cabbage/aishare
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Compact MathML cleanup

Browse Source
This commit is contained in:
Codex
2026-07-08 09:48:46 +00:00
parent 65759088bd
commit 9ddc369536
1 changed files with 0 additions and 69 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/capture.ts
-69
View File
@@ -279,13 +279,10 @@ function normalizeCopiedHtml(html: string): string {
);
});
normalizeMathMl($);
$(":empty")
.not("br,img,hr")
.filter((_, element) => $(element).parents("math, .katex, .katex-display").length === 0)
.remove();
normalizeMathMl($);
const finalDocument = cheerio.load($.html(), {}, false);
normalizeMathMl(finalDocument);
@@ -476,72 +473,6 @@ async function extractConversation(sourceUrl: string): Promise<ExtractedConversa
});
});
function comparableMathText(value: string): string {
return value.replace(/\\div/g, "÷").replace(/\s+/g, "").trim();
}
function trimDuplicateMathFallback(fallback: string, expression: string): string | null {
const target = comparableMathText(expression);
if (!target) return null;
let consumed = "";
let index = 0;
while (index < fallback.length && consumed.length < target.length) {
const next = fallback.startsWith("\\div", index) ? "÷" : fallback[index];
const width = fallback.startsWith("\\div", index) ? 4 : 1;
index += width;
if (/\s/.test(next)) continue;
consumed += next;
}
if (consumed === target) return fallback.slice(index);
const leading = fallback.match(/^\s*/)?.[0] || "";
const rest = fallback.slice(leading.length);
const fallbackRun = rest.match(/^[^\s<]+/)?.[0] || "";
const mathOperators = /[=∑∏√±×⋅*/^_≥≤<>−+÷]/;
if (fallbackRun.length >= 3 && mathOperators.test(fallbackRun) && mathOperators.test(target)) {
return `${leading}${rest.slice(fallbackRun.length)}`;
}
const firstLine = rest.split(/\n/, 1)[0] || "";
const compactLine = comparableMathText(firstLine);
if (compactLine.length >= 8 && mathOperators.test(compactLine) && mathOperators.test(target)) {
const available = new Map<string, number>();
for (const char of target) available.set(char, (available.get(char) || 0) + 1);
let shared = 0;
for (const char of compactLine) {
const count = available.get(char) || 0;
if (count <= 0) continue;
available.set(char, count - 1);
shared += 1;
}
if (shared / Math.min(target.length, compactLine.length) >= 0.6) {
return `${leading}${rest.slice(firstLine.length)}`;
}
}
return null;
}
function trimMathFallbackNode(node: ChildNode | null, expression: string): void {
if (node?.nodeType === Node.TEXT_NODE) {
const trimmed = trimDuplicateMathFallback(node.textContent || "", expression);
if (trimmed !== null) node.textContent = trimmed;
return;
}
if (!(node instanceof HTMLElement) || node.tagName.toLowerCase() !== "span") return;
if (node.childNodes.length !== 1 || node.firstChild?.nodeType !== Node.TEXT_NODE) return;
const trimmed = trimDuplicateMathFallback(node.textContent || "", expression);
if (trimmed !== null) node.textContent = trimmed;
}
clone.querySelectorAll("math").forEach((math) => {
const visible = math.cloneNode(true) as Element;
visible.querySelectorAll("annotation, annotation-xml").forEach((node) => node.remove());
const visibleText = visible.textContent || "";
math.querySelectorAll("annotation, annotation-xml").forEach((node) => node.remove());
trimMathFallbackNode(math.nextSibling, visibleText);
});
return clone.innerHTML.trim();
}