1/* eslint-disable no-irregular-whitespace */
2import { describe, it, expect, test } from 'vitest';
3import { maskInlineLaTeX, preprocessLaTeX } from '$lib/utils/latex-protection';
4
5describe('maskInlineLaTeX', () => {
6 it('should protect LaTeX $x + y$ but not money $3.99', () => {
7 const latexExpressions: string[] = [];
8 const input = 'I have $10, $3.99 and $x + y$ and $100x$. The amount is $2,000.';
9 const output = maskInlineLaTeX(input, latexExpressions);
10
11 expect(output).toBe('I have $10, $3.99 and <<LATEX_0>> and <<LATEX_1>>. The amount is $2,000.');
12 expect(latexExpressions).toEqual(['$x + y$', '$100x$']);
13 });
14
15 it('should ignore money like $5 and $12.99', () => {
16 const latexExpressions: string[] = [];
17 const input = 'Prices are $12.99 and $5. Tax?';
18 const output = maskInlineLaTeX(input, latexExpressions);
19
20 expect(output).toBe('Prices are $12.99 and $5. Tax?');
21 expect(latexExpressions).toEqual([]);
22 });
23
24 it('should protect inline math $a^2 + b^2$ even after text', () => {
25 const latexExpressions: string[] = [];
26 const input = 'Pythagorean: $a^2 + b^2 = c^2$.';
27 const output = maskInlineLaTeX(input, latexExpressions);
28
29 expect(output).toBe('Pythagorean: <<LATEX_0>>.');
30 expect(latexExpressions).toEqual(['$a^2 + b^2 = c^2$']);
31 });
32
33 it('should not protect math that has letter after closing $ (e.g. units)', () => {
34 const latexExpressions: string[] = [];
35 const input = 'The cost is $99 and change.';
36 const output = maskInlineLaTeX(input, latexExpressions);
37
38 expect(output).toBe('The cost is $99 and change.');
39 expect(latexExpressions).toEqual([]);
40 });
41
42 it('should allow $x$ followed by punctuation', () => {
43 const latexExpressions: string[] = [];
44 const input = 'We know $x$, right?';
45 const output = maskInlineLaTeX(input, latexExpressions);
46
47 expect(output).toBe('We know <<LATEX_0>>, right?');
48 expect(latexExpressions).toEqual(['$x$']);
49 });
50
51 it('should work across multiple lines', () => {
52 const latexExpressions: string[] = [];
53 const input = `Emma buys cupcakes for $3 each.\nHow much is $x + y$?`;
54 const output = maskInlineLaTeX(input, latexExpressions);
55
56 expect(output).toBe(`Emma buys cupcakes for $3 each.\nHow much is <<LATEX_0>>?`);
57 expect(latexExpressions).toEqual(['$x + y$']);
58 });
59
60 it('should not protect $100 but protect $matrix$', () => {
61 const latexExpressions: string[] = [];
62 const input = '$100 and $\\mathrm{GL}_2(\\mathbb{F}_7)$ are different.';
63 const output = maskInlineLaTeX(input, latexExpressions);
64
65 expect(output).toBe('$100 and <<LATEX_0>> are different.');
66 expect(latexExpressions).toEqual(['$\\mathrm{GL}_2(\\mathbb{F}_7)$']);
67 });
68
69 it('should skip if $ is followed by digit and alphanumeric after close (money)', () => {
70 const latexExpressions: string[] = [];
71 const input = 'I paid $5 quickly.';
72 const output = maskInlineLaTeX(input, latexExpressions);
73
74 expect(output).toBe('I paid $5 quickly.');
75 expect(latexExpressions).toEqual([]);
76 });
77
78 it('should protect LaTeX even with special chars inside', () => {
79 const latexExpressions: string[] = [];
80 const input = 'Consider $\\alpha_1 + \\beta_2$ now.';
81 const output = maskInlineLaTeX(input, latexExpressions);
82
83 expect(output).toBe('Consider <<LATEX_0>> now.');
84 expect(latexExpressions).toEqual(['$\\alpha_1 + \\beta_2$']);
85 });
86
87 it('short text', () => {
88 const latexExpressions: string[] = ['$0$'];
89 const input = '$a$\n$a$ and $b$';
90 const output = maskInlineLaTeX(input, latexExpressions);
91
92 expect(output).toBe('<<LATEX_1>>\n<<LATEX_2>> and <<LATEX_3>>');
93 expect(latexExpressions).toEqual(['$0$', '$a$', '$a$', '$b$']);
94 });
95
96 it('empty text', () => {
97 const latexExpressions: string[] = [];
98 const input = '$\n$$\n';
99 const output = maskInlineLaTeX(input, latexExpressions);
100
101 expect(output).toBe('$\n$$\n');
102 expect(latexExpressions).toEqual([]);
103 });
104
105 it('LaTeX-spacer preceded by backslash', () => {
106 const latexExpressions: string[] = [];
107 const input = `\\[
108\\boxed{
109\\begin{aligned}
110N_{\\text{att}}^{\\text{(MHA)}} &=
111h \\bigl[\\, d_{\\text{model}}\\;d_{k} + d_{\\text{model}}\\;d_{v}\\, \\bigr] && (\\text{Q,K,V の重み})\\\\
112&\\quad+ h(d_{k}+d_{k}+d_{v}) && (\\text{バイアス Q,K,V）}\\\\[4pt]
113&\\quad+ (h d_{v})\\, d_{\\text{model}} && (\\text{出力射影 }W^{O})\\\\
114&\\quad+ d_{\\text{model}} && (\\text{バイアス }b^{O})
115\\end{aligned}}
116\\]`;
117 const output = maskInlineLaTeX(input, latexExpressions);
118
119 expect(output).toBe(input);
120 expect(latexExpressions).toEqual([]);
121 });
122});
123
124describe('preprocessLaTeX', () => {
125 test('converts inline \\( ... \\) to $...$', () => {
126 const input =
127 '\\( \\mathrm{GL}_2(\\mathbb{F}_7) \\): Group of invertible matrices with entries in \\(\\mathbb{F}_7\\).';
128 const output = preprocessLaTeX(input);
129 expect(output).toBe(
130 '$ \\mathrm{GL}_2(\\mathbb{F}_7) $: Group of invertible matrices with entries in $\\mathbb{F}_7$.'
131 );
132 });
133
134 test("don't inline \\\\( ... \\) to $...$", () => {
135 const input =
136 'Chapter 20 of The TeXbook, in source "Definitions\\\\(also called Macros)", containst the formula \\((x_1,\\ldots,x_n)\\).';
137 const output = preprocessLaTeX(input);
138 expect(output).toBe(
139 'Chapter 20 of The TeXbook, in source "Definitions\\\\(also called Macros)", containst the formula $(x_1,\\ldots,x_n)$.'
140 );
141 });
142
143 test('preserves display math \\[ ... \\] and protects adjacent text', () => {
144 const input = `Some kernel of \\(\\mathrm{SL}_2(\\mathbb{F}_7)\\):
145 \\[
146 \\left\\{ \\begin{pmatrix} 1 & 0 \\\\ 0 & 1 \\end{pmatrix}, \\begin{pmatrix} -1 & 0 \\\\ 0 & -1 \\end{pmatrix} \\right\\} = \\{\\pm I\\}
147 \\]`;
148 const output = preprocessLaTeX(input);
149
150 expect(output).toBe(`Some kernel of $\\mathrm{SL}_2(\\mathbb{F}_7)$:
151 $$
152 \\left\\{ \\begin{pmatrix} 1 & 0 \\\\ 0 & 1 \\end{pmatrix}, \\begin{pmatrix} -1 & 0 \\\\ 0 & -1 \\end{pmatrix} \\right\\} = \\{\\pm I\\}
153 $$`);
154 });
155
156 test('handles standalone display math equation', () => {
157 const input = `Algebra:
158\\[
159x = \\frac{-b \\pm \\sqrt{\\,b^{2}-4ac\\,}}{2a}
160\\]`;
161 const output = preprocessLaTeX(input);
162
163 expect(output).toBe(`Algebra:
164$$
165x = \\frac{-b \\pm \\sqrt{\\,b^{2}-4ac\\,}}{2a}
166$$`);
167 });
168
169 test('does not interpret currency values as LaTeX', () => {
170 const input = 'I have $10, $3.99 and $x + y$ and $100x$. The amount is $2,000.';
171 const output = preprocessLaTeX(input);
172
173 expect(output).toBe('I have \\$10, \\$3.99 and $x + y$ and $100x$. The amount is \\$2,000.');
174 });
175
176 test('ignores dollar signs followed by digits (money), but keeps valid math $x + y$', () => {
177 const input = 'I have $10, $3.99 and $x + y$ and $100x$. The amount is $2,000.';
178 const output = preprocessLaTeX(input);
179
180 expect(output).toBe('I have \\$10, \\$3.99 and $x + y$ and $100x$. The amount is \\$2,000.');
181 });
182
183 test('handles real-world word problems with amounts and no math delimiters', () => {
184 const input =
185 'Emma buys 2 cupcakes for $3 each and 1 cookie for $1.50. How much money does she spend in total?';
186 const output = preprocessLaTeX(input);
187
188 expect(output).toBe(
189 'Emma buys 2 cupcakes for \\$3 each and 1 cookie for \\$1.50. How much money does she spend in total?'
190 );
191 });
192
193 test('handles decimal amounts in word problem correctly', () => {
194 const input =
195 'Maria has $20. She buys a notebook for $4.75 and a pack of pencils for $3.25. How much change does she receive?';
196 const output = preprocessLaTeX(input);
197
198 expect(output).toBe(
199 'Maria has \\$20. She buys a notebook for \\$4.75 and a pack of pencils for \\$3.25. How much change does she receive?'
200 );
201 });
202
203 test('preserves display math with surrounding non-ASCII text', () => {
204 const input = `1 kg の質量は
205 \\[
206 E = (1\\ \\text{kg}) \\times (3.0 \\times 10^8\\ \\text{m/s})^2 \\approx 9.0 \\times 10^{16}\\ \\text{J}
207 \\]
208 というエネルギーに相当します。これは約 21 百万トンの TNT が爆発したときのエネルギーに匹敵します。`;
209 const output = preprocessLaTeX(input);
210
211 expect(output).toBe(
212 `1 kg の質量は
213 $$
214 E = (1\\ \\text{kg}) \\times (3.0 \\times 10^8\\ \\text{m/s})^2 \\approx 9.0 \\times 10^{16}\\ \\text{J}
215 $$
216 というエネルギーに相当します。これは約 21 百万トンの TNT が爆発したときのエネルギーに匹敵します。`
217 );
218 });
219
220 test('LaTeX-spacer preceded by backslash', () => {
221 const input = `\\[
222\\boxed{
223\\begin{aligned}
224N_{\\text{att}}^{\\text{(MHA)}} &=
225h \\bigl[\\, d_{\\text{model}}\\;d_{k} + d_{\\text{model}}\\;d_{v}\\, \\bigr] && (\\text{Q,K,V の重み})\\\\
226&\\quad+ h(d_{k}+d_{k}+d_{v}) && (\\text{バイアス Q,K,V）}\\\\[4pt]
227&\\quad+ (h d_{v})\\, d_{\\text{model}} && (\\text{出力射影 }W^{O})\\\\
228&\\quad+ d_{\\text{model}} && (\\text{バイアス }b^{O})
229\\end{aligned}}
230\\]`;
231 const output = preprocessLaTeX(input);
232 expect(output).toBe(
233 `$$
234\\boxed{
235\\begin{aligned}
236N_{\\text{att}}^{\\text{(MHA)}} &=
237h \\bigl[\\, d_{\\text{model}}\\;d_{k} + d_{\\text{model}}\\;d_{v}\\, \\bigr] && (\\text{Q,K,V の重み})\\\\
238&\\quad+ h(d_{k}+d_{k}+d_{v}) && (\\text{バイアス Q,K,V）}\\\\[4pt]
239&\\quad+ (h d_{v})\\, d_{\\text{model}} && (\\text{出力射影 }W^{O})\\\\
240&\\quad+ d_{\\text{model}} && (\\text{バイアス }b^{O})
241\\end{aligned}}
242$$`
243 );
244 });
245
246 test('converts \\[ ... \\] even when preceded by text without space', () => {
247 const input = 'Some line ...\nAlgebra: \\[x = \\frac{-b \\pm \\sqrt{\\,b^{2}-4ac\\,}}{2a}\\]';
248 const output = preprocessLaTeX(input);
249
250 expect(output).toBe(
251 'Some line ...\nAlgebra: \n$$x = \\frac{-b \\pm \\sqrt{\\,b^{2}-4ac\\,}}{2a}$$\n'
252 );
253 });
254
255 test('converts \\[ ... \\] in table-cells', () => {
256 const input = `| ID | Expression |\n| #1 | \\[
257 x = \\frac{-b \\pm \\sqrt{\\,b^{2}-4ac\\,}}{2a}
258\\] |`;
259 const output = preprocessLaTeX(input);
260
261 expect(output).toBe(
262 '| ID | Expression |\n| #1 | $x = \\frac{-b \\pm \\sqrt{\\,b^{2}-4ac\\,}}{2a}$ |'
263 );
264 });
265
266 test('escapes isolated $ before digits ($5 → \\$5), but not valid math', () => {
267 const input = 'This costs $5 and this is math $x^2$. $100 is money.';
268 const output = preprocessLaTeX(input);
269
270 expect(output).toBe('This costs \\$5 and this is math $x^2$. \\$100 is money.');
271 // Note: Since $x^2$ is detected as valid LaTeX, it's preserved.
272 // $5 becomes \$5 only *after* real math is masked — but here it's correct because the masking logic avoids treating $5 as math.
273 });
274
275 test('display with LaTeX-line-breaks', () => {
276 const input = String.raw`- Algebraic topology, Homotopy Groups of $\mathbb{S}^3$:
277$$\pi_n(\mathbb{S}^3) = \begin{cases}
278\mathbb{Z} & n = 3 \\
2790 & n > 3, n \neq 4 \\
280\mathbb{Z}_2 & n = 4 \\
281\end{cases}$$`;
282 const output = preprocessLaTeX(input);
283 // If the formula contains '\\' the $$-delimiters should be in their own line.
284 expect(output).toBe(`- Algebraic topology, Homotopy Groups of $\\mathbb{S}^3$:
285$$\n\\pi_n(\\mathbb{S}^3) = \\begin{cases}
286\\mathbb{Z} & n = 3 \\\\
2870 & n > 3, n \\neq 4 \\\\
288\\mathbb{Z}_2 & n = 4 \\\\
289\\end{cases}\n$$`);
290 });
291
292 test('handles mhchem notation safely if present', () => {
293 const input = 'Chemical reaction: \\( \\ce{H2O} \\) and $\\ce{CO2}$';
294 const output = preprocessLaTeX(input);
295
296 expect(output).toBe('Chemical reaction: $ \\ce{H2O} $ and $\\ce{CO2}$');
297 });
298
299 test('preserves code blocks', () => {
300 const input = 'Inline code: `sum $total` and block:\n```\ndollar $amount\n```\nEnd.';
301 const output = preprocessLaTeX(input);
302
303 expect(output).toBe(input); // Code blocks prevent misinterpretation
304 });
305
306 test('preserves backslash parentheses in code blocks (GitHub issue)', () => {
307 const input = '```python\nfoo = "\\(bar\\)"\n```';
308 const output = preprocessLaTeX(input);
309
310 expect(output).toBe(input); // Code blocks should not have LaTeX conversion applied
311 });
312
313 test('preserves backslash brackets in code blocks', () => {
314 const input = '```python\nfoo = "\\[bar\\]"\n```';
315 const output = preprocessLaTeX(input);
316
317 expect(output).toBe(input); // Code blocks should not have LaTeX conversion applied
318 });
319
320 test('preserves backslash parentheses in inline code', () => {
321 const input = 'Use `foo = "\\(bar\\)"` in your code.';
322 const output = preprocessLaTeX(input);
323
324 expect(output).toBe(input);
325 });
326
327 test('escape backslash in mchem ce', () => {
328 const input = 'mchem ce:\n$\\ce{2H2(g) + O2(g) -> 2H2O(l)}$';
329 const output = preprocessLaTeX(input);
330
331 // mhchem-escape would insert a backslash here.
332 expect(output).toBe('mchem ce:\n$\\ce{2H2(g) + O2(g) -> 2H2O(l)}$');
333 });
334
335 test('escape backslash in mchem pu', () => {
336 const input = 'mchem pu:\n$\\pu{-572 kJ mol^{-1}}$';
337 const output = preprocessLaTeX(input);
338
339 // mhchem-escape would insert a backslash here.
340 expect(output).toBe('mchem pu:\n$\\pu{-572 kJ mol^{-1}}$');
341 });
342
343 test('LaTeX in blockquotes with display math', () => {
344 const input =
345 '> **Definition (limit):** \n> \\[\n> \\lim_{x\\to a} f(x) = L\n> \\]\n> means that as \\(x\\) gets close to \\(a\\).';
346 const output = preprocessLaTeX(input);
347
348 // Blockquote markers should be preserved, LaTeX should be converted
349 expect(output).toContain('> **Definition (limit):**');
350 expect(output).toContain('$$');
351 expect(output).toContain('$x$');
352 expect(output).not.toContain('\\[');
353 expect(output).not.toContain('\\]');
354 expect(output).not.toContain('\\(');
355 expect(output).not.toContain('\\)');
356 });
357
358 test('LaTeX in blockquotes with inline math', () => {
359 const input =
360 "> The derivative \\(f'(x)\\) at point \\(x=a\\) measures slope.\n> Formula: \\(f'(a)=\\lim_{h\\to 0}\\frac{f(a+h)-f(a)}{h}\\)";
361 const output = preprocessLaTeX(input);
362
363 // Blockquote markers should be preserved, inline LaTeX converted to $...$
364 expect(output).toContain("> The derivative $f'(x)$ at point $x=a$ measures slope.");
365 expect(output).toContain("> Formula: $f'(a)=\\lim_{h\\to 0}\\frac{f(a+h)-f(a)}{h}$");
366 });
367
368 test('Mixed content with blockquotes and regular text', () => {
369 const input =
370 'Regular text with \\(x^2\\).\n\n> Quote with \\(y^2\\).\n\nMore text with \\(z^2\\).';
371 const output = preprocessLaTeX(input);
372
373 // All LaTeX should be converted, blockquote markers preserved
374 expect(output).toBe('Regular text with $x^2$.\n\n> Quote with $y^2$.\n\nMore text with $z^2$.');
375 });
376});