test(security): close tar remediation evidence gaps

test/dockerfile-run-commands.test.ts

@@ -2,7 +2,10 @@
 // SPDX-License-Identifier: Apache-2.0
 
 import { describe, expect, it } from "vitest";
-import { requireSingleReviewedDockerfileRunCommand } from "./helpers/dockerfile-run-commands";
+import {
+  dockerfileRunCommandPositions,
+  requireSingleReviewedDockerfileRunCommand,
+} from "./helpers/dockerfile-run-commands";
 
 const command = "node --experimental-strip-types /scripts/patch-bundled-npm-tar.mts";
 const corporateCaPath = "/usr/local/share/nemoclaw/corporate-ca.pem";
@@ -11,6 +14,60 @@ const invocation = [command, ...requiredArguments].join(" ");
 const splicedCommand = command.replace("strip-types", "strip-\\\ntypes");
 
 describe("Dockerfile RUN command discovery", () => {
+  it("finds only executable unquoted npm command words in RUN instructions (#9933)", () => {
+    const source = [
+      "# npm install",
+      'LABEL example="npm install"',
+      'RUN echo "npm install"',
+      "RUN echo npm install # npm ci",
+      "RUN true && \\",
+      "    # install from the reviewed lock",
+      "    npm --prefix /runtime ci",
+      "RUN if true; then npm --prefix=/runtime install; fi",
+      "RUN true && NPM_CONFIG_OFFLINE=true OTHER=value npm ci --prefix /runtime",
+      "RUN if npm ci --prefix /if; then true; fi",
+      "RUN if false; then true; elif npm install --prefix /elif; then true; fi",
+      "RUN while npm ci --prefix /while; do true; done",
+      "RUN until npm install --prefix /until; do true; done",
+      "RUN ( npm ci --prefix /subshell )",
+      "RUN { npm install --prefix /group; }",
+      "RUN case value in value) npm ci --prefix /case ;; esac",
+      "RUN ! npm install --prefix /negated",
+      'RUN NPM_CONFIG_CACHE="/tmp/npm cache" npm ci --prefix /quoted-assignment',
+      "RUN NPM_CONFIG_CACHE=/tmp/npm\\ cache npm install --prefix /escaped-assignment",
+      "",
+    ].join("\n");
+
+    expect(dockerfileRunCommandPositions(source, "npm")).toEqual([
+      source.indexOf("npm --prefix /runtime"),
+      source.indexOf("npm --prefix=/runtime"),
+      source.indexOf("npm ci --prefix"),
+      source.indexOf("npm ci --prefix /if"),
+      source.indexOf("npm install --prefix /elif"),
+      source.indexOf("npm ci --prefix /while"),
+      source.indexOf("npm install --prefix /until"),
+      source.indexOf("npm ci --prefix /subshell"),
+      source.indexOf("npm install --prefix /group"),
+      source.indexOf("npm ci --prefix /case"),
+      source.indexOf("npm install --prefix /negated"),
+      source.indexOf("npm ci --prefix /quoted-assignment"),
+      source.indexOf("npm install --prefix /escaped-assignment"),
+    ]);
+  });
+
+  it("ignores npm assignment values, redirection operands, and here-document delimiters (#9933)", () => {
+    const source = [
+      "RUN VALUE=npm ci",
+      "RUN echo ok > npm",
+      "RUN cat <<npm",
+      "payload",
+      "npm",
+      "",
+    ].join("\n");
+
+    expect(dockerfileRunCommandPositions(source, "npm")).toEqual([]);
+  });
+
   it("ignores command text in comments, strings, and non-RUN instructions", () => {
     const source = [
       `# ${command}`,

test/helpers/dockerfile-run-commands.ts

@@ -47,14 +47,15 @@ export function dockerfileInstructions(source: string): DockerfileInstruction[]
     }
 
     let end = endOfFirstLine;
-    let currentLine = firstLine;
-    while (continuesInstruction(currentLine)) {
+    let continues = continuesInstruction(firstLine);
+    while (continues) {
       if (end >= source.length) {
         throw new Error(`Dockerfile ends inside the ${instructionMatch[1]} instruction`);
       }
       const nextEnd = lineEnd(source, end);
-      currentLine = source.slice(end, nextEnd);
+      const currentLine = source.slice(end, nextEnd);
       end = nextEnd;
+      continues = /^[ \t]*#/u.test(currentLine) || continuesInstruction(currentLine);
     }
 
     const bodyStart = offset + instructionMatch[0].length;
@@ -134,6 +135,98 @@ function unquotedTextIndexes(source: string, text: string): number[] {
   return indexes;
 }
 
+function shellCommandPrefixWords(source: string, end: number): string[] {
+  const words: string[] = [];
+  let wordStart: number | undefined;
+  let quote: "'" | '"' | "`" | null = null;
+  let comment = false;
+
+  const finishWord = (wordEnd: number): void => {
+    if (wordStart === undefined) return;
+    words.push(source.slice(wordStart, wordEnd));
+    wordStart = undefined;
+  };
+
+  for (let index = 0; index < end; index += 1) {
+    const character = source[index]!;
+    if (comment) {
+      if (character === "\n") {
+        comment = false;
+        words.length = 0;
+      }
+      continue;
+    }
+    if (quote !== null) {
+      if (character === "\\" && quote !== "'") {
+        index += 1;
+      } else if (character === quote) {
+        quote = null;
+      }
+      continue;
+    }
+    if (character === "'" || character === '"' || character === "`") {
+      wordStart ??= index;
+      quote = character;
+      continue;
+    }
+    if (character === "\\") {
+      wordStart ??= index;
+      index += 1;
+      continue;
+    }
+    if (character === "#" && (index === 0 || /[\s;&|(){}]/u.test(source[index - 1]!))) {
+      finishWord(index);
+      comment = true;
+      continue;
+    }
+    if (/[ \t\r]/u.test(character)) {
+      finishWord(index);
+      continue;
+    }
+    if (character === "\n" || ";&|({)".includes(character)) {
+      finishWord(index);
+      words.length = 0;
+      continue;
+    }
+    wordStart ??= index;
+  }
+
+  finishWord(end);
+  return words;
+}
+
+function followsShellCommandSeparator(source: string, index: number): boolean {
+  return shellCommandPrefixWords(source, index).every(
+    (word) =>
+      ["!", "do", "elif", "else", "if", "then", "until", "while"].includes(word) ||
+      /^[A-Za-z_][A-Za-z0-9_]*=.*$/u.test(word),
+  );
+}
+
+function startsShellWord(source: string, index: number): boolean {
+  const previousCharacter = source[index - 1];
+  return previousCharacter === undefined || /[\t\r\n &|();<>]/u.test(previousCharacter);
+}
+
+export function dockerfileRunCommandPositions(source: string, command: string): number[] {
+  const positions: number[] = [];
+  for (const instruction of dockerfileInstructions(source)) {
+    if (instruction.keyword !== "RUN") continue;
+    const collapsed = collapseDockerfileContinuations(instruction.body);
+    for (const index of unquotedTextIndexes(collapsed.text, command)) {
+      const afterCommand = collapsed.text[index + command.length];
+      if (
+        startsShellWord(collapsed.text, index) &&
+        followsShellCommandSeparator(collapsed.text, index) &&
+        (afterCommand === undefined || /[ \t\r\n;&|(){}<>]/u.test(afterCommand))
+      ) {
+        positions.push(instruction.bodyStart + collapsed.originalIndexes[index]!);
+      }
+    }
+  }
+  return positions;
+}
+
 function normalizedInstructionBody(source: string): string {
   return source
     .replace(/\\\r?\n/gu, " ")