@stupify/cli 0.0.3 → 0.0.4

package/src/render.ts

@@ -1,44 +1,42 @@
 import { VERSION } from "./constants.ts";
-import type { AnalysisReport, AnalyzeCommand } from "./types.ts";
+import type { SearchCommand, SearchRunJson } from "./types.ts";
 
-export function renderReport(report: AnalysisReport, command: AnalyzeCommand): string {
-  if (command.json) {
-    return JSON.stringify({
-      schemaVersion: "0.4",
-      model: { id: report.run.modelId },
-      checks: report.run.checkIds,
-      run: report.run,
-      findings: report.result.findings,
-      summary: report.result.summary,
-    }, null, 2);
+export function renderSearchRun(run: SearchRunJson, command: SearchCommand): string {
+  if (command.json) return JSON.stringify(run, null, 2);
+
+  if (run.stats.skipped && run.stats.skipReason === "input_too_large") {
+    return `🧙 stupify 🪄
+Search input is too large for precise local search.
+Size:
+~${run.stats.inputTokens ?? "unknown"} tokens
+Limit:
+${run.stats.inputTokenCap ?? "unknown"} tokens
+Stupify skipped the search rather than review truncated context.
+Nothing was blocked.
+Try:
+stupify ${sourceHint(command)} --max-search-input-tokens ${Math.max((run.stats.inputTokens ?? 12_000) + 1, (run.stats.inputTokenCap ?? 12_000) * 2)}`;
+  }
+
+  if (run.stats.skipped && run.stats.skipReason === "no_candidates") {
+    return `🧙 stupify 🪄
+Search complete.
+Patterns: ${run.patterns.join(", ")}
+No search targets found.`;
   }
 
-  if (report.run.engine === "sem") {
-    return `Search:
-  ${report.run.entitiesScanned} entities scanned
-  ${report.run.candidateCount} candidate entities found
-Audit:
-  ${report.run.auditedCandidateCount} candidates inspected
-  ${report.result.findings.length} findings
-${renderAuditStats(report)}
-${renderWarnings(report)}
-Findings:
-${renderFindings(report)}
-Timing:
-  total_ms=${report.run.timingsMs.total} entity_diff_ms=${report.run.timingsMs.diff} model_ms=${report.run.timingsMs.modelLoad} scout_ms=${report.run.timingsMs.search} context_audit_ms=${report.run.timingsMs.audit}`;
+  if (run.matches.length === 0) {
+    return `🧙 stupify 🪄
+Search complete.
+Patterns: ${run.patterns.join(", ")}
+No judgment-offload signals found.`;
   }
 
-  return `Search:
-  ${report.run.batchesScanned} batches scanned
-  ${report.run.candidateCount} candidate regions found
-Audit:
-  ${report.run.auditedCandidateCount} candidates inspected
-  ${report.result.findings.length} findings
-${renderWarnings(report)}
-Findings:
-${renderFindings(report)}
-Timing:
-  total_ms=${report.run.timingsMs.total} diff_ms=${report.run.timingsMs.diff} model_ms=${report.run.timingsMs.modelLoad} search_ms=${report.run.timingsMs.search} audit_ms=${report.run.timingsMs.audit}`;
+  return `🧙 stupify 🪄
+Possible judgment-offload detected:
+${run.matches.map((match, index) => `${index + 1}. ${match.patternId}
+   ${match.reason}
+   Proof: ${match.proof}`).join("\n")}
+Search mode is warn-only.`;
 }
 
 export function helpText(): string {
@@ -49,59 +47,50 @@ Usage:
   stupify --since "2 weeks ago"
   stupify --commit <commit>
   stupify --commits <count>
-  stupify experiment <config.json>
+  stupify --staged
+  stupify --mode search --staged
+  stupify hook install|uninstall|status
+  stupify doctor
+  stupify bench search experiments/search-bench.json
   git diff HEAD~1..HEAD | stupify --stdin
 
 Options:
-  --since <date>        Analyze the net diff from the first commit before this git date to HEAD.
-  --commit <commit>     Analyze one commit as a net diff.
-  --commits <count>     Analyze the net diff across the last N non-merge commits.
-  --stdin               Read a git diff from stdin.
-  --engine <engine>     raw-diff or sem. Default: raw-diff.
-  --scout <mode>        llm or counter for --engine sem. Default: counter.
-  --audit-context <mode>
-                         none or repomix for --engine sem. Default: repomix.
-  --audit-prompt <name> strict or high_bar for --engine sem. Default: high_bar.
-  --debug-sem           Print sem commands and stderr.
-  --debug-targets       Include audited sem target details in JSON output.
-  --max-candidates <n>  Max semantic candidates for --engine sem. Default: 10.
-  --audit-batch-size <n>
-                         Semantic candidates per audit model call. Default: 25.
-  --max-audit-input-tokens <n>
-                         Max findings-audit input tokens before splitting. Default: 20000.
-  --audit-concurrency <n>
-                         Parallel findings-audit model calls. Default: 2.
-  --checks <ids>        Comma-separated check ids.
-  --model <id>          gemma-4-e2b, gemma-4-e4b, gemma-4-26b-a4b, qwen3-4b-magicquant, qwen2.5-coder-1.5b, qwen2.5-coder-7b, or qwen2.5-coder-32b.
-  --json                Print JSON only.
+  --staged             Search staged changes.
+  --mode <mode>        search. Search is the only analysis mode.
+  --since <date>       Search the net diff from the first commit before this git date to HEAD.
+  --commit <commit>    Search one commit as a net diff.
+  --commits <count>    Search the net diff across the last N non-merge commits.
+  --stdin              Read a git diff from stdin.
+  --debug-sem          Print sem commands and stderr.
+  --max-candidates <n> Max semantic search targets. Default: 10.
+  --max-search-input-tokens <n>
+                        Max search input tokens before skipping. Default: 12000.
+  --checks <ids>       Comma-separated pattern ids.
+  --model <id>         gemma-4-e2b, gemma-4-e4b, gemma-4-26b-a4b, qwen3-4b-magicquant, qwen2.5-coder-1.5b, qwen2.5-coder-7b, or qwen2.5-coder-32b.
+  --search-profile <path>
+                       Dev/bench-only search profile override.
+  --include-counter-reason-in-prompt
+                       Debug/bench-only: include counter reason in the model prompt.
+  --json               Print JSON only.
+
+Diagnostics:
+  stupify doctor       Check local setup, hook status, and privacy boundary.
 
 Default:
   stupify is equivalent to stupify --since "2 weeks ago".
 
-Not included:
-  Baselines, sharing, hosted server calls, Ollama, GitHub, dashboards, or repo-wide scanning.
-`;
-}
-
-function renderFindings(report: AnalysisReport): string {
-  if (report.result.findings.length === 0) return "  None.";
+Pipeline:
+  sem diff -> counter scout -> Repomix context -> local search model.
 
-  return report.result.findings
-    .map((finding) => `- ${finding.checkId}
-  ${finding.why}
-  Proof: ${finding.proof}`)
-    .join("\n");
-}
-
-function renderWarnings(report: AnalysisReport): string {
-  if (report.run.warnings.length === 0) return "";
-  return `Warnings:
-${report.run.warnings.map((warning) => `  ${warning}`).join("\n")}
+Not included:
+  Findings audit, validators, judges, baselines, sharing, hosted server calls, GitHub, dashboards, or repo-wide crawling.
 `;
 }
 
-function renderAuditStats(report: AnalysisReport): string {
-  const stats = report.run.auditStats;
-  if (!stats) return "";
-  return `  ${stats.totalTargets} targets reviewed: ${stats.finding} finding, ${stats.uncertain} uncertain, ${stats.clean} clean, ${stats.invalid} invalid`;
+function sourceHint(command: SearchCommand): string {
+  if (command.kind === "staged") return "--staged";
+  if (command.kind === "since") return `--since "${command.since}"`;
+  if (command.kind === "commit") return `--commit ${command.commit}`;
+  if (command.kind === "commits") return `--commits ${command.count}`;
+  return "--stdin";
 }

package/src/repomix-provider.ts

@@ -2,18 +2,20 @@ import { mkdtemp, readFile, rm, stat } from "node:fs/promises";
 import { tmpdir } from "node:os";
 import path from "node:path";
 import { pack, setLogLevel } from "repomix";
-import type { SemCandidate, SemChange, SemContext, SemContextPack } from "./types.ts";
+import type { RepomixSearchConfig, SemCandidate, SemChange, SemContext, SemContextPack } from "./types.ts";
 
 const MAX_PACK_FILE_SIZE_BYTES = 48 * 1024;
 const MAX_PACK_TOTAL_SIZE_BYTES = 128 * 1024;
 
 export function emptyContextPack(): SemContextPack {
+  const config = repomixSearchConfig();
   return {
     provider: "repomix",
     filePaths: [],
     totalCharacters: 0,
     totalTokens: 0,
     text: "",
+    config,
   };
 }
 
@@ -21,10 +23,14 @@ export async function repomixContextPack(
   cwd: string,
   contexts: readonly SemContext[],
   changes: readonly SemChange[],
+  config = repomixSearchConfig(),
 ): Promise<SemContextPack> {
-  const filePaths = await candidateFilePaths(cwd, contexts, changes);
+  const filePaths = await candidateFilePaths(cwd, contexts, changes, config);
   if (filePaths.length === 0) {
-    return emptyContextPack();
+    return {
+      ...emptyContextPack(),
+      config,
+    };
   }
 
   setLogLevel(-1);
@@ -35,7 +41,7 @@ export async function repomixContextPack(
       [cwd],
       {
         cwd,
-        input: { maxFileSize: MAX_PACK_FILE_SIZE_BYTES },
+        input: { maxFileSize: config.maxFileSizeBytes },
         output: {
           filePath: outputPath,
           style: "xml",
@@ -44,10 +50,10 @@ export async function repomixContextPack(
           directoryStructure: false,
           files: true,
           removeComments: false,
-          removeEmptyLines: true,
-          compress: true,
+          removeEmptyLines: config.removeEmptyLines,
+          compress: config.compress,
           topFilesLength: 0,
-          showLineNumbers: true,
+          showLineNumbers: config.showLineNumbers,
           truncateBase64: true,
           copyToClipboard: false,
           includeFullDirectoryStructure: false,
@@ -65,7 +71,7 @@ export async function repomixContextPack(
           useGitignore: true,
           useDotIgnore: true,
           useDefaultPatterns: true,
-          customPatterns: [],
+          customPatterns: [...config.ignorePatterns],
         },
         security: { enableSecurityCheck: false },
         tokenCount: { encoding: "o200k_base" },
@@ -80,6 +86,7 @@ export async function repomixContextPack(
       totalCharacters: result.totalCharacters,
       totalTokens: result.totalTokens,
       text: await readFile(outputPath, "utf8"),
+      config,
     };
   } finally {
     await rm(tempDir, { recursive: true, force: true });
@@ -120,6 +127,7 @@ async function candidateFilePaths(
   cwd: string,
   contexts: readonly SemContext[],
   changes: readonly SemChange[],
+  config: RepomixSearchConfig,
 ): Promise<readonly string[]> {
   const byEntityId = new Map(changes.map((change) => [change.entityId, change.filePath]));
   const paths = contexts.flatMap((context) => context.filePath ?? byEntityId.get(context.entityId) ?? []);
@@ -127,15 +135,63 @@ async function candidateFilePaths(
   const selected = [];
   let totalBytes = 0;
   for (const filePath of safePaths) {
+    if (matchesAnyPattern(filePath, config.ignorePatterns)) continue;
     const bytes = await fileSize(cwd, filePath);
-    if (bytes === null || bytes > MAX_PACK_FILE_SIZE_BYTES) continue;
-    if (totalBytes + bytes > MAX_PACK_TOTAL_SIZE_BYTES) continue;
+    if (bytes === null || bytes > config.maxFileSizeBytes) continue;
+    if (totalBytes + bytes > config.maxTotalSizeBytes) continue;
     totalBytes += bytes;
     selected.push(filePath);
   }
   return selected;
 }
 
+export function repomixSearchConfig(): RepomixSearchConfig {
+  return {
+    compress: envBoolean("STUPIFY_REPOMIX_COMPRESS", true),
+    showLineNumbers: envBoolean("STUPIFY_REPOMIX_SHOW_LINE_NUMBERS", true),
+    removeEmptyLines: envBoolean("STUPIFY_REPOMIX_REMOVE_EMPTY_LINES", true),
+    maxFileSizeBytes: envInteger("STUPIFY_REPOMIX_MAX_FILE_BYTES", MAX_PACK_FILE_SIZE_BYTES),
+    maxTotalSizeBytes: envInteger("STUPIFY_REPOMIX_MAX_TOTAL_BYTES", MAX_PACK_TOTAL_SIZE_BYTES),
+    ignorePatterns: envList("STUPIFY_REPOMIX_IGNORE_PATTERNS"),
+  };
+}
+
+function envBoolean(name: string, fallback: boolean): boolean {
+  const value = process.env[name];
+  if (value === undefined || value === "") return fallback;
+  return /^(1|true|yes|on)$/i.test(value);
+}
+
+function envInteger(name: string, fallback: number): number {
+  const value = Number(process.env[name]);
+  return Number.isInteger(value) && value > 0 ? value : fallback;
+}
+
+function envList(name: string): readonly string[] {
+  return (process.env[name] ?? "")
+    .split(",")
+    .map((item) => item.trim())
+    .filter(Boolean);
+}
+
+function matchesAnyPattern(filePath: string, patterns: readonly string[]): boolean {
+  return patterns.some((pattern) => matchesPattern(filePath, pattern));
+}
+
+function matchesPattern(filePath: string, pattern: string): boolean {
+  if (pattern === filePath) return true;
+  if (!pattern.includes("*")) return false;
+  const escaped = pattern
+    .split("*")
+    .map(escapeRegExp)
+    .join(".*");
+  return new RegExp(`^${escaped}$`).test(filePath);
+}
+
+function escapeRegExp(value: string): string {
+  return value.replace(/[.*+?^${}()|[\]\\]/g, "\\$&");
+}
+
 function isSafeRelativeFilePath(value: string): boolean {
   if (!value || path.isAbsolute(value)) return false;
   const normalized = path.normalize(value);