Text Filename Similarity Filtering Tool

Groups TXT/HTML files based on filename similarity and automatically keeps the ideal file in each group according to content metrics such as word count and sentence count, enabling efficient deduplication.

Bug report

Browser execution mode: Your data is processed in your browser and is not uploaded to the server.

Speed and Stability: Processing speed depends on your device and browser. For large batch work, the desktop version may be more stable.

Loading tool, please wait...

If the online tool fails to load or run, try the desktop tool.https://tools.yikeaigc.com/

Tool Usage

Return to old version

Processing tips:Before filtering, confirm the similarity threshold, algorithm, and keep rules in “Filter Settings”. Files are read and packaged only on this page and are not uploaded to the server.

1Select text files

Click to select, or drag TXT / HTML files here

Batch selection supported. Only the first 20 are previewed, but all files will be processed.

2Sample Data

Selected files

Total size

0 KB

Text type

Preview count

3File Preview

No files selected. You can load sample data first to check the settings before processing your own files.

1Grouping Rules

Filename similarity threshold

76%

Higher values require filenames to be more similar to enter the same group.

Similarity algorithm

The combined algorithm works well for Chinese, English, and mixed filenames with numbers.

Grouping method

Chain grouping is looser; representative grouping is more stable.

2Keep rule

Keep by within group

Metrics are compared only within similar groups and do not change original files.

Tie handling

Used when keep scores are tied.

Empty content files

Good for excluding placeholder files with empty text after reading.

3Read & Normalize

Text Encoding

Auto detect prioritizes BOM and common Chinese encodings.

ZIP Filename

Leave blank to use the default filename.

Preview Rows

Only affects page display, not the packaged result.

Ignore extensions when comparing Ignore case Normalize spaces, hyphens, and punctuation Downplay duplicate, copy, and version markers Group directly when content is identical Include CSV and JSON reports in ZIP

1Progress & Results

Waiting. Please select files and set options first. 0%

After processing, groups, kept files, filtered files, and download buttons will appear here.

2Similar Group Preview

No group results yet.

3Filtered File Preview

No filtered files yet.

4Processing Log

Waiting to start.

', 86400000 * 5), createFile('product_detail_page 副本.htm', '

商品详情页

页面包含规格参数和购买须知。

', 86400000 * 4), createFile('product-detail-page-2026.html', '

商品详情页

页面包含规格参数、适用场景、售后说明、购买须知、常见问题和配送说明。

', 86400000 * 2), createFile('about-us.html', '

关于我们

公司简介、服务范围和联系方式。

', 86400000) ]; const mixed = [ createFile('Report_A_2026_Final.txt', 'Report A. Revenue: 126000. Cost: 84200. Notes: checked by finance team.', 86400000 * 8), createFile('report-a-2026-final copy.txt', 'Report A. Revenue: 126000. Cost: 84200.', 86400000 * 7), createFile('Report A 2026 v3.txt', 'Report A. Revenue: 126000. Cost: 84200. Notes: checked by finance team. Added channel split and summary.', 86400000 * 2), createFile('客户名单_北京.txt', '张三\n李四\n王五\n赵六\n钱七', 86400000 * 3), createFile('客户名单-北京-副本.txt', '张三\n李四\n王五', 86400000 * 2), createFile('客户名单上海.txt', '陈一\n周二\n吴三\n郑四', 86400000) ]; if (kind === 'html') setFiles(html); else if (kind === 'mixed') setFiles(mixed); else setFiles(article); appendLog('已载入示例数据，可以切换到筛选设置进行试算。'); switchTab('tf-tab-settings'); } function normalizeFilename(name, options) { let value = options.ignoreExt ? fileBase(name) : name; try { value = value.normalize('NFKC'); } catch (e) { // Older browsers may not support normalize; keep original text. } if (options.normalizeCase) value = value.toLowerCase(); if (options.removeCopy) { value = value .replace(/\b(copy|copied|final|draft|new|old)\b/gi, ' ') .replace(/(副本|复件|拷贝|复制|修订版|定稿|草稿|新版|旧版)/g, ' ') .replace(/[\s_-]*(v|ver|version|版本)\s*\d+(\.\d+)?/gi, ' ') .replace(/[\s_-]*$?\d{4}[-_.年]?\d{1,2}?[-_.月]?\d{0,2}日?$?/g, ' ') .replace(/[\s_-]*[$（]\d+[$）]/g, ' '); } if (options.normalizeSymbols) { value = value .replace(/[·•,，.。;；:：!！?？'"“”‘’`~@#$%^&*+=|\\/<>《》【】\[\]{}()（）]/g, ' ') .replace(/[-_—–]+/g, ' ') .replace(/\s+/g, ''); } else { value = value.replace(/\s+/g, ' ').trim(); } return value.trim(); } function levenshteinSimilarity(a, b) { if (a === b) return 100; const lenA = a.length; const lenB = b.length; if (!lenA || !lenB) return lenA === lenB ? 100 : 0; let previous = new Array(lenB + 1); let current = new Array(lenB + 1); for (let j = 0; j <= lenB; j++) previous[j] = j; for (let i = 1; i <= lenA; i++) { current[0] = i; for (let j = 1; j <= lenB; j++) { const cost = a[i - 1] === b[j - 1] ? 0 : 1; current[j] = Math.min(previous[j] + 1, current[j - 1] + 1, previous[j - 1] + cost); } const temp = previous; previous = current; current = temp; } return (1 - previous[lenB] / Math.max(lenA, lenB)) * 100; } function lcsSimilarity(a, b) { if (a === b) return 100; const lenA = a.length; const lenB = b.length; if (!lenA || !lenB) return lenA === lenB ? 100 : 0; let prev = new Array(lenB + 1).fill(0); let curr = new Array(lenB + 1).fill(0); for (let i = 1; i <= lenA; i++) { for (let j = 1; j <= lenB; j++) { curr[j] = a[i - 1] === b[j - 1] ? prev[j - 1] + 1 : Math.max(prev[j], curr[j - 1]); } const temp = prev; prev = curr; curr = temp.fill(0); } return (2 * prev[lenB] / (lenA + lenB)) * 100; } function charSetSimilarity(a, b) { const setA = new Set(Array.from(a)); const setB = new Set(Array.from(b)); if (!setA.size && !setB.size) return 100; let intersection = 0; setA.forEach((char) => { if (setB.has(char)) intersection++; }); const union = new Set([...setA, ...setB]).size; return union ? intersection / union * 100 : 0; } function cosineSimilarity(a, b) { const count = (text) => { const map = new Map(); Array.from(text).forEach((char) => map.set(char, (map.get(char) || 0) + 1)); return map; }; const mapA = count(a); const mapB = count(b); const chars = new Set([...mapA.keys(), ...mapB.keys()]); if (!chars.size) return 100; let dot = 0; let magA = 0; let magB = 0; chars.forEach((char) => { const av = mapA.get(char) || 0; const bv = mapB.get(char) || 0; dot += av * bv; magA += av * av; magB += bv * bv; }); if (!magA || !magB) return 0; return dot / (Math.sqrt(magA) * Math.sqrt(magB)) * 100; } function filenameSimilarity(a, b, algorithm) { if (a === b) return 100; if (!a || !b) return 0; if (algorithm === 'levenshtein') return levenshteinSimilarity(a, b); if (algorithm === 'sequence') return lcsSimilarity(a, b); if (algorithm === 'jaccard') return charSetSimilarity(a, b); if (algorithm === 'cosine') return cosineSimilarity(a, b); const lev = levenshteinSimilarity(a, b); const seq = lcsSimilarity(a, b); const jac = charSetSimilarity(a, b); const cos = cosineSimilarity(a, b); return lev * 0.42 + seq * 0.28 + jac * 0.14 + cos * 0.16; } function hashText(text) { let hash = 5381; for (let i = 0; i < text.length; i++) { hash = ((hash << 5) + hash) ^ text.charCodeAt(i); hash >>>= 0; } return hash.toString(16); } function stripHtml(raw, name) { const looksHtml = /\.html?$/i.test(name) || /<\/?[a-z][\s\S]*>/i.test(raw); if (!looksHtml) return raw; try { const doc = new DOMParser().parseFromString(raw, 'text/html'); doc.querySelectorAll('script,style,noscript,template,svg,canvas').forEach((node) => node.remove()); return (doc.body ? doc.body.textContent : doc.documentElement.textContent) || raw.replace(/<[^>]*>/g, ' '); } catch (e) { return raw.replace(/<[^>]*>/g, ' '); } } function detectEncoding(buffer, preferred) { if (preferred && preferred !== 'auto') return preferred; const bytes = new Uint8Array(buffer); if (bytes[0] === 0xef && bytes[1] === 0xbb && bytes[2] === 0xbf) return 'utf-8'; if (bytes[0] === 0xff && bytes[1] === 0xfe) return 'utf-16le'; if (bytes[0] === 0xfe && bytes[1] === 0xff) return 'utf-16be'; if (window.jschardet && bytes.length > 0) { try { let probe = ''; const limit = Math.min(bytes.length, 8000); for (let i = 0; i < limit; i++) probe += String.fromCharCode(bytes[i]); const detected = window.jschardet.detect(probe); const name = String(detected.encoding || '').toLowerCase(); if (/gb|gbk|gb2312|gb18030/.test(name)) return 'gb18030'; if (/big5/.test(name)) return 'big5'; if (/utf-16le/.test(name)) return 'utf-16le'; if (/utf-16be/.test(name)) return 'utf-16be'; if (/utf-8|ascii/.test(name)) return 'utf-8'; } catch (e) { return 'utf-8'; } } return 'utf-8'; } function decodeBuffer(buffer, encoding) { const candidates = [encoding, 'utf-8', 'gb18030']; for (const item of candidates) { try { return new TextDecoder(item, { fatal: false }).decode(buffer); } catch (e) { // Try next candidate. } } return new TextDecoder().decode(buffer); } function countHtmlTagRatio(raw) { if (!raw) return 0; const matches = raw.match(/<[^>]+>/g) || []; const tagLength = matches.reduce((sum, item) => sum + item.length, 0); return tagLength / raw.length; } function analyzeText(raw, file) { const clean = stripHtml(raw, file.name) .replace(/\u00a0/g, ' ') .replace(/\r\n?/g, '\n') .trim(); const noPunc = clean.replace(/[\s.,;:!?，。！？；：“”‘’"'、()（）《》<>【】\[\]{}\-_=+\\/|`~@#$%^&*…·]/g, ''); const charCount = Array.from(noPunc).length; const cjkUnits = clean.match(/[\u3400-\u9fff]{1,2}/g) || []; const latinWords = clean.match(/[A-Za-z0-9]+(?:[._-][A-Za-z0-9]+)*/g) || []; const sentenceParts = clean.split(/[。！？!?；;\n]+/).map((s) => s.trim()).filter(Boolean); const paragraphs = clean.split(/\n\s*\n+/).map((s) => s.trim()).filter(Boolean); const digits = clean.match(/\d/g) || []; const wordCount = cjkUnits.length + latinWords.length; return { rawLength: raw.length, cleanText: clean, char_count_no_punc: charCount, word_count_segmented: wordCount, sentence_count: sentenceParts.length, paragraph_count: paragraphs.length || (clean ? 1 : 0), avg_sentence_length: sentenceParts.length ? charCount / sentenceParts.length : 0, digit_ratio: charCount ? digits.length / charCount : 0, html_tag_ratio: countHtmlTagRatio(raw), content_hash: hashText(clean.replace(/\s+/g, '')) }; } async function readAndAnalyzeFile(file, options, index) { const buffer = await file.arrayBuffer(); const encoding = detectEncoding(buffer, options.encoding); const raw = decodeBuffer(buffer, encoding); const metrics = analyzeText(raw, file); return { index, file, name: file.name, ext: fileExt(file.name), size: file.size || 0, lastModified: file.lastModified || 0, encoding, normalizedName: normalizeFilename(file.name, options), metrics }; } function enrichQualityScores(filesInfo) { const max = { chars: Math.max(1, ...filesInfo.map((info) => info.metrics.char_count_no_punc)), words: Math.max(1, ...filesInfo.map((info) => info.metrics.word_count_segmented)), sentences: Math.max(1, ...filesInfo.map((info) => info.metrics.sentence_count)), paragraphs: Math.max(1, ...filesInfo.map((info) => info.metrics.paragraph_count)) }; filesInfo.forEach((info) => { const m = info.metrics; const contentPart = (m.char_count_no_punc / max.chars) * 42 + (m.word_count_segmented / max.words) * 24 + (m.sentence_count / max.sentences) * 18 + (m.paragraph_count / max.paragraphs) * 10; const htmlPenalty = Math.min(10, m.html_tag_ratio * 20); const emptyPenalty = m.char_count_no_punc === 0 ? 24 : 0; m.quality_score = Math.max(0, contentPart + 6 - htmlPenalty - emptyPenalty); }); } function sleepFrame() { return new Promise((resolve) => requestAnimationFrame(resolve)); } function makePairKey(a, b) { return a < b ? `${a}:${b}` : `${b}:${a}`; } function getSimilarityValue(infoA, infoB, options, pairScores) { const key = makePairKey(infoA.index, infoB.index); if (pairScores && pairScores.has(key)) return pairScores.get(key); if (options.exactContent && infoA.metrics.content_hash === infoB.metrics.content_hash && infoA.metrics.cleanText) { if (pairScores) pairScores.set(key, 100); return 100; } const score = filenameSimilarity(infoA.normalizedName, infoB.normalizedName, options.algorithm); if (pairScores) pairScores.set(key, score); return score; } async function buildConnectedGroups(filesInfo, options) { const n = filesInfo.length; const adj = Array.from({ length: n }, () => []); const pairScores = new Map(); const total = n * (n - 1) / 2 || 1; let done = 0; for (let i = 0; i < n; i++) { for (let j = i + 1; j < n; j++) { if (cancelRequested) throw new Error('处理已取消'); const score = getSimilarityValue(filesInfo[i], filesInfo[j], options, pairScores); if (score >= options.threshold) { adj[i].push(j); adj[j].push(i); } done++; if (done % 120 === 0 || done === total) { setProgress(45 + done / total * 35, `正在比较文件名相似度：${done}/${total}`); await sleepFrame(); } } } const visited = new Array(n).fill(false); const groups = []; for (let start = 0; start < n; start++) { if (visited[start]) continue; const queue = [start]; const members = []; visited[start] = true; while (queue.length) { const item = queue.shift(); members.push(item); adj[item].forEach((next) => { if (!visited[next]) { visited[next] = true; queue.push(next); } }); } groups.push({ memberIndexes: members }); } return { groups, pairScores }; } async function buildRepresentativeGroups(filesInfo, options) { const pairScores = new Map(); const groups = []; let done = 0; const total = Math.max(1, filesInfo.length * Math.max(1, filesInfo.length - 1) / 2); for (let i = 0; i < filesInfo.length; i++) { if (cancelRequested) throw new Error('处理已取消'); let placed = false; for (const group of groups) { const rep = filesInfo[group.memberIndexes[0]]; const score = getSimilarityValue(filesInfo[i], rep, options, pairScores); done++; if (score >= options.threshold) { group.memberIndexes.push(i); placed = true; break; } } if (!placed) groups.push({ memberIndexes: [i] }); if (i % 20 === 0 || i === filesInfo.length - 1) { setProgress(45 + Math.min(35, done / total * 35), `正在按代表文件分组：${i + 1}/${filesInfo.length}`); await sleepFrame(); } } return { groups, pairScores }; } function parseRetention(value) { const direction = value.endsWith('_asc') ? 'asc' : 'desc'; const metric = value.replace(/_(asc|desc)$/, ''); return { metric, direction }; } function metricValue(info, metric) { if (metric === 'file_size') return info.size || 0; return Number(info.metrics[metric] || 0); } function tieBreakCompare(a, b, options) { if (options.tiebreak === 'modified_new') return (b.lastModified || 0) - (a.lastModified || 0); if (options.tiebreak === 'modified_old') return (a.lastModified || 0) - (b.lastModified || 0); if (options.tiebreak === 'name_short') return a.name.length - b.name.length; if (options.tiebreak === 'size_large') return (b.size || 0) - (a.size || 0); return a.index - b.index; } function chooseKeptFile(memberInfos, options) { const parsed = parseRetention(options.retention); const sorted = [...memberInfos].sort((a, b) => { const av = metricValue(a, parsed.metric); const bv = metricValue(b, parsed.metric); if (Math.abs(av - bv) > 0.000001) { return parsed.direction === 'asc' ? av - bv : bv - av; } return tieBreakCompare(a, b, options); }); return sorted[0]; } function averageSimilarity(memberInfos, pairScores) { if (memberInfos.length <= 1) return 100; let sum = 0; let count = 0; for (let i = 0; i < memberInfos.length; i++) { for (let j = i + 1; j < memberInfos.length; j++) { const score = pairScores.get(makePairKey(memberInfos[i].index, memberInfos[j].index)); if (typeof score === 'number') { sum += score; count++; } } } return count ? sum / count : 0; } function decideRetention(filesInfo, rawGroups, pairScores, options) { const groups = []; const retained = []; const discarded = []; let groupNumber = 1; rawGroups.forEach((rawGroup) => { const memberInfos = rawGroup.memberIndexes.map((idx) => filesInfo[idx]); const emptyMembers = options.emptyAction === 'discard' ? memberInfos.filter((info) => info.metrics.char_count_no_punc === 0) : []; const eligibleMembers = options.emptyAction === 'discard' ? memberInfos.filter((info) => info.metrics.char_count_no_punc > 0) : memberInfos; let keep = null; if (eligibleMembers.length) { keep = chooseKeptFile(eligibleMembers, options); retained.push({ info: keep, groupId: groupNumber }); } memberInfos.forEach((info) => { if (!keep || info.index !== keep.index) { const emptyReason = info.metrics.char_count_no_punc === 0 && options.emptyAction === 'discard'; discarded.push({ info, groupId: groupNumber, reason: emptyReason ? '空内容排除' : '同组筛选' }); } }); groups.push({ id: groupNumber, members: memberInfos, keep, discarded: memberInfos.filter((info) => !keep || info.index !== keep.index), avgSimilarity: averageSimilarity(memberInfos, pairScores) }); groupNumber++; }); return { groups, retained, discarded }; } function uniqueZipName(originalName, usedNames) { const sanitized = String(originalName || 'file.txt').replace(/[\\/:*?"<>|]+/g, '_'); const dot = sanitized.lastIndexOf('.'); const base = dot > 0 ? sanitized.slice(0, dot) : sanitized; const ext = dot > 0 ? sanitized.slice(dot) : ''; let candidate = sanitized || 'file.txt'; let counter = 2; while (usedNames.has(candidate.toLowerCase())) { candidate = `${base}_${counter}${ext}`; counter++; } usedNames.add(candidate.toLowerCase()); return candidate; } function buildReportRows(result) { const rows = []; result.groups.forEach((group) => { group.members.forEach((info) => { const kept = group.keep && info.index === group.keep.index; rows.push({ group_id: group.id, status: kept ? '保留' : '筛选', filename: info.name, normalized_name: info.normalizedName, group_size: group.members.length, avg_similarity: group.avgSimilarity.toFixed(2), size: info.size, encoding: info.encoding, chars_no_punc: info.metrics.char_count_no_punc, words: info.metrics.word_count_segmented, sentences: info.metrics.sentence_count, avg_sentence_length: info.metrics.avg_sentence_length.toFixed(2), digit_ratio: (info.metrics.digit_ratio * 100).toFixed(2) + '%', html_tag_ratio: (info.metrics.html_tag_ratio * 100).toFixed(2) + '%', quality_score: info.metrics.quality_score.toFixed(2) }); }); }); return rows; } function rowsToCsv(rows) { const headers = [ 'group_id', 'status', 'filename', 'normalized_name', 'group_size', 'avg_similarity', 'size', 'encoding', 'chars_no_punc', 'words', 'sentences', 'avg_sentence_length', 'digit_ratio', 'html_tag_ratio', 'quality_score' ]; const lines = [headers.join(',')]; rows.forEach((row) => { lines.push(headers.map((key) => escapeCsv(row[key])).join(',')); }); return '\ufeff' + lines.join('\n'); } function buildJsonReport(result) { return JSON.stringify({ generatedAt: new Date().toISOString(), options: result.options, summary: result.summary, groups: result.groups.map((group) => ({ id: group.id, avgSimilarity: Number(group.avgSimilarity.toFixed(2)), kept: group.keep ? group.keep.name : null, members: group.members.map((info) => ({ filename: info.name, status: group.keep && info.index === group.keep.index ? 'keep' : 'discard', metrics: { charsNoPunc: info.metrics.char_count_no_punc, words: info.metrics.word_count_segmented, sentences: info.metrics.sentence_count, avgSentenceLength: Number(info.metrics.avg_sentence_length.toFixed(2)), digitRatio: Number((info.metrics.digit_ratio * 100).toFixed(2)), htmlTagRatio: Number((info.metrics.html_tag_ratio * 100).toFixed(2)), qualityScore: Number(info.metrics.quality_score.toFixed(2)) } })) })) }, null, 2); } function renderEmptyResult() { latestResult = null; setProgress(0, '等待处理。请先选择文件并设置参数。'); els.resultPlaceholder.style.display = 'block'; els.resultContent.style.display = 'none'; els.groupsEmpty.style.display = 'block'; els.groupsTable.style.display = 'none'; els.groupsBody.innerHTML = ''; els.groupsMore.textContent = ''; els.discardEmpty.style.display = 'block'; els.discardTable.style.display = 'none'; els.discardBody.innerHTML = ''; els.discardMore.textContent = ''; els.downloadZip.disabled = true; els.downloadCsv.disabled = true; els.downloadJson.disabled = true; } function renderResults(result, mode) { const options = result.options; const similarGroups = result.groups.filter((group) => group.members.length > 1); const ungrouped = result.groups.filter((group) => group.members.length === 1).length; const previewLimit = options.previewLimit; els.resultPlaceholder.style.display = 'none'; els.resultContent.style.display = 'block'; els.resTotal.textContent = result.summary.totalFiles; els.resGroups.textContent = similarGroups.length; els.resKeep.textContent = result.retained.length; els.resDrop.textContent = result.discarded.length; els.resTime.textContent = `${(result.summary.elapsedMs / 1000).toFixed(2)}s`; els.resultPills.innerHTML = ''; [ `算法：${els.algorithm.options[els.algorithm.selectedIndex].text}`, `阈值：${options.threshold}%`, `保留：${metricLabels[options.retention] || '自定义指标'}`, `单文件组：${ungrouped}` ].forEach((text) => { const pill = document.createElement('span'); pill.className = 'tf-pill'; pill.textContent = text; els.resultPills.appendChild(pill); }); els.downloadZip.disabled = mode === 'preview' || result.retained.length === 0; els.downloadCsv.disabled = false; els.downloadJson.disabled = false; const shownGroups = result.groups.slice(0, previewLimit); els.groupsBody.innerHTML = ''; if (shownGroups.length) { shownGroups.forEach((group) => { const tr = document.createElement('tr'); const keepName = group.keep ? group.keep.name : '无保留文件'; const cells = [ group.id, keepName, group.members.length, `${group.avgSimilarity.toFixed(1)}%`, metricLabels[options.retention] || options.retention ]; cells.forEach((value, idx) => { const td = document.createElement('td'); td.textContent = value; if (idx === 1) td.className = 'tf-mono'; tr.appendChild(td); }); els.groupsBody.appendChild(tr); }); els.groupsEmpty.style.display = 'none'; els.groupsTable.style.display = 'block'; els.groupsMore.textContent = result.groups.length > previewLimit ? `还有 ${result.groups.length - previewLimit} 个分组未在页面展示，可在报告中查看。` : ''; } else { els.groupsEmpty.style.display = 'block'; els.groupsTable.style.display = 'none'; els.groupsMore.textContent = ''; } const shownDiscarded = result.discarded.slice(0, previewLimit); els.discardBody.innerHTML = ''; if (shownDiscarded.length) { shownDiscarded.forEach((item) => { const tr = document.createElement('tr'); const nameTd = document.createElement('td'); nameTd.textContent = item.info.name; nameTd.className = 'tf-mono'; const groupTd = document.createElement('td'); groupTd.textContent = item.groupId; const statusTd = document.createElement('td'); const badge = document.createElement('span'); badge.className = 'tf-badge-drop'; badge.textContent = item.reason; statusTd.appendChild(badge); tr.append(nameTd, groupTd, statusTd); els.discardBody.appendChild(tr); }); els.discardEmpty.style.display = 'none'; els.discardTable.style.display = 'block'; els.discardMore.textContent = result.discarded.length > previewLimit ? `还有 ${result.discarded.length - previewLimit} 个筛选文件未在页面展示，可在报告中查看。` : ''; } else { els.discardEmpty.style.display = 'block'; els.discardTable.style.display = 'none'; els.discardMore.textContent = ''; } } async function runProcessing(mode) { if (busy) return; if (!selectedFiles.length) { appendLog('请先选择文件或加载示例数据。'); switchTab('tf-tab-files'); return; } if (selectedFiles.length < 2) { appendLog('文件数量不足，至少需要 2 个文件才能进行相似度筛选。'); return; } const options = getOptions(); const startTime = performance.now(); cancelRequested = false; setBusy(true); latestResult = null; els.log.textContent = ''; switchTab('tf-tab-results'); setProgress(2, '开始读取文件内容...'); appendLog(`准备处理 ${selectedFiles.length} 个文件。`); try { const filesInfo = []; for (let i = 0; i < selectedFiles.length; i++) { if (cancelRequested) throw new Error('处理已取消'); const info = await readAndAnalyzeFile(selectedFiles[i], options, i); filesInfo.push(info); setProgress(5 + (i + 1) / selectedFiles.length * 35, `正在读取：${info.name}`); if (i % 8 === 0) await sleepFrame(); } enrichQualityScores(filesInfo); appendLog('文件内容统计完成。'); setProgress(43, '开始计算文件名相似度...'); const groupBuild = options.groupMode === 'representative' ? await buildRepresentativeGroups(filesInfo, options) : await buildConnectedGroups(filesInfo, options); if (cancelRequested) throw new Error('处理已取消'); setProgress(83, '正在按保留规则筛选...'); const decided = decideRetention(filesInfo, groupBuild.groups, groupBuild.pairScores, options); const elapsedMs = performance.now() - startTime; const reportRows = buildReportRows(decided); const result = { ...decided, options, reportRows, csv: rowsToCsv(reportRows), json: '', summary: { totalFiles: selectedFiles.length, retainedFiles: decided.retained.length, discardedFiles: decided.discarded.length, similarGroups: decided.groups.filter((group) => group.members.length > 1).length, elapsedMs } }; result.json = buildJsonReport(result); latestResult = result; setProgress(100, mode === 'preview' ? '分组预览完成。' : '筛选完成，可以下载结果。'); appendLog(`处理完成：保留 ${result.retained.length} 个文件，筛选 ${result.discarded.length} 个文件。`); renderResults(result, mode); } catch (error) { setProgress(0, error.message === '处理已取消' ? '处理已取消。' : '处理失败。'); appendLog(error.message || '处理失败。'); } finally { setBusy(false); } } async function downloadZip() { if (!latestResult || !latestResult.retained.length) return; if (!window.JSZip) { appendLog('JSZip 组件未加载，暂时无法打包下载。'); return; } els.downloadZip.disabled = true; appendLog('正在生成 ZIP 文件...'); const zip = new JSZip(); const usedNames = new Set(); latestResult.retained.forEach((item) => { const outputName = uniqueZipName(item.info.name, usedNames); zip.file(outputName, item.info.file); }); if (latestResult.options.includeReport) { zip.file('筛选报告.csv', latestResult.csv); zip.file('筛选报告.json', latestResult.json); } try { const blob = await zip.generateAsync({ type: 'blob' }, (metadata) => { setProgress(metadata.percent, `正在打包 ZIP：${metadata.percent.toFixed(1)}%`); }); let name = latestResult.options.zipName || '文本文件名相似度筛选结果.zip'; if (!/\.zip$/i.test(name)) name += '.zip'; downloadBlob(blob, name); setProgress(100, 'ZIP 已生成。'); appendLog('ZIP 下载已触发。'); } catch (error) { appendLog(`ZIP 生成失败：${error.message || error}`); } finally { els.downloadZip.disabled = false; } } function downloadCsv() { if (!latestResult) return; downloadBlob(new Blob([latestResult.csv], { type: 'text/csv;charset=utf-8' }), '文本文件名相似度筛选报告.csv'); } function downloadJson() { if (!latestResult) return; downloadBlob(new Blob([latestResult.json], { type: 'application/json;charset=utf-8' }), '文本文件名相似度筛选报告.json'); } els.tabs.forEach((tab) => tab.addEventListener('click', () => switchTab(tab.dataset.tab))); els.selectFiles.addEventListener('click', () => els.fileInput.click()); els.dropzone.addEventListener('click', () => els.fileInput.click()); els.dropzone.addEventListener('keydown', (event) => { if (event.key === 'Enter' || event.key === ' ') { event.preventDefault(); els.fileInput.click(); } }); els.fileInput.addEventListener('change', () => setFiles(els.fileInput.files)); els.clearFiles.addEventListener('click', () => { selectedFiles = []; els.fileInput.value = ''; renderFilePreview(); renderEmptyResult(); appendLog('已清空文件列表。'); }); els.toSettings.addEventListener('click', () => switchTab('tf-tab-settings')); els.backFiles.addEventListener('click', () => switchTab('tf-tab-files')); els.goSettings.addEventListener('click', () => switchTab('tf-tab-settings')); els.threshold.addEventListener('input', () => { els.thresholdValue.textContent = els.threshold.value; }); els.previewGroups.addEventListener('click', () => runProcessing('preview')); els.start.addEventListener('click', () => runProcessing('process')); els.cancel.addEventListener('click', () => { cancelRequested = true; appendLog('正在尝试取消处理...'); }); els.downloadZip.addEventListener('click', downloadZip); els.downloadCsv.addEventListener('click', downloadCsv); els.downloadJson.addEventListener('click', downloadJson); $$('.tf-sample-btn').forEach((button) => { button.addEventListener('click', () => loadSample(button.dataset.sample)); }); ['dragenter', 'dragover'].forEach((eventName) => { els.dropzone.addEventListener(eventName, (event) => { event.preventDefault(); event.stopPropagation(); els.dropzone.classList.add('dragover'); }); }); ['dragleave', 'drop'].forEach((eventName) => { els.dropzone.addEventListener(eventName, (event) => { event.preventDefault(); event.stopPropagation(); els.dropzone.classList.remove('dragover'); }); }); els.dropzone.addEventListener('drop', (event) => { const files = event.dataTransfer ? event.dataTransfer.files : []; setFiles(files); }); renderFilePreview(); renderEmptyResult(); })();

Instructions

Software Usage Instructions

Select files or load an example: Click “Select Files” to batch select text files such as .txt, .html, and .htm, or use the sample data for testing first. The page previews only the first 20 files, but all files will be included during processing.
Set similarity grouping:
- Similarity threshold: The higher the value, the more similar file names must be to be placed in the same group.
- Similarity algorithm: You can choose the combined algorithm, Levenshtein edit distance, continuous match ratio, Jaccard, or character frequency cosine.
- Grouping method: Similar chain grouping is suitable for files with continuous version changes; representative file grouping is suitable for organizing files around a baseline name.
Choose the retention basis: You can decide which file to keep within a group based on overall content quality score, plain text character count, word count, sentence count, average sentence length, number ratio, HTML tag ratio, or file size.
Configure reading and output: You can set the text encoding, ZIP file name, and number of preview lines, and choose whether to ignore extensions, case, copy markers, and version number markers, as well as whether to include CSV and JSON reports in the ZIP.
Start filtering: After clicking “Start Filtering,” the tool will read the files, extract the main text, calculate file name similarity, generate groups, and keep files according to the rules.
View and download: After processing is complete, you can view the group preview, filtered file preview, and processing log. Click “Download Kept Files ZIP” to get the results. If files with the same name appear in the results, numbers will be appended to the file names to distinguish them.

FAQ

A: It is recommended to start testing with the default threshold. When the threshold is higher, only files with similar names will be grouped together; when the threshold is lower, files with larger naming differences may also be placed in the same group. You can use “Preview Groups” first to check the effect before running the final filtering.

A: Levenshtein mainly measures the edit distance between file names and is suitable for files with minor name changes. The combined algorithm combines edit distance, continuous matching, character sets, and character frequency, making it more flexible for file names that mix Chinese, English, serial numbers, and copy markers.

A: This keeps the page display clean and the experience smooth. The preview count does not affect the processing scope; all selected files will be included in the similarity calculation and ZIP packaging.

A: Kept files will use their original file names whenever possible. If files with the same name appear in the ZIP, the system will append a number to the original file name, such as 文章_2.txt, to avoid overwriting.

A: HTML files first have their visible body text extracted, with scripts, styles, and other content removed, before metrics such as character count, word count, and sentence count are calculated. The report also records the percentage of HTML tags to help assess page content density.

A: No. File reading, content statistics, similarity grouping, and ZIP packaging are all completed locally in the page. The server will not receive your file contents.

A: No. Original files will not be moved, renamed, or deleted. Filtering only determines which files are included in the download package; files that are not kept will be listed in the report and page preview.