@1aboveio/skills 1.0.21 → 1.0.23

package/skills/risk-compliance/portfolio-payment-analysis/references/workflow-and-contracts.md

@@ -0,0 +1,71 @@
+# Workflow and module contracts
+
+The skill serves payment portfolios whose records can be attributed to submerchants.
+Worldpay CSV and DART-340 are the bundled source adapters. Other sources require an
+explicit field/identity mapping into the same event layers; a similar descriptor is
+not such a mapping. Keep source-specific interpretation in preparation, financial
+arithmetic in calculation, report selection in construction, and browser behavior in
+rendering.
+
+## Stages and completion criteria
+
+| Stage | Input → output | Complete when |
+| --- | --- | --- |
+| Scope | Request and authorized files → dates, identities, coverage | Main, authorization and risk periods are explicit; missing layers and identity conflicts are recorded. |
+| Prepare | Original or selected sources → verified event layers | Source hashes and selected-row accounting reconcile; authorization, sale/refund and dispute grains are distinct; unknown identities remain unknown. |
+| Calculate | Event layers and daily FX → `facts.json` and business CSVs | Merchant numerators, denominators, event dates, currency conversions, nulls and observed zeros are reproducible without a chart library. |
+| Construct | Frozen facts and merchant selection → `presentation.json` | Both report sections and each selected merchant's charts reference the correct evidence and independent cutoff; exclusions apply to every business output. |
+| Review | Saved presentation and evidence → reviewed presentation | Findings, coverage and terminology are checked; required language checks are recorded honestly, with no changes to financial facts. |
+| Render | Reviewed presentation → HTML and Markdown companion | HTML embeds ECharts and notices, loads offline, and shows all tabs and supported chart interactions. Rendering does not load payment sources or recalculate metrics. |
+| Verify | Saved outputs → verification record | Numeric series, missing values, cutoffs, issuer ranking, actual tooltips, tab resizing and printing have evidence; incomplete checks are named. |
+
+## Report structure
+
+The default compact merchant report contains findings followed by a merchant comparison
+and merchant-detail tabs. The legacy visible headings `1. 子商户对比` and
+`3. 重点子商户详情` are retained for compatibility; their numbering does not imply
+hidden chapters. All verified merchants remain reachable. An initial-tab selection is
+navigation, while exclusion changes the report population. Unknown identities stay in
+reconciliation, outside merchant tabs.
+
+Comparison: main-period attempts, approvals, approval rate, approved authorization USD,
+refunds, dispute observations and daily attempt-count sparklines. Details: up to eight
+observed authorization weeks; two-week country, product, amount, authentication and
+response comparisons; twelve-month/eight-week risk histories. Exact chart and metric
+contracts are in [metrics and reporting](metrics-and-reporting.md) and
+[risk history](risk-history.md).
+
+## Dates and missing data
+
+Main comparison and risk history share the earliest relevant layer's final event date.
+Authorization-only details and trends use the earlier of requested end and final
+authorization date. Paired authorization periods match weekdays across adjacent weeks;
+missing dates retain positions. No full period or continuous coverage is inferred from
+minimum/maximum timestamps. A null is unavailable, not zero. A zero approval numerator
+with positive attempts is 0%; no attempts has no defined approval rate. Missing FX
+invalidates the affected amount, not its count.
+
+## Script responsibilities
+
+- `analyze.py` owns source loading, verified merchant mapping and reconciliation; `report_data.py`
+  derives reusable aggregate evidence from those event layers.
+- `payment_amounts.py` owns shared complete-or-unavailable daily USD sums.
+- Preparation accepts calculation functions explicitly and returns JSON-compatible
+  aggregates. It does not import the report builder or construct chart containers.
+- `report_model.py` accepts frozen aggregate evidence, explicit risk denominators
+  and dispute counts. It does not retain mutable process-wide calculation state.
+- `build_report.py` owns findings, selections and business exports; its model module
+  defines the report structure and chart declarations.
+- `render.py` is the stable CLI. `render_base.py` and `render_custom.py` compose
+  the document and tabs; `echarts_report.py` creates chart declarations and embeds
+  `scripts/assets/charts.js`, `charts.css` and `tabs.js`. The HTML shell receives chart and cell rendering
+  functions explicitly. ECharts reads only chart declarations and serializes safely;
+  JavaScript chart code and styles are packaged assets, not payment calculators.
+
+## Extension policy
+
+Add a source adapter when a supported source needs different parsing or identity rules.
+Add a metric only with its population, date basis, numerator, denominator, currency and
+availability rule. Add a chart only after its declared series exist in frozen facts.
+Do not add arbitrary report configuration for a hypothetical use case. Keep the
+validated default report and CLI stable while internal modules become reusable.

package/skills/risk-compliance/portfolio-payment-analysis/scripts/analyze.py

@@ -597,7 +597,16 @@ def analyze(root, start, end, fxpath):
                       'timezone': 'Source timestamps are naive; timezone is unknown'}
     merchant_risk_history, reconciliation['risk_history'] = risk_history.build(
         root, end, comparison_merchants, layers['settlement'], layers['dispute'])
-    return {'schema_version': 1, 'metadata': {'main_start': start.isoformat(), 'main_end': end.isoformat(),
+    import report_data, sys
+    interactive_merchants=list(merchants)
+    for mid in all_ids:
+        if mid in ids:
+            continue
+        label=sorted(names[mid],key=lambda x:(-names[mid][x],x))[0] if names[mid] else 'Unknown'
+        interactive_merchants.append({'id':mid or None,'name':label,**summary([],[],[],rates,present)})
+    auth_end=min(requested_end,date.fromisoformat(last_dates['auth']))
+    interactive = report_data.calculate(layers, end, interactive_merchants, rates, sys.modules[__name__],auth_end=auth_end)
+    return {'schema_version': 1, 'interactive_report': interactive, 'metadata': {'main_start': start.isoformat(), 'main_end': end.isoformat(),
               'requested_main_end': requested_end.isoformat(), 'layers_present': present, 'layer_last_dates': last_dates,
               'comparison_start': previous_start.isoformat(), 'comparison_end': previous_end.isoformat(),
               'auth_date': 'Timestamp', 'settlement_date': 'Settlement date', 'dispute_denominator_date': 'Timestamp (transaction/capture date)', 'fx_date': 'Timestamp', 'rate_unit': 'fraction'},

package/skills/risk-compliance/portfolio-payment-analysis/scripts/assets/charts.css

@@ -0,0 +1 @@
+.ec-block{margin:28px 0;min-width:0}.ec-block h3{margin-bottom:10px}.ec-control{display:block;margin:12px 0}.ec-control select{font:inherit;min-height:44px;padding:6px 10px}.ec-scroll{overflow-x:auto}.ec-host{width:100%;height:430px;min-width:640px}.ec-spark{width:168px}.ec-spark .ec-host{width:168px;min-width:168px;height:44px}.ec-spark-dates{font-size:12px;color:#536770}.ec-tooltip{font-variant-numeric:tabular-nums;line-height:1.7}.ec-tooltip b{font-weight:650}.ec-block .ec-host:focus-visible{outline:2px solid #c16613}.ec-tooltip-row{white-space:normal}@media(max-width:700px){.ec-host{min-width:760px}.ec-spark .ec-host{min-width:168px}.ec-block[data-ec-pie] .ec-host{min-width:550px}}@media print{.ec-control{display:none}.ec-scroll{overflow:visible}.ec-host{min-width:0!important}.ec-block{break-inside:avoid}}

package/skills/risk-compliance/portfolio-payment-analysis/scripts/assets/charts.js

@@ -0,0 +1,60 @@
+(()=>{
+const colors=['#087b71','#7761ad','#a65721','#4f7489','#8b7135','#6e8178'];const states=[];
+const esc=x=>String(x??'').replace(/[&<>"']/g,c=>({'&':'&amp;','<':'&lt;','>':'&gt;','"':'&quot;',"'":'&#39;'}[c]));
+const value=v=>v===null||v===undefined||v===''?null:Number(v);
+const fmt=(v,f='count')=>v===null||v===undefined?'—':f==='percent'?(Number(v)*100).toFixed(2)+'%':(f==='usd'?'USD ':'')+Number(v).toLocaleString('en-US',{minimumFractionDigits:f==='usd'?2:0,maximumFractionDigits:f==='usd'?2:0});
+const tooltip={trigger:'axis',showDelay:0,hideDelay:0,transitionDuration:0,appendTo:()=>document.body,confine:false,backgroundColor:'#142c36',borderColor:'#36545b',textStyle:{color:'#fff',fontSize:13},extraCssText:'max-width:min(460px,calc(100vw - 24px));white-space:normal;line-height:1.7;z-index:10000;',axisPointer:{type:'cross'},className:'ec-tooltip'};
+const wrap=(head,lines)=>'<b>'+esc(head)+'</b><br>'+lines.map(esc).join('<br>');
+const base=()=>({animation:false,color:colors,aria:{enabled:true},textStyle:{fontFamily:"-apple-system,BlinkMacSystemFont,'PingFang SC',sans-serif"},legend:{type:'scroll',top:0,textStyle:{color:'#344a51'}},grid:{left:80,right:90,top:78,bottom:62},tooltip:{...tooltip},xAxis:{type:'category',axisLabel:{hideOverlap:true,color:'#536770'}},yAxis:{type:'value',min:0,axisLabel:{color:'#536770'},splitLine:{lineStyle:{color:'#e2eae6'}}}});
+function option(state){
+ const d=state.data,o=base();state.rows=[];
+ if(d.type==='sparkline'){
+  o.legend={show:false};o.grid={left:4,right:4,top:4,bottom:4};o.xAxis={type:'category',show:false,boundaryGap:false,data:d.points.map(p=>p.date)};o.yAxis={type:'value',show:false,min:0,max:d.max||1};o.series=[{name:'每日授权笔数',type:'line',connectNulls:false,symbolSize:3,lineStyle:{width:2},data:d.points.map(p=>p.attempts)}];o.tooltip={...tooltip,axisPointer:{type:'line'},formatter:ps=>{const p=d.points[ps[0]?.dataIndex];return p?wrap(p.date,['授权尝试：'+fmt(p.attempts)+' 笔']):'';}};return o;
+ }
+ if(d.type==='risk_dual_axis'){
+  o.xAxis.data=d.points.map(p=>p.label+(p.partial?'*':''));o.yAxis=[{type:'value',name:'数量（笔）',min:0,minInterval:1,splitLine:{lineStyle:{color:'#e2eae6'}}},{type:'value',name:'比率 %',min:0,axisLabel:{formatter:v=>fmt(v,'percent')},splitLine:{show:false}}];
+  o.series=d.count_series.map((s,i)=>({name:s.label+' · 数量',type:'bar',yAxisIndex:0,barMaxWidth:26,itemStyle:{color:colors[i%colors.length],opacity:.55},data:d.points.map(p=>p.counts[s.key])})).concat(d.rate_series.map((s,i)=>({name:s.label+' · 比率',type:'line',yAxisIndex:1,connectNulls:false,symbolSize:6,itemStyle:{color:colors[i%colors.length]},data:d.points.map(p=>{const v=p.rates[s.key];return v!==null&&v>0?v:null;})})));
+  o.tooltip.formatter=ps=>{const p=d.points[ps[0]?.dataIndex];if(!p)return '';const lines=d.count_series.map(s=>s.label+'：'+fmt(p.counts[s.key])+' 笔');d.rate_series.forEach(s=>{const v=p.rates[s.key],detail=p.details[s.key]||{};lines.push(s.label+'：'+fmt(v,'percent')+(detail.numerator!==null&&detail.numerator!==undefined&&detail.denominator?' ('+fmt(detail.numerator)+' / '+fmt(detail.denominator)+')':''));});return wrap(p.start+' 至 '+p.end+(p.partial?'（不完整期间）':''),lines);};return o;
+ }
+ if(d.type==='weekly_transaction_trend'){
+  o.xAxis.data=d.points.map(p=>p.start.slice(5)+' 至 '+p.end.slice(5));o.yAxis=[{type:'value',min:0,name:'授权尝试金额 USD',axisLabel:{formatter:v=>v>=1000?(v/1000).toLocaleString()+'k':v},splitLine:{lineStyle:{color:'#e2eae6'}}},{type:'value',min:0,minInterval:1,name:'授权尝试笔数',splitLine:{show:false}}];o.series=[{name:'授权尝试金额 USD',type:'bar',yAxisIndex:0,barMaxWidth:50,data:d.points.map(p=>value(p.amount_usd))},{name:'授权尝试笔数',type:'line',yAxisIndex:1,connectNulls:false,data:d.points.map(p=>p.count)}];o.tooltip.formatter=ps=>{const p=d.points[ps[0]?.dataIndex];return p?wrap(p.start+' 至 '+p.end,['授权尝试金额：'+fmt(p.amount_usd,'usd'),'授权尝试笔数：'+fmt(p.count)+' 笔']):'';};if(!d.points.length)o.graphic={type:'text',left:'center',top:'middle',style:{text:'最近 8 周未观察到数据',fill:'#536770',fontSize:15}};return o;
+ }
+ const metric=d.metrics.find(m=>m.key===state.metric)||d.metrics[0];
+ if(d.layout==='pie_pair'){
+  state.el.setAttribute('data-ec-pie','true');delete o.grid;delete o.xAxis;delete o.yAxis;o.legend={type:'scroll',bottom:0,data:d.categories.map(c=>c.label)};o.title=d.series.map((s,i)=>({text:s.label,left:i?'55%':'5%',top:12,textStyle:{fontSize:13,color:'#344a51'}}));o.series=[];o.graphic=[];
+  d.series.forEach((s,i)=>{const values=d.categories.map(c=>value(c.values[s.key]?.[metric.key]));if(values.some(v=>v===null||v<0)||!values.some(v=>v>0)){o.graphic.push({type:'text',left:i?'63%':'13%',top:'43%',style:{text:'暂无完整可绘制数据',fill:'#536770',fontSize:14}});return;}o.series.push({name:s.label,type:'pie',radius:['0%','48%'],center:[i?'75%':'25%','48%'],label:{show:false},data:d.categories.map((c,j)=>({name:c.label,value:values[j]}))});});o.tooltip={...tooltip,trigger:'item',formatter:p=>wrap(p.seriesName,[p.name+'：'+fmt(p.value,metric.format)])};return o;
+ }
+ if(d.layout==='histogram_pair'||d.layout==='bars_pair'){
+  const histogram=d.layout==='histogram_pair',cats=d.categories.filter(c=>!histogram||!c.is_unknown),all=cats.flatMap(c=>d.series.map(s=>value(c.values[s.key]?.[metric.key]))).filter(v=>v!==null),max=metric.format==='percent'?1:Math.max(...all,1);
+  if(!histogram)state.host.style.height=Math.max(350,cats.length*40+135)+'px';
+  o.grid=histogram?[{left:75,right:'54%',top:70,bottom:100},{left:'57%',right:40,top:70,bottom:100}]:[{left:260,right:'53%',top:60,bottom:40},{left:'60%',right:85,top:60,bottom:40}];
+  o.xAxis=[0,1].map(i=>histogram?{type:'category',gridIndex:i,data:cats.map(c=>c.label),axisLabel:{rotate:30,hideOverlap:false,fontSize:11}}:{type:'value',gridIndex:i,min:0,max,axisLabel:{formatter:v=>fmt(v,metric.format)}});
+  o.yAxis=[0,1].map(i=>histogram?{type:'value',gridIndex:i,min:0,max,axisLabel:{formatter:v=>fmt(v,metric.format)}}:{type:'category',gridIndex:i,inverse:true,data:cats.map(c=>c.label),axisLabel:{show:i===0,width:240,overflow:'truncate'},axisTick:{show:false},axisLine:{show:false}});
+  o.series=d.series.map((s,i)=>({name:s.label,type:'bar',xAxisIndex:i,yAxisIndex:i,barMaxWidth:histogram?35:15,itemStyle:{color:i?'#087b71':'#727c80'},data:cats.map(c=>value(c.values[s.key]?.[metric.key]))}));o.tooltip.formatter=ps=>{const c=cats[ps[0]?.dataIndex];return c?wrap(c.label,d.series.map(s=>s.label+'：'+fmt(c.values[s.key]?.[metric.key],metric.format))):'';};
+  if(histogram){const unknown=d.categories.filter(c=>c.is_unknown);if(unknown.length)o.graphic={type:'text',left:12,bottom:0,style:{text:unknown.map(c=>c.label+'：'+d.series.map(s=>s.label+' '+fmt(c.values[s.key]?.[metric.key],metric.format)).join('；')).join('\\n'),fill:'#536770',fontSize:11,width:700,overflow:'break'}};}
+  return o;
+ }
+ const rows=[];d.categories.forEach((c,i)=>{rows.push({c,index:i,child:false});if(state.expanded.has(i))(c.children||[]).forEach(c=>rows.push({c,index:i,child:true}));});state.rows=rows;
+ const label=r=>(r.child?'    ':r.c.children?.length?(state.expanded.has(r.index)?'▾ ':'▸ '):'')+r.c.label;
+ state.host.style.height=Math.max(320,rows.length*52+125)+'px';o.grid={left:285,right:175,top:60,bottom:38};o.legend.data=d.series.map(s=>s.label);o.xAxis={type:'value',min:0,max:metric.format==='percent'?1:undefined,axisLabel:{formatter:v=>fmt(v,metric.format)},splitLine:{lineStyle:{color:'#e2eae6'}}};o.yAxis={type:'category',inverse:true,triggerEvent:true,data:rows.map(label),axisLabel:{width:260,overflow:'truncate',color:'#344a51'},axisTick:{show:false},axisLine:{show:false}};
+ const get=(c,s)=>metric.key==='auth_rate'&&!(Number(c.values[s.key]?.count)>0)?null:value(c.values[s.key]?.[metric.key]);
+ const show=(c,s)=>{const values=c.values[s.key]||{},v=get(c,s);return fmt(v,metric.format)+(metric.key==='auth_rate'&&v!==null?' ('+fmt(values.approvals)+'/'+fmt(values.count)+')':'');};
+ o.series=d.series.map((s,i)=>({name:s.label,type:'bar',barMaxWidth:12,itemStyle:{color:i?'#087b71':'#727c80'},data:rows.map(r=>get(r.c,s)),label:{show:true,position:'right',color:i?'#087b71':'#536770',fontSize:12,formatter:p=>show(rows[p.dataIndex].c,s)}}));o.tooltip.axisPointer={type:'shadow'};o.tooltip.formatter=ps=>{const row=rows[ps[0]?.dataIndex];if(!row)return '';return wrap(row.c.label,d.series.map(s=>s.label+'：'+show(row.c,s)));};return o;
+}
+function decorateCountry(state){
+ if(!state.data.issuer_drilldown)return;
+ state.host.querySelectorAll('svg text').forEach(node=>{const index=state.rows.findIndex(r=>!r.child&&r.c.children?.length&&((state.expanded.has(r.index)?'▾ ':'▸ ')+r.c.label)===node.textContent);if(index<0)return;const row=state.rows[index];node.setAttribute('role','button');node.setAttribute('tabindex','0');node.setAttribute('aria-expanded',String(state.expanded.has(row.index)));node.setAttribute('data-ec-country-index',row.index);node.onkeydown=event=>{if(event.key==='Enter'||event.key===' '){event.preventDefault();state.expanded.has(row.index)?state.expanded.delete(row.index):state.expanded.add(row.index);state.dirty=true;render(state);requestAnimationFrame(()=>state.host.querySelector('[data-ec-country-index="'+row.index+'"]')?.focus());}};});
+}
+function render(state,force=false){
+ if(!force&&(state.host.closest('[role=tabpanel][hidden]')||state.host.getBoundingClientRect().width===0))return;
+ if(state.chart&&!state.dirty){const width=state.host.clientWidth,height=state.host.clientHeight;if(width>0&&height>0&&(state.chart.getWidth()!==width||state.chart.getHeight()!==height))state.chart.resize({width,height});return;}
+ const opt=option(state);
+ for(const axisKey of ['xAxis','yAxis']){const axes=Array.isArray(opt[axisKey])?opt[axisKey]:opt[axisKey]?[opt[axisKey]]:[];axes.forEach((axis,index)=>{if(axis.type!=='value')return;let format='count';if(state.data.type==='risk_dual_axis'&&axisKey==='yAxis'&&index===1)format='percent';else if(state.data.type==='weekly_transaction_trend'&&axisKey==='yAxis'&&index===0)format='usd';else if(state.data.type==='comparison_chart')format=(state.data.metrics.find(m=>m.key===state.metric)||state.data.metrics[0]).format;axis.axisPointer={label:{formatter:p=>fmt(p.value,format)}};});}
+ if(!state.chart){state.chart=echarts.init(state.host,null,{renderer:'svg',width:force?(state.host.closest('main')?.clientWidth||1100):undefined});state.chart.on('finished',()=>decorateCountry(state));state.chart.on('click',p=>{if(state.data.type!=='comparison_chart'||!state.data.issuer_drilldown)return;let index=p.componentType==='yAxis'?state.rows.findIndex(r=>((r.child?'    ':r.c.children?.length?(state.expanded.has(r.index)?'▾ ':'▸ '):'')+r.c.label)===p.value):p.dataIndex;const row=state.rows[index];if(!row||row.child||!row.c.children?.length)return;state.expanded.has(row.index)?state.expanded.delete(row.index):state.expanded.add(row.index);state.dirty=true;render(state);});}
+ const previousSelection=state.chart.getOption()?.legend?.[0]?.selected;if(previousSelection&&opt.legend)opt.legend.selected=previousSelection;
+ state.chart.setOption(opt,{notMerge:true});state.chart.resize({width:state.host.clientWidth||1100,height:state.host.clientHeight||430});state.dirty=false;decorateCountry(state);state.el.dataset.ecReady='true';
+}
+document.querySelectorAll('[data-echarts-block]').forEach(el=>{const host=el.querySelector('.ec-host'),data=JSON.parse(el.querySelector('[data-ec-data]').textContent),state={el,host,data,metric:data.default_metric,expanded:new Set(),chart:null,dirty:true};states.push(state);el.querySelector('[data-ec-metric]')?.addEventListener('change',e=>{state.metric=e.target.value;state.dirty=true;render(state);});});
+window.reportECharts={states,renderVisible:()=>states.forEach(s=>render(s)),version:echarts.version};
+const schedule=()=>requestAnimationFrame(()=>states.forEach(s=>render(s)));new ResizeObserver(schedule).observe(document.querySelector('main'));document.querySelectorAll('[role=tab]').forEach(tab=>tab.addEventListener('click',schedule));new MutationObserver(schedule).observe(document.querySelector('main'),{subtree:true,attributes:true,attributeFilter:['hidden']});window.addEventListener('resize',schedule);window.addEventListener('beforeprint',()=>states.forEach(s=>render(s,true)));window.addEventListener('afterprint',schedule);schedule();
+})();

package/skills/risk-compliance/portfolio-payment-analysis/scripts/assets/tabs.js

@@ -0,0 +1,2 @@
+document.getElementById('print-report').addEventListener('click',()=>window.print());
+document.querySelectorAll('[role=tablist]').forEach(list=>{const tabs=[...list.querySelectorAll('[role=tab]')];function activate(t){tabs.forEach(x=>{const yes=x===t;x.setAttribute('aria-selected',String(yes));x.tabIndex=yes?0:-1;document.getElementById(x.getAttribute('aria-controls')).hidden=!yes;});}tabs.forEach((t,i)=>{t.addEventListener('click',()=>activate(t));t.addEventListener('keydown',e=>{let j;if(e.key==='ArrowRight')j=(i+1)%tabs.length;else if(e.key==='ArrowLeft')j=(i+tabs.length-1)%tabs.length;else if(e.key==='Home')j=0;else if(e.key==='End')j=tabs.length-1;else return;e.preventDefault();activate(tabs[j]);tabs[j].focus();});});});

package/skills/risk-compliance/portfolio-payment-analysis/scripts/build_report.py

@@ -6,6 +6,7 @@ Portfolio fields remain calculation/reconciliation evidence; this report never r
 Edit presentation.json for narrative review, then rerender with render.py.
 """
 import argparse
+import csv
 from csv_export import safe_writer
 import json
 from datetime import date
@@ -45,49 +46,6 @@ def table(headers, rows, **extra):
     return {'type':'table','headers':headers,'rows':rows,**extra}
 
 
-def top_reasons(merchant):
-    return sorted(merchant.get('decline_reasons',[]),key=lambda r:(-r['count'],r['code'],r['message']))[:5]
-
-
-def weekly_blocks(facts, merchants, selected):
-    series={entry.get('id'):entry for entry in facts.get('merchant_weeks',[])}
-    blocks=[]
-    if not series:
-        return blocks+[para('尚未生成子商户周数据，请先运行最新计算脚本。','callout')]
-    grid=next(iter(series.values()))['weeks']
-    def label(week):
-        return week['start']+' 至 '+week['end']+('（不完整周）' if week['partial_week'] else '（完整周）')
-    rows=[]
-    for merchant in merchants:
-        weeks=series.get(merchant.get('id'),{}).get('weeks',[])
-        cells=[]
-        for week in weeks:
-            metrics=week['metrics']
-            if not metrics.get('layer_observed',{}).get('auth',False):
-                cells.append('')
-            else:
-                successful_usd=amount(metrics,'approved_amount')
-                cells.append(number(metrics['approvals'])+' / '+number(metrics['attempts'])+' · '+percent(metrics['auth_rate'])+(' · 授权成功 USD '+successful_usd if successful_usd else ''))
-        rows.append([name(merchant),merchant.get('id') or '未归属',*cells])
-    blocks.append(para('周汇总格里的美元是授权成功金额，不是销售金额。结算层未观察到时，销售金额留空，授权成功金额仍可出现。','note'))
-    blocks.append(table(['子商户','商户 ID',*[label(w) for w in grid]],rows,wrap_headers=True))
-    chosen={m.get('id') for m in selected}
-    for merchant in merchants:
-        weeks=series.get(merchant.get('id'),{}).get('weeks',[])
-        metric_labels=['授权尝试','授权成功笔数','授权成功率','授权成功金额 USD','较上周（百分点）','尝试笔数周环比','销售金额 USD','销售金额周环比','退款笔数','退款金额 USD','争议快照记录']
-        weekly_columns=[]
-        for week in weeks:
-            metric=week['metrics'];comparison=week['comparison'];observed=metric.get('layer_observed',{}).get('auth',False)
-            pp=comparison.get('auth_rate_pp')
-            weekly_columns.append([number(metric['attempts']) if observed else '',number(metric['approvals']) if observed else '',percent(metric['auth_rate']),amount(metric,'approved_amount'),('' if pp is None else f'{pp:+.2f}'),percent(comparison.get('attempts_pct')),amount(metric,'sales'),percent(comparison.get('sales_usd_pct')),number(metric['refund_count']) if metric['layer_observed']['settlement'] else '',amount(metric,'refunds',True),number(metric['dispute_snapshots']) if metric['layer_observed']['dispute'] else ''])
-        weekly_rows=[[metric_label,*[column[i] for column in weekly_columns]] for i,metric_label in enumerate(metric_labels)]
-        blocks.append(table(['指标',*[label(w) for w in weeks]],weekly_rows,wrap_headers=True,collapsed=merchant.get('id') not in chosen,caption=name(merchant)+' · '+str(merchant.get('id') or '未归属')+' · 每周明细'))
-        if merchant.get('id') in chosen:
-            # Visible tables need a merchant heading; collapsed ones have a summary caption.
-            blocks.insert(len(blocks)-1,para(name(merchant)+' · '+str(merchant.get('id') or '未归属')+' · 每周明细','heading'))
-    return blocks
-
-
 def write_weekly_csv(facts, output):
     with (output/'merchant-weekly.csv').open('w',encoding='utf-8',newline='') as handle:
         writer=safe_writer(handle,numeric_columns={11,13,15})
@@ -134,10 +92,6 @@ def segment_charts(merchant, periods):
     return blocks
 
 
-def merchant_detail_blocks(merchant, periods):
-    return segment_charts(merchant,periods)
-
-
 def risk_charts(history):
     if not history:
         return []
@@ -174,16 +128,17 @@ def write_risk_csv(facts, output):
     with (output/'merchant-risk-history.csv').open('w',encoding='utf-8-sig',newline='') as handle:
         writer=safe_writer(handle,numeric_columns={7,10,11})
         writer.writerow(['merchant_id','merchant_name','metric','frequency','start','end','series','value','date_basis','status','numerator','denominator','denominator_start','denominator_end','formula','reason'])
+        import report_model, sys
         for merchant in facts.get('merchant_risk_history',[]):
-            for metric in ('disputes','fraud','vamp'):
-                history=merchant[metric]
-                for frequency in ('monthly','weekly'):
-                    for point in history.get(frequency,[]):
-                        for series,value in point['values'].items():
-                            writer.writerow([merchant.get('id'),merchant.get('name'),metric,frequency,point['start'],point['end'],series,value,history.get('basis',''),history.get('status',''),'','','','','',''])
-                        for series,value in point.get('rates',{}).items():
-                            detail=point.get('rate_details',{}).get(series,{})
-                            writer.writerow([merchant.get('id'),merchant.get('name'),metric+'_rate',frequency,point['start'],point['end'],series,value,detail.get('date_basis',''),detail.get('status',''),detail.get('numerator'),detail.get('denominator'),detail.get('denominator_start'),detail.get('denominator_end'),detail.get('formula'),detail.get('reason')])
+            for chart in report_model.risk_charts(facts,merchant,sys.modules[__name__]):
+                metric='vamp' if 'VAMP' in chart['title'] else 'fraud' if '欺诈报告' in chart['title'] else 'disputes'
+                frequency='weekly' if '周度' in chart['title'] else 'monthly'
+                for point in chart['points']:
+                    for series,value in point['counts'].items():
+                        writer.writerow([merchant.get('id'),merchant.get('name'),metric,frequency,point['start'],point['end'],series,value,chart.get('date_basis',''),'extract_observation','','','','','',''])
+                    for series,value in point['rates'].items():
+                        detail=point['details'].get(series,{})
+                        writer.writerow([merchant.get('id'),merchant.get('name'),metric+'_rate',frequency,point['start'],point['end'],series,value,detail.get('date_basis',chart.get('date_basis','')),detail.get('status',''),detail.get('numerator'),detail.get('denominator'),detail.get('denominator_start'),detail.get('denominator_end'),detail.get('formula'),detail.get('reason')])
 
 
 def write_segment_csv(facts, output):
@@ -203,7 +158,8 @@ def write_detail_csv(facts, output):
     with (output/'merchant-detail-comparison.csv').open('w',encoding='utf-8',newline='') as handle:
         writer=safe_writer(handle,numeric_columns={13,14})
         writer.writerow(['merchant_id','merchant_name','dimension','population','group_key','group_label','period','start','end','count','approvals','auth_rate','share','attempt_amount_usd','approved_amount_usd'])
-        for merchant in facts.get('merchant_detail_comparison',{}).get('merchants',[]):
+        comparison=facts.get('interactive_report') or facts.get('merchant_detail_comparison',{})
+        for merchant in comparison.get('merchants',[]):
             for dimension in merchant['dimensions']:
                 for group in dimension['groups']:
                     for period in group['periods']:
@@ -223,6 +179,12 @@ def without_merchants(facts, excluded):
     out['merchant_breakdowns']=[m for m in facts.get('merchant_breakdowns',[]) if keep(m)]
     detail=facts.get('merchant_detail_comparison') or {}
     out['merchant_detail_comparison']={**detail,'merchants':[m for m in detail.get('merchants',[]) if keep(m)]}
+    interactive=facts.get('interactive_report')
+    if interactive:
+        out['interactive_report']={**interactive,'merchants':[m for m in interactive.get('merchants',[]) if keep(m)],'risk_sales':[r for r in interactive.get('risk_sales',[]) if r.get('id') not in excluded]}
+        opened=interactive.get('opened_disputes')
+        if opened is not None and 'merchant_month_counts' in opened:
+            out['interactive_report']['opened_disputes']={**opened,'merchant_month_counts':[r for r in opened['merchant_month_counts'] if r['merchant_id'] not in excluded]}
     return out
 
 
@@ -235,14 +197,16 @@ def filter_existing_csv(path, excluded):
         return
     header,body=rows[0],rows[1:]
     id_at=next((i for i,cell in enumerate(header) if cell in ('merchant_id','id')),None)
-    kept=[row for row in body if (row[id_at] if id_at is not None and id_at<len(row) else None) not in excluded and not any(cell in excluded for cell in row)]
+    if id_at is None:
+        raise ValueError('Business CSV is missing a merchant ID column: '+path.name)
+    kept=[row for row in body if (row[id_at] if id_at<len(row) else None) not in excluded]
     with path.open('w',newline='',encoding='utf-8') as handle:
         csv.writer(handle).writerows([header,*kept])
 
 
 def build(facts, detail_merchants=None, excluded_ids=None):
     excluded=set(excluded_ids or [])
-    known={m.get('id') for m in facts.get('merchants',[])}
+    known={m.get('id') for m in facts.get('merchants',[])} | {m.get('id') for m in facts.get('interactive_report',{}).get('merchants',[])}
     unknown=excluded-known
     if unknown:
         raise ValueError('Unknown excluded merchant IDs: '+', '.join(sorted(unknown)))
@@ -283,84 +247,11 @@ def build(facts, detail_merchants=None, excluded_ids=None):
         'summary_title':'子商户核查重点','summary':findings,'sections':[],
         'links':[{'label':'商户风险历史 CSV','file':'merchant-risk-history.csv'},{'label':'本周与上周同期细分 CSV','file':'merchant-detail-comparison.csv'},{'label':'子商户每周指标 CSV','file':'merchant-weekly.csv'},{'label':'子商户指标 CSV','file':'merchant-metrics.csv'},{'label':'商户归属 CSV','file':'merchant-mapping.csv'},{'label':'归属与核验记录 JSON','file':'coverage-and-reconciliation.json'},{'label':'Markdown 报告','file':'overview.zh.md'}],
         'footer':'来源：Worldpay 授权、结算、争议及 DART-340 欺诈文件。'}
-    sections=model['sections']
-    comparison=[]
-    for m in merchants:
-        status='本期有授权记录' if m['attempts'] else '本期无授权记录'
-        if 0<m['attempts']<30:status='样本少于 30 笔'
-        observed=m.get('layer_observed',{}).get('auth',False)
-        comparison.append([name(m),m.get('id') or '未归属',number(m['attempts']) if observed else '',number(m['approvals']) if observed else '',percent(m['auth_rate']),amount(m,'sales'),number(m['refund_count']),amount(m,'refunds',True),number(m['dispute_snapshots']),status])
-    sections.append({'title':'1. 子商户对比','nav':'子商户对比','blocks':[
-        table(['子商户','商户 ID','授权尝试','授权成功','授权成功率','销售金额 USD','退款笔数','退款金额 USD','争议快照记录','样本说明'],comparison,filter=True),
-    ]})
-    sections.append({'title':'2. 子商户逐周对比','nav':'逐周对比','blocks':weekly_blocks(facts,merchants,selected)})
-    detail_groups=[]
-    comparison_data=facts.get('merchant_detail_comparison',{})
-    periods=comparison_data.get('periods',[])
-    comparisons={item.get('id'):item for item in comparison_data.get('merchants',[])}
-    histories={item.get('id'):item for item in facts.get('merchant_risk_history',[])}
-    expanded={m.get('id') for m in selected}
-    for m in merchants:
-        comparison=comparisons.get(m.get('id'))
-        blocks=merchant_detail_blocks(comparison,periods) if comparison else []
-        blocks.extend(risk_charts(histories.get(m.get('id'))))
-        detail_groups.append({'type':'group','caption':name(m)+' · '+str(m.get('id') or '未归属'),'collapsed':m.get('id') not in expanded,'blocks':blocks})
-    sections.append({'title':'3. 重点子商户详情','nav':'商户详情','blocks':detail_groups})
-    action_rows=[]
-    for m in selected:
-        reasons=top_reasons(m)
-        if reasons:
-            reason=reasons[0]
-            evidence=f"{reason['message']}：{number(reason['count'])} 笔；授权成功率 {percent(m['auth_rate'])}"
-            action='核对该响应在该商户下的处理商含义，再按发卡行国家和认证状态细分检查。'
-        else:
-            evidence=f"授权成功 {number(m['approvals'])} / {number(m['attempts'])}；缺少可展开的拒绝响应"
-            action='核对该商户的响应字段与来源覆盖，避免根据缺失原因直接调整策略。'
-        action_rows.append([name(m),m.get('id') or '未归属',evidence,action,'不将观察差异视为已确认原因，不估算可挽回金额。'])
-    for m in followups:
-        action_rows.append([name(m),m.get('id') or '未归属',f"未观察到本期授权；退款 {number(m['refund_count'])} 笔，争议快照记录 {number(m['dispute_snapshots'])} 条",'核对原始销售所属商户、退款处理及争议责任。','无本期授权不代表停业；争议记录不等于新增案件。'])
-    if not action_rows:
-        action_rows=[['未归属或缺少观察','','缺少支持逐商户判断的记录','先补齐商户身份及来源覆盖。','暂不形成商户表现结论。']]
-    sections.append({'title':'4. 逐商户核查清单','nav':'核查清单','blocks':[table(['子商户','商户 ID','观察依据','下一步核查','解释边界'],action_rows)]})
-    coverage=[table(['子商户','商户 ID','授权记录','结算记录','争议记录'],[[name(m),m.get('id') or '未归属',*['✓' if m.get('layer_observed',{}).get(layer,False) else '' for layer in ('auth','settlement','dispute')]] for m in merchants])]
-    audit=facts.get('quality',{}).get('risk_history',{})
-    if audit:
-        attribution=audit.get('fraud_attribution',{})
-        files=audit.get('files',[])
-        invalid=[item.get('date',item['file']) for item in files if item.get('status')=='invalid']
-        future=[item.get('date',item['file']) for item in files if item.get('status')=='future_not_parsed']
-        unmatched=attribution.get('unmatched')
-        amounts=audit.get('unmatched_amounts_by_currency') or {}
-        rows=[
-            ['欺诈记录归属',str(attribution.get('attributed',0))+' / '+str(audit.get('distinct_exact_rows',0))],
-            ['无效欺诈文件日期','、'.join(invalid) or '无'],
-            ['数据截至',meta['main_end']]
-        ]
-        if future:
-            rows.append(['晚于截止日、未解析的欺诈文件','、'.join(future)])
-        if unmatched:
-            amount_text='、'.join(currency+' '+money(value) for currency,value in amounts.items())
-            rows.append(['未归属欺诈记录',str(unmatched)+' 条'+(('，金额 '+amount_text) if amount_text else '')+'，未计入商户笔数'])
-        coverage.append(table(['数据范围','观察值'],rows))
-        notes=[]
-        if any(item.get('error')=='missing_header' for item in files):
-            notes.append('缺少表头的欺诈文件未解析，因此没有进入笔数。')
-        if future:
-            notes.append('晚于截止日的欺诈文件未解析。')
-        if unmatched:
-            notes.append('未能归属的欺诈记录不计入商户笔数。')
-        definitions_note=''.join(notes)
-    else:
-        definitions_note=''
-    definitions=[
-        para('空白为不可用或未观察到的数据，不作为 0；0 仅表示提供的记录中观察到的零值。'),
-        para('授权、销售与退款按各自事件日期统计。授权成功金额与结算销售金额不同，退款金额以绝对值展示。周汇总格里的美元是授权成功金额，不是销售金额。销售金额 0.00 是结算层观察到的零值；没有结算层记录时销售金额留空。'),
-        para('争议趋势按同一阶段最早观察到的争议日期统计案件，状态更新不重复计数；并非新增案件。欺诈按 340FRD 文件生成日作为报告日，完全重复记录取最早报告日。'),
-        para('比率按卡组织月度时间口径作提取数据估算，并非正式监控结果。Visa 使用当月分母，Mastercard 拒付与 EFM 使用上月分母；EFM 分子是 4837/4863 欺诈拒付，不是 340FRD 的全部 Mastercard 报告。周图仅展示笔数；预争议比率为描述性指标，不纳入 VAMP。'),
-        para('340FRD 报告日已按文件生成日确认。'+definitions_note+'VAMP 仍需核验 TC15/TC05 处理日、事件身份、适用范围及排除标记。')
-    ]
-    coverage.append({'type':'group','caption':'数据口径','collapsed':True,'blocks':definitions})
-    sections.append({'title':'5. 商户归属与覆盖','nav':'归属与覆盖','blocks':coverage})
+    import report_model, sys
+    model=report_model.build_model(facts,model,sys.modules[__name__])
+    if detail_merchants:
+        tabs=model['sections'][1]['blocks'][0]['tabs']
+        tabs.sort(key=lambda tab:(tab.get('id') not in set(detail_merchants),list(detail_merchants).index(tab.get('id')) if tab.get('id') in detail_merchants else 0))
     return model
 
 

package/skills/risk-compliance/portfolio-payment-analysis/scripts/echarts_report.py

@@ -0,0 +1,43 @@
+"""Self-contained ECharts renderer; calculations remain in presentation.json."""
+import json
+from pathlib import Path
+import render_custom as incumbent
+base=incumbent.base
+ROOT=Path(__file__).parent
+
+def payload(value):return json.dumps(value,ensure_ascii=False).replace('<','\\u003c').replace('>','\\u003e').replace('&','\\u0026')
+
+def chart_html(b,identifier,spark=False):
+ title=b.get('title','每日授权笔数')
+ data={**b,'type':'sparkline'} if spark else b
+ hp=('<div class="ec-spark"' if spark else '<div class="ec-block"')+' data-echarts-block="true">'
+ if not spark:
+  hp+='<h3>'+base.esc(title)+'</h3>'
+  note=b.get('dual_note') or b.get('description') or ''
+  if b['type']=='weekly_transaction_trend':note='柱状：授权尝试金额 USD（左轴）；折线：授权尝试笔数（右轴）。最近八周内展示有数据的周。'
+  if note:hp+='<p class="note">'+base.esc(note)+'</p>'
+  if b['type']=='comparison_chart' and not b.get('approval_rate_only') and len(b['metrics'])>1:
+   hp+='<label class="ec-control">选择指标 <select data-ec-metric>'+''.join('<option value="'+base.esc(m['key'])+'"'+(' selected' if m['key']==b['default_metric'] else '')+'>'+base.esc(m['label'])+'</option>' for m in b['metrics'])+'</select></label>'
+ hp+='<div class="ec-scroll"><div class="ec-host" role="img" aria-label="'+base.esc(title)+'"></div></div><script type="application/json" data-ec-data>'+payload(data)+'</script></div>'
+ return hp
+
+def spark(cell):
+ hp=chart_html(cell,'spark',True);points=cell['points'];desc='；'.join(p['date']+' '+(f"{p['attempts']:,} 笔" if p['attempts'] is not None else '未观察到数据') for p in points)
+ return hp+'<small class="ec-spark-dates">'+points[0]['date'][5:]+' — '+points[-1]['date'][5:]+'</small>',desc
+
+def output(b,identifier,level=3):
+ if b['type'] in ('risk_dual_axis','weekly_transaction_trend','comparison_chart'):
+  return chart_html(b,identifier),'### '+base.mdtext(b['title'])+'\n\n'+base.mdtext(b.get('dual_note') or b.get('description') or '')
+ return incumbent.render_block(b,identifier,level,output,spark)
+
+CSS=(ROOT/'assets/charts.css').read_text(encoding='utf-8')
+
+JS=(ROOT/'assets/charts.js').read_text(encoding='utf-8')
+
+TAB_JS=(ROOT/'assets/tabs.js').read_text(encoding='utf-8')
+
+def render(model,output_dir):
+ base.render(model,output_dir,block_renderer=output,extra_css=incumbent.CSS+CSS,script=TAB_JS,comparison_js='')
+ path=Path(output_dir)/'overview.zh.html';page=path.read_text();bundle=(ROOT/'vendor/echarts.min.js').read_text().replace('</script','<\\/script')
+ notice=(ROOT/'vendor/NOTICE').read_text();license_text=(ROOT/'vendor/LICENSE').read_text();assert '-->' not in notice+license_text
+ page=page.replace('</body>','<!-- Apache ECharts 6.1.0 third-party notices\n'+notice+'\n'+license_text+'\n--><script>'+bundle+'</script><script>'+JS+'</script></body>');path.write_text(page)

package/skills/risk-compliance/portfolio-payment-analysis/scripts/interactive_report.py

@@ -0,0 +1,3 @@
+"""Compatibility imports for saved integrations; implementation has explicit stages."""
+from report_data import calculate
+from report_model import build_model

package/skills/risk-compliance/portfolio-payment-analysis/scripts/issuer_data.py

@@ -0,0 +1,52 @@
+"""Merchant-scoped issuing-bank drilldown aggregates; no raw rows are exported."""
+from collections import Counter
+from payment_amounts import attempt_amount_usd
+
+def attach(blocks,dimensions,auth,periods,calc,rates=None):
+ country=next((d for d in dimensions if d['key']=='bin_country'),None)
+ chart=next((b for b in blocks if b.get('title')==(country or {}).get('label')),None)
+ if not country or not chart:return []
+ groups={g['label']:g for g in country['groups']};audit=[]
+ current=next(p for p in periods if p['key']=='current')
+ for category in chart['categories']:
+  group=groups.get(category['label'])
+  if group is None and category.get('is_unknown'):group=next((g for g in country['groups'] if g['key']=='["unknown"]'),None)
+  if group is None:continue
+  import json
+  key=json.loads(group['key'])
+  if key[0]=='other':continue
+  rows=[r for r in auth if (r.get('Issuing bank country','').strip()==key[1] if key[0]=='value' else not r.get('Issuing bank country','').strip())]
+  ranking_rows=[r for r in rows if current['start']<=calc.day(r['Timestamp'])<=current['end']]
+  ranking=Counter(r.get('Issuing bank','').strip() or '未知发卡行' for r in ranking_rows)
+  banks=sorted((name for name,n in ranking.items() if n>0),key=lambda name:(-ranking[name],name))[:5];children=[]
+  for bank in banks:
+   values={}
+   for p in periods:
+    country_rows=[r for r in rows if p['start']<=calc.day(r['Timestamp'])<=p['end']]
+    all_rows=[r for r in auth if p['start']<=calc.day(r['Timestamp'])<=p['end']]
+    rr=[r for r in country_rows if (r.get('Issuing bank','').strip() or '未知发卡行')==bank]
+    n=len(rr);ap=sum(r.get('Response type','').lower()=='approval' for r in rr)
+    def amount(selected):
+     return attempt_amount_usd(selected,rates or {},calc.day)
+    values[p['key']]={'count':n if country_rows else None,'approvals':ap if country_rows else None,'auth_rate':ap/n if n else None,'share':n/len(all_rows) if country_rows and all_rows else None,'attempt_amount':amount(rr) if country_rows else None,'approved_amount':amount([r for r in rr if r.get('Response type','').lower()=='approval']) if country_rows else None}
+   children.append({'label':bank,'values':values,'ranking_count':ranking[bank]})
+  category['children']=children
+  audit.append({'country':category['label'],'banks':children,'country_count':len(ranking_rows),'ranking_period':{'start':current['start'],'end':current['end']},'top5_count':sum(ranking[b] for b in banks)})
+ chart['issuer_drilldown']=True
+ chart['description']='点击国家展开或折叠 Top 5 发卡行。按本周（'+current['start']+' 至 '+current['end']+'）授权尝试笔数降序排名（包括授权拒绝，不是按成功笔数）；上周比较同一组发卡行。授权成功率显示成功笔数/尝试笔数；尝试为 0 时留空。银行比例以该商户当周全部授权尝试为分母；Top 5 不代表该国全部银行。'
+ return audit
+
+
+def calculate(dimensions,auth,periods,calc,rates=None):
+ """Return issuer groups using source grouping keys, without a chart builder."""
+ country=next((d for d in dimensions if d['key']=='bin_country'),None)
+ if not country:return None
+ import json
+ categories=[]
+ for group in country['groups']:
+  key=json.loads(group['key'])
+  label='其他（合并）' if key==['other'] else '未知（缺失）' if key==['unknown'] else group['label']
+  categories.append({'label':label,'is_unknown':key==['unknown']})
+ block={'title':country['label'],'categories':categories}
+ attach([block],dimensions,auth,periods,calc,rates)
+ return block

package/skills/risk-compliance/portfolio-payment-analysis/scripts/payment_amounts.py

@@ -0,0 +1,20 @@
+"""Shared daily-USD amount calculation for authorization aggregates."""
+from decimal import Decimal, InvalidOperation
+
+def attempt_amount_usd(rows, rates, day):
+    """Return a complete decimal sum or None; never a partial converted total."""
+    total = Decimal(0)
+    for row in rows:
+        try:
+            amount = Decimal(row.get('Transaction amount', ''))
+            currency = (row.get('Transaction amount currency') or '').strip().upper()
+            rate = Decimal(1) if currency == 'USD' else rates.get((day(row['Timestamp']), currency))
+            if not amount.is_finite() or amount < 0 or rate is None:
+                return None
+            rate = Decimal(str(rate))
+            if not rate.is_finite() or rate <= 0:
+                return None
+            total += amount * rate
+        except (InvalidOperation, TypeError, KeyError, ValueError):
+            return None
+    return str(total)