fmode-voc-projects/scripts/tools/短剧变现/analyze.js

const d = require('./docs/短剧变现/raw/_merged.json');

const hypotheses = ['H1', 'H2', 'H3', 'H4', 'H5', 'H6', 'H7', 'H8'];

function groupByHypothesis(items) {
  const map = {};
  for (const h of hypotheses) map[h] = [];
  for (const item of items) {
    for (const h of (item.hypothesis || [])) {
      if (map[h]) map[h].push(item);
    }
  }
  return map;
}

function sentimentDist(items) {
  const dist = {};
  for (const item of items) {
    const s = item.sentiment || 'unknown';
    dist[s] = (dist[s] || 0) + 1;
  }
  return dist;
}

function topTags(items, n) {
  n = n || 20;
  const counts = {};
  for (const item of items) {
    for (const tag of (item.tags || [])) {
      counts[tag] = (counts[tag] || 0) + 1;
    }
  }
  return Object.entries(counts)
    .sort(function(a, b) { return b[1] - a[1]; })
    .slice(0, n);
}

function truncate(str, len) {
  if (!str) return '';
  return str.length > len ? str.slice(0, len) + '...' : str;
}

var grouped = groupByHypothesis(d.items);
var overallTags = topTags(d.items, 20);
var overallSentiment = sentimentDist(d.items);

console.log('================================================================================');
console.log('  短剧变现数据分析报告');
console.log('================================================================================');
console.log('总条目数: ' + d.items.length);
console.log('采集时间: ' + d.meta.collectedAt);
console.log('');

console.log('------------------------------------------------------------');
console.log('  整体情感分布:');
console.log('------------------------------------------------------------');
var ks = Object.keys(overallSentiment).sort(function(a, b) { return overallSentiment[b] - overallSentiment[a]; });
for (var i = 0; i < ks.length; i++) {
  var s = ks[i];
  var c = overallSentiment[s];
  var pct = (c / d.items.length * 100).toFixed(1);
  console.log('  ' + s + ': ' + c + ' (' + pct + '%)');
}
console.log('');

console.log('------------------------------------------------------------');
console.log('  全量 Top 20 标签:');
console.log('------------------------------------------------------------');
for (var i = 0; i < overallTags.length; i++) {
  var tag = overallTags[i][0];
  var count = overallTags[i][1];
  console.log('  ' + pad(i + 1, 2) + '. ' + pad(tag, 18) + ' ' + count);
}
console.log('');

for (var hi = 0; hi < hypotheses.length; hi++) {
  var h = hypotheses[hi];
  var items = grouped[h];
  if (items.length === 0) continue;

  var sorted = items.slice().sort(function(a, b) { return b.likes - a.likes; });
  var sent = sentimentDist(items);
  var topN = sorted.slice(0, 8);

  console.log('------------------------------------------------------------');
  console.log('  ' + h + ' -- 共 ' + items.length + ' 条');
  console.log('------------------------------------------------------------');

  var sentParts = [];
  var sk = Object.keys(sent).sort(function(a, b) { return sent[b] - sent[a]; });
  for (var si = 0; si < sk.length; si++) {
    var ss = sk[si];
    var sc = sent[ss];
    sentParts.push(ss + ': ' + sc + ' (' + (sc / items.length * 100).toFixed(1) + '%)');
  }
  console.log('  情感分布: ' + sentParts.join(' | '));
  console.log('');

  console.log('  Top 8 高点赞条目:');
  console.log('  ' + repeat('-', 74));

  for (var ti = 0; ti < topN.length; ti++) {
    var item = topN[ti];
    var tagStr = (item.tags || []).slice(0, 3).join(', ') || '(无)';
    console.log('  ' + (ti + 1) + '. [' + pad(item.platform, 6) + '] Likes:' + pad(String(item.likes), 7) + ' | ' + pad(item.keyword, 12) + ' | ' + pad(item.sentiment || '', 9) + ' | ' + tagStr);
    console.log('     ' + truncate(item.content, 80));
    console.log('');
  }

  console.log('');
}

console.log('================================================================================');
console.log('  全量 Top 20 标签汇总:');
console.log('================================================================================');
var maxCount = overallTags[0] ? overallTags[0][1] : 1;
for (var i = 0; i < overallTags.length; i++) {
  var tag = overallTags[i][0];
  var count = overallTags[i][1];
  var bar = repeat('|', Math.max(1, Math.round(count / maxCount * 30)));
  console.log('  ' + pad(i + 1, 2) + '. ' + pad(tag, 18) + ' ' + pad(String(count), 4) + '  ' + bar);
}

function pad(s, n) {
  s = String(s);
  while (s.length < n) s = s + ' ';
  return s;
}

function repeat(ch, n) {
  var r = '';
  for (var i = 0; i < n; i++) r += ch;
  return r;
}