circle-ir 3.181.0 → 3.185.0

package/dist/browser/circle-ir.js

@@ -10020,6 +10020,10 @@ function processGoBlock(node, defs, uses, scopeStack, counters) {
     } else if (child.type === "for_statement") {
       const rangeClause = findChildByTypeGo(child, "range_clause");
       if (rangeClause) {
+        const right = rangeClause.childForFieldName("right");
+        if (right) {
+          extractGoUses(right, uses, scopeStack);
+        }
         const left = rangeClause.childForFieldName("left");
         if (left) {
           extractGoLhsDefs(left, defs, scopeStack, child.startPosition.row + 1);
@@ -10027,6 +10031,7 @@ function processGoBlock(node, defs, uses, scopeStack, counters) {
       }
     } else if (child.type === "call_expression") {
       extractGoUses(child, uses, scopeStack);
+      recordGoOpaqueCodecDestDef(child, defs, scopeStack);
     } else if (child.type === "return_statement") {
       for (let i2 = 0; i2 < child.childCount; i2++) {
         const expr = child.child(i2);
@@ -10119,6 +10124,63 @@ function findChildByTypeGo(node, type) {
   }
   return null;
 }
+var GO_OPAQUE_CODEC_METHODS = /* @__PURE__ */ new Set([
+  "Unmarshal",
+  // json/xml/yaml/toml/gob
+  "Decode",
+  // json.NewDecoder(r).Decode(&dest), gob.Decoder.Decode
+  "NewDecoder",
+  // wrapper; handled via chained Decode above
+  "UnmarshalYAML",
+  "UnmarshalJSON",
+  "UnmarshalText",
+  "UnmarshalBinary"
+]);
+function recordGoOpaqueCodecDestDef(call, defs, scopeStack) {
+  const fn = call.childForFieldName("function");
+  if (!fn || fn.type !== "selector_expression") return;
+  const fieldNode = fn.childForFieldName("field");
+  if (!fieldNode) return;
+  const method = getNodeText(fieldNode);
+  if (!GO_OPAQUE_CODEC_METHODS.has(method)) return;
+  const argsNode = call.childForFieldName("arguments");
+  if (!argsNode) return;
+  const args2 = [];
+  for (let i2 = 0; i2 < argsNode.childCount; i2++) {
+    const c = argsNode.child(i2);
+    if (!c) continue;
+    if (c.type === "," || c.type === "(" || c.type === ")") continue;
+    args2.push(c);
+  }
+  if (args2.length === 0) return;
+  const destArg = args2.length >= 2 ? args2[1] : args2[0];
+  const destName = extractGoAddressableVarName(destArg);
+  if (!destName || destName === "_") return;
+  const line = call.startPosition.row + 1;
+  const def = {
+    id: defs.length + 1,
+    variable: destName,
+    kind: "local",
+    line
+  };
+  defs.push(def);
+  currentScope(scopeStack).set(destName, def.id);
+}
+function extractGoAddressableVarName(node) {
+  if (node.type === "unary_expression") {
+    const operand = node.childForFieldName("operand") ?? node.child(1) ?? null;
+    if (operand) return extractGoAddressableVarName(operand);
+    return null;
+  }
+  if (node.type === "identifier") {
+    return getNodeText(node);
+  }
+  if (node.type === "selector_expression") {
+    const field = node.childForFieldName("field");
+    if (field) return getNodeText(field);
+  }
+  return null;
+}
 
 // src/core/extractors/runtime-registrations.ts
 var HTTP_VERB_METHODS = /* @__PURE__ */ new Set([
@@ -11264,6 +11326,49 @@ var DEFAULT_SOURCES = [
   // Server-side interceptor receives `Metadata headers`; `headers.get(KEY)`
   // returns caller-supplied header values.
   { method: "get", class: "Metadata", type: "http_header", severity: "high", return_tainted: true, languages: ["java"] },
+  // --- WebSocket transport channels (cognium-dev #213 second slice) ---
+  //
+  // Server-side WebSocket handlers receive attacker-authored frames on
+  // every call to a receive-shaped method. Untrusted the moment they
+  // return, regardless of any application-level auth on the socket.
+  //
+  // Python — FastAPI / Starlette (`from fastapi import WebSocket`):
+  //   data = await websocket.receive_text()
+  //   data = await websocket.receive_bytes()
+  //   data = await websocket.receive_json()   # parsed dict/list
+  //   data = await websocket.receive()        # {'type', 'text'|'bytes'}
+  { method: "receive_text", class: "WebSocket", type: "network_input", severity: "high", return_tainted: true, languages: ["python"] },
+  { method: "receive_bytes", class: "WebSocket", type: "network_input", severity: "high", return_tainted: true, languages: ["python"] },
+  { method: "receive_json", class: "WebSocket", type: "http_body", severity: "high", return_tainted: true, languages: ["python"] },
+  //
+  // `receive` intentionally class-scoped to WebSocket (Starlette pattern).
+  // Broadening to unqualified would collide with queue/signal `.receive()`.
+  { method: "receive", class: "WebSocket", type: "network_input", severity: "high", return_tainted: true, languages: ["python"] },
+  // Django Channels `AsyncJsonWebsocketConsumer` / `WebsocketConsumer`
+  // expose the same receive_* names on `self`; the class filter matches
+  // `WebSocket` only, so add unqualified fallbacks for the receive_json /
+  // receive_text convention (broad — no class filter possible without
+  // hardcoding Django's consumer names).
+  { method: "receive_json", type: "http_body", severity: "high", return_tainted: true, languages: ["python"] },
+  { method: "receive_text", type: "network_input", severity: "high", return_tainted: true, languages: ["python"] },
+  { method: "receive_bytes", type: "network_input", severity: "high", return_tainted: true, languages: ["python"] },
+  // Go — gorilla/websocket (`github.com/gorilla/websocket`):
+  //   messageType, message, err := conn.ReadMessage()
+  //   _, r, err := conn.NextReader()
+  { method: "ReadMessage", class: "Conn", type: "network_input", severity: "high", return_tainted: true, languages: ["go"] },
+  { method: "NextReader", class: "Conn", type: "network_input", severity: "high", return_tainted: true, languages: ["go"] },
+  // nhooyr.io/websocket exposes a `Read` method on `*websocket.Conn`;
+  // signature is `func (c *Conn) Read(ctx) (MessageType, []byte, error)`.
+  { method: "Read", class: "Conn", type: "network_input", severity: "high", return_tainted: true, languages: ["go"] },
+  // Java — Jakarta / Java WebSocket API (`javax.websocket` / `jakarta.websocket`):
+  //   session.getBasicRemote().sendText(input);      // OUT (not a source)
+  //   @OnMessage public void onMessage(String msg) — `msg` is a callback
+  //   param, not a return; handled via param_tainted when annotation is set.
+  { annotation: "OnMessage", type: "network_input", severity: "high", param_tainted: true, languages: ["java"] },
+  // Spring `@MessageMapping` STOMP-over-WebSocket handlers receive
+  // untrusted payloads on every dispatched frame.
+  { annotation: "MessageMapping", type: "http_body", severity: "high", param_tainted: true, languages: ["java"] },
+  { annotation: "SubscribeMapping", type: "http_body", severity: "high", param_tainted: true, languages: ["java"] },
   // --- Cache reads (second-order taint — Redis / Memcached / Django cache) ---
   // The cache round-trip is a canonical second-order sink: whatever was
   // written previously (potentially attacker-controlled) resurfaces on read.
@@ -13336,6 +13441,60 @@ var DEFAULT_SANITIZERS = [
   // Python Type coercion
   { method: "int", removes: ["sql_injection", "command_injection", "xss"] },
   { method: "float", removes: ["sql_injection", "command_injection"] },
+  // Python URL encoding (cognium-dev #213 fifth slice).
+  //
+  //   urllib.parse.quote(x)      — RFC-3986 percent-encode; safe for URL path
+  //   urllib.parse.quote_plus(x) — same + space→+; safe for query string
+  //   urllib.parse.urlencode(d)  — encode a dict of pairs
+  //
+  // Bounded to URL-context sinks. Class-scoped to the specific
+  // `urllib.parse` receiver to avoid colliding with unrelated bare
+  // `quote(...)` calls in other libraries.
+  { method: "quote", class: "urllib.parse", removes: ["ssrf", "open_redirect", "xss", "path_traversal"] },
+  { method: "quote_plus", class: "urllib.parse", removes: ["ssrf", "open_redirect", "xss", "path_traversal"] },
+  { method: "urlencode", class: "urllib.parse", removes: ["ssrf", "open_redirect", "xss"] },
+  // Bare aliases — `from urllib.parse import quote` then unqualified.
+  { method: "quote_plus", removes: ["ssrf", "open_redirect", "xss", "path_traversal"] },
+  { method: "urlencode", removes: ["ssrf", "open_redirect", "xss"] },
+  // `quote` bare is intentionally NOT registered — it collides with
+  // shlex.quote (bare-imported) which is a command_injection sanitizer,
+  // not a URL sanitizer. The class-scoped variant above catches the
+  // qualified `urllib.parse.quote(...)` shape; unqualified callers who
+  // want URL-context credit should use `quote_plus` (which does not
+  // collide with any command_injection sanitizer).
+  // Python XSS — additional common sanitizers.
+  //
+  //   bleach.clean(...)   — already covered above (line 2898)
+  //   bleach.linkify(...) — turns URLs into anchor tags; sanitizes as well
+  //   django.utils.html.escape / strip_tags — Django's XSS escape helpers
+  //   jinja2.escape       — Jinja2's Markup escape (aliased from markupsafe)
+  //   flask.escape        — Flask re-export of markupsafe.escape
+  //   saxutils.escape     — stdlib xml.sax.saxutils.escape for XML docs
+  { method: "linkify", class: "bleach", removes: ["xss"] },
+  // Bare `linkify(...)` alias — `from bleach import linkify`.
+  { method: "linkify", removes: ["xss"] },
+  { method: "escape", class: "django.utils.html", removes: ["xss"] },
+  { method: "strip_tags", class: "django.utils.html", removes: ["xss"] },
+  { method: "escape", class: "jinja2", removes: ["xss"] },
+  { method: "escape", class: "flask", removes: ["xss"] },
+  { method: "escape", class: "saxutils", removes: ["xss"] },
+  { method: "escape", class: "xml.sax.saxutils", removes: ["xss"] },
+  { method: "quoteattr", class: "saxutils", removes: ["xss"] },
+  { method: "quoteattr", class: "xml.sax.saxutils", removes: ["xss"] },
+  // Python ReDoS — `re.escape(user)` when building a regex from user input
+  // strips regex metacharacters. Downstream `re.compile / re.match` cannot
+  // interpret user-supplied alternations or quantifiers. Also covers the
+  // `code_injection` categorization that `re.compile` currently emits
+  // (a re.escape-wrapped pattern cannot execute anything, so both are safe).
+  { method: "escape", class: "re", removes: ["redos", "code_injection"] },
+  // Python SQLAlchemy — `text(...).bindparams(...)` binds params safely.
+  // The `bindparams` call is the sanitizer; the parent `text` wraps the
+  // template. Also add `expression.literal` for explicit SQL literals.
+  { method: "bindparams", removes: ["sql_injection"] },
+  // psycopg2 sql-composition helpers
+  { method: "Identifier", class: "sql", removes: ["sql_injection"] },
+  { method: "Literal", class: "sql", removes: ["sql_injection"] },
+  { method: "Placeholder", class: "sql", removes: ["sql_injection"] },
   // =========================================================================
   // Rust Sanitizers
   // =========================================================================
@@ -15259,7 +15418,14 @@ function matchesSanitizerPattern(call, pattern) {
       return false;
     }
     if (pattern.class) {
-      if (!call.receiver || !receiverMightBeClass(call.receiver, pattern.class)) {
+      if (!call.receiver) {
+        const target = call.resolution?.target;
+        const expectedTail = `${pattern.class}.${pattern.method}`;
+        if (target && (target === expectedTail || target.endsWith("." + expectedTail))) {
+        } else {
+          return false;
+        }
+      } else if (!receiverMightBeClass(call.receiver, pattern.class)) {
         return false;
       }
     }
@@ -17002,15 +17168,21 @@ function propagateTaint(graphOrDfg, callsOrSources, sourcesOrSinks, sinksOrSanit
     const callsAtSink = callsByLine.get(sink.line) ?? [];
     for (const call of callsAtSink) {
       for (const arg of call.arguments) {
-        if (arg.variable) {
-          if (sink.argPositions && sink.argPositions.length > 0) {
-            if (!sink.argPositions.includes(arg.position)) {
-              continue;
-            }
+        if (sink.argPositions && sink.argPositions.length > 0) {
+          if (!sink.argPositions.includes(arg.position)) {
+            continue;
           }
-          for (const use of usesAtSink) {
-            if (use.variable === arg.variable && use.def_id !== null) {
+        }
+        const candidateUses = arg.variable ? usesAtSink.filter((u) => u.variable === arg.variable) : usesAtSink;
+        {
+          for (const use of candidateUses) {
+            if (use.def_id !== null) {
               if (allTaintedDefIds.has(use.def_id)) {
+                if (!arg.variable) {
+                  if (typeof arg.expression !== "string" || arg.expression.length === 0) continue;
+                  const re = new RegExp(`(?:^|[^A-Za-z0-9_$])${use.variable.replace(/[.*+?^${}()|[\]\\]/g, "\\$&")}(?:[^A-Za-z0-9_$]|$)`);
+                  if (!re.test(arg.expression)) continue;
+                }
                 const taintInfo = taintByDefId.get(use.def_id);
                 if (taintInfo) {
                   const isSanitized = checkSanitized(
@@ -17048,7 +17220,15 @@ function propagateTaint(graphOrDfg, callsOrSources, sourcesOrSinks, sinksOrSanit
       }
     }
   }
-  return { taintedVars, flows, reachableSinks };
+  const seen = /* @__PURE__ */ new Set();
+  const deduped = [];
+  for (const f of flows) {
+    const key = `${f.source.line}|${f.sink.line}|${f.sink.type}|${f.path.map((p) => p.variable).join(">")}`;
+    if (seen.has(key)) continue;
+    seen.add(key);
+    deduped.push(f);
+  }
+  return { taintedVars, flows: deduped, reachableSinks };
 }
 function findInitialTaint(sources, callsByLine, defsByLine) {
   const tainted = [];
@@ -25571,7 +25751,22 @@ var PYTHON_TAINTED_PATTERNS2 = [
   // configs/sources/python.json but was not in the forward-taint regex
   // registry, so `name = input()` was not added to pyTaintedVars. Closes
   // the deferred `getattr(obj, input())()` reflection-invocation shape.
-  { pattern: /\binput\s*\(/, type: "io_input" }
+  { pattern: /\binput\s*\(/, type: "io_input" },
+  // WebSocket transport channels (cognium-dev #213 second slice).
+  // FastAPI / Starlette (`from fastapi import WebSocket`) and Django
+  // Channels consumers both expose `receive_*` methods that return
+  // untrusted frame payloads on every call. Registered in
+  // config-loader.ts as return-tainted; the forward-taint regex here
+  // enables `data = await websocket.receive_text()` → `data` tainted,
+  // which the DFG-less Python path needs so downstream sinks that
+  // consume `data` are flagged.
+  //
+  // Deliberately not adding a bare `.receive\s*\(` here — that would
+  // match too many unrelated APIs (queue receive, signal receive, etc.)
+  // and produce spurious sources on every `x = q.receive()` in the wild.
+  { pattern: /\.receive_text\s*\(/, type: "network_input" },
+  { pattern: /\.receive_bytes\s*\(/, type: "network_input" },
+  { pattern: /\.receive_json\s*\(/, type: "http_body" }
 ];
 var LanguageSourcesPass = class {
   name = "language-sources";
@@ -26263,6 +26458,31 @@ function findJavaScriptAssignmentSources(sourceCode, language) {
       }
     }
   }
+  sources.push(...findJavaScriptCallbackParamSources(sourceCode));
+  return sources;
+}
+function findJavaScriptCallbackParamSources(sourceCode) {
+  const sources = [];
+  const CB_EVENT_RE = /\.on\s*\(\s*['"](?:message|text|binary)['"]\s*,\s*(?:async\s+)?(?:function\s*\*?\s*(?:\w+\s*)?\(\s*(\w+)|\(\s*(\w+)\s*(?:[,):]|$)|(\w+)\s*=>)/;
+  const lines = sourceCode.split("\n");
+  for (let i2 = 0; i2 < lines.length; i2++) {
+    const line = lines[i2];
+    const trimmed = line.trimStart();
+    if (trimmed.startsWith("//") || trimmed.startsWith("*")) continue;
+    const m = line.match(CB_EVENT_RE);
+    if (!m) continue;
+    const paramName = m[1] ?? m[2] ?? m[3];
+    if (!paramName) continue;
+    if (paramName === "err" || paramName === "error") continue;
+    sources.push({
+      type: "network_input",
+      location: `WebSocket .on(...) callback param '${paramName}' at line ${i2 + 1}`,
+      severity: "high",
+      line: i2 + 1,
+      confidence: 0.95,
+      variable: paramName
+    });
+  }
   return sources;
 }
 function findPythonAssignmentSources(sourceCode, language) {
@@ -26545,6 +26765,14 @@ function buildJavaScriptTaintedVars(sourceCode, language) {
   if (!["javascript", "typescript"].includes(language)) return /* @__PURE__ */ new Map();
   const tainted = /* @__PURE__ */ new Map();
   const lines = sourceCode.split("\n");
+  const CB_EVENT_RE = /\.on\s*\(\s*['"](?:message|text|binary)['"]\s*,\s*(?:async\s+)?(?:function\s*\*?\s*(?:\w+\s*)?\(\s*(\w+)|\(\s*(\w+)\s*(?:[,):]|$)|(\w+)\s*=>)/;
+  for (let i2 = 0; i2 < lines.length; i2++) {
+    const m = lines[i2].match(CB_EVENT_RE);
+    if (!m) continue;
+    const paramName = m[1] ?? m[2] ?? m[3];
+    if (!paramName || paramName === "err" || paramName === "error") continue;
+    tainted.set(paramName, i2 + 1);
+  }
   for (let i2 = 0; i2 < lines.length; i2++) {
     const line = lines[i2];
     const trimmed = line.trimStart();
@@ -26741,6 +26969,79 @@ function findBashTaintSources(sourceCode, dfg) {
         });
       }
     }
+    const readMatch = /^read\s*\(/.test(trimmed) ? null : trimmed.match(/^read\b([^#(]*)$/);
+    if (readMatch) {
+      const argStr = readMatch[1].trim();
+      const noRedirect = argStr.replace(/\s*<[^<].*$/, "").trim();
+      const tokens = noRedirect.split(/\s+/);
+      const varNames = [];
+      for (let ti = 0; ti < tokens.length; ti++) {
+        const tok = tokens[ti];
+        if (!tok) continue;
+        if (tok.startsWith("-")) {
+          if (/^-[antpidu]$|^-N$/.test(tok)) ti++;
+          continue;
+        }
+        if (/^[A-Za-z_][\w]*$/.test(tok)) varNames.push(tok);
+      }
+      if (varNames.length === 0) varNames.push("REPLY");
+      for (const v of varNames) {
+        const already = sources.some((s) => s.line === lineNumber && s.variable === v);
+        if (already) continue;
+        sources.push({
+          type: "io_input",
+          location: `read \u2192 $${v} (stdin)`,
+          severity: "high",
+          line: lineNumber,
+          confidence: 0.9,
+          variable: v
+        });
+      }
+    }
+    const mapfileMatch = /^(?:mapfile|readarray)\s*\(/.test(trimmed) ? null : trimmed.match(/^(?:mapfile|readarray)\b([^#(]*)$/);
+    if (mapfileMatch) {
+      const argStr = mapfileMatch[1].trim();
+      const tokens = argStr.split(/\s+/);
+      const varNames = [];
+      for (let ti = 0; ti < tokens.length; ti++) {
+        const tok = tokens[ti];
+        if (!tok) continue;
+        if (tok.startsWith("-")) {
+          if (/^-[cCnOsud]$/.test(tok)) ti++;
+          continue;
+        }
+        if (/^[A-Za-z_][\w]*$/.test(tok)) varNames.push(tok);
+      }
+      if (varNames.length === 0) varNames.push("MAPFILE");
+      for (const v of varNames) {
+        const already = sources.some((s) => s.line === lineNumber && s.variable === v);
+        if (already) continue;
+        sources.push({
+          type: "io_input",
+          location: `mapfile \u2192 $${v} (stdin array)`,
+          severity: "high",
+          line: lineNumber,
+          confidence: 0.9,
+          variable: v
+        });
+      }
+    }
+    const getoptsMatch = trimmed.match(/\bgetopts\s+["'][^"']+["']\s+(\w+)/);
+    if (getoptsMatch) {
+      const flagVar = getoptsMatch[1];
+      for (const v of [flagVar, "OPTARG"]) {
+        const already = sources.some((s) => s.line === lineNumber && s.variable === v);
+        if (already) continue;
+        sources.push({
+          type: "io_input",
+          location: `getopts \u2192 $${v} (CLI arg)`,
+          severity: "high",
+          line: lineNumber,
+          confidence: 0.9,
+          variable: v
+        });
+      }
+    }
     const envRe = /\$([A-Z][A-Z0-9_]{2,})|\$\{([A-Z][A-Z0-9_]{2,})\}/g;
     let em;
     while ((em = envRe.exec(line)) !== null) {
@@ -33641,6 +33942,18 @@ var TaintPropagationPass = class {
       if (isFP) continue;
       pushIfNew(f);
     }
+    if (ctx.language === "go" && typeof ctx.code === "string") {
+      const goGlobalFlows = detectGoPackageGlobalFlows(
+        ctx.code,
+        calls,
+        sources,
+        sinks,
+        constProp.unreachableLines
+      ) ?? [];
+      for (const f of goGlobalFlows) {
+        pushIfNew(f);
+      }
+    }
     const sanitizedNames = constProp.sanitizedVars;
     let finalFlows = sanitizedNames.size === 0 ? flows : flows.filter((f) => {
       if (f.path.length === 0) return true;
@@ -33855,6 +34168,96 @@ function isInJavaSanitizedMethod(code, types, sinkLine, sinkType) {
   }
   return false;
 }
+function detectGoPackageGlobalFlows(code, _calls, sources, sinks, unreachableLines) {
+  const flows = [];
+  if (!code || sinks.length === 0) return flows;
+  const lines = code.split("\n");
+  const packageVars = /* @__PURE__ */ new Set();
+  let inVarBlock = false;
+  for (const line of lines) {
+    const stripped = line.replace(/\s+$/, "");
+    if (!stripped) continue;
+    if (/^[ \t]/.test(line)) {
+      if (inVarBlock) {
+        const m = stripped.match(/^\s*([A-Za-z_][A-Za-z0-9_]*)\s+/);
+        if (m) packageVars.add(m[1]);
+        if (/^\s*\)/.test(stripped)) inVarBlock = false;
+      }
+      continue;
+    }
+    if (/^var\s*\(/.test(stripped)) {
+      inVarBlock = true;
+      continue;
+    }
+    if (inVarBlock && /^\)/.test(stripped)) {
+      inVarBlock = false;
+      continue;
+    }
+    const varOne = stripped.match(/^var\s+([A-Za-z_][A-Za-z0-9_]*)\b/);
+    if (varOne) packageVars.add(varOne[1]);
+  }
+  if (packageVars.size === 0) return flows;
+  const sourceVarNames = /* @__PURE__ */ new Set();
+  for (const s of sources) {
+    if (typeof s.variable === "string" && s.variable.length > 0) {
+      sourceVarNames.add(s.variable);
+    }
+  }
+  const GO_SOURCE_RE = /\b(r\.(?:URL\.Query|Form|PostForm|Header|Cookie|Body)|mux\.Vars|c\.(?:Query|Param|PostForm|GetHeader|Cookie)|ctx\.(?:Query|Param|PostForm|GetHeader|Cookie))\b/;
+  const writes = [];
+  for (let i2 = 0; i2 < lines.length; i2++) {
+    const line = lines[i2];
+    if (unreachableLines.has(i2 + 1)) continue;
+    if (/:=/.test(line)) continue;
+    const m = line.match(/^\s*([A-Za-z_][A-Za-z0-9_]*)\s*=\s*(.+)$/);
+    if (!m) continue;
+    const [, lhs, rhs] = m;
+    if (!packageVars.has(lhs)) continue;
+    let taintedBy = null;
+    if (GO_SOURCE_RE.test(rhs)) {
+      const src = sources.find((s) => s.line === i2 + 1);
+      taintedBy = src ? { line: src.line, type: src.type } : { line: i2 + 1, type: "http_param" };
+    } else {
+      for (const v of sourceVarNames) {
+        const re = new RegExp(`(?<![A-Za-z0-9_$])${v.replace(/[.*+?^${}()|[\]\\]/g, "\\$&")}(?![A-Za-z0-9_$])`);
+        if (re.test(rhs)) {
+          const src = sources.find((s) => s.variable === v);
+          if (src) {
+            taintedBy = { line: src.line, type: src.type };
+            break;
+          }
+        }
+      }
+    }
+    if (!taintedBy) continue;
+    writes.push({ varName: lhs, line: i2 + 1, sourceLine: taintedBy.line, sourceType: taintedBy.type });
+  }
+  if (writes.length === 0) return flows;
+  for (const sink of sinks) {
+    if (unreachableLines.has(sink.line)) continue;
+    const sinkLineText = lines[sink.line - 1] ?? "";
+    if (!sinkLineText) continue;
+    for (const w of writes) {
+      if (w.line >= sink.line) continue;
+      const re = new RegExp(`(?<![A-Za-z0-9_$])${w.varName.replace(/[.*+?^${}()|[\]\\]/g, "\\$&")}(?![A-Za-z0-9_$])`);
+      if (!re.test(sinkLineText)) continue;
+      flows.push({
+        source_line: w.sourceLine,
+        sink_line: sink.line,
+        source_type: w.sourceType,
+        sink_type: sink.type,
+        path: [
+          { variable: w.varName, line: w.sourceLine, type: "source" },
+          { variable: w.varName, line: w.line, type: "assignment" },
+          { variable: w.varName, line: sink.line, type: "sink" }
+        ],
+        confidence: 0.9,
+        sanitized: false
+      });
+    }
+  }
+  return flows;
+}
 function detectCollectionFlows(calls, sources, sinks, taintedVars, unreachableLines, code, types) {
   const flows = [];
   const callsByLine = /* @__PURE__ */ new Map();