkit-border-horizontal-spacing: 2px; '
        '-webkit-border-vertical-spacing: 2px;"><br /></span></span>',
    ]
    for s in figure_slides:
        n_imgs = len(s['images'])
        # Build caption from body texts
        raw_caption = ' '.join(s['bodies'])
        caption     = apply_latex(raw_caption, do_latex)
        caption     = caption.replace('\n', ' ').strip()

        for j in range(n_imgs):
            png_name = fig_map.get((s['num'], j), '')
            if not png_name:
                continue
            lines.append('')
            lines.append(twiki_figure_block(png_name, caption))

    # References section
    ref_slides = [s for s in classified if s['kind'] == 'references']
    if ref_slides:
        lines += ['', f'---++ {ref_sec_num}. References']
        for s in ref_slides:
            raw = '\n'.join(s['bodies'])
            lines.append(format_references(raw))

    # Glossary section
    glos_slides = [s for s in classified if s['kind'] == 'glossary']
    if glos_slides:
        lines += ['', f'---++ {glos_sec_num}. Glossary', '']
        for s in glos_slides:
            raw = '\n'.join(s['bodies'])
            lines.append(format_glossary(raw))

    lines.append('')  # trailing newline

    # ---- Write TWiki file ----
    pptx_stem  = Path(pptx_path).stem
    twiki_file = out_path / f'{slugify(pptx_stem)}_twiki.txt'
    with open(twiki_file, 'w', encoding='utf-8') as fh:
        fh.write('\n'.join(lines))

    # ---- Summary ----
    n_figs = len(fig_map)
    print(f'Done.')
    print(f'  TWiki source : {twiki_file}')
    print(f'  Figures      : {figs_path}  ({n_figs} PNG files)')
    print()
    print('To publish:')
    print(f'  1. Create CMSPublic.DP{dp_number.replace("/", "")} and paste the TWiki source.')
    print(f'  2. Upload all {n_figs} PNG files from figures/ as attachments.')


 # ---------------------------------------------------------------------------
 # CLI entry point
 # ---------------------------------------------------------------------------

 def main():
    parser = argparse.ArgumentParser(
        description='Convert a CMS DP-note PPTX file to TWiki source + PNG figures.',
        formatter_class=argparse.RawDescriptionHelpFormatter,
        epilog=textwrap.dedent("""\
            Examples
            --------
            # Basic conversion
            python pptx_to_twiki.py "HLT TOPO DP Note 2026 published.pptx"

            # Specify DP number and CDS record
            python pptx_to_twiki.py note.pptx --dp-number 2026/042 --cds-record 2951848

            # Custom output directory
            python pptx_to_twiki.py note.pptx --out-dir ./my_twiki_output/
        """),
    )
    parser.add_argument('pptx', help='Path to the input .pptx file')
    parser.add_argument('--out-dir',         default=None,
                        help='Output directory (default: <stem>_twiki/)')
    parser.add_argument('--dp-number',       default='2026/XXX',
                        help='DP note number, e.g. 2026/042')
    parser.add_argument('--cds-record',      default='XXXXXXX',
                        help='CDS record number or full URL')
    parser.add_argument('--title',           default=None,
                        help='Override document title')
    parser.add_argument('--figures-section', default=None, type=int,
                        help='Force the Figures section to be section N')
    parser.add_argument('--no-latex',        action='store_true',
                        help='Disable automatic physics-notation LaTeX substitutions')

    args = parser.parse_args()

    if not os.path.isfile(args.pptx):
        sys.exit(f'ERROR: File not found: {args.pptx}')

    out_dir = args.out_dir or (Path(args.pptx).stem + '_twiki')

    convert(
        pptx_path       = args.pptx,
        out_dir         = out_dir,
        dp_number       = args.dp_number,
        cds_record      = args.cds_record,
        title_override  = args.title,
        figures_section = args.figures_section,
        do_latex        = not args.no_latex,
    )


 if __name__ == '__main__':
    main()
	#!/usr/bin/env python3
	"""
	pptx_to_twiki.py
	================
	Convert a CMS DP-note PowerPoint file into a TWiki source file and a folder
	of individual plot PNG images, following the style of CMS public DP note pages
	(e.g. https://twiki.cern.ch/twiki/bin/view/CMSPublic/DP2026001).

	Usage
	-----
	python pptx_to_twiki.py input.pptx [options]

	Options
	-------
	--out-dir DIR Directory to write outputs (default: <pptx_stem>_twiki/)
	--dp-number STR DP note number, e.g. 2026/042 (default: 2026/XXX)
	--cds-record STR CDS record number or full URL (default: XXXXXXX)
	--title STR Override document title (default: taken from slide 1)
	--figures-section N Place the Figures section as section N (default: auto,
	= last text section + 1)
	--no-latex Skip automatic physics-notation LaTeX substitutions

	Requirements
	------------
	pip install python-pptx Pillow
	"""

	import argparse
	import os
	import re
	import shutil
	import sys
	import textwrap
	from pathlib import Path

	try:
	from pptx import Presentation
	from pptx.enum.shapes import MSO_SHAPE_TYPE
	except ImportError:
	sys.exit("ERROR: python-pptx is required. Run: pip install python-pptx")


	# ---------------------------------------------------------------------------
	# Physics-notation LaTeX substitutions
	# Applied to all text before writing to the TWiki source.
	# Order matters: longer patterns before shorter ones.
	# ---------------------------------------------------------------------------
	LATEX_SUBS = [
	# Arrows and Greek in decay chains
	(r'→', r'\\rightarrow '),
	(r'->', r'\\rightarrow '),
	(r'τh', r'\\tau_h'),
	(r'τµ', r'\\tau_\\mu'),
	(r'τ', r'\\tau'),
	(r'µ', r'\\mu'),
	(r'η', r'\\eta'),
	# Superscripts / subscripts written as plain text
	(r'fb-1', r'fb%$^{-1}$%'),
	(r'fb−1', r'fb%$^{-1}$%'),
	# Kinematic variables
	(r'\bpT\b', r'%$p_T$%'),
	(r'\bp_T\b', r'%$p_T$%'),
	(r'\bHT\b', r'%$H_T$%'),
	(r'\bH_T\b', r'%$H_T$%'),
	(r'\bmHH\b', r'%$m_{HH}$%'),
	(r'\bm_HH\b', r'%$m_{HH}$%'),
	(r'\bm_{HH}\b', r'%$m_{HH}$%'),
	(r'\bsqrt\(s\)', r'%$\\sqrt{s}$%'),
	(r'√s', r'%$\\sqrt{s}$%'),
	# tt-bar
	(r'tt̄', r'%$t\\bar{t}$%'),
	(r'tt-bar', r'%$t\\bar{t}$%'),
	# Delta
	(r'\bΔR\b', r'%$\\Delta R$%'),
	(r'\bΔη\b', r'%$\\Delta\\eta$%'),
	# Wrap bare numbers with units that follow GeV
	(r'(\d+)\s*GeV', r'%$\1$% GeV'),
	(r'(\d+)\s*TeV', r'%$\1$% TeV'),
	# Subscript formatting for L1HT
	(r'\bL1HT\b', r'L1%$H_T$%'),
	]

	# Slide titles that trigger special handling instead of going into the
	# generic Figures section or text sections.
	REFERENCES_TITLES = {'references', 'bibliography'}
	GLOSSARY_TITLES = {'glossary', 'abbreviations'}
	SUMMARY_TITLES = {'summary', 'conclusions', 'conclusion'}
	SKIP_TITLES = {'title', ''} # slide 1 (title slide) is skipped


	# ---------------------------------------------------------------------------
	# Helpers
	# ---------------------------------------------------------------------------

	def slugify(text: str, max_len: int = 50) -> str:
	"""Turn arbitrary text into a filesystem-safe slug."""
	text = text.strip()
	text = re.sub(r'[^\w\s\-]', '', text)
	text = re.sub(r'\s+', '_', text)
	text = re.sub(r'_+', '_', text)
	return text[:max_len].strip('_')


	def apply_latex(text: str, do_latex: bool) -> str:
	"""Apply physics-notation LaTeX substitutions."""
	if not do_latex:
	return text
	for pattern, replacement in LATEX_SUBS:
	try:
	text = re.sub(pattern, replacement, text)
	except re.error:
	text = text.replace(pattern, replacement)
	return text


	def wrap_math(text: str) -> str:
	"""Wrap bare \\frac{} etc that aren't already in %$...$% markers."""
	# Ensure any remaining bare \command{} blocks are wrapped
	text = re.sub(r'(?<!%\$)(\\[a-zA-Z]+\{[^}]*\})(?!\$%)', r'%$\1$%', text)
	return text


	def get_slide_title(slide) -> str:
	"""Return the title text of a slide (first placeholder of type TITLE or
	the first text box whose text is short enough to be a title)."""
	for shape in slide.placeholders:
	if shape.placeholder_format.idx == 0: # idx 0 = title
	return shape.text.strip()
	# Fallback: first short text shape
	for shape in slide.shapes:
	if hasattr(shape, 'text') and shape.text.strip():
	t = shape.text.strip().split('\n')[0]
	if len(t) < 120:
	return t
	return ''


	def get_slide_body_texts(slide) -> list[str]:
	"""Return all non-title, non-page-number text blocks from a slide."""
	texts = []
	title = get_slide_title(slide)
	for shape in slide.shapes:
	if not hasattr(shape, 'text'):
	continue
	t = shape.text.strip()
	if not t or t == '‹#›' or t == title:
	continue
	# Skip tiny label-only shapes (likely legend labels etc.)
	if len(t) < 4:
	continue
	texts.append(t)
	return texts


	def get_visible_images(slide, slide_w: float, slide_h: float) -> list[dict]:
	"""Return list of dicts for each image fully or partly on the slide canvas."""
	images = []
	for shape in slide.shapes:
	if shape.shape_type != 13: # MSO_SHAPE_TYPE.PICTURE = 13
	continue
	try:
	rId = shape._element.blipFill.blip.rEmbed
	rel = slide.part.rels[rId]
	fname = os.path.basename(rel.target_ref)
	img = shape.image
	except Exception:
	continue
	w = shape.width.inches
	h = shape.height.inches
	l = shape.left.inches
	t = shape.top.inches
	# Skip if entirely outside the slide canvas
	if l >= slide_w or t >= slide_h or l + w <= 0 or t + h <= 0:
	continue
	# Skip tiny images (logos, watermarks, labels) — heuristic: < 0.5" in
	# either dimension
	if w < 0.5 or h < 0.5:
	continue
	images.append({
	'fname': fname,
	'left': l,
	'top': t,
	'width': w,
	'height': h,
	'blob': img.blob,
	'ext': img.ext, # 'png', 'jpeg', …
	})
	# Sort left-to-right
	images.sort(key=lambda x: x['left'])
	return images


	def figure_filename(slide_num: int, sub_idx: int, title_slug: str,
	n_images: int) -> str:
	"""Build a figure PNG filename."""
	prefix = f'Figure_{slide_num:03d}'
	if n_images > 1:
	suffix = chr(ord('a') + sub_idx) # a, b, c, …
	return f'{prefix}-{suffix}_{title_slug}.png'
	return f'{prefix}_{title_slug}.png'


	def twiki_figure_block(png_name: str, caption: str) -> str:
	"""Render the standard CMS TWiki figure table block."""
	return (
	'\| Figure \| Caption \|\|\|\n'
	f'\| [[%ATTACHURLPATH%/{png_name}]'
	f'[<img alt="" src="%ATTACHURL%/{png_name}" height="300/" />]] '
	f'\| <p> {caption} </br> '
	f'[[[%ATTACHURL%/{png_name}][Get png version]]] <br> </p> \|\|\|\n'
	)


	def format_references(text: str) -> str:
	"""Convert a slide reference list into TWiki numbered list with links."""
	lines = []
	for line in text.split('\n'):
	line = line.strip()
	if not line:
	continue
	# Try to extract a URL and linkify it
	url_match = re.search(r'(https?://\S+)', line)
	if url_match:
	url = url_match.group(1).rstrip('.,;)')
	line = line.replace(url_match.group(1), f'[[{url}][{url}]]')
	lines.append(f' 1. {line}')
	return '\n'.join(lines)


	def format_glossary(text: str) -> str:
	"""Convert a bullet-list glossary slide into TWiki bullet list."""
	lines = []
	for line in text.split('\n'):
	line = line.strip().lstrip('•*–-').strip()
	if not line:
	continue
	lines.append(f' * {line}')
	return '\n'.join(lines)


	# ---------------------------------------------------------------------------
	# Main converter
	# ---------------------------------------------------------------------------

	def convert(pptx_path: str, out_dir: str, dp_number: str, cds_record: str,
	title_override: str \| None, figures_section: int \| None,
	do_latex: bool) -> None:

	prs = Presentation(pptx_path)
	slide_w = prs.slide_width.inches
	slide_h = prs.slide_height.inches

	out_path = Path(out_dir)
	figs_path = out_path / 'figures'
	out_path.mkdir(parents=True, exist_ok=True)
	figs_path.mkdir(parents=True, exist_ok=True)

	# ---- Pass 1: classify every slide ----
	classified = [] # list of dicts
	for i, slide in enumerate(prs.slides):
	slide_num = i + 1
	title = get_slide_title(slide)
	tl = title.lower().strip()
	bodies = get_slide_body_texts(slide)
	images = get_visible_images(slide, slide_w, slide_h)

	if slide_num == 1:
	kind = 'title'
	elif tl in REFERENCES_TITLES:
	kind = 'references'
	elif tl in GLOSSARY_TITLES:
	kind = 'glossary'
	elif tl in SUMMARY_TITLES:
	kind = 'summary'
	else:
	kind = 'figures' if images else 'text'

	classified.append({
	'num': slide_num,
	'title': title,
	'bodies': bodies,
	'images': images,
	'kind': kind,
	})

	# ---- Determine document title ----
	doc_title = title_override or classified[0]['title'] if classified else 'CMS DP Note'

	# ---- Determine auto section numbering ----
	text_slides = [s for s in classified if s['kind'] in ('text', 'summary')]
	figure_slides = [s for s in classified if s['kind'] == 'figures']

	n_text_sections = len(text_slides)
	if figures_section is None:
	fig_sec_num = n_text_sections + 1
	else:
	fig_sec_num = figures_section
	ref_sec_num = fig_sec_num + 1
	glos_sec_num = fig_sec_num + 2

	# ---- Pass 2: extract PNG files ----
	# Map (slide_num, sub_idx) -> png filename
	fig_map = {}
	for s in figure_slides:
	n_imgs = len(s['images'])
	title_slug = slugify(s['title'])
	for j, img in enumerate(s['images']):
	png_name = figure_filename(s['num'], j, title_slug, n_imgs)
	dest = figs_path / png_name
	with open(dest, 'wb') as fh:
	fh.write(img['blob'])
	fig_map[(s['num'], j)] = png_name

	# ---- Pass 3: build TWiki source ----
	lines = []

	# Header
	lines += [
	'<noautolink />',
	'<div style="text-align: justify;">',
	f'---+ {doc_title} (CMS-DP-{dp_number})',
	f'Status: Results are published in the DPS note '
	f'[[https://cds.cern.ch/record/{cds_record}][DP-{dp_number}]]',
	]

	# Text sections
	sec_num = 0
	for s in classified:
	if s['kind'] not in ('text', 'summary'):
	continue
	sec_num += 1
	title_tw = apply_latex(s['title'], do_latex)
	lines += ['', f'---++ {sec_num}. {title_tw}']
	for body in s['bodies']:
	body_tw = apply_latex(body, do_latex)
	# Split long body into paragraphs at double newlines
	paras = re.split(r'\n{2,}', body_tw)
	for para in paras:
	para = para.strip()
	if para:
	lines += [para, '']

	# Figures section
	lines += [
	'',
	f'---++ {fig_sec_num}. Figures',
	'<span style="font-size: small;">'
	'<span style="-webkit-border-horizontal-spacing: 2px; '
	'-webkit-border-vertical-spacing: 2px;"><br /></span></span>',
	]
	for s in figure_slides:
	n_imgs = len(s['images'])
	# Build caption from body texts
	raw_caption = ' '.join(s['bodies'])
	caption = apply_latex(raw_caption, do_latex)
	caption = caption.replace('\n', ' ').strip()

	for j in range(n_imgs):
	png_name = fig_map.get((s['num'], j), '')
	if not png_name:
	continue
	lines.append('')
	lines.append(twiki_figure_block(png_name, caption))

	# References section
	ref_slides = [s for s in classified if s['kind'] == 'references']
	if ref_slides:
	lines += ['', f'---++ {ref_sec_num}. References']
	for s in ref_slides:
	raw = '\n'.join(s['bodies'])
	lines.append(format_references(raw))

	# Glossary section
	glos_slides = [s for s in classified if s['kind'] == 'glossary']
	if glos_slides:
	lines += ['', f'---++ {glos_sec_num}. Glossary', '']
	for s in glos_slides:
	raw = '\n'.join(s['bodies'])
	lines.append(format_glossary(raw))

	lines.append('') # trailing newline

	# ---- Write TWiki file ----
	pptx_stem = Path(pptx_path).stem
	twiki_file = out_path / f'{slugify(pptx_stem)}_twiki.txt'
	with open(twiki_file, 'w', encoding='utf-8') as fh:
	fh.write('\n'.join(lines))

	# ---- Summary ----
	n_figs = len(fig_map)
	print(f'Done.')
	print(f' TWiki source : {twiki_file}')
	print(f' Figures : {figs_path} ({n_figs} PNG files)')
	print()
	print('To publish:')
	print(f' 1. Create CMSPublic.DP{dp_number.replace("/", "")} and paste the TWiki source.')
	print(f' 2. Upload all {n_figs} PNG files from figures/ as attachments.')


	# ---------------------------------------------------------------------------
	# CLI entry point
	# ---------------------------------------------------------------------------

	def main():
	parser = argparse.ArgumentParser(
	description='Convert a CMS DP-note PPTX file to TWiki source + PNG figures.',
	formatter_class=argparse.RawDescriptionHelpFormatter,
	epilog=textwrap.dedent("""\
	Examples
	--------
	# Basic conversion
	python pptx_to_twiki.py "HLT TOPO DP Note 2026 published.pptx"

	# Specify DP number and CDS record
	python pptx_to_twiki.py note.pptx --dp-number 2026/042 --cds-record 2951848

	# Custom output directory
	python pptx_to_twiki.py note.pptx --out-dir ./my_twiki_output/
	"""),
	)
	parser.add_argument('pptx', help='Path to the input .pptx file')
	parser.add_argument('--out-dir', default=None,
	help='Output directory (default: <stem>_twiki/)')
	parser.add_argument('--dp-number', default='2026/XXX',
	help='DP note number, e.g. 2026/042')
	parser.add_argument('--cds-record', default='XXXXXXX',
	help='CDS record number or full URL')
	parser.add_argument('--title', default=None,
	help='Override document title')
	parser.add_argument('--figures-section', default=None, type=int,
	help='Force the Figures section to be section N')
	parser.add_argument('--no-latex', action='store_true',
	help='Disable automatic physics-notation LaTeX substitutions')

	args = parser.parse_args()

	if not os.path.isfile(args.pptx):
	sys.exit(f'ERROR: File not found: {args.pptx}')

	out_dir = args.out_dir or (Path(args.pptx).stem + '_twiki')

	convert(
	pptx_path = args.pptx,
	out_dir = out_dir,
	dp_number = args.dp_number,
	cds_record = args.cds_record,
	title_override = args.title,
	figures_section = args.figures_section,
	do_latex = not args.no_latex,
	)


	if __name__ == '__main__':
	main()
No results found