opencode-workflow-guard 1.19.0 → 1.19.1

package/src/lib/shell.ts

@@ -1,3 +1,6 @@
+import { homedir, userInfo } from "node:os";
+import { dirname, join, resolve } from "node:path";
+
 function asRecord(value: unknown): Record<string, unknown> | undefined {
 	if (typeof value !== "object" || value === null || Array.isArray(value)) {
 		return undefined;
@@ -193,6 +196,290 @@ function envSplitWords(value: string): string[] {
 	return words;
 }
 
+export interface HeredocRange {
+	/** Offset of the `<<` that introduces the heredoc. */
+	start: number;
+	/** Offset of the body's first byte. */
+	bodyStart: number;
+	/** Offset of the terminator line, or end of input when unterminated. */
+	bodyEnd: number;
+	/** Offset of the terminator line's newline, or end of input. */
+	end: number;
+}
+
+interface HeredocHeader {
+	start: number;
+	/** Offset just past the delimiter word. */
+	end: number;
+	delimiter: string;
+	stripTabs: boolean;
+}
+
+function heredocHeaderAt(command: string, start: number): HeredocHeader | undefined {
+	// `<<<` is a herestring, whose operand is a single word on the same line.
+	if (command[start + 2] === "<") return undefined;
+	const header = /^<<(-?)[ \t]*(?:'([^'\n]+)'|"([^"\n]+)"|([^\s'";|&()<>\n]+))/.exec(command.slice(start));
+	if (!header) return undefined;
+	const delimiter = header[2] ?? header[3] ?? header[4];
+	if (!delimiter) return undefined;
+	return { start, end: start + header[0].length, delimiter, stripTabs: header[1] === "-" };
+}
+
+function heredocBodyAt(
+	command: string,
+	from: number,
+	delimiter: string,
+	stripTabs: boolean,
+): { bodyStart: number; bodyEnd: number; end: number } {
+	let cursor = from;
+	while (cursor <= command.length) {
+		const nextEnd = command.indexOf("\n", cursor);
+		const line = command.slice(cursor, nextEnd < 0 ? command.length : nextEnd);
+		const comparable = stripTabs ? line.replace(/^\t+/, "") : line;
+		if (comparable === delimiter) {
+			// Stop at the terminator's own newline and let the caller split on it,
+			// so the command that follows the heredoc starts its own segment.
+			return { bodyStart: from, bodyEnd: cursor, end: nextEnd < 0 ? command.length : nextEnd };
+		}
+		if (nextEnd < 0) break;
+		cursor = nextEnd + 1;
+	}
+	// A missing terminator runs to end of input, which is where the shell would
+	// stop reading too. Bash still runs the body it collected.
+	return { bodyStart: from, bodyEnd: command.length, end: command.length };
+}
+
+/**
+ * Every heredoc introduced on the line that starts at `start`.
+ *
+ * One header line can introduce several bodies (`cat <<A <<B`), and the shell
+ * reads them one after another. Ignoring the later ones read a second body as
+ * top-level commands.
+ */
+export function heredocRunAt(command: string, start: number): HeredocRange[] | undefined {
+	const headers: HeredocHeader[] = [];
+	let cursor = start;
+	for (;;) {
+		const header = heredocHeaderAt(command, cursor);
+		if (!header) break;
+		headers.push(header);
+		let next = header.end;
+		while (next < command.length && /[ \t]/.test(command[next]!)) next += 1;
+		if (command[next] !== "<" || command[next + 1] !== "<") break;
+		cursor = next;
+	}
+	if (headers.length === 0) return undefined;
+	const lastHeader = headers[headers.length - 1]!;
+	const lineEnd = command.indexOf("\n", lastHeader.end);
+	if (lineEnd < 0) return undefined;
+	const ranges: HeredocRange[] = [];
+	let bodyStart = lineEnd + 1;
+	for (const header of headers) {
+		const body = heredocBodyAt(command, bodyStart, header.delimiter, header.stripTabs);
+		ranges.push({ start: header.start, ...body });
+		if (body.end >= command.length) break;
+		bodyStart = body.end + 1;
+	}
+	return ranges;
+}
+
+/**
+ * Every heredoc in `command`, found with the same quote tracking the segment
+ * splitter uses.
+ *
+ * A `<<` inside a quoted span opens no heredoc. Scanning for `<<` without that
+ * check read a quoted `"<<"` as a header and masked the code that followed it.
+ */
+export function heredocRanges(command: string): HeredocRange[] {
+	const ranges: HeredocRange[] = [];
+	let quote: "'" | '"' | "ansi" | undefined;
+	let escaped = false;
+	for (let i = 0; i < command.length; i += 1) {
+		const char = command[i]!;
+		if (escaped) {
+			escaped = false;
+			continue;
+		}
+		if (!quote && char === "$" && command[i + 1] === "'") {
+			quote = "ansi";
+			i += 1;
+			continue;
+		}
+		if (char === "\\" && quote !== "'") {
+			escaped = true;
+			continue;
+		}
+		if (quote === "ansi" && char === "'") {
+			quote = undefined;
+			continue;
+		}
+		if (char === "'" || char === '"') {
+			if (!quote) quote = char;
+			else if (quote === char) quote = undefined;
+			continue;
+		}
+		if (quote || char !== "<" || command[i + 1] !== "<") continue;
+		const run = heredocRunAt(command, i);
+		if (!run) continue;
+		ranges.push(...run);
+		// Lines inside a body are data, so a `<<` there opens nothing.
+		i = run[run.length - 1]!.end - 1;
+	}
+	return ranges;
+}
+
+/**
+ * The shells and interpreters that run a heredoc body as code.
+ *
+ * `source` and the `.` builtin execute a body they read from stdin, so their
+ * bodies are code too.
+ */
+const HEREDOC_EXECUTOR_RE = /\b(?:python3?|node|perl|ruby|osascript|bash|sh|zsh|dash|ksh|source)\b|^\s*\.\s/i;
+
+/**
+ * Commands that read a heredoc body as data.
+ *
+ * The body of these is never executed, so its lines are prose. An unrecognized
+ * command is treated as an executor instead: guessing "data" there would let a
+ * body a shell reads as code escape the scan.
+ *
+ * `sed`, `awk`, and `sort` are deliberately absent. Each can run a command named
+ * inside its own arguments (`sed 'e'`, `awk '{system($0)}'`,
+ * `sort --compress-program`), so a body behind one of them is executed text.
+ */
+const DATA_HEREDOC_CONSUMERS = new Set([
+	"cat",
+	"tee",
+	"echo",
+	"printf",
+	"grep",
+	"egrep",
+	"fgrep",
+	"uniq",
+	"wc",
+	"head",
+	"tail",
+	"tr",
+	"cut",
+	"patch",
+	"diff",
+	"xxd",
+	"od",
+	"base64",
+	"sha1sum",
+	"sha256sum",
+	"md5sum",
+	"jq",
+	"column",
+	"less",
+	"more",
+]);
+
+/*
+ * A program that can run a command named in its own arguments is not a data
+ * consumer. Testing the whole header line matched a word inside a quoted
+ * argument, so `sed -n '1e echo hi'` read as data because of the `echo` in its
+ * script.
+ */
+function consumerWords(before: string): string[] {
+	// Only the command the heredoc belongs to reads the body, so take the last
+	// segment. Earlier commands on the same line never see it.
+	const segments = splitShellSegments(before);
+	return unwrapShellWords(segments[segments.length - 1] ?? "");
+}
+
+/** The program that reads the body, or `undefined` when no word names one. */
+function consumerProgram(before: string): { head: string; sub?: string } | undefined {
+	const words = consumerWords(before);
+	const head = words[0]?.toLowerCase();
+	if (!head) return undefined;
+	return { head, sub: words[1]?.toLowerCase() };
+}
+
+/** Whether a program reads a heredoc body as data rather than executing it. */
+function readsBodyAsData(program: { head: string; sub?: string }): boolean {
+	if (program.head === "git") return program.sub === "commit" || program.sub === "tag";
+	return DATA_HEREDOC_CONSUMERS.has(program.head);
+}
+
+/**
+ * The command word of every stage a header line pipes the body into.
+ *
+ * A pipeline hands the body to every stage, so `| grep x | sh` still ends in a
+ * shell. Only the first stage used to be read, and that masked the body. A `||`
+ * is a logical separator and runs nothing on success, so it is skipped, and a
+ * quoted `|` is an argument character rather than a pipe. `|&` is a real pipe
+ * that also redirects stderr, so only the first of its two characters is a pipe
+ * separator.
+ */
+function pipedStageHeads(text: string): string[] {
+	const heads: string[] = [];
+	let quote: "'" | '"' | undefined;
+	let escaped = false;
+	for (let i = 0; i < text.length; i += 1) {
+		const char = text[i]!;
+		if (escaped) {
+			escaped = false;
+			continue;
+		}
+		if (char === "\\" && quote !== "'") {
+			escaped = true;
+			continue;
+		}
+		if (char === "'" || char === '"') {
+			if (!quote) quote = char;
+			else if (quote === char) quote = undefined;
+			continue;
+		}
+		if (quote || char !== "|") continue;
+		// `||` runs the right side only on failure and hands it no body. `|&` is a
+		// real pipe that also redirects stderr, so it is read as one pipe stage.
+		if (text[i + 1] === "|") {
+			i += 1;
+			continue;
+		}
+		const stageStart = text[i + 1] === "&" ? i + 2 : i + 1;
+		heads.push(shellWords(text.slice(stageStart))[0] ?? "");
+	}
+	return heads;
+}
+
+/** Whether the command on a heredoc's header line executes its body. */
+export function heredocBodyIsCode(command: string, start: number): boolean {
+	const lineStart = command.lastIndexOf("\n", start - 1) + 1;
+	const lineEnd = command.indexOf("\n", start);
+	const line = command.slice(lineStart, lineEnd < 0 ? command.length : lineEnd);
+	const before = command.slice(lineStart, start);
+	if (HEREDOC_EXECUTOR_RE.test(before)) return true;
+	// Every piped stage is a consumer. Treating an unknown one as a filter let
+	// `cat <<EOF | ash` and `cat <<EOF | grep x | sh` run unread.
+	for (const head of pipedStageHeads(line.slice(start - lineStart))) {
+		if (!DATA_HEREDOC_CONSUMERS.has(head.toLowerCase())) return true;
+	}
+	const program = consumerProgram(before);
+	return !program || !readsBodyAsData(program);
+}
+
+/**
+ * `command` with every heredoc body that no command executes replaced by
+ * spaces.
+ *
+ * A body feeding `cat` is prose, so a line in it that looks like a command must
+ * not be judged as one. Newlines are preserved so line-oriented callers still
+ * see the same line structure.
+ */
+export function maskDataHeredocBodies(command: string): string {
+	let out = "";
+	let cursor = 0;
+	for (const range of heredocRanges(command)) {
+		if (heredocBodyIsCode(command, range.start)) continue;
+		out += command.slice(cursor, range.bodyStart);
+		out += command.slice(range.bodyStart, range.bodyEnd).replace(/[^\n]/g, " ");
+		cursor = range.bodyEnd;
+	}
+	return out + command.slice(cursor);
+}
+
 export function splitShellSegments(command: string): string[] {
 	const segments: string[] = [];
 	let current = "";
@@ -227,6 +514,26 @@ export function splitShellSegments(command: string): string[] {
 			current += char;
 			continue;
 		}
+		// A heredoc body belongs to the segment that introduces it. Splitting it
+		// on its newlines made data read as code, so documentation quoting a
+		// blocked command shape was refused, and a code body's own lines looked
+		// like top-level commands.
+		//
+		// One header line can introduce several heredocs (`cat <<A <<B`), and the
+		// shell reads their bodies one after another. Consume the whole run here,
+		// so a later body is not read as top-level commands.
+		if (!quote && char === "<" && command[i + 1] === "<") {
+			const run = heredocRunAt(command, i);
+			if (run) {
+				let cursor = i;
+				for (const range of run) {
+					current += command.slice(cursor, range.end);
+					cursor = range.end;
+				}
+				i = cursor - 1;
+				continue;
+			}
+		}
 		if (!quote && /[\n|;&]/.test(char)) {
 			if (current.trim()) segments.push(current.trim());
 			current = "";
@@ -238,7 +545,7 @@ export function splitShellSegments(command: string): string[] {
 	return segments;
 }
 
-function unwrapShellWordsImpl(command: string): { words: string[]; changesCwd: boolean } {
+function unwrapShellWordsImpl(command: string): { words: string[]; changesCwd: boolean; cwdTarget?: string; assignments: string[] } {
 	let cleanCmd = decodeShellEscapes(command.trim());
 	// Strip outer grouping parentheses/braces if present e.g. ( git commit ) or { git commit; }
 	while (
@@ -252,6 +559,16 @@ function unwrapShellWordsImpl(command: string): { words: string[]; changesCwd: b
 	const words = shellWords(cleanCmd);
 	let i = 0;
 	let changesCwd = false;
+	// Directory a wrapper (`env -C`, `sudo -D`) runs its command in. Recorded
+	// so a caller can judge a nested command in the tree it actually runs in
+	// instead of the shell's initial workdir.
+	let cwdTarget: string | undefined;
+	// Environment assignments consumed while unwrapping wrappers. They apply to
+	// the command that follows, so callers must not read them out of the command
+	// arguments instead: `git commit -m GIT_DIR=/repo` is message text, not an
+	// assignment, and treating it as one would claim a repository that the
+	// invocation never named.
+	const assignments: string[] = [];
 	while (i < words.length) {
 		const rawWord = words[i]!;
 		const word = rawWord.replace(/^[\({]+/, "").replace(/[\)};]+$/, "");
@@ -337,7 +654,12 @@ function unwrapShellWordsImpl(command: string): { words: string[]; changesCwd: b
 			]);
 			while (i < words.length && words[i]!.startsWith("-")) {
 				const option = words[i++]!;
-				if (option === "-D" || (option.startsWith("-D") && option !== "-D") || option === "--chdir" || option.startsWith("--chdir=")) changesCwd = true;
+				if (option === "-D" || (option.startsWith("-D") && option !== "-D") || option === "--chdir" || option.startsWith("--chdir=")) {
+					changesCwd = true;
+					// The target is the next word unless it was attached (-Ddir).
+					if (option === "-D" || option === "--chdir") cwdTarget = words[i];
+					else cwdTarget = option.replace(/^-D|^--chdir=/, "");
+				}
 				if (valueOptions.has(option) && i < words.length) i++;
 			}
 			continue;
@@ -360,6 +682,7 @@ function unwrapShellWordsImpl(command: string): { words: string[]; changesCwd: b
 					break;
 				}
 				if (/^[A-Za-z_][A-Za-z0-9_]*=/.test(w)) {
+					assignments.push(w);
 					i++;
 					continue;
 				}
@@ -369,7 +692,8 @@ function unwrapShellWordsImpl(command: string): { words: string[]; changesCwd: b
 				if (clustered?.[1] === "C") {
 					changesCwd = true;
 					i++;
-					if (!clustered[2] && i < words.length) i++;
+					if (clustered[2]) cwdTarget = clustered[2];
+					else if (i < words.length) cwdTarget = words[i++];
 					continue;
 				}
 				if (clustered?.[1] === "S") {
@@ -397,17 +721,24 @@ function unwrapShellWordsImpl(command: string): { words: string[]; changesCwd: b
 				}
 				if (w.startsWith("-C") && w !== "-C") {
 					changesCwd = true;
+					cwdTarget = w.slice(2);
 					i++;
 					continue;
 				}
 				if (!w.startsWith("-")) break;
 				i++;
-				if (w === "-C" || w === "--chdir" || w.startsWith("--chdir=")) changesCwd = true;
+				if (w === "-C" || w === "--chdir" || w.startsWith("--chdir=")) {
+					changesCwd = true;
+					if (w === "--chdir") cwdTarget = words[i];
+					else if (w.startsWith("--chdir=")) cwdTarget = w.slice("--chdir=".length);
+					else cwdTarget = words[i];
+				}
 				if (valueOptions.has(w) && i < words.length) i++;
 			}
 			continue;
 		}
 		if (/^[A-Za-z_][A-Za-z0-9_]*=/.test(word)) {
+			assignments.push(word);
 			i++;
 			continue;
 		}
@@ -416,13 +747,379 @@ function unwrapShellWordsImpl(command: string): { words: string[]; changesCwd: b
 	return { words: words
 		.slice(i)
 		.map((w) => w.replace(/^[\({]+/, "").replace(/[\)};]+$/, ""))
-		.filter(Boolean), changesCwd };
+		.filter(Boolean), changesCwd, cwdTarget, assignments };
 }
 
 export function unwrapShellWords(command: string): string[] {
 	return unwrapShellWordsImpl(command).words;
 }
 
+/**
+ * The directory a wrapper (`env -C`, `sudo -D`) runs its command in, if it
+ * names one. The target is left unexpanded: the caller knows the directory it
+ * is relative to.
+ */
+export function shellWrapperCwd(command: string): string | undefined {
+	return unwrapShellWordsImpl(command).cwdTarget;
+}
+
+/**
+ * One shell segment and the directory it runs in.
+ *
+ * `root` is `undefined` when the walk cannot determine it: an unresolvable
+ * `cd` target (`cd $DIR`, `cd -`), a `pushd` that swaps the directory stack,
+ * or a `popd` with no tracked push. Callers treat that as indeterminate
+ * instead of assuming the session's checkout, because that assumption is how a
+ * `cd` into another tree used to hide a mutation from the boundary policy.
+ */
+export interface ShellSegmentPlan {
+	segment: string;
+	root: string | undefined;
+}
+
+const CWD_WORDS = new Set(["cd", "pushd", "popd"]);
+
+/**
+ * The directory a git repository selector names.
+ *
+ * `git -C ~/repo` reaches a directory through the shell's own expansion, so a
+ * literal `~` must not be read as a workspace-relative path. Only the forms a
+ * shell expands deterministically are expanded here. Anything else keeps its
+ * literal text, so an unresolvable `$VARIABLE` still fails closed through
+ * `hasUnresolvableVariable`.
+ */
+export function resolveGitRepoSelector(baseDir: string, value: string): string {
+	return resolve(baseDir, expandCwdTarget(value) ?? value);
+}
+
+/** Expands only the directory forms a shell expands deterministically. */
+export function expandCwdTarget(target: string): string | undefined {
+	const home = process.env.HOME || homedir();
+	if (target === "~") return home;
+	if (target.startsWith("~/")) return join(home, target.slice(2));
+	// `cd ~user` resolves to that user's home, never to a literal `~user`
+	// directory. Reading it as a relative path put a `cd ~csh` walk inside the
+	// workspace while the shell reached the real home. The path after the name
+	// is part of the target, so dropping it judged the wrong directory.
+	const named = /^~([^/\s]+)(\/.*)?$/.exec(target);
+	if (named) {
+		const user = named[1]!;
+		const tail = named[2] ?? "";
+		let base: string;
+		try {
+			// The walk cannot read another user's home portably, so it joins the name
+			// onto this home's parent. That is the same heuristic `expandShellTargetPath`
+			// uses, which keeps both scanners agreeing on one directory. A guess that
+			// lands inside the workspace arrives here only when the workspace already
+			// spans the home parent, the same exposure the boundary scan has.
+			base = user === userInfo().username ? home : join(dirname(home), user);
+		} catch {
+			return undefined;
+		}
+		return tail === "" ? base : join(base, tail.slice(1));
+	}
+	if (/^\$(?:HOME|\{HOME\})(?:$|[/\\])/.test(target)) {
+		return target.replace(/^\$(?:HOME|\{HOME\})/, home);
+	}
+	if (target.includes("$")) return undefined;
+	return target;
+}
+
+/** Words that carry no directory of their own; a control-flow body follows them. */
+const TRANSPARENT_PREFIX = new Set([
+	"builtin",
+	"command",
+	"then",
+	"else",
+	"elif",
+	"do",
+	"if",
+	"while",
+	"until",
+	"for",
+	"!",
+	"time",
+]);
+
+/**
+ * A `pushd` operand that rotates or only pushes the stack.
+ *
+ * `pushd +N`/`-N` swaps entries, and `-n` pushes without changing directory.
+ * The walk tracks a single saved directory, so it cannot model either. Both are
+ * reported as an unchecked change, which fails closed for a mutation.
+ */
+function pushdRotation(operand: string): boolean {
+	return /^[+-]\d+$/.test(operand) || operand === "-n" || /^-n./.test(operand);
+}
+
+function cwdChangeIn(segment: string, insideCase = false): { word: string; target?: string; unmodeled?: boolean } | undefined {
+	const words = unwrapShellWords(segment);
+	let i = 0;
+	if (insideCase) {
+		// `case x in a) cd /other;; esac` arrives as `a) cd /other`, and
+		// `unwrapShellWords` has already stripped the pattern's closing paren.
+		// The pattern word runs nothing, so the command after it is the body.
+		while (i < words.length && !CWD_WORDS.has(words[i]!) && !TRANSPARENT_PREFIX.has(words[i]!) && words[i] !== "case") i += 1;
+	}
+	while (i < words.length) {
+		const word = words[i]!;
+		if (TRANSPARENT_PREFIX.has(word)) {
+			i += 1;
+			continue;
+		}
+		// `case` opens an arm wherever it appears, including after `then`, `do`,
+		// or `else`: the pattern is one word after `in`, and the body follows the
+		// pattern. Treating only a leading `case` as an arm hid `then case x in x)
+		// cd <other>`, so the `cd` was never seen.
+		if (word === "case") {
+			const inIndex = words.indexOf("in", i);
+			i = inIndex >= 0 ? inIndex + 2 : words.length;
+			continue;
+		}
+		if (CWD_WORDS.has(word)) break;
+		// A case pattern closes with `)`, and the command after it runs here.
+		if (word.endsWith(")") || word === "}") {
+			i += 1;
+			continue;
+		}
+		return undefined;
+	}
+	const word = words[i];
+	if (!word || !CWD_WORDS.has(word)) return undefined;
+	if (word === "popd") {
+		// `popd +N`/`-N` rotate the stack instead of popping it, which this walk
+		// does not model. A plain `popd` consumes the top entry.
+		return words.slice(i + 1).some((operand) => pushdRotation(operand)) ? { word, unmodeled: true } : { word };
+	}
+	let target: string | undefined;
+	let unmodeled = false;
+	for (let j = i + 1; j < words.length; j += 1) {
+		const candidate = words[j]!;
+		if (candidate === "--") continue;
+		if (word === "pushd" && pushdRotation(candidate)) {
+			unmodeled = true;
+			continue;
+		}
+		// `cd -` moves to OLDPWD, which this walk does not track.
+		if (candidate === "-") return { word, target: undefined };
+		if (candidate.startsWith("-")) continue;
+		target = candidate;
+		break;
+	}
+	if (unmodeled) return { word, unmodeled: true };
+	if (target !== undefined) return { word, target };
+	// A bare `cd` moves to $HOME, a known directory. A bare `pushd` swaps the
+	// top two entries of the directory stack, which is not tracked here.
+	return { word, target: word === "cd" ? "~" : undefined };
+}
+
+/**
+ * Unquoted parenthesis counts in a segment.
+ *
+ * `splitShellSegments` splits on `&&`/`;`/`|` but not on grouping, so a
+ * subshell arrives as segments that still carry its parens. A `cd` inside one
+ * must not move the directory of the segments after the subshell closes.
+ *
+ * Inside a `case` body a pattern's own `)` is not a grouping close, so it is
+ * not counted while a case construct is open.
+ */
+function groupParensIn(segment: string, insideCase = false): { opens: number; closes: number } {
+	// Which characters sit outside quotes, so a `)` inside a string is never a
+	// group or pattern terminator.
+	const unquoted: boolean[] = [];
+	let quote: "'" | '"' | undefined;
+	let escaped = false;
+	for (let i = 0; i < segment.length; i += 1) {
+		const char = segment[i]!;
+		let bare = false;
+		if (escaped) escaped = false;
+		else if (char === "\\" && quote !== "'") escaped = true;
+		else if (char === "'" || char === '"') {
+			if (!quote) quote = char;
+			else if (quote === char) quote = undefined;
+		} else if (!quote) bare = true;
+		unquoted.push(bare);
+	}
+	// A `case` arm pattern ends with `)`, which closes no group. An arm can open
+	// in this segment (`case x in x) …`) or arrive already open (`x) …`), so both
+	// forms are recognised. Missing the in-segment form let a pattern's paren pop
+	// a subshell frame, which then swallowed the real closing paren.
+	const patternCloses = new Set<number>();
+	const spelled = unquoted.map((bare, i) => (bare ? segment[i]! : " ")).join("");
+	if (/[^\s]/.test(spelled)) {
+		for (const head of spelled.matchAll(/\bcase\b[^;|&]*?\bin\b/g)) {
+			let from = (head.index ?? 0) + head[0].length;
+			while (from < spelled.length) {
+				const close = spelled.indexOf(")", from);
+				if (close < 0) break;
+				patternCloses.add(close);
+				const nextArm = spelled.indexOf(";;", close);
+				if (nextArm < 0) break;
+				from = nextArm + 2;
+			}
+		}
+	}
+	let opens = 0;
+	let closes = 0;
+	// At segment start, `esac )` carries a real group close, not an arm pattern.
+	// Reading it as a pattern left the surrounding subshell open, so the segment
+	// after it inherited the arm's directory instead of the outer one.
+	const startsAtClose = /^\s*(?:esac|fi|done)\b/.test(segment);
+	for (let i = 0; i < segment.length; i += 1) {
+		if (!unquoted[i]) continue;
+		const char = segment[i]!;
+		if (char === "(") opens += 1;
+		else if (char === ")") {
+			if (patternCloses.has(i)) continue;
+			// An arm pattern that was already open when this segment began ends
+			// here. It closes no group.
+			if (insideCase && !startsAtClose && closes + 1 > opens) continue;
+			closes += 1;
+		}
+	}
+	return { opens, closes };
+}
+
+/**
+ * Which construct boundary a segment carries, if any.
+ *
+ * Only a keyword in command position counts: `echo done` and
+ * `git commit -m "if"` carry no construct boundary.
+ */
+function constructKeywordIn(segment: string): "open" | "close" | "case-open" | "case-close" | undefined {
+	const words = unwrapShellWords(segment);
+	let i = 0;
+	// Wrappers and negation can precede the keyword: `time if …`, `! case …`.
+	while (i < words.length && (words[i] === "builtin" || words[i] === "command" || words[i] === "time" || words[i] === "!")) i += 1;
+	// A body opener can precede a nested construct: `then case …`, `do case …`.
+	// Without this the nested `case` opened no frame, so its `esac` popped the
+	// enclosing `if`/`for` frame instead and the arm's `cd` was never recorded.
+	// `if`/`while`/`until`/`for` stay out of this list, because they are the
+	// openers this function reports.
+	while (i < words.length && (words[i] === "then" || words[i] === "else" || words[i] === "do" || words[i] === "elif")) i += 1;
+	const word = words[i];
+	if (word === "if" || word === "while" || word === "until" || word === "for") return "open";
+	if (word === "case") return "case-open";
+	if (word === "esac") return "case-close";
+	if (word === "fi" || word === "done") return "close";
+	return undefined;
+}
+
+/**
+ * Whether a segment runs `eval` as its command word.
+ *
+ * Only a leading `eval` runs its body in this shell. A prose mention
+ * (`Use eval "…" to show the bypass`) names no command, and forwarding it made
+ * a documentation heredoc look like a directory change. The wrappers that do not
+ * fork (`builtin`, `command`, `time`) and a leading assignment (`A=1 eval …`)
+ * both still reach `eval`, so they are skipped here.
+ */
+export function isLeadingEvalSegment(segment: string): boolean {
+	const words = shellWords(segment);
+	let i = 0;
+	while (i < words.length) {
+		const word = words[i]!;
+		if (word === "builtin" || word === "command" || word === "time") {
+			i += 1;
+			continue;
+		}
+		if (/^[A-Za-z_][A-Za-z0-9_]*=/.test(word)) {
+			i += 1;
+			continue;
+		}
+		break;
+	}
+	return words[i] === "eval";
+}
+
+/**
+ * Walk a multi-segment command and resolve the directory each segment runs in.
+ *
+ * Later segments run in the directory an earlier `cd`/`pushd` selected, and a
+ * wrapper `-C`/`-D` moves only its own command. Judging every segment against
+ * the shell's initial `workdir` let `cd <other-tree> && git commit` be checked
+ * as if it happened in the session's own checkout.
+ *
+ * A `cd` inside a grouping subshell is scoped to it, so the segments after the
+ * closing paren keep the outer directory. Without that, `( cd /other && ls )
+ * && rm -rf dist` would read the removal as happening in `/other`.
+ *
+ * A `cd` inside a conditional or loop body is NOT scoped: that body runs in the
+ * current shell. Which body runs depends on a condition this walk cannot
+ * evaluate, so a construct whose bodies change directory leaves the later
+ * directory undetermined rather than letting the last body parsed win.
+ */
+export function planShellSegments(command: string, baseDir: string | undefined, evalBody?: (segment: string) => string | undefined): ShellSegmentPlan[] {
+	const plan: ShellSegmentPlan[] = [];
+	let cwd = baseDir;
+	const dirStack: Array<string | undefined> = [];
+	// Each open grouping paren saves the outer directory and the outer stack.
+	// A `pushd` inside `( … )` must not leave an entry a later outer `popd`
+	// consumes: bash discards the subshell's stack with it.
+	const groupStack: Array<{ cwd: string | undefined; dirStack: Array<string | undefined> }> = [];
+	// One frame per open `if`/`while`/`until`/`for`/`case`, recording whether any
+	// body changed directory.
+	const constructStack: Array<{ changed: boolean }> = [];
+	let caseDepth = 0;
+	for (const segment of splitShellSegments(command)) {
+		const { opens, closes } = groupParensIn(segment, caseDepth > 0);
+		for (let i = 0; i < opens; i += 1) groupStack.push({ cwd, dirStack: [...dirStack] });
+		const wrapperCwd = shellWrapperCwd(segment);
+		let root = cwd;
+		if (wrapperCwd !== undefined) {
+			const expanded = expandCwdTarget(wrapperCwd);
+			root = expanded === undefined || cwd === undefined ? undefined : resolve(cwd, expanded);
+		}
+		plan.push({ segment, root });
+		const keyword = constructKeywordIn(segment);
+		if (keyword === "open") constructStack.push({ changed: false });
+		if (keyword === "case-open") {
+			constructStack.push({ changed: false });
+			caseDepth += 1;
+		}
+		// `eval` runs its body in the current shell, so a `cd` inside it moves the
+		// directory of later segments. The body is supplied by the caller because
+		// extracting it lives in the interpreter policy, not here.
+		const evaluated = evalBody?.(segment);
+		const change = cwdChangeIn(segment, caseDepth > 0) ?? (evaluated === undefined ? undefined : cwdChangeIn(evaluated, caseDepth > 0));
+		if (change) {
+			for (const frame of constructStack) frame.changed = true;
+			if (change.unmodeled) {
+				cwd = undefined;
+			} else if (change.word === "popd") {
+				// `popd` with nothing to pop fails in bash and leaves the
+				// directory alone, so it must not clear a tracked directory.
+				cwd = dirStack.length > 0 ? dirStack.pop() : cwd;
+			} else {
+				if (change.word === "pushd") dirStack.push(cwd);
+				const expanded = change.target === undefined ? undefined : expandCwdTarget(change.target);
+				cwd = cwd === undefined || expanded === undefined ? undefined : resolve(cwd, expanded);
+			}
+		}
+		if (keyword === "case-close") caseDepth = Math.max(0, caseDepth - 1);
+		if (keyword === "close" || keyword === "case-close") {
+			const frame = constructStack.pop();
+			// A body that changed directory only sometimes leaves the later
+			// directory dependent on a condition, which fails closed.
+			if (frame?.changed) cwd = undefined;
+		}
+		for (let i = 0; i < closes; i += 1) {
+			const outer = groupStack.pop();
+			if (outer) {
+				cwd = outer.cwd;
+				dirStack.length = 0;
+				dirStack.push(...outer.dirStack);
+			}
+		}
+	}
+	return plan;
+}
+
+/** Environment assignments that precede the command word, after unwrapping. */
+export function shellPrefixAssignments(command: string): string[] {
+	return unwrapShellWordsImpl(command).assignments;
+}
+
 export function shellWrappersChangeCwd(command: string): boolean {
 	return unwrapShellWordsImpl(command).changesCwd;
 }