agentfootprint 11.4.0 → 12.0.0

package/dist/adapters/code/fileName.js

@@ -0,0 +1,100 @@
+"use strict";
+/**
+ * The one file-name encoder both code adapters stage inputs under.
+ *
+ * A staged input's name is caller data landing in a filesystem — a local temp
+ * directory for `localCodeRunner`, the sandbox's own disk for
+ * `agentCoreCodeRunner`. Both turn it into ONE inert segment through this
+ * function, so the same name lands under the same file name on either backend
+ * and two names never become one file on either.
+ */
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.safeFileName = void 0;
+const sha256_js_1 = require("../../lib/sha256.js");
+/** Longest file name this adapter will write. Well inside every POSIX
+ *  `NAME_MAX`, and the cap this adapter has always used. */
+const FILE_NAME_MAX = 120;
+/** Marks a name this adapter had to rewrite. Nothing arm A returns may start
+ *  with it. */
+const ENCODED_FILE_PREFIX = '_enc_';
+/** Already one inert segment: no separator, no escape-looking bytes. */
+const FILE_NAME_ALREADY_LEGAL = /^[A-Za-z0-9._-]+$/;
+/** `.`, `..`, `...` — a name that is nothing but dots. Not a hop once the
+ *  separators are gone, but not a name either, so it goes to arm B. */
+const DOTS_ONLY = /^\.+$/;
+/** Legal AND not the escape character — the set {@link encodeFileName} passes
+ *  through untouched. */
+const FILE_CHAR_PASSES_THROUGH = /[A-Za-z0-9.-]/;
+/**
+ * One caller-supplied name → one inert file-name segment, **without ever
+ * folding two names into one file**.
+ *
+ * Separators cannot survive as a parent hop and `a/b` cannot become two
+ * segments — the same law `artifacts/scopePath` states for scope tuples,
+ * applied to the one other place caller strings become file names in this
+ * package. But the old version got there by REPLACING every illegal character
+ * with `_`, which meant `a/b.csv` and `a_b.csv` became the same file: the
+ * second write clobbered the first's bytes, `stageInputs` answered two manifest
+ * entries pointing at one path, and the model's code opened `a/b.csv` and read
+ * whatever `a_b.csv` contained. Nothing failed. The computation was simply run
+ * against the wrong dataset — which is the worst shape a data bug can take,
+ * because the answer still looks like an answer. `.slice(0, 120)` folded the
+ * same way for any two names agreeing on their first 120 characters.
+ *
+ * So this ENCODES instead of sanitising, in two arms whose outputs cannot meet
+ * (the shape `safeSessionId` uses in `adapters/hosting/agentcore.ts`):
+ *
+ *   A. already one legal segment, not claiming to be an encoded name, and not
+ *      dots-only → returned BYTE FOR BYTE. This is what keeps the fix from
+ *      renaming anything that works today: `dataset.json`, `a_b.json`, every
+ *      name `codeRunnerTool` derives from a declared argument, all land at
+ *      exactly the path they landed at before.
+ *   B. anything else → `_enc_` + an escaped form, where `_` is the escape
+ *      character (`_` → `__`, any other illegal UTF-16 unit → `_uXXXX`). A
+ *      decoder is a left inverse of that, which is what makes it injective.
+ *
+ * Arm A never starts with `_enc_` (excluded by construction) and every arm B
+ * output does, so no name from one arm can collide with a name from the other.
+ * Both arms produce ASCII with no `/`, no `\` and no NUL, so the path-hop law
+ * the old version enforced is unchanged.
+ */
+function safeFileName(raw) {
+    if (raw.length <= FILE_NAME_MAX &&
+        FILE_NAME_ALREADY_LEGAL.test(raw) &&
+        !raw.startsWith(ENCODED_FILE_PREFIX) &&
+        !DOTS_ONLY.test(raw)) {
+        return raw;
+    }
+    const encoded = ENCODED_FILE_PREFIX + encodeFileName(raw);
+    // Strictly less, so the two shapes below are told apart by LENGTH alone: a
+    // digested name is always exactly FILE_NAME_MAX and an escaped one never is.
+    if (encoded.length < FILE_NAME_MAX)
+        return encoded;
+    // Past the cap no mapping can stay injective — there are more names than
+    // there are file names — so the tail becomes a digest of the WHOLE raw name.
+    // SHA-256 rather than the package's 32-bit FNV-1a: a name reaching this
+    // adapter is caller data, and a 32-bit digest is a collision somebody can go
+    // and find, which for a file the model then reads as data is the same defect
+    // this function exists to close. `lib/sha256.ts` — the library's own,
+    // checked against `node:crypto` — so `agentCoreCodeRunner`, which may run
+    // where there is no `node:crypto`, encodes a name exactly as the local runner
+    // does.
+    const digest = (0, sha256_js_1.sha256Hex)(raw).slice(0, 32);
+    return `${encoded.slice(0, FILE_NAME_MAX - digest.length - 2)}_z${digest}`;
+}
+exports.safeFileName = safeFileName;
+/** `_` → `__`, anything else illegal → `_uXXXX`. Injective by construction. */
+function encodeFileName(raw) {
+    let out = '';
+    for (let i = 0; i < raw.length; i++) {
+        const ch = raw[i];
+        if (ch === '_')
+            out += '__';
+        else if (FILE_CHAR_PASSES_THROUGH.test(ch))
+            out += ch;
+        else
+            out += `_u${raw.charCodeAt(i).toString(16).padStart(4, '0')}`;
+    }
+    return out;
+}
+//# sourceMappingURL=fileName.js.map

package/dist/adapters/code/fileName.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"fileName.js","sourceRoot":"","sources":["../../../src/adapters/code/fileName.ts"],"names":[],"mappings":";AAAA;;;;;;;;GAQG;;;AAEH,mDAAgD;AAEhD;4DAC4D;AAC5D,MAAM,aAAa,GAAG,GAAG,CAAC;AAE1B;eACe;AACf,MAAM,mBAAmB,GAAG,OAAO,CAAC;AAEpC,wEAAwE;AACxE,MAAM,uBAAuB,GAAG,mBAAmB,CAAC;AAEpD;uEACuE;AACvE,MAAM,SAAS,GAAG,OAAO,CAAC;AAE1B;yBACyB;AACzB,MAAM,wBAAwB,GAAG,eAAe,CAAC;AAEjD;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAgCG;AACH,SAAgB,YAAY,CAAC,GAAW;IACtC,IACE,GAAG,CAAC,MAAM,IAAI,aAAa;QAC3B,uBAAuB,CAAC,IAAI,CAAC,GAAG,CAAC;QACjC,CAAC,GAAG,CAAC,UAAU,CAAC,mBAAmB,CAAC;QACpC,CAAC,SAAS,CAAC,IAAI,CAAC,GAAG,CAAC,EACpB,CAAC;QACD,OAAO,GAAG,CAAC;IACb,CAAC;IACD,MAAM,OAAO,GAAG,mBAAmB,GAAG,cAAc,CAAC,GAAG,CAAC,CAAC;IAC1D,2EAA2E;IAC3E,6EAA6E;IAC7E,IAAI,OAAO,CAAC,MAAM,GAAG,aAAa;QAAE,OAAO,OAAO,CAAC;IACnD,yEAAyE;IACzE,6EAA6E;IAC7E,wEAAwE;IACxE,6EAA6E;IAC7E,6EAA6E;IAC7E,sEAAsE;IACtE,0EAA0E;IAC1E,8EAA8E;IAC9E,QAAQ;IACR,MAAM,MAAM,GAAG,IAAA,qBAAS,EAAC,GAAG,CAAC,CAAC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;IAC3C,OAAO,GAAG,OAAO,CAAC,KAAK,CAAC,CAAC,EAAE,aAAa,GAAG,MAAM,CAAC,MAAM,GAAG,CAAC,CAAC,KAAK,MAAM,EAAE,CAAC;AAC7E,CAAC;AAxBD,oCAwBC;AAED,+EAA+E;AAC/E,SAAS,cAAc,CAAC,GAAW;IACjC,IAAI,GAAG,GAAG,EAAE,CAAC;IACb,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,GAAG,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACpC,MAAM,EAAE,GAAG,GAAG,CAAC,CAAC,CAAW,CAAC;QAC5B,IAAI,EAAE,KAAK,GAAG;YAAE,GAAG,IAAI,IAAI,CAAC;aACvB,IAAI,wBAAwB,CAAC,IAAI,CAAC,EAAE,CAAC;YAAE,GAAG,IAAI,EAAE,CAAC;;YACjD,GAAG,IAAI,KAAK,GAAG,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,EAAE,CAAC,CAAC,QAAQ,CAAC,CAAC,EAAE,GAAG,CAAC,EAAE,CAAC;IACrE,CAAC;IACD,OAAO,GAAG,CAAC;AACb,CAAC"}