| 1 | """Markdown to plain text conversion and notes slide XML generation.""" |
| 2 | |
| 3 | from __future__ import annotations |
| 4 | |
| 5 | import re |
| 6 | |
| 7 | from language_tags import normalize_language_tag |
| 8 | |
| 9 | from ..drawingml.utils import ( |
| 10 | detect_text_lang, |
| 11 | text_has_rtl_characters, |
| 12 | text_uses_rtl, |
| 13 | ) |
| 14 | |
| 15 | |
| 16 | def markdown_to_plain_text(md_content: str) -> str: |
| 17 | """Convert Markdown notes to plain text for PPTX notes. |
| 18 | |
| 19 | Args: |
| 20 | md_content: Markdown formatted notes content. |
| 21 | |
| 22 | Returns: |
| 23 | Plain text content. |
| 24 | """ |
| 25 | def strip_inline_bold(text: str) -> str: |
| 26 | text = re.sub(r'\*\*(.+?)\*\*', r'\1', text) |
| 27 | text = re.sub(r'__(.+?)__', r'\1', text) |
| 28 | return text |
| 29 | |
| 30 | lines: list[str] = [] |
| 31 | for line in md_content.split('\n'): |
| 32 | if line.startswith('#'): |
| 33 | text = re.sub(r'^#+\s*', '', line).strip() |
| 34 | text = strip_inline_bold(text) |
| 35 | if text: |
| 36 | lines.append(text) |
| 37 | lines.append('') |
| 38 | elif line.strip().startswith('- '): |
| 39 | item_text = line.strip()[2:] |
| 40 | item_text = strip_inline_bold(item_text) |
| 41 | lines.append('• ' + item_text) |
| 42 | elif line.strip(): |
| 43 | text = strip_inline_bold(line.strip()) |
| 44 | lines.append(text) |
| 45 | else: |
| 46 | lines.append('') |
| 47 | |
| 48 | # Merge consecutive empty lines |
| 49 | result: list[str] = [] |
| 50 | is_prev_empty = False |
| 51 | for line in lines: |
| 52 | if line == '': |
| 53 | if not is_prev_empty: |
| 54 | result.append(line) |
| 55 | is_prev_empty = True |
| 56 | else: |
| 57 | result.append(line) |
| 58 | is_prev_empty = False |
| 59 | |
| 60 | return '\n'.join(result).strip() |
| 61 | |
| 62 | |
| 63 | def create_notes_slide_xml( |
| 64 | slide_num: int, |
| 65 | notes_text: str, |
| 66 | primary_language: str | None = None, |
| 67 | ) -> str: |
| 68 | """Create notes slide XML. |
| 69 | |
| 70 | Args: |
| 71 | slide_num: Slide number. |
| 72 | notes_text: Notes text in plain text format. |
| 73 | primary_language: Canonical BCP-47 deck language, when available. |
| 74 | |
| 75 | Returns: |
| 76 | Notes slide XML string. |
| 77 | """ |
| 78 | primary_language = ( |
| 79 | normalize_language_tag(primary_language) |
| 80 | if primary_language is not None |
| 81 | else None |
| 82 | ) |
| 83 | default_language = primary_language or 'en-US' |
| 84 | notes_text = (notes_text |
| 85 | .replace('&', '&') |
| 86 | .replace('<', '<') |
| 87 | .replace('>', '>')) |
| 88 | |
| 89 | paragraphs: list[str] = [] |
| 90 | for para in notes_text.split('\n'): |
| 91 | if para.strip(): |
| 92 | lang = detect_text_lang(para, primary_language) |
| 93 | paragraph_rtl = ' rtl="1"' if text_uses_rtl( |
| 94 | para, |
| 95 | primary_language, |
| 96 | ) else '' |
| 97 | run_rtl = ( |
| 98 | '<a:rtl val="1"/>' |
| 99 | if text_has_rtl_characters(para) |
| 100 | else '' |
| 101 | ) |
| 102 | paragraphs.append(f'''<a:p> |
| 103 | <a:pPr{paragraph_rtl}/> |
| 104 | <a:r> |
| 105 | <a:rPr lang="{lang}" dirty="0">{run_rtl}</a:rPr> |
| 106 | <a:t>{para}</a:t> |
| 107 | </a:r> |
| 108 | </a:p>''') |
| 109 | else: |
| 110 | paragraph_rtl = ( |
| 111 | ' rtl="1"' |
| 112 | if primary_language and text_uses_rtl('', primary_language) |
| 113 | else '' |
| 114 | ) |
| 115 | paragraphs.append( |
| 116 | f'<a:p><a:pPr{paragraph_rtl}/>' |
| 117 | f'<a:endParaRPr lang="{default_language}" dirty="0"/></a:p>' |
| 118 | ) |
| 119 | |
| 120 | paragraphs_xml = ( |
| 121 | '\n '.join(paragraphs) |
| 122 | if paragraphs |
| 123 | else ( |
| 124 | f'<a:p><a:endParaRPr lang="{default_language}" dirty="0"/></a:p>' |
| 125 | ) |
| 126 | ) |
| 127 | |
| 128 | return f'''<?xml version="1.0" encoding="UTF-8" standalone="yes"?> |
| 129 | <p:notes xmlns:a="http://schemas.openxmlformats.org/drawingml/2006/main" |
| 130 | xmlns:r="http://schemas.openxmlformats.org/officeDocument/2006/relationships" |
| 131 | xmlns:p="http://schemas.openxmlformats.org/presentationml/2006/main"> |
| 132 | <p:cSld> |
| 133 | <p:spTree> |
| 134 | <p:nvGrpSpPr> |
| 135 | <p:cNvPr id="1" name=""/> |
| 136 | <p:cNvGrpSpPr/> |
| 137 | <p:nvPr/> |
| 138 | </p:nvGrpSpPr> |
| 139 | <p:grpSpPr> |
| 140 | <a:xfrm> |
| 141 | <a:off x="0" y="0"/> |
| 142 | <a:ext cx="0" cy="0"/> |
| 143 | <a:chOff x="0" y="0"/> |
| 144 | <a:chExt cx="0" cy="0"/> |
| 145 | </a:xfrm> |
| 146 | </p:grpSpPr> |
| 147 | <p:sp> |
| 148 | <p:nvSpPr> |
| 149 | <p:cNvPr id="2" name="Slide Image Placeholder 1"/> |
| 150 | <p:cNvSpPr> |
| 151 | <a:spLocks noGrp="1" noRot="1" noChangeAspect="1"/> |
| 152 | </p:cNvSpPr> |
| 153 | <p:nvPr> |
| 154 | <p:ph type="sldImg" idx="2"/> |
| 155 | </p:nvPr> |
| 156 | </p:nvSpPr> |
| 157 | <p:spPr/> |
| 158 | </p:sp> |
| 159 | <p:sp> |
| 160 | <p:nvSpPr> |
| 161 | <p:cNvPr id="3" name="Notes Placeholder 2"/> |
| 162 | <p:cNvSpPr> |
| 163 | <a:spLocks noGrp="1"/> |
| 164 | </p:cNvSpPr> |
| 165 | <p:nvPr> |
| 166 | <p:ph type="body" sz="quarter" idx="3"/> |
| 167 | </p:nvPr> |
| 168 | </p:nvSpPr> |
| 169 | <p:spPr/> |
| 170 | <p:txBody> |
| 171 | <a:bodyPr/> |
| 172 | <a:lstStyle/> |
| 173 | {paragraphs_xml} |
| 174 | </p:txBody> |
| 175 | </p:sp> |
| 176 | </p:spTree> |
| 177 | </p:cSld> |
| 178 | <p:clrMapOvr> |
| 179 | <a:masterClrMapping/> |
| 180 | </p:clrMapOvr> |
| 181 | </p:notes>''' |
| 182 | |
| 183 | |
| 184 | def create_notes_slide_rels_xml(slide_num: int) -> str: |
| 185 | """Create notes slide relationship file XML. |
| 186 | |
| 187 | Args: |
| 188 | slide_num: Slide number. |
| 189 | |
| 190 | Returns: |
| 191 | Relationship file XML string. |
| 192 | """ |
| 193 | return f'''<?xml version="1.0" encoding="UTF-8" standalone="yes"?> |
| 194 | <Relationships xmlns="http://schemas.openxmlformats.org/package/2006/relationships"> |
| 195 | <Relationship Id="rId1" Type="http://schemas.openxmlformats.org/officeDocument/2006/relationships/notesMaster" Target="../notesMasters/notesMaster1.xml"/> |
| 196 | <Relationship Id="rId2" Type="http://schemas.openxmlformats.org/officeDocument/2006/relationships/slide" Target="../slides/slide{slide_num}.xml"/> |
| 197 | </Relationships>''' |
| 198 | |
| 199 | |
| 200 | def create_notes_master_xml(primary_language: str | None = None) -> str: |
| 201 | """Create a minimal PowerPoint-compatible notes master XML.""" |
| 202 | language = ( |
| 203 | normalize_language_tag(primary_language) |
| 204 | if primary_language is not None |
| 205 | else 'en-US' |
| 206 | ) |
| 207 | paragraph_rtl = ' rtl="1"' if text_uses_rtl('', language) else '' |
| 208 | return f'''<?xml version="1.0" encoding="UTF-8" standalone="yes"?> |
| 209 | <p:notesMaster xmlns:a="http://schemas.openxmlformats.org/drawingml/2006/main" |
| 210 | xmlns:r="http://schemas.openxmlformats.org/officeDocument/2006/relationships" |
| 211 | xmlns:p="http://schemas.openxmlformats.org/presentationml/2006/main"> |
| 212 | <p:cSld> |
| 213 | <p:spTree> |
| 214 | <p:nvGrpSpPr> |
| 215 | <p:cNvPr id="1" name=""/> |
| 216 | <p:cNvGrpSpPr/> |
| 217 | <p:nvPr/> |
| 218 | </p:nvGrpSpPr> |
| 219 | <p:grpSpPr> |
| 220 | <a:xfrm> |
| 221 | <a:off x="0" y="0"/> |
| 222 | <a:ext cx="0" cy="0"/> |
| 223 | <a:chOff x="0" y="0"/> |
| 224 | <a:chExt cx="0" cy="0"/> |
| 225 | </a:xfrm> |
| 226 | </p:grpSpPr> |
| 227 | <p:sp> |
| 228 | <p:nvSpPr> |
| 229 | <p:cNvPr id="2" name="Header Placeholder 1"/> |
| 230 | <p:cNvSpPr><a:spLocks noGrp="1"/></p:cNvSpPr> |
| 231 | <p:nvPr><p:ph type="hdr" sz="quarter"/></p:nvPr> |
| 232 | </p:nvSpPr> |
| 233 | <p:spPr> |
| 234 | <a:xfrm> |
| 235 | <a:off x="0" y="0"/> |
| 236 | <a:ext cx="2971800" cy="457200"/> |
| 237 | </a:xfrm> |
| 238 | <a:prstGeom prst="rect"><a:avLst/></a:prstGeom> |
| 239 | </p:spPr> |
| 240 | <p:txBody><a:bodyPr/><a:lstStyle/><a:p><a:pPr{paragraph_rtl}/><a:endParaRPr lang="{language}"/></a:p></p:txBody> |
| 241 | </p:sp> |
| 242 | <p:sp> |
| 243 | <p:nvSpPr> |
| 244 | <p:cNvPr id="3" name="Date Placeholder 2"/> |
| 245 | <p:cNvSpPr><a:spLocks noGrp="1"/></p:cNvSpPr> |
| 246 | <p:nvPr><p:ph type="dt" sz="half" idx="1"/></p:nvPr> |
| 247 | </p:nvSpPr> |
| 248 | <p:spPr> |
| 249 | <a:xfrm> |
| 250 | <a:off x="3884613" y="0"/> |
| 251 | <a:ext cx="2971800" cy="457200"/> |
| 252 | </a:xfrm> |
| 253 | <a:prstGeom prst="rect"><a:avLst/></a:prstGeom> |
| 254 | </p:spPr> |
| 255 | <p:txBody><a:bodyPr/><a:lstStyle/><a:p><a:pPr{paragraph_rtl}/><a:endParaRPr lang="{language}"/></a:p></p:txBody> |
| 256 | </p:sp> |
| 257 | <p:sp> |
| 258 | <p:nvSpPr> |
| 259 | <p:cNvPr id="4" name="Slide Image Placeholder 3"/> |
| 260 | <p:cNvSpPr><a:spLocks noGrp="1" noRot="1" noChangeAspect="1"/></p:cNvSpPr> |
| 261 | <p:nvPr><p:ph type="sldImg" idx="2"/></p:nvPr> |
| 262 | </p:nvSpPr> |
| 263 | <p:spPr> |
| 264 | <a:xfrm> |
| 265 | <a:off x="1143000" y="685800"/> |
| 266 | <a:ext cx="4572000" cy="3429000"/> |
| 267 | </a:xfrm> |
| 268 | <a:prstGeom prst="rect"><a:avLst/></a:prstGeom> |
| 269 | </p:spPr> |
| 270 | </p:sp> |
| 271 | <p:sp> |
| 272 | <p:nvSpPr> |
| 273 | <p:cNvPr id="5" name="Notes Placeholder 4"/> |
| 274 | <p:cNvSpPr><a:spLocks noGrp="1"/></p:cNvSpPr> |
| 275 | <p:nvPr><p:ph type="body" sz="quarter" idx="3"/></p:nvPr> |
| 276 | </p:nvSpPr> |
| 277 | <p:spPr> |
| 278 | <a:xfrm> |
| 279 | <a:off x="685800" y="4343400"/> |
| 280 | <a:ext cx="5486400" cy="4114800"/> |
| 281 | </a:xfrm> |
| 282 | <a:prstGeom prst="rect"><a:avLst/></a:prstGeom> |
| 283 | </p:spPr> |
| 284 | <p:txBody><a:bodyPr/><a:lstStyle/><a:p><a:pPr{paragraph_rtl}/><a:endParaRPr lang="{language}"/></a:p></p:txBody> |
| 285 | </p:sp> |
| 286 | <p:sp> |
| 287 | <p:nvSpPr> |
| 288 | <p:cNvPr id="6" name="Footer Placeholder 5"/> |
| 289 | <p:cNvSpPr><a:spLocks noGrp="1"/></p:cNvSpPr> |
| 290 | <p:nvPr><p:ph type="ftr" sz="quarter" idx="4"/></p:nvPr> |
| 291 | </p:nvSpPr> |
| 292 | <p:spPr> |
| 293 | <a:xfrm> |
| 294 | <a:off x="0" y="8685213"/> |
| 295 | <a:ext cx="2971800" cy="457200"/> |
| 296 | </a:xfrm> |
| 297 | <a:prstGeom prst="rect"><a:avLst/></a:prstGeom> |
| 298 | </p:spPr> |
| 299 | <p:txBody><a:bodyPr/><a:lstStyle/><a:p><a:pPr{paragraph_rtl}/><a:endParaRPr lang="{language}"/></a:p></p:txBody> |
| 300 | </p:sp> |
| 301 | <p:sp> |
| 302 | <p:nvSpPr> |
| 303 | <p:cNvPr id="7" name="Slide Number Placeholder 6"/> |
| 304 | <p:cNvSpPr><a:spLocks noGrp="1"/></p:cNvSpPr> |
| 305 | <p:nvPr><p:ph type="sldNum" sz="quarter" idx="5"/></p:nvPr> |
| 306 | </p:nvSpPr> |
| 307 | <p:spPr> |
| 308 | <a:xfrm> |
| 309 | <a:off x="3884613" y="8685213"/> |
| 310 | <a:ext cx="2971800" cy="457200"/> |
| 311 | </a:xfrm> |
| 312 | <a:prstGeom prst="rect"><a:avLst/></a:prstGeom> |
| 313 | </p:spPr> |
| 314 | <p:txBody><a:bodyPr/><a:lstStyle/><a:p><a:pPr{paragraph_rtl}/><a:endParaRPr lang="{language}"/></a:p></p:txBody> |
| 315 | </p:sp> |
| 316 | </p:spTree> |
| 317 | </p:cSld> |
| 318 | <p:clrMap bg1="lt1" tx1="dk1" bg2="lt2" tx2="dk2" |
| 319 | accent1="accent1" accent2="accent2" accent3="accent3" |
| 320 | accent4="accent4" accent5="accent5" accent6="accent6" |
| 321 | hlink="hlink" folHlink="folHlink"/> |
| 322 | <p:hf/> |
| 323 | <p:notesStyle> |
| 324 | <a:lvl1pPr marL="0" algn="l"{paragraph_rtl}> |
| 325 | <a:defRPr sz="1200" lang="{language}"/> |
| 326 | </a:lvl1pPr> |
| 327 | </p:notesStyle> |
| 328 | </p:notesMaster>''' |
| 329 | |
| 330 | |
| 331 | def create_notes_master_rels_xml() -> str: |
| 332 | """Create notes master relationships XML.""" |
| 333 | return '''<?xml version="1.0" encoding="UTF-8" standalone="yes"?> |
| 334 | <Relationships xmlns="http://schemas.openxmlformats.org/package/2006/relationships"> |
| 335 | <Relationship Id="rId1" Type="http://schemas.openxmlformats.org/officeDocument/2006/relationships/theme" Target="../theme/theme2.xml"/> |
| 336 | </Relationships>''' |
| 337 |