@goodandready/dsh-approval-gate 0.1.4 → 0.1.6

package/lib/inspect.js

@@ -1,4 +1,10 @@
-import { MESSAGES } from "./messages.js";
+import { DYNAMIC_MARKER, HEREDOC_OPS, REDIR_OPS, SPLIT_OPS, tokenize, readCommandSubstitution, readBacktick, readParameterExpansion } from "./tokenizer.js";
+import { expandHome } from "./paths.js";
+import { askMessage, denyMessage, sessionTag } from "./messages.js";
+
+export { DYNAMIC_MARKER, tokenize, splitSubcommands } from "./tokenizer.js";
+export { askMessage, denyMessage, sessionTag } from "./messages.js";
+export { expandHome } from "./paths.js";
 
 // Pure command/path inspector for dsh-approval-gate.
 // No Cordis imports: unit tests run this without the harness.
@@ -13,10 +19,6 @@ export const DEFAULT_FILE_WRITE_TOOLS = Object.freeze([
   "apply_patch",
 ]);
 
-const SPLIT_OPS = new Set(["|", "|&", "||", "&&", ";", "&"]);
-const REDIR_OPS = new Set([">", ">>", ">|", "<>", "2>", "2>>", "&>", "&>>"]);
-const HEREDOC_OPS = new Set(["<<", "<<-"]);
-const DYNAMIC_MARKER = "__shell_dynamic__";
 const PREFIXES = new Set(["then", "do"]);
 const SHELLS = new Set(["bash", "sh", "dash", "zsh", "ksh", "ash"]);
 const INTERPRETERS = new Set(["python", "python2", "python3", "node", "nodejs", "perl", "ruby", "php"]);
@@ -36,335 +38,23 @@ export function basename(path) {
   return parts[parts.length - 1] || "";
 }
 
-export function isProtectedPath(path) {
+export function isProtectedPath(path, env = process.env) {
   if (typeof path !== "string" || path.trim() === "") return false;
-  const base = basename(path);
+  const expanded = expandHome(path, env);
+  const normalized = expanded.replace(/\\/g, "/");
+  if (normalized.includes("/etc/sudoers.d/")) return true;
+  const base = basename(normalized);
   if (base === ".env" || base.startsWith(".env.")) return true;
   if (/^credentials\.ya?ml$/i.test(base)) return true;
   if (/^settings\.ya?ml$/i.test(base)) return true;
   if (base === "cordis.patch.yml") return true;
   if (/\.(pem|key|crt|p12|pfx)$/i.test(base)) return true;
   if (/^(id_rsa|id_ed25519|id_ecdsa)$/.test(base)) return true;
+  if (/^(shadow|gshadow|shadow-|gshadow-|sudoers|authorized_keys2?)$/i.test(base)) return true;
   if (/(?:^|[._-])(?:secret|token|credential|passwd|api[_-]?key)(?:$|[._-])/i.test(base)) return true;
   return false;
 }
 
-function fail(reason, snippet) {
-  return { ok: false, reason: reason, snippet: snippet };
-}
-
-function opAt(s, i) {
-  const rest = s.slice(i);
-  if (rest.startsWith("<(") || rest.startsWith(">(")) return null;
-  const candidates = ["2>&", "2>>", "&>>", "<<<", "<<-", "2>", ">>", ">|", "|&", "&&", "||", "&>", ">&", "<&", "<>", "<<", ">", "<", "|", ";", "&"];
-  for (const op of candidates) {
-    if (rest.startsWith(op)) return op;
-  }
-  return null;
-}
-
-function readCommandSubstitution(src, start) {
-  let depth = 1;
-  let quote = "";
-  let escaped = false;
-  for (let i = start + 2; i < src.length; i += 1) {
-    const c = src[i];
-    if (escaped) { escaped = false; continue; }
-    if (c === "\\") { escaped = true; continue; }
-    if (quote) {
-      if (c === quote) quote = "";
-      continue;
-    }
-    if (c === "'" || c === '"') { quote = c; continue; }
-    if (c === String.fromCharCode(96)) {
-      const end = readBacktick(src, i);
-      if (!end) return null;
-      i = end.end;
-      continue;
-    }
-    if (src.startsWith("$(", i)) { depth += 1; i += 1; continue; }
-    if (c === "(") { depth += 1; continue; }
-    if (c === ")") {
-      depth -= 1;
-      if (depth === 0) return { script: src.slice(start + 2, i), end: i + 1 };
-    }
-  }
-  return null;
-}
-
-function readBacktick(src, start) {
-  let escaped = false;
-  for (let i = start + 1; i < src.length; i += 1) {
-    if (escaped) { escaped = false; continue; }
-    if (src[i] === "\\") { escaped = true; continue; }
-    if (src[i] === String.fromCharCode(96)) return { script: src.slice(start + 1, i), end: i + 1 };
-  }
-  return null;
-}
-
-function readParameterExpansion(src, start) {
-  if (src[start + 1] !== "{") {
-    let i = start + 1;
-    if (/[A-Za-z_]/.test(src[i] || "")) {
-      i += 1;
-      while (i < src.length && /[A-Za-z0-9_]/.test(src[i])) i += 1;
-    } else if (i < src.length) i += 1;
-    return { end: i };
-  }
-  let depth = 1;
-  let quote = "";
-  for (let i = start + 2; i < src.length; i += 1) {
-    const c = src[i];
-    if (c === "\\") { i += 1; continue; }
-    if (quote) {
-      if (c === quote) quote = "";
-      continue;
-    }
-    if (c === "'" || c === '"') { quote = c; continue; }
-    if (c === String.fromCharCode(96)) {
-      const nested = readBacktick(src, i);
-      if (!nested) return null;
-      i = nested.end - 1;
-      continue;
-    }
-    if (src.startsWith("$" + "(", i)) {
-      const nested = readCommandSubstitution(src, i);
-      if (!nested) return null;
-      i = nested.end - 1;
-      continue;
-    }
-    if (src.startsWith("$" + "{", i)) {
-      const nested = readParameterExpansion(src, i);
-      if (!nested) return null;
-      i = nested.end - 1;
-      continue;
-    }
-    if (c === "{") depth += 1;
-    if (c === "}") {
-      depth -= 1;
-      if (depth === 0) return { end: i + 1, body: src.slice(start + 2, i) };
-    }
-  }
-  return null;
-}
-
-function lexFail(src, index, reason, tokens) {
-  return { ok: false, reason: reason, snippet: src.slice(Math.max(0, index - 32), index + 96), tokens: tokens };
-}
-
-function readHeredocDelimiter(src, index) {
-  let i = index;
-  while (i < src.length && (src[i] === " " || src[i] === "\t")) i += 1;
-  if (i >= src.length || src[i] === "\n" || src[i] === "\r") return null;
-  let delimiter = "";
-  let quote = "";
-  let started = false;
-  let quoted = false;
-  while (i < src.length && !/\s/.test(src[i]) && !opAt(src, i)) {
-    const c = src[i];
-    if (c === "'" || c === '"') {
-      if (!quote) { quote = c; quoted = true; started = true; i += 1; continue; }
-      if (quote === c) { quote = ""; i += 1; continue; }
-    }
-    if (c === "\\") {
-      if (i + 1 >= src.length) return null;
-      quoted = true;
-      delimiter += src[i + 1];
-      started = true;
-      i += 2;
-      continue;
-    }
-    delimiter += c;
-    started = true;
-    i += 1;
-  }
-  if (quote || !started || !delimiter) return null;
-  return { delimiter: delimiter, end: i, quoted: quoted };
-}
-
-export function tokenize(src) {
-  if (typeof src !== "string") return { ok: false, reason: "not-a-string", snippet: "", tokens: [] };
-  const tokens = [];
-  const pendingHeredocs = [];
-  let i = 0;
-  const n = src.length;
-
-  function flushWord(value, started, dynamic, substitutions) {
-    if (started) tokens.push({ kind: "word", value: value, dynamic: dynamic, substitutions: substitutions });
-  }
-
-  while (i < n) {
-    if (src.startsWith("\\\n", i)) { i += 2; continue; }
-    if (src[i] === "\r" || src[i] === "\n") {
-      if (src[i] === "\r" && src[i + 1] === "\n") i += 2;
-      else i += 1;
-      if (pendingHeredocs.length) {
-        for (const heredoc of pendingHeredocs) {
-          let body = "";
-          let found = false;
-          while (i <= n) {
-            const lineEnd = src.indexOf("\n", i);
-            const hasNewline = lineEnd !== -1;
-            let line = src.slice(i, hasNewline ? lineEnd : n);
-            if (line.endsWith("\r")) line = line.slice(0, -1);
-            const compare = heredoc.stripTabs ? line.replace(/^\t+/, "") : line;
-            if (compare === heredoc.delimiter) {
-              i = hasNewline ? lineEnd + 1 : n;
-              found = true;
-              break;
-            }
-            body += (heredoc.stripTabs ? line.replace(/^\t+/, "") : line) + (hasNewline ? "\n" : "");
-            if (!hasNewline) { i = n; break; }
-            i = lineEnd + 1;
-          }
-          if (!found) return lexFail(src, i, "unterminated-heredoc", tokens);
-          tokens.push({ kind: "heredoc", body: body, quoted: heredoc.quoted });
-        }
-        pendingHeredocs.length = 0;
-        tokens.push({ kind: "op", value: ";" });
-      } else {
-        tokens.push({ kind: "op", value: ";" });
-      }
-      continue;
-    }
-    if (/\s/.test(src[i])) { i += 1; continue; }
-    if (src[i] === "#" && (tokens.length === 0 || tokens[tokens.length - 1].kind === "op")) {
-      while (i < n && src[i] !== "\n") i += 1;
-      continue;
-    }
-
-    const op = opAt(src, i);
-    if (op) {
-      tokens.push({ kind: "op", value: op });
-      i += op.length;
-      if (HEREDOC_OPS.has(op)) {
-        const parsed = readHeredocDelimiter(src, i);
-        if (!parsed) return lexFail(src, i, "invalid-heredoc-delimiter", tokens);
-        tokens.push({ kind: "heredoc-delimiter", value: parsed.delimiter });
-        pendingHeredocs.push({ delimiter: parsed.delimiter, stripTabs: op === "<<-", quoted: parsed.quoted });
-        i = parsed.end;
-      }
-      continue;
-    }
-
-    let word = "";
-    let started = false;
-    let dynamic = false;
-    const substitutions = [];
-    while (i < n && !/\s/.test(src[i]) && !opAt(src, i)) {
-      const c = src[i];
-      if (src.startsWith("<(", i) || src.startsWith(">(", i)) {
-        const sub = readCommandSubstitution(src, i);
-        if (!sub) return lexFail(src, i, "unclosed-process-substitution", tokens);
-        substitutions.push({ kind: "process-substitution", script: sub.script });
-        word += DYNAMIC_MARKER;
-        dynamic = true;
-        started = true;
-        i = sub.end;
-        continue;
-      }
-      if (c === "(" || c === ")") return lexFail(src, i, "unsupported-shell-grouping", tokens);
-      if (c === "\\") {
-        if (i + 1 >= n) return lexFail(src, i, "dangling-backslash", tokens);
-        word += src[i + 1];
-        started = true;
-        i += 2;
-        continue;
-      }
-      if (c === "'") {
-        const j = src.indexOf("'", i + 1);
-        if (j < 0) return lexFail(src, i, "unclosed-single-quote", tokens);
-        word += src.slice(i + 1, j);
-        started = true;
-        i = j + 1;
-        continue;
-      }
-      if (c === '"') {
-        started = true;
-        i += 1;
-        let closed = false;
-        while (i < n) {
-          if (src[i] === '"') { i += 1; closed = true; break; }
-          if (src[i] === "\\") {
-            if (i + 1 >= n) return lexFail(src, i, "dangling-backslash", tokens);
-            word += src[i + 1];
-            i += 2;
-            continue;
-          }
-          if (src.startsWith("$(", i)) {
-            const sub = readCommandSubstitution(src, i);
-            if (!sub) return lexFail(src, i, "unclosed-command-substitution", tokens);
-            substitutions.push({ kind: "command-substitution", script: sub.script });
-            word += DYNAMIC_MARKER;
-            dynamic = true;
-            i = sub.end;
-            continue;
-          }
-          if (src[i] === String.fromCharCode(96)) {
-            const sub = readBacktick(src, i);
-            if (!sub) return lexFail(src, i, "unclosed-backtick-substitution", tokens);
-            substitutions.push({ kind: "backtick-substitution", script: sub.script });
-            word += DYNAMIC_MARKER;
-            dynamic = true;
-            i = sub.end;
-            continue;
-          }
-          if (src[i] === "$") {
-            const expansion = readParameterExpansion(src, i);
-            if (!expansion) return lexFail(src, i, "unclosed-parameter-expansion", tokens);
-            word += DYNAMIC_MARKER;
-            dynamic = true;
-            if (expansion.body !== undefined) substitutions.push({ kind: "parameter-expansion", script: expansion.body });
-            i = expansion.end;
-            continue;
-          }
-          word += src[i];
-          i += 1;
-        }
-        if (!closed) return lexFail(src, i, "unclosed-double-quote", tokens);
-        continue;
-      }
-      if (src.startsWith("$(", i)) {
-        const sub = readCommandSubstitution(src, i);
-        if (!sub) return lexFail(src, i, "unclosed-command-substitution", tokens);
-        substitutions.push({ kind: "command-substitution", script: sub.script });
-        word += DYNAMIC_MARKER;
-        dynamic = true;
-        started = true;
-        i = sub.end;
-        continue;
-      }
-      if (c === String.fromCharCode(96)) {
-        const sub = readBacktick(src, i);
-        if (!sub) return lexFail(src, i, "unclosed-backtick-substitution", tokens);
-        substitutions.push({ kind: "backtick-substitution", script: sub.script });
-        word += DYNAMIC_MARKER;
-        dynamic = true;
-        started = true;
-        i = sub.end;
-        continue;
-      }
-      if (c === "$") {
-        const expansion = readParameterExpansion(src, i);
-        if (!expansion) return lexFail(src, i, "unclosed-parameter-expansion", tokens);
-        word += DYNAMIC_MARKER;
-        dynamic = true;
-        started = true;
-        if (expansion.body !== undefined) substitutions.push({ kind: "parameter-expansion", script: expansion.body });
-        i = expansion.end;
-        continue;
-      }
-      if (c === "*" || c === "?") dynamic = true;
-      word += c;
-      started = true;
-      i += 1;
-    }
-    flushWord(word, started, dynamic, substitutions);
-  }
-  if (pendingHeredocs.length) return lexFail(src, n, "unterminated-heredoc", tokens);
-  return { ok: true, tokens: tokens };
-}
-
 function wordsOf(seg) {
   return seg.filter(function (t) { return t.kind === "word"; }).map(function (t) { return t.value; });
 }
@@ -510,7 +200,7 @@ function inspectArgv(argv, depth, hasHeredoc) {
 
   if (cmd === "eval") return deny("eval", argv.join(" "));
   if (KILL_CMDS.has(cmd)) return deny("process-signal", argv.join(" "));
-  if (cmd === "rm" && (hasLetterFlag(argv, "r") || rest.indexOf("--recursive") !== -1)) {
+  if (cmd === "rm" && (hasLetterFlag(argv, "r") || hasLetterFlag(argv, "R") || rest.indexOf("--recursive") !== -1)) {
     return deny("recursive-rm", argv.join(" "));
   }
   if (cmd === "systemctl") {
@@ -543,17 +233,24 @@ function inspectArgv(argv, depth, hasHeredoc) {
       if (a.startsWith("of=") && (isDevicePath(a.slice(3)) || isProtectedPath(a.slice(3)))) return deny("device-write", a);
     }
   }
-  if (cmd === "tee" || cmd === "install" || cmd === "cp" || cmd === "mv" || cmd === "truncate") {
+  if (cmd === "tee" || cmd === "install" || cmd === "cp" || cmd === "mv" || cmd === "truncate" || cmd === "chmod" || cmd === "chown") {
     for (const a of rest) {
       if (a[0] !== "-" && isProtectedPath(a)) return deny("secret-write", argv.join(" "));
     }
+    if ((cmd === "chmod" || cmd === "chown") && (hasLetterFlag(argv, "R") || hasLetterFlag(argv, "r") || rest.indexOf("--recursive") !== -1)) {
+      for (const a of rest) {
+        if (a === "/" || a === "/etc" || a === "/var" || a === "/usr" || a === "/root" || a === "/boot") {
+          return deny("device-write", argv.join(" "));
+        }
+      }
+    }
   }
-  if (cmd === "sed" && (hasLetterFlag(argv, "i") || rest.indexOf("--in-place") !== -1)) {
+  if (cmd === "sed" && (hasLetterFlag(argv, "i") || rest.some(function (a) { return a === "--in-place" || a.startsWith("--in-place="); }))) {
     for (const a of rest) {
       if (a[0] !== "-" && isProtectedPath(a)) return deny("secret-write", argv.join(" "));
     }
   }
-  if (SHELLS.has(cmd)) {
+  if (SHELLS.has(cmd) || cmd === "source" || cmd === ".") {
     const script = scriptAfterDashC(rest);
     if (typeof script === "string") {
       if (depth > 4) return deny("nested-shell", script);
@@ -583,7 +280,7 @@ function hasDangerousPipeline(tokens) {
   for (const token of tokens) {
     if (token.kind === "op" && SPLIT_OPS.has(token.value)) {
       const cmd = inspectSegment();
-      if (inPipeline && SHELLS.has(cmd) && pipelineHasDownloader) return true;
+      if (inPipeline && (SHELLS.has(cmd) || INTERPRETERS.has(cmd)) && pipelineHasDownloader) return true;
       const continues = token.value === "|" || token.value === "|&";
       if (continues) {
         pipelineHasDownloader = pipelineHasDownloader || cmd === "curl" || cmd === "wget";
@@ -598,7 +295,33 @@ function hasDangerousPipeline(tokens) {
     segment.push(token);
   }
   const cmd = inspectSegment();
-  return inPipeline && pipelineHasDownloader && SHELLS.has(cmd);
+  return inPipeline && pipelineHasDownloader && (SHELLS.has(cmd) || INTERPRETERS.has(cmd));
+}
+
+function hasDestructiveSqlPipeline(tokens) {
+  let pipelineHasDestructiveSql = false;
+  let inPipeline = false;
+  let segment = [];
+  for (const token of tokens) {
+    if (token.kind === "op" && SPLIT_OPS.has(token.value)) {
+      const cmd = commandNameFromSegment(segment);
+      if (inPipeline && DB_CLIENTS.has(cmd) && pipelineHasDestructiveSql) return true;
+      const continues = token.value === "|" || token.value === "|&";
+      if (continues) {
+        const segText = wordsOf(segment).join(" ");
+        pipelineHasDestructiveSql = pipelineHasDestructiveSql || DESTRUCTIVE_SQL.test(segText);
+        inPipeline = true;
+      } else {
+        pipelineHasDestructiveSql = false;
+        inPipeline = false;
+      }
+      segment = [];
+      continue;
+    }
+    segment.push(token);
+  }
+  const cmd = commandNameFromSegment(segment);
+  return inPipeline && pipelineHasDestructiveSql && DB_CLIENTS.has(cmd);
 }
 
 function executableTokenOf(seg) {
@@ -727,6 +450,7 @@ export function inspectBashCommand(command, depth) {
     if (knownDanger) return knownDanger;
   }
   if (hasDangerousPipeline(tokens)) return deny("curl-pipe", command);
+  if (hasDestructiveSqlPipeline(tokens)) return deny("destructive-sql", command);
   const pipelineHeredocHit = inspectPipelineHeredocs(tokens, depth);
   if (pipelineHeredocHit.deny || pipelineHeredocHit.ask) return pipelineHeredocHit;
 
@@ -780,6 +504,12 @@ export function inspectBashCommand(command, depth) {
           const nested = inspectInterpreterPayload(commandName, heredoc.body);
           if (nested.deny || nested.ask) return nested;
         }
+      } else if (DB_CLIENTS.has(commandName)) {
+        for (const heredoc of heredocs) {
+          if (DESTRUCTIVE_SQL.test(heredoc.body)) {
+            return deny("destructive-sql", heredoc.body);
+          }
+        }
       }
     }
 
@@ -822,97 +552,3 @@ export function inspectExecution(execution, config) {
   return allow();
 }
 
-export function sessionTag(execution) {
-  const agent = execution && execution.agent;
-  if (typeof agent === "string") return " (session " + agent + ")";
-  if (agent && agent.session) return " (session " + String(agent.session) + ")";
-  return "";
-}
-
-function safeSnippet(value) {
-  return String(value || "")
-    .replace(/[\r\n\t\u0000-\u001f]+/g, " ")
-    .replace(/(authorization\s*:\s*(?:bearer|token)\s+)\S+/gi, "$1[redacted]")
-    .replace(/((?:password|passwd|secret|token|api[_-]?key)\s*[:=]\s*)[^\s,;]+/gi, "$1[redacted]")
-    .slice(0, 120);
-}
-
-function translated(translate, key, logger) {
-  if (typeof translate === "function") {
-    try {
-      const value = translate(key);
-      if (value && value !== key) return value;
-    } catch {
-      if (logger && typeof logger.warn === "function") {
-        logger.warn("[dsh-approval-gate] localization lookup failed; using English fallback");
-      }
-    }
-  }
-  return MESSAGES.en[key] || key;
-}
-
-const RULE_KEYS = Object.freeze({
-  "recursive-rm": "ruleRecursiveRm",
-  "process-signal": "ruleProcessSignal",
-  "service-control": "ruleServiceControl",
-  "git-clean": "ruleGitClean",
-  "git-reset-hard": "ruleGitResetHard",
-  "find-delete": "ruleFindDelete",
-  "destructive-sql": "ruleDestructiveSql",
-  "secret-write": "ruleSecretWrite",
-  "device-write": "ruleDeviceWrite",
-  "mkfs-device": "ruleMkfsDevice",
-  "curl-pipe": "ruleCurlPipe",
-  "eval": "ruleEval",
-  "nested-shell": "ruleNestedShell",
-  "interpreter-payload": "ruleInterpreterPayload",
-});
-
-const ASK_KEYS = Object.freeze({
-  unparseable: "askUnparseable",
-  "unknown-format": "askUnknownFormat",
-  "unknown-write-target": "askUnknownWriteTarget",
-  "dynamic-command": "askDynamicCommand",
-  "dynamic-arguments": "askDynamicArguments",
-  "dynamic-redirect": "askDynamicRedirect",
-  "script-file": "askScriptFile",
-  "shell-stdin": "askShellStdin",
-  "interpreter-payload": "askInterpreterPayload",
-  "credential-in-argv": "askCredentialInArgv",
-});
-
-const DETAIL_KEYS = Object.freeze({
-  "process-substitution": "detailProcessSubstitution",
-  "unclosed-process-substitution": "detailUnclosedProcessSubstitution",
-  "unclosed-command-substitution": "detailUnclosedCommandSubstitution",
-  "unclosed-backtick-substitution": "detailUnclosedBacktick",
-  "unclosed-parameter-expansion": "detailUnclosedParameterExpansion",
-  "unclosed-single-quote": "detailUnclosedSingleQuote",
-  "unclosed-double-quote": "detailUnclosedDoubleQuote",
-  "dangling-backslash": "detailDanglingEscape",
-  "invalid-heredoc-delimiter": "detailInvalidHeredoc",
-  "unterminated-heredoc": "detailUnterminatedHeredoc",
-  "unsupported-shell-grouping": "detailUnsupportedGrouping",
-});
-
-export function denyMessage(hit, who, translate, logger) {
-  const key = RULE_KEYS[hit && hit.reason] || "ruleUnknown";
-  const snippet = safeSnippet(hit && hit.snippet);
-  const parts = [translated(translate, "blockedPrefix", logger), translated(translate, key, logger)];
-  if (snippet) parts.push(translated(translate, "detectedLabel", logger) + ": " + snippet);
-  if (who) parts.push(who.trim());
-  return parts.join(" — ") + ".";
-}
-
-export function askMessage(hit, who, translate, logger) {
-  const key = ASK_KEYS[hit && hit.reason] || "askUnknown";
-  const snippet = safeSnippet(hit && hit.snippet);
-  const parts = [translated(translate, "approvalRequired", logger), translated(translate, key, logger)];
-  if (hit && hit.reason === "unparseable") {
-    const detailKey = DETAIL_KEYS[hit.detail] || "detailUnsupported";
-    parts.push(translated(translate, "parseReasonLabel", logger) + ": " + translated(translate, detailKey, logger));
-  }
-  if (snippet) parts.push(translated(translate, "commandLabel", logger) + ": " + snippet);
-  if (who) parts.push(who.trim());
-  return parts.join(" — ") + ".";
-}