import argparse import os import re from pptx import Presentation from lxml import etree # 名前空間の定義 NAMESPACES = { 'p': 'http://schemas.openxmlformats.org/presentationml/2006/main', 'a': 'http://schemas.openxmlformats.org/drawingml/2006/main', 'm': 'http://schemas.openxmlformats.org/officeDocument/2006/math', 'r': 'http://schemas.openxmlformats.org/officeDocument/2006/relationships', 'a14': 'http://schemas.microsoft.com/office/drawing/2010/main' } # (中略: MATH_UNICODE_MAP, NARY_TO_LATEX などの定数定義は元のまま) MATH_UNICODE_MAP = { '𝑷': 'P', '𝑽': 'V', '𝑹': 'R', '𝑻': 'T', '𝒂': 'a', '𝒃': 'b', '𝒄': 'c', '𝒅': '\\mathrm{d}', '𝒆': 'e', '𝒇': 'f', '𝒈': 'g', '𝒉': 'h', '𝒊': 'i', '𝒋': 'j', '𝒌': 'k', '𝒍': 'l', '𝒎': 'm', '𝒏': 'n', '𝒙': 'x', '𝒚': 'y', '𝒛': 'z', '𝟐': '2', '𝟏': '1', '𝟎': '0', '−': '-', '＋': '+', '÷': '/', '×': '*', '⋅': '\\cdot', '…': '...', '∞': '\\infty', '∑': '\\sum', '∏': '\\prod', '∫': '\\int', '∬': '\\iint', '∭': '\\iiint', '∮': '\\oint', 'α': '\\alpha', 'β': '\\beta', 'γ': '\\gamma', 'δ': '\\delta', 'ε': '\\epsilon', 'ζ': '\\zeta', 'η': '\\eta', 'θ': '\\theta', 'ι': '\\iota', 'κ': '\\kappa', 'λ': '\\lambda', 'μ': '\\mu', 'ν': '\\nu', 'ξ': '\\xi', 'π': '\\pi', 'ρ': '\\rho', 'σ': '\\sigma', 'τ': '\\tau', 'υ': '\\upsilon', 'φ': '\\phi', 'χ': '\\chi', 'ψ': '\\psi', 'ω': '\\omega', 'Γ': '\\Gamma', 'Δ': '\\Delta', 'Θ': '\\Theta', 'Λ': '\\Lambda', 'Ξ': '\\Xi', 'Π': '\\Pi', 'Σ': '\\Sigma', 'Φ': '\\Phi', 'Ψ': '\\Psi', 'Ω': '\\Omega', '′': "'", '°': '\\degree', '℃': '\\degree C' } NARY_TO_LATEX = { '∑': '\\sum', '∏': '\\prod', '∫': '\\int', '∬': '\\iint', '∭': '\\iiint', '∮': '\\oint', '⋀': '\\bigwedge', '⋁': '\\bigvee', '⋂': '\\bigcap', '⋃': '\\bigcup', } OPERATOR_CHARS = set(['∑', '∏', '∫', '∬', '∭', '∮']) pause = 0 def terminate(): if pause: input("\nPress ENTER to terminate\n") exit() def initialize(): parser = argparse.ArgumentParser(description="PowerPointファイルからテキスト、数式、図を抽出し、Markdownに出力します。") parser.add_argument("-i", "--input", required=True, help="入力するPowerPointファイル名") parser.add_argument("-o", "--output", required=True, help="出力するMarkdownファイル名") parser.add_argument("--xml", action="store_true", help="数式の元のOMML XMLを出力します。") parser.add_argument("--imagedir", default="images", help="画像ディレクトリ") parser.add_argument("--pause", type=int, default=0, help="終了時待機") args = parser.parse_args() return args def get_slide_title(slide): """スライドからタイトルを抽出します。""" if slide.shapes.title: return slide.shapes.title.text.strip() return "無題のスライド" def _safe_text_replace_math_unicode(text: str) -> str: if not text: return "" for u, ltx in MATH_UNICODE_MAP.items(): text = text.replace(u, ltx) return text def _find_first(element, candidates): for cand in candidates: found = element.find(cand, NAMESPACES) if found is not None: return found return None def _detect_nary_op_char(element): op_tag = element.find('m:naryPr/m:chr', NAMESPACES) if op_tag is not None: val = op_tag.get(f"{{{NAMESPACES['m']}}}val", "") if val: return val ts = element.xpath('.//m:t[not(ancestor::m:e) and not(ancestor::m:sub) and not(ancestor::m:sup)]', namespaces=NAMESPACES) for t in ts: s = t.text or "" for ch in s: if ch in OPERATOR_CHARS: return ch return "" def omml_to_latex(element): """(元のコードの omml_to_latex 関数をそのまま使用)""" tag = etree.QName(element).localname if tag in ('oMath', 'oMathPara'): return "".join(omml_to_latex(child) for child in element) elif tag == 'f': num = element.find('m:num', NAMESPACES); den = element.find('m:den', NAMESPACES) return f"\\frac{{{omml_to_latex(num)}}}{{{omml_to_latex(den)}}}" elif tag == 'rad': deg = element.find('m:deg', NAMESPACES); e = element.find('m:e', NAMESPACES) if deg is not None: return f"\\sqrt[{omml_to_latex(deg)}]{{{omml_to_latex(e)}}}" return f"\\sqrt{{{omml_to_latex(e)}}}" elif tag == 'sSup': e = element.find('m:e', NAMESPACES); sup = element.find('m:sup', NAMESPACES) return f"{omml_to_latex(e)}^{{{omml_to_latex(sup)}}}" elif tag == 'sSub': e = element.find('m:e', NAMESPACES); sub = element.find('m:sub', NAMESPACES) return f"{omml_to_latex(e)}_{{{omml_to_latex(sub)}}}" elif tag == 'sSubSup': e = element.find('m:e', NAMESPACES); sub = element.find('m:sub', NAMESPACES); sup = element.find('m:sup', NAMESPACES) return f"{omml_to_latex(e)}_{{{omml_to_latex(sub)}}}^{{{omml_to_latex(sup)}}}" elif tag == 'd': beg_chr = element.find('m:dPr/m:begChr', NAMESPACES); end_chr = element.find('m:dPr/m:endChr', NAMESPACES) beg = beg_chr.get(f"{{{NAMESPACES['m']}}}val") if beg_chr is not None else "(" end = end_chr.get(f"{{{NAMESPACES['m']}}}val") if end_chr is not None else ")" content = "".join(omml_to_latex(child) for child in element if etree.QName(child).localname != 'dPr') return f"{beg}{content}{end}" elif tag == 'r': return "".join(omml_to_latex(child) for child in element) elif tag == 't': return _safe_text_replace_math_unicode(element.text or "") elif tag == 'limLow': base = omml_to_latex(element.find('m:e', NAMESPACES)); low = omml_to_latex(element.find('m:lim', NAMESPACES)) return f"{base}_{{{low}}}" elif tag == 'limUpp': base = omml_to_latex(element.find('m:e', NAMESPACES)); upp = omml_to_latex(element.find('m:lim', NAMESPACES)) return f"{base}^{{{upp}}}" elif tag == 'int': lower = _find_first(element, ['m:sub', 'm:low']); upper = _find_first(element, ['m:sup', 'm:up']); e = element.find('m:e', NAMESPACES) l_ltx = omml_to_latex(lower) if lower is not None else ""; u_ltx = omml_to_latex(upper) if upper is not None else ""; e_ltx = omml_to_latex(e) if e is not None else "" return f"\\int_{{{l_ltx}}}^{{{u_ltx}}}{e_ltx}" if (l_ltx or u_ltx) else f"\\int {e_ltx}" elif tag == 'nary': op_char = _detect_nary_op_char(element); lower_tag = _find_first(element, ['m:sub', 'm:low']); upper_tag = _find_first(element, ['m:sup', 'm:up']); content_tag = element.find('m:e', NAMESPACES) l_ltx = omml_to_latex(lower_tag) if lower_tag is not None else ''; u_ltx = omml_to_latex(upper_tag) if upper_tag is not None else ''; c_ltx = omml_to_latex(content_tag) if content_tag is not None else '' op_ltx = NARY_TO_LATEX.get(op_char, _safe_text_replace_math_unicode(op_char)) if lower_tag is not None or upper_tag is not None: return f"{op_ltx}_{{{l_ltx}}}^{{{u_ltx}}}{c_ltx}" return f"{op_ltx} {c_ltx}".rstrip() return "".join(omml_to_latex(child) for child in element) def split_latex_blocks(s: str): if not s: return [] parts = [p.strip() for p in re.split(r'\\\\', s) if p.strip()] return parts if parts else [s.strip()] def extract_content_to_markdown(input_pptx, output_md, image_dir, include_xml=False): try: presentation = Presentation(input_pptx) except Exception as e: print(f"エラー: {e}") return os.makedirs(image_dir, exist_ok=True) markdown_output = "" for i, slide in enumerate(presentation.slides): title = get_slide_title(slide) markdown_output += f"# スライド {i + 1}　{title}\n\n" # スライドのXMLルートを取得 slide_xml = slide.part.blob root = etree.fromstring(slide_xml) # --- 1. テキストセクション (全テキスト・改行・箇条書きの完全復元) --- markdown_output += "## テキスト\n\n" # 段落 (a:p) を順番にすべて取得する paragraphs = root.xpath('//a:p', namespaces=NAMESPACES) has_text = False for p in paragraphs: para_text_parts = [] # 箇条書きレベルの取得 lvl_attr = p.xpath('./a:pPr/@lvl', namespaces=NAMESPACES) level = int(lvl_attr[0]) if lvl_attr else 0 indent = " " * level # 段落内の「直下の子要素」だけを順番に走査する（入れ子を深く見すぎない） for child in p.iterchildren(): local_name = etree.QName(child).localname # 1. 通常のテキストラン (a:r) if local_name == 'r': # 数式の中の a:r ではなく、段落直下の a:r だけを処理 for t in child.findall('.//a:t', NAMESPACES): if t.text: para_text_parts.append(t.text) # 2. 改行 (a:br) elif local_name == 'br': para_text_parts.append("\n" + indent + " ") # 3. 数式 (m:oMath, m:oMathPara) elif local_name in ('oMath', 'oMathPara'): latex = omml_to_latex(child) if latex.strip(): para_text_parts.append(f" ${latex}$ ") full_para_text = "".join(para_text_parts).strip() # ... (以下、Markdownへの書き出し処理) if full_para_text: markdown_output += f"{indent}* {full_para_text}\n" has_text = True if not has_text: markdown_output += "（テキストなし）\n\n" else: markdown_output += "\n" # --- 2. 数式セクション (元のXPathロジックをそのまま使用) --- math_elements = root.xpath( '//m:oMathPara | //m:oMath[not(ancestor::m:oMathPara)]', namespaces=NAMESPACES ) if math_elements: markdown_output += "## 数式\n\n" seen_omml = set() for math_elem in math_elements: omml_string = etree.tostring(math_elem, encoding='unicode') if omml_string in seen_omml: continue seen_omml.add(omml_string) latex_code = omml_to_latex(math_elem) for one_line in split_latex_blocks(latex_code): one_line = one_line.replace('\\mathrm{d}', '\\,\\mathrm{d}') markdown_output += f"$$ {one_line} $$\n\n" # --- 3. 図セクション (元のXPathロジックをそのまま使用) --- image_elements = root.xpath('//a:blip[@r:embed]', namespaces=NAMESPACES) if image_elements: markdown_output += "## 図\n\n" for j, image_elem in enumerate(image_elements): r_id = image_elem.get('{http://schemas.openxmlformats.org/officeDocument/2006/relationships}embed') try: image_part = slide.part.rels[r_id].target_part ext = image_part.content_type.split('/')[-1].replace('x-', '') image_filename = f"slide{i+1}_image{j+1}.{ext}" image_path = os.path.join(image_dir, image_filename) with open(image_path, 'wb') as f: f.write(image_part.blob) markdown_output += f"![スライド{i+1}の図{j+1}]({image_dir}/{image_filename})\n\n" except: continue markdown_output += "---\n\n" with open(output_md, "w", encoding="utf-8") as f: f.write(markdown_output) print(f"変換完了: {output_md}") def main(): args = initialize() global pause pause = args.pause if not os.path.exists(args.input): print("入力ファイルが見つかりません"); return extract_content_to_markdown(args.input, args.output, args.imagedir, include_xml=args.xml) if __name__ == "__main__": main() terminate()