superlab 0.1.9 → 0.1.10

package/lib/context.cjs

@@ -28,7 +28,7 @@ function readWorkflowLanguage(targetDir) {
 function extractValue(text, labels) {
   for (const label of labels) {
     const escaped = label.replace(/[.*+?^${}()|[\]\\]/g, "\\$&");
-    const regex = new RegExp(`^\\s*(?:[-\\d.]+\\s*)?${escaped}:\\s*(.+?)\\s*$`, "im");
+    const regex = new RegExp(`^\\s*(?:[-\\d.]+\\s*)?${escaped}:[ \\t]*([^\\n\\r]+?)[ \\t]*$`, "im");
     const match = text.match(regex);
     if (match && match[1]) {
       return match[1].trim();
@@ -53,6 +53,10 @@ function extractOpenQuestion(text) {
   return "";
 }
 
+function joinNonEmpty(parts, separator = "; ") {
+  return parts.filter(Boolean).join(separator);
+}
+
 function extractClaim(text) {
   const blocks = text
     .split(/\n(?=\d+\.\s)/)
@@ -67,6 +71,10 @@ function extractClaim(text) {
   return "";
 }
 
+function labelValue(text, englishLabels, chineseLabels = []) {
+  return extractValue(text, [...englishLabels, ...chineseLabels]);
+}
+
 function renderSummary(lang, data) {
   if (lang === "zh") {
     return `# 研究摘要
@@ -76,6 +84,15 @@ function renderSummary(lang, data) {
 - Mission: ${data.problem || "待补充"}
 - Approved direction: ${data.direction || "待补充"}
 - Active stage: ${data.stage || "待补充"}
+- Dataset package: ${data.datasetPackage || "待补充"}
+- Dataset years: ${data.datasetYears || "待补充"}
+- Benchmark role: ${data.benchmarkRole || "待补充"}
+- Benchmark rationale: ${data.benchmarkRationale || "待补充"}
+- Comparison suite: ${data.comparisonMethods || "待补充"}
+- Comparison rationale: ${data.comparisonRationale || "待补充"}
+- Comparison evidence: ${data.comparisonEvidence || "待补充"}
+- Source basis: ${data.datasetSources || "待补充"}
+- Representative papers: ${data.datasetPapers || "待补充"}
 
 ## 最强证据
 
@@ -98,6 +115,15 @@ function renderSummary(lang, data) {
 - Mission: ${data.problem || "TBD"}
 - Approved direction: ${data.direction || "TBD"}
 - Active stage: ${data.stage || "TBD"}
+- Dataset package: ${data.datasetPackage || "TBD"}
+- Dataset years: ${data.datasetYears || "TBD"}
+- Benchmark role: ${data.benchmarkRole || "TBD"}
+- Benchmark rationale: ${data.benchmarkRationale || "TBD"}
+- Comparison suite: ${data.comparisonMethods || "TBD"}
+- Comparison rationale: ${data.comparisonRationale || "TBD"}
+- Comparison evidence: ${data.comparisonEvidence || "TBD"}
+- Source basis: ${data.datasetSources || "TBD"}
+- Representative papers: ${data.datasetPapers || "TBD"}
 
 ## Strongest Evidence
 
@@ -175,6 +201,15 @@ ${data.problem || "待补充"}
 
 - Approved direction: ${data.direction || "待补充"}
 - Why this is the active path: ${data.why || "当前已批准方向"}
+- Dataset package: ${data.datasetPackage || "待补充"}
+- Dataset years: ${data.datasetYears || "待补充"}
+- Benchmark role: ${data.benchmarkRole || "待补充"}
+- Benchmark rationale: ${data.benchmarkRationale || "待补充"}
+- Comparison suite: ${data.comparisonMethods || "待补充"}
+- Comparison rationale: ${data.comparisonRationale || "待补充"}
+- Comparison evidence: ${data.comparisonEvidence || "待补充"}
+- Source basis: ${data.datasetSources || "待补充"}
+- Representative papers: ${data.datasetPapers || "待补充"}
 
 ## 主要风险
 
@@ -208,6 +243,15 @@ ${data.problem || "TBD"}
 
 - Approved direction: ${data.direction || "TBD"}
 - Why this is the active path: ${data.why || "This is the approved direction"}
+- Dataset package: ${data.datasetPackage || "TBD"}
+- Dataset years: ${data.datasetYears || "TBD"}
+- Benchmark role: ${data.benchmarkRole || "TBD"}
+- Benchmark rationale: ${data.benchmarkRationale || "TBD"}
+- Comparison suite: ${data.comparisonMethods || "TBD"}
+- Comparison rationale: ${data.comparisonRationale || "TBD"}
+- Comparison evidence: ${data.comparisonEvidence || "TBD"}
+- Source basis: ${data.datasetSources || "TBD"}
+- Representative papers: ${data.datasetPapers || "TBD"}
 
 ## Main Risk
 
@@ -232,6 +276,62 @@ function buildContextSnapshot(targetDir) {
   const state = readFileIfExists(contextFile(targetDir, "state.md"));
   const evidence = readFileIfExists(contextFile(targetDir, "evidence-index.md"));
   const questions = readFileIfExists(contextFile(targetDir, "open-questions.md"));
+  const dataDecisions = readFileIfExists(contextFile(targetDir, "data-decisions.md"));
+  const classicBenchmarks = labelValue(
+    dataDecisions,
+    ["Classic public benchmarks", "Classic benchmarks"],
+    ["经典公开 benchmark", "经典 benchmark"]
+  );
+  const recentBenchmarks = labelValue(
+    dataDecisions,
+    ["Recent strong public benchmarks", "Latest benchmarks"],
+    ["近期强公开 benchmark", "最新 benchmark"]
+  );
+  const claimSpecificBenchmarks = labelValue(
+    dataDecisions,
+    ["Claim-specific benchmarks"],
+    ["专项 benchmark"]
+  );
+  const canonicalComparisons = labelValue(
+    dataDecisions,
+    ["Canonical baselines", "Classic comparison methods"],
+    ["规范基线", "经典对比方法"]
+  );
+  const strongHistoricalComparisons = labelValue(
+    dataDecisions,
+    ["Strong historical baselines"],
+    ["强历史基线"]
+  );
+  const recentComparisons = labelValue(
+    dataDecisions,
+    ["Recent strong public methods", "Recent strong or SOTA comparison methods"],
+    ["近期强公开方法", "近期强基线或 SOTA 对比方法"]
+  );
+  const closestPriorWork = labelValue(
+    dataDecisions,
+    ["Closest prior work"],
+    ["最接近的前作"]
+  );
+  const canonicalComparisonPapers = labelValue(
+    dataDecisions,
+    ["Representative papers for canonical baselines"],
+    ["规范基线的代表论文"]
+  );
+  const strongHistoricalComparisonPapers = labelValue(
+    dataDecisions,
+    ["Representative papers for strong historical baselines"],
+    ["强历史基线的代表论文"]
+  );
+  const recentComparisonPapers = labelValue(
+    dataDecisions,
+    ["Representative papers for recent strong public methods"],
+    ["近期强公开方法的代表论文"]
+  );
+  const closestPriorPapers = labelValue(
+    dataDecisions,
+    ["Representative papers for closest prior work"],
+    ["最接近前作的代表论文"]
+  );
 
   return {
     problem: extractValue(mission, ["One-sentence problem", "一句话问题"]),
@@ -246,6 +346,128 @@ function buildContextSnapshot(targetDir) {
     claim: extractClaim(evidence),
     question: extractOpenQuestion(questions),
     risk: extractValue(questions, ["Why it matters", "为什么重要"]) || extractValue(state, ["Current blocker", "当前阻塞"]),
+    datasetPackage:
+      extractValue(dataDecisions, ["Approved dataset package", "已批准数据集包"]) ||
+      extractValue(dataDecisions, ["Approved datasets", "已批准数据集"]),
+    datasetYears: extractValue(dataDecisions, ["Dataset years", "数据集年份"]),
+    datasetPapers: extractValue(dataDecisions, ["Papers that used the approved datasets", "使用过已批准数据集的论文", "使用过该数据集的论文"]),
+    datasetSources: joinNonEmpty(
+      [
+        labelValue(dataDecisions, ["Why the selected download sources are trusted"], ["为什么这些下载来源可信"]),
+        labelValue(dataDecisions, ["Official benchmark or dataset pages"], ["官方 benchmark 或数据集页面"]),
+      ],
+      " | "
+    ),
+    benchmarkRationale: joinNonEmpty(
+      [
+        (() => {
+          const value = labelValue(
+            dataDecisions,
+            ["Why classic public benchmarks are included", "Why classic benchmarks are included"],
+            ["为什么纳入经典公开 benchmark", "为什么纳入经典 benchmark"]
+          );
+          return value ? `Classic public: ${value}` : "";
+        })(),
+        (() => {
+          const value = labelValue(
+            dataDecisions,
+            ["Why recent strong public benchmarks are included", "Why latest benchmarks are included"],
+            ["为什么纳入近期强公开 benchmark", "为什么纳入最新 benchmark"]
+          );
+          return value ? `Recent strong public: ${value}` : "";
+        })(),
+        (() => {
+          const value = labelValue(
+            dataDecisions,
+            ["Why claim-specific benchmarks are included or why none are needed"],
+            ["为什么纳入专项 benchmark 或为什么不需要"]
+          );
+          return value ? `Claim-specific: ${value}` : "";
+        })(),
+        (() => {
+          const value = labelValue(
+            dataDecisions,
+            ["Why this benchmark mix is representative"],
+            ["为什么这个 benchmark 组合具有代表性"]
+          );
+          return value ? `Mix: ${value}` : "";
+        })(),
+      ],
+      " | "
+    ),
+    comparisonMethods: joinNonEmpty(
+      [
+        canonicalComparisons ? `Canonical: ${canonicalComparisons}` : "",
+        strongHistoricalComparisons ? `Strong historical: ${strongHistoricalComparisons}` : "",
+        recentComparisons ? `Recent strong public: ${recentComparisons}` : "",
+        closestPriorWork ? `Closest prior: ${closestPriorWork}` : "",
+      ],
+      "; "
+    ),
+    comparisonEvidence: joinNonEmpty(
+      [
+        canonicalComparisonPapers ? `Canonical baselines: ${canonicalComparisonPapers}` : "",
+        strongHistoricalComparisonPapers
+          ? `Strong historical baselines: ${strongHistoricalComparisonPapers}`
+          : "",
+        recentComparisonPapers ? `Recent strong public methods: ${recentComparisonPapers}` : "",
+        closestPriorPapers ? `Closest prior work: ${closestPriorPapers}` : "",
+      ],
+      " | "
+    ),
+    comparisonRationale: joinNonEmpty(
+      [
+        (() => {
+          const value = labelValue(
+            dataDecisions,
+            ["Why these canonical baselines are included", "Why these classic baselines are included"],
+            ["为什么纳入这些规范基线", "为什么纳入这些经典基线"]
+          );
+          return value ? `Canonical baselines: ${value}` : "";
+        })(),
+        (() => {
+          const value = labelValue(
+            dataDecisions,
+            ["Why these strong historical baselines are included"],
+            ["为什么纳入这些强历史基线"]
+          );
+          return value ? `Strong historical baselines: ${value}` : "";
+        })(),
+        (() => {
+          const value = labelValue(
+            dataDecisions,
+            ["Why these recent strong public methods are included", "Why these recent or SOTA baselines are included"],
+            ["为什么纳入这些近期强公开方法", "为什么纳入这些近期或 SOTA 基线"]
+          );
+          return value ? `Recent strong public methods: ${value}` : "";
+        })(),
+        (() => {
+          const value = labelValue(
+            dataDecisions,
+            ["Why this closest prior work is included or why none qualifies"],
+            ["为什么纳入这个最接近的前作或为什么没有合格项"]
+          );
+          return value ? `Closest prior work: ${value}` : "";
+        })(),
+        (() => {
+          const value = labelValue(
+            dataDecisions,
+            ["Why these comparison methods are fair"],
+            ["为什么这些对比方法是公平的"]
+          );
+          return value ? `Fairness: ${value}` : "";
+        })(),
+      ],
+      " | "
+    ),
+    benchmarkRole: joinNonEmpty(
+      [
+        classicBenchmarks ? `Classic public: ${classicBenchmarks}` : "",
+        recentBenchmarks ? `Recent strong public: ${recentBenchmarks}` : "",
+        claimSpecificBenchmarks ? `Claim-specific: ${claimSpecificBenchmarks}` : "",
+      ],
+      "; "
+    ),
   };
 }