cognium-dev 3.181.0 → 3.185.0

package/dist/cli.js

@@ -9440,6 +9440,10 @@ 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);
@@ -9447,6 +9451,7 @@ 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);
@@ -9546,6 +9551,70 @@ function findChildByTypeGo(node, type) {
   }
   return null;
 }
+var GO_OPAQUE_CODEC_METHODS = new Set([
+  "Unmarshal",
+  "Decode",
+  "NewDecoder",
+  "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;
+}
 // ../circle-ir/dist/core/extractors/runtime-registrations.js
 var HTTP_VERB_METHODS = new Set([
   "get",
@@ -10622,6 +10691,19 @@ var DEFAULT_SOURCES = [
   { method: "getMap", class: "Metadata", type: "http_header", severity: "high", return_tainted: true, languages: ["javascript", "typescript"] },
   { method: "FromIncomingContext", class: "metadata", type: "http_header", severity: "high", return_tainted: true, languages: ["go"] },
   { method: "get", class: "Metadata", type: "http_header", severity: "high", return_tainted: true, languages: ["java"] },
+  { 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"] },
+  { method: "receive", class: "WebSocket", type: "network_input", severity: "high", return_tainted: true, languages: ["python"] },
+  { 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"] },
+  { 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"] },
+  { method: "Read", class: "Conn", type: "network_input", severity: "high", return_tainted: true, languages: ["go"] },
+  { annotation: "OnMessage", type: "network_input", severity: "high", param_tainted: true, languages: ["java"] },
+  { annotation: "MessageMapping", type: "http_body", severity: "high", param_tainted: true, languages: ["java"] },
+  { annotation: "SubscribeMapping", type: "http_body", severity: "high", param_tainted: true, languages: ["java"] },
   { method: "get", class: "Redis", type: "db_input", severity: "medium", return_tainted: true, languages: ["python"] },
   { method: "hget", class: "Redis", type: "db_input", severity: "medium", return_tainted: true, languages: ["python"] },
   { method: "mget", class: "Redis", type: "db_input", severity: "medium", return_tainted: true, languages: ["python"] },
@@ -11913,6 +11995,26 @@ var DEFAULT_SANITIZERS = [
   { method: "resolve", class: "Path", removes: ["path_traversal"] },
   { method: "int", removes: ["sql_injection", "command_injection", "xss"] },
   { method: "float", removes: ["sql_injection", "command_injection"] },
+  { 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"] },
+  { method: "quote_plus", removes: ["ssrf", "open_redirect", "xss", "path_traversal"] },
+  { method: "urlencode", removes: ["ssrf", "open_redirect", "xss"] },
+  { method: "linkify", class: "bleach", removes: ["xss"] },
+  { 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"] },
+  { method: "escape", class: "re", removes: ["redos", "code_injection"] },
+  { method: "bindparams", removes: ["sql_injection"] },
+  { method: "Identifier", class: "sql", removes: ["sql_injection"] },
+  { method: "Literal", class: "sql", removes: ["sql_injection"] },
+  { method: "Placeholder", class: "sql", removes: ["sql_injection"] },
   { method: "query!", removes: ["sql_injection"] },
   { method: "query_as!", removes: ["sql_injection"] },
   { method: "query_scalar!", removes: ["sql_injection"] },
@@ -13836,7 +13938,13 @@ 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;
       }
     }
@@ -16432,15 +16540,23 @@ 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(taintInfo.line, sink.line, sink.type, sanitizersByLine, {
@@ -16468,7 +16584,16 @@ function propagateTaint(graphOrDfg, callsOrSources, sourcesOrSinks, sinksOrSanit
       }
     }
   }
-  return { taintedVars, flows, reachableSinks };
+  const seen = 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 = [];
@@ -24602,7 +24727,10 @@ var PYTHON_TAINTED_PATTERNS2 = [
   { pattern: /\bget_query_parameter\s*\(/, type: "http_param" },
   { pattern: /\bget_header_value\s*\(/, type: "http_header" },
   { pattern: /\bget_cookie_value\s*\(/, type: "http_cookie" },
-  { pattern: /\binput\s*\(/, type: "io_input" }
+  { pattern: /\binput\s*\(/, type: "io_input" },
+  { pattern: /\.receive_text\s*\(/, type: "network_input" },
+  { pattern: /\.receive_bytes\s*\(/, type: "network_input" },
+  { pattern: /\.receive_json\s*\(/, type: "http_body" }
 ];
 
 class LanguageSourcesPass {
@@ -25246,6 +25374,36 @@ 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*(?:[,):]|$)|(\w+)\s*=>)/;
+  const lines = sourceCode.split(`
+`);
+  for (let i2 = 0;i2 < lines.length; i2++) {
+    const line = lines[i2];
+    const trimmed = line.trimStart();
+    if (trimmed.startsWith("//") || trimmed.startsWith("*"))
+      continue;
+    const m = line.match(CB_EVENT_RE);
+    if (!m)
+      continue;
+    const paramName = m[1] ?? m[2] ?? m[3];
+    if (!paramName)
+      continue;
+    if (paramName === "err" || paramName === "error")
+      continue;
+    sources.push({
+      type: "network_input",
+      location: `WebSocket .on(...) callback param '${paramName}' at line ${i2 + 1}`,
+      severity: "high",
+      line: i2 + 1,
+      confidence: 0.95,
+      variable: paramName
+    });
+  }
   return sources;
 }
 function findPythonAssignmentSources(sourceCode, language) {
@@ -25578,6 +25736,16 @@ function buildJavaScriptTaintedVars(sourceCode, language) {
   const tainted = new Map;
   const lines = sourceCode.split(`
 `);
+  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();
@@ -25784,6 +25952,90 @@ 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 → $${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 → $${v} (stdin array)`,
+          severity: "high",
+          line: lineNumber,
+          confidence: 0.9,
+          variable: v
+        });
+      }
+    }
+    const getoptsMatch = trimmed.match(/\bgetopts\s+["'][^"']+["']\s+(\w+)/);
+    if (getoptsMatch) {
+      const flagVar = getoptsMatch[1];
+      for (const v of [flagVar, "OPTARG"]) {
+        const already = sources.some((s) => s.line === lineNumber && s.variable === v);
+        if (already)
+          continue;
+        sources.push({
+          type: "io_input",
+          location: `getopts → $${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) {
@@ -33393,6 +33645,12 @@ class TaintPropagationPass {
         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)
@@ -33649,6 +33907,113 @@ 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(`
+`);
+  const packageVars = 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 = 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 = new Map;
@@ -45451,7 +45816,7 @@ var colors = {
 };
 
 // src/version.ts
-var version = "3.181.0";
+var version = "3.185.0";
 
 // src/formatters.ts
 var LIBRARY_API_SURFACE_TAG2 = "library-api-surface:caller-responsibility";

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "cognium-dev",
-  "version": "3.181.0",
+  "version": "3.185.0",
   "description": "Static Application Security Testing CLI for detecting security vulnerabilities via taint tracking",
   "main": "dist/index.js",
   "types": "dist/index.d.ts",
@@ -66,7 +66,7 @@
   },
   "dependencies": {
     "@cognium/project-profile-detect": "^1.1.0",
-    "circle-ir": "^3.181.0"
+    "circle-ir": "^3.185.0"
   },
   "devDependencies": {
     "@types/node": "^25.5.0",