|
| 1 | +// mathmlToLatex.js — Convert MathML DOM or string to LaTeX |
| 2 | + |
| 3 | +import { SYMBOLS, FUNCTIONS, LARGE_OPS } from './data.js'; |
| 4 | + |
| 5 | +// Build reverse lookup tables: Unicode char → LaTeX command |
| 6 | +const CHAR_TO_LATEX = {}; |
| 7 | +for (const [cmd, ch] of Object.entries(SYMBOLS)) { |
| 8 | + // Prefer shorter command names when duplicates exist |
| 9 | + if (!CHAR_TO_LATEX[ch] || cmd.length < CHAR_TO_LATEX[ch].length) { |
| 10 | + CHAR_TO_LATEX[ch] = cmd; |
| 11 | + } |
| 12 | +} |
| 13 | +for (const [cmd, ch] of Object.entries(LARGE_OPS)) { |
| 14 | + if (!CHAR_TO_LATEX[ch] || cmd.length < CHAR_TO_LATEX[ch].length) { |
| 15 | + CHAR_TO_LATEX[ch] = cmd; |
| 16 | + } |
| 17 | +} |
| 18 | + |
| 19 | +/** |
| 20 | + * Convert a MathML string or DOM element to LaTeX. |
| 21 | + * @param {string|Element} input — MathML markup string or a DOM <math> element |
| 22 | + * @returns {string} LaTeX string |
| 23 | + */ |
| 24 | +export function mathmlToLatex(input) { |
| 25 | + let root; |
| 26 | + if (typeof input === 'string') { |
| 27 | + const doc = new DOMParser().parseFromString(input, 'application/xml'); |
| 28 | + root = doc.documentElement; |
| 29 | + // Check for parse errors |
| 30 | + if (root.tagName === 'parsererror' || root.querySelector('parsererror')) { |
| 31 | + throw new Error('Invalid MathML: ' + root.textContent); |
| 32 | + } |
| 33 | + } else { |
| 34 | + root = input; |
| 35 | + } |
| 36 | + |
| 37 | + return convertNode(root).trim(); |
| 38 | +} |
| 39 | + |
| 40 | +function convertNode(node) { |
| 41 | + if (node.nodeType === 3) { // text node |
| 42 | + return node.textContent.trim(); |
| 43 | + } |
| 44 | + if (node.nodeType !== 1) return ''; // skip non-element nodes |
| 45 | + |
| 46 | + const tag = localName(node); |
| 47 | + |
| 48 | + switch (tag) { |
| 49 | + case 'math': |
| 50 | + return convertChildren(node); |
| 51 | + |
| 52 | + case 'semantics': { |
| 53 | + // If there's a LaTeX annotation, prefer it |
| 54 | + const annotation = node.querySelector('annotation[encoding="LaTeX"], annotation[encoding="application/x-tex"]'); |
| 55 | + if (annotation) return annotation.textContent; |
| 56 | + // Otherwise convert the presentation MathML (first child) |
| 57 | + return node.firstElementChild ? convertNode(node.firstElementChild) : ''; |
| 58 | + } |
| 59 | + |
| 60 | + case 'mrow': |
| 61 | + return convertChildren(node); |
| 62 | + |
| 63 | + case 'mn': |
| 64 | + return node.textContent.trim(); |
| 65 | + |
| 66 | + case 'mi': { |
| 67 | + const text = node.textContent.trim(); |
| 68 | + const variant = node.getAttribute('mathvariant'); |
| 69 | + // Function names rendered as upright |
| 70 | + if (variant === 'normal' && FUNCTIONS.has(text)) { |
| 71 | + return '\\' + text; |
| 72 | + } |
| 73 | + // Check for special symbols (Greek, etc.) |
| 74 | + if (text.length === 1 && CHAR_TO_LATEX[text]) { |
| 75 | + return CHAR_TO_LATEX[text]; |
| 76 | + } |
| 77 | + return text; |
| 78 | + } |
| 79 | + |
| 80 | + case 'mo': { |
| 81 | + const text = node.textContent.trim(); |
| 82 | + if (CHAR_TO_LATEX[text]) return CHAR_TO_LATEX[text]; |
| 83 | + // Stretchy delimiters handled by parent (mrow with fences) |
| 84 | + return text; |
| 85 | + } |
| 86 | + |
| 87 | + case 'mtext': |
| 88 | + return '\\text{' + node.textContent + '}'; |
| 89 | + |
| 90 | + case 'mspace': |
| 91 | + return ''; |
| 92 | + |
| 93 | + case 'mfrac': { |
| 94 | + const children = elements(node); |
| 95 | + if (children.length < 2) return '\\frac{}{}'; |
| 96 | + const num = convertNode(children[0]); |
| 97 | + const den = convertNode(children[1]); |
| 98 | + const linethickness = node.getAttribute('linethickness'); |
| 99 | + if (linethickness === '0') { |
| 100 | + return '\\binom{' + num + '}{' + den + '}'; |
| 101 | + } |
| 102 | + return '\\frac{' + num + '}{' + den + '}'; |
| 103 | + } |
| 104 | + |
| 105 | + case 'msqrt': { |
| 106 | + const body = convertChildren(node); |
| 107 | + return '\\sqrt{' + body + '}'; |
| 108 | + } |
| 109 | + |
| 110 | + case 'mroot': { |
| 111 | + const children = elements(node); |
| 112 | + if (children.length < 2) return '\\sqrt{}'; |
| 113 | + const body = convertNode(children[0]); |
| 114 | + const index = convertNode(children[1]); |
| 115 | + return '\\sqrt[' + index + ']{' + body + '}'; |
| 116 | + } |
| 117 | + |
| 118 | + case 'msup': { |
| 119 | + const children = elements(node); |
| 120 | + if (children.length < 2) return ''; |
| 121 | + const base = convertNode(children[0]); |
| 122 | + const sup = convertNode(children[1]); |
| 123 | + return wrapBase(base) + '^{' + sup + '}'; |
| 124 | + } |
| 125 | + |
| 126 | + case 'msub': { |
| 127 | + const children = elements(node); |
| 128 | + if (children.length < 2) return ''; |
| 129 | + const base = convertNode(children[0]); |
| 130 | + const sub = convertNode(children[1]); |
| 131 | + return wrapBase(base) + '_{' + sub + '}'; |
| 132 | + } |
| 133 | + |
| 134 | + case 'msubsup': { |
| 135 | + const children = elements(node); |
| 136 | + if (children.length < 3) return ''; |
| 137 | + const base = convertNode(children[0]); |
| 138 | + const sub = convertNode(children[1]); |
| 139 | + const sup = convertNode(children[2]); |
| 140 | + return wrapBase(base) + '_{' + sub + '}^{' + sup + '}'; |
| 141 | + } |
| 142 | + |
| 143 | + case 'mover': { |
| 144 | + const children = elements(node); |
| 145 | + if (children.length < 2) return ''; |
| 146 | + const body = convertNode(children[0]); |
| 147 | + const accent = children[1].textContent.trim(); |
| 148 | + const cmd = OVER_ACCENTS[accent]; |
| 149 | + if (cmd) return cmd + '{' + body + '}'; |
| 150 | + return '\\overset{' + accent + '}{' + body + '}'; |
| 151 | + } |
| 152 | + |
| 153 | + case 'munder': { |
| 154 | + const children = elements(node); |
| 155 | + if (children.length < 2) return ''; |
| 156 | + const body = convertNode(children[0]); |
| 157 | + const accent = children[1].textContent.trim(); |
| 158 | + const cmd = UNDER_ACCENTS[accent]; |
| 159 | + if (cmd) return cmd + '{' + body + '}'; |
| 160 | + return '\\underset{' + accent + '}{' + body + '}'; |
| 161 | + } |
| 162 | + |
| 163 | + case 'mtable': |
| 164 | + return convertTable(node); |
| 165 | + |
| 166 | + case 'mtr': |
| 167 | + return elements(node).map(convertNode).join(' & '); |
| 168 | + |
| 169 | + case 'mtd': |
| 170 | + return convertChildren(node); |
| 171 | + |
| 172 | + case 'mpadded': |
| 173 | + case 'mstyle': |
| 174 | + case 'merror': |
| 175 | + return convertChildren(node); |
| 176 | + |
| 177 | + case 'annotation': |
| 178 | + case 'annotation-xml': |
| 179 | + return ''; // skip annotations (handled in semantics) |
| 180 | + |
| 181 | + default: |
| 182 | + // Unknown element — just convert children |
| 183 | + return convertChildren(node); |
| 184 | + } |
| 185 | +} |
| 186 | + |
| 187 | +const OVER_ACCENTS = { |
| 188 | + '\u20D7': '\\vec', // combining right arrow above |
| 189 | + '\u2192': '\\vec', // rightwards arrow |
| 190 | + '\u0302': '\\hat', // combining circumflex |
| 191 | + '\u005E': '\\hat', // circumflex |
| 192 | + '^': '\\hat', |
| 193 | + '\u0303': '\\tilde', // combining tilde |
| 194 | + '~': '\\tilde', |
| 195 | + '\u0304': '\\bar', // combining macron |
| 196 | + '\u00AF': '\\bar', // macron |
| 197 | + '\u0305': '\\overline', // combining overline |
| 198 | + '\u0307': '\\dot', // combining dot above |
| 199 | + '\u0308': '\\ddot', // combining diaeresis |
| 200 | + '¯': '\\overline', |
| 201 | +}; |
| 202 | + |
| 203 | +const UNDER_ACCENTS = { |
| 204 | + '\u0332': '\\underline', |
| 205 | + '_': '\\underline', |
| 206 | +}; |
| 207 | + |
| 208 | +function convertTable(node) { |
| 209 | + const rows = elements(node).filter(el => localName(el) === 'mtr'); |
| 210 | + const body = rows.map(row => { |
| 211 | + const cells = elements(row).filter(el => localName(el) === 'mtd'); |
| 212 | + return cells.map(convertNode).join(' & '); |
| 213 | + }).join(' \\\\ '); |
| 214 | + return '\\begin{matrix} ' + body + ' \\end{matrix}'; |
| 215 | +} |
| 216 | + |
| 217 | +/** Convert all child elements, concatenated */ |
| 218 | +function convertChildren(node) { |
| 219 | + let result = ''; |
| 220 | + for (const child of node.childNodes) { |
| 221 | + result += convertNode(child); |
| 222 | + } |
| 223 | + return result; |
| 224 | +} |
| 225 | + |
| 226 | +/** Get element children only (skip text/comment nodes) */ |
| 227 | +function elements(node) { |
| 228 | + return Array.from(node.children || []); |
| 229 | +} |
| 230 | + |
| 231 | +/** Get local name without namespace prefix */ |
| 232 | +function localName(el) { |
| 233 | + return (el.localName || el.tagName || '').replace(/^.*:/, '').toLowerCase(); |
| 234 | +} |
| 235 | + |
| 236 | +/** Wrap multi-char base in braces for sup/sub */ |
| 237 | +function wrapBase(base) { |
| 238 | + // Single char or single command doesn't need braces |
| 239 | + if (base.length <= 1) return base; |
| 240 | + if (base.startsWith('\\') && !base.includes('{')) return base; |
| 241 | + return base; |
| 242 | +} |
0 commit comments