@enricai/barnacle 1.12.80 → 1.12.82

package/dist/scripts/recon-generate.js

@@ -82,7 +82,6 @@ exports.getScanPrimaryCandidateGroupsCallCountForTest = getScanPrimaryCandidateG
 exports.resetScanPrimaryCandidateGroupsCallCountForTest = resetScanPrimaryCandidateGroupsCallCountForTest;
 exports.detectDrillDownFoldPlan = detectDrillDownFoldPlan;
 exports.parseFoldReturnSpec = parseFoldReturnSpec;
-exports.parseBrowserFlowOnlyOptIn = parseBrowserFlowOnlyOptIn;
 exports.parseFallbackGateSpec = parseFallbackGateSpec;
 exports.readComparableBodyLeaf = readComparableBodyLeaf;
 exports.collectRequestValuesIncludingHeaders = collectRequestValuesIncludingHeaders;
@@ -96,6 +95,7 @@ exports.emitContractTs = emitContractTs;
 exports.unreferencedRequiredUrlFields = unreferencedRequiredUrlFields;
 exports.assertRequiredUrlFieldsReferenced = assertRequiredUrlFieldsReferenced;
 exports.unreferencedDeclaredPayloadFields = unreferencedDeclaredPayloadFields;
+exports.unreadPayloadFieldChecklistLines = unreadPayloadFieldChecklistLines;
 exports.buildStepInstructionExpr = buildStepInstructionExpr;
 exports.emitConfigManifest = emitConfigManifest;
 exports.emitBrowserFlowTs = emitBrowserFlowTs;
@@ -481,7 +481,7 @@ function walkNavigateToFacets(flowSteps, env) {
                     origin: "step",
                     literals: collectStepFacetLiterals(step.step, env),
                     kind: deriveFacetKind(step.payloadField),
-                    kindHints: [],
+                    kindHints: collectStepFacetKindHints(step.step, env),
                 });
             }
             continue;
@@ -523,6 +523,45 @@ function collectStepFacetLiterals(instruction, env) {
     const primary = extractStepPersonaValue(instruction, env);
     return uniqueLiterals(primary, findInstructionSpans(instruction).map((span) => span.text));
 }
+const STEP_KIND_STOP_WORDS = new Set([
+    "facet",
+    "filter",
+    "field",
+    "value",
+    "option",
+    "selection",
+    "the",
+    "a",
+    "an",
+    "dropdown",
+    "drop-down",
+    "selector",
+    "menu",
+    "list",
+    "widget",
+    "question",
+]);
+/**
+ * Element kinds a click step's own prose names: the noun phrase trailing the
+ * value span ("select the 'Lisbon' city option" names `city`), so a facet whose
+ * field name differs from the page's vocabulary can still own the element of
+ * that kind. Reads only the generic instruction grammar, no site vocabulary.
+ */
+function collectStepFacetKindHints(instruction, env) {
+    const value = extractStepPersonaValue(instruction, env);
+    const span = findInstructionSpans(instruction).find((candidate) => candidate.text === value);
+    if (span === undefined)
+        return [];
+    const phrase = instruction
+        .slice(span.end)
+        .split(/[.,;:]|\b(?:from|in|on|for|with|and|then|to)\b/i)[0];
+    const words = phrase
+        .split(/\s+/)
+        .map((word) => word.replace(/^['"]+|['"]+$/g, ""))
+        .filter((word) => /^[A-Za-z][A-Za-z0-9_-]*$/.test(word));
+    const kindWords = words.filter((word) => !STEP_KIND_STOP_WORDS.has(word.toLowerCase()));
+    return [...new Set([...(kindWords.length > 1 ? [kindWords.join(" ")] : []), ...kindWords])];
+}
 function collectHashFacetLiterals(url, previousHash) {
     const primary = extractNavigateToHashFragmentValue(url, previousHash);
     const hashIndex = url.indexOf("#");
@@ -6800,18 +6839,34 @@ function resolveRecurringNavigateToFacets(declared, actions) {
     // recurs in one vocabulary (a hash slug, a quoted label) says nothing about
     // the same facet's element in another (`WW;filterType=ship`).
     const claimedLiterals = new Set();
-    const aliased = declared.flatMap((facet) => {
-        const literal = findKindKeyedLiteral(facet, actions);
-        if (literal === undefined || declared.some(({ value }) => value === literal))
+    const pairsByField = new Map();
+    const aliased = declared
+        .flatMap((facet) => findKindKeyedLiterals(facet, actions).flatMap((literal) => {
+        if (declared.some(({ value }) => value === literal))
             return [];
         if (claimedLiterals.has(literal))
             return [];
         claimedLiterals.add(literal);
         const known = [facet.value, ...(facet.literals ?? [])];
+        // A case-only variant is already owned by the declared facet, whose
+        // renderer re-cases any caller value to the captured element's casing.
         if (known.some((candidate) => candidate.toLowerCase() === literal.toLowerCase()))
             return [];
-        const valueMap = new Map(facet.value.length > 0 ? [[facet.value, literal]] : []);
-        return [{ ...facet, value: literal, literals: [literal], keyedByKind: true, valueMap }];
+        // The traffic carries codes, not which declared value each belongs to;
+        // pair only where the code abbreviates the value, never by position.
+        const sameField = declared.filter(({ field, value }) => field === facet.field && value.length > 0);
+        const abbreviated = sameField.filter(({ value }) => abbreviates(literal, value));
+        const owners = abbreviated.length > 0 || sameField.length !== 1 ? abbreviated : sameField;
+        if (owners.length > 0)
+            pairsByField.set(facet.field, [
+                ...(pairsByField.get(facet.field) ?? []),
+                ...owners.map((owner) => [owner.value, literal]),
+            ]);
+        return [{ ...facet, value: literal, literals: [literal], keyedByKind: true }];
+    }))
+        .map((alias) => {
+        const pairs = pairsByField.get(alias.field) ?? [];
+        return pairs.length === 0 ? alias : { ...alias, valueMap: new Map(pairs) };
     });
     if (unreachable.length === 0 && aliased.length === 0)
         return resolvedDeclared;
@@ -6837,13 +6892,15 @@ function resolveRecurringNavigateToFacets(declared, actions) {
     return [...withAliases, ...extras];
 }
 /**
- * The literal a facet takes in the code vocabulary: the captured request
+ * The literals a facet takes in the code vocabulary: the captured request
  * element (query value or body string leaf) whose `<literal><delim>key=<kind>`
  * suffix names the facet's own kind. Derived from the traffic itself, so it
  * never depends on how many other facets or transitions the flow declared.
- * Returns `undefined` unless exactly one distinct literal carries the kind.
+ * A kind carrying several distinct literals across captures yields each of
+ * them, so every element of the kind is owned instead of none; the month
+ * evidence narrows them only when it singles out some of them.
  */
-function findKindKeyedLiteral(facet, actions) {
+function findKindKeyedLiterals(facet, actions) {
     const byTier = new Map();
     const sameMonth = new Set();
     const facetMonths = new Set([facet.value, ...(facet.literals ?? [])]
@@ -6874,10 +6931,10 @@ function findKindKeyedLiteral(facet, actions) {
         }
     }
     const closest = [...byTier.entries()].sort(([a], [b]) => a - b)[0]?.[1];
-    if (closest !== undefined && closest.size === 1)
-        return [...closest][0];
-    const sameMonthInKind = [...(closest ?? sameMonth)].filter((literal) => sameMonth.has(literal));
-    return sameMonthInKind.length === 1 ? sameMonthInKind[0] : undefined;
+    if (closest === undefined)
+        return sameMonth.size === 1 ? [...sameMonth] : [];
+    const sameMonthInKind = [...closest].filter((literal) => sameMonth.has(literal));
+    return sameMonthInKind.length > 0 && closest.size > 1 ? sameMonthInKind : [...closest];
 }
 /**
  * Resolves each zero-recurrence navigateTo facet (in flow order) to the real
@@ -8460,6 +8517,20 @@ function facetCaseMethod(captured, facet) {
         return "toLowerCase";
     return undefined;
 }
+/** True when `code`'s letters occur in order within `value`, starting at its first letter (`LIS` in `Lisbon`). */
+function abbreviates(code, value) {
+    const letters = code.toLowerCase().replace(/[^a-z]/g, "");
+    const target = value.toLowerCase();
+    if (letters.length === 0 || letters[0] !== target[0])
+        return false;
+    let from = 0;
+    for (const letter of letters) {
+        from = target.indexOf(letter, from) + 1;
+        if (from === 0)
+            return false;
+    }
+    return true;
+}
 /**
  * JS expression for the value a facet's payload field carries in the code
  * vocabulary: a kind-keyed alias translates the caller's declared literal
@@ -10013,23 +10084,6 @@ function parseFoldReturnSpec(flowFileContents) {
         return null;
     }
 }
-/**
- * Reads the flow file's `browserFlowOnly: true` opt-in, which forces the
- * browser-flow-only emit for a submission flow whose HTTP analysis would
- * otherwise produce an executeHttp path. Anything but a literal `true` (or an
- * unreadable/legacy bare-array file) is "not opted in".
- */
-function parseBrowserFlowOnlyOptIn(flowFileContents) {
-    try {
-        const raw = JSON.parse(flowFileContents);
-        if (raw === null || typeof raw !== "object" || Array.isArray(raw))
-            return false;
-        return raw.browserFlowOnly === true;
-    }
-    catch {
-        return false;
-    }
-}
 /**
  * Parses `browserFallbackGate`/`httpTimeoutMs` out of an object-form
  * recon-flow.json, mirroring {@link parseFoldReturnSpec}'s null-safe pattern:
@@ -11917,7 +11971,7 @@ function emitContractTs(opts) {
         browserFlowOnly: omitExecuteHttp,
         payloadFieldNames,
     });
-    const isReservedByApplicantContactSchema = (name) => usesApplicantContactSchema && applicantContactFieldNames.has(name);
+    const isReservedByApplicantContactSchema = (name) => usesApplicantContactSchema && applicantContactFieldNames.has(canonicalPayloadField(name));
     // A missing inputBody, or one with no ApplicantContactSchema evidence,
     // means this is not a job-application submission flow. The former keeps
     // its own read-flow `{ query }` contract untouched; the latter (a body WAS
@@ -12045,7 +12099,7 @@ function emitContractTs(opts) {
         // The browser-flow-only emit canonicalises its own splices already; this
         // covers a caller handing over vocabulary-row names directly, so one never
         // becomes a required key beside the base schema's own equivalent.
-        const canonicaliseNames = usesApplicantContactSchema && omitExecuteHttp && isSubmissionFlow;
+        const canonicaliseNames = usesApplicantContactSchema;
         for (const rawName of [...payloadFieldNames].sort()) {
             const name = canonicaliseNames ? canonicalPayloadField(rawName) : rawName;
             if (isReservedByApplicantContactSchema(name))
@@ -12967,6 +13021,13 @@ function unreferencedDeclaredPayloadFields(contractCode, browserFlowCode, flowSt
 function unreadDeclaredPayloadFields(contractCode, browserFlowCode, flowSteps) {
     return new Set(unreferencedDeclaredPayloadFields(contractCode, contractCode.includes("httpClient(") ? "" : browserFlowCode, flowSteps));
 }
+/**
+ * Review-checklist lines for declared payload fields no emitted request reads,
+ * so a reviewer sees the silently-optionalized field outside the log stream.
+ */
+function unreadPayloadFieldChecklistLines(fields) {
+    return [...fields].map((field) => `Declared payloadField ${field} is read by no emitted request and was emitted optional — confirm it should affect the request, or drop it`);
+}
 /**
  * Escape a literal string segment so it is safe INSIDE a JS backtick template
  * literal — backslashes, backticks, and `${` interpolation starts must all be
@@ -13099,7 +13160,15 @@ function lastFlowStepDeclaresNonSubmit(flowSteps) {
  */
 function emitConfigManifest(opts) {
     const { siteId, displayName, baseUrl, flowSteps, vocabulary, inputBody, recoveredFields, httpModulePath, isSubmissionFlow = false, hasMultipartStep = false, browserFlowOnly = false, env = process.env, } = opts;
-    const canonicalise = isSubmissionFlow && browserFlowOnly;
+    // Raw (uncanonicalised) splice names are the evidence for a body-less flow;
+    // the decision must be made on them before canonicalising, or it is circular.
+    const canonicalise = contractUsesApplicantContactSchema({
+        inputBody,
+        isSubmissionFlow,
+        browserFlowOnly,
+        payloadFieldNames: computeFlowPayloadFieldNames(flowSteps, vocabulary, env, false)
+            .payloadFieldNames,
+    });
     const payloadFieldNames = new Set();
     const knownFieldValues = buildKnownFieldValues(flowSteps, vocabulary ?? vocabulary_1.EMPTY_VOCABULARY, env, canonicalise);
     const steps = flowSteps.map((step) => {
@@ -13303,20 +13372,20 @@ function computeFlowPayloadFieldNames(flowSteps, vocabulary, env, canonicalise =
 }
 function emitBrowserFlowTs(opts) {
     const { siteId, pascal, flowSteps, isSubmissionFlow, hasMultipartStep = false, browserFlowOnly = false, inputBody, vocabulary, frameSelector, capturedFallbackFields = new Set(), env = process.env, } = opts;
-    const canonicalise = isSubmissionFlow && browserFlowOnly;
+    const canonicalise = contractUsesApplicantContactSchema({
+        inputBody,
+        isSubmissionFlow,
+        browserFlowOnly,
+        payloadFieldNames: computeFlowPayloadFieldNames(flowSteps, vocabulary, env, false)
+            .payloadFieldNames,
+    });
     const { payloadFieldNames, optionalPayloadFieldNames } = computeFlowPayloadFieldNames(flowSteps, vocabulary, env, canonicalise);
     for (const field of capturedFallbackFields) {
         if (payloadFieldNames.has(field))
             optionalPayloadFieldNames.add(field);
     }
     const hasUploadStep = flowSteps.some((s) => typeof s !== "string" && s.upload === true);
-    const contractCarriesResume = hasMultipartStep ||
-        contractUsesApplicantContactSchema({
-            inputBody,
-            isSubmissionFlow,
-            browserFlowOnly,
-            payloadFieldNames,
-        });
+    const contractCarriesResume = hasMultipartStep || canonicalise;
     const knownFieldValues = buildKnownFieldValues(flowSteps, vocabulary ?? vocabulary_1.EMPTY_VOCABULARY, env, canonicalise);
     let usesThrowawayPassword = false;
     let usesEmailStep = false;
@@ -13886,14 +13955,7 @@ async function main() {
             };
         }
     })();
-    const browserFlowOnlyOptIn = (() => {
-        try {
-            return parseBrowserFlowOnlyOptIn((0, node_fs_1.readFileSync)(flowFile, "utf8"));
-        }
-        catch {
-            return false;
-        }
-    })();
+    const browserFlowOnlyOptIn = (0, node_fs_1.existsSync)(flowFile) ? (0, recon_shared_1.readBrowserFlowOnly)(flowFile) : false;
     // Flow-declared signals that isolate the submission POSTs from same-origin
     // page chrome. Threaded into action-sequence extraction and header derivation
     // so both draw from the real submission, not incidental widget/chatbot POSTs.
@@ -13930,8 +13992,14 @@ async function main() {
      * already written its output and there is nothing left for main() to do.
      */
     function generateFromCaptures(activeCaptures) {
+        if (activeCaptures.length === 0 && !allowEmptyCapture) {
+            throw new Error("empty capture pool: no captures remain to derive a base URL from (a placeholder origin would leak into the contract)");
+        }
+        // An opted-in empty capture has no origin to derive; "" leaves the
+        // operator's siteBaseUrls config as the only source instead of a
+        // placeholder origin leaking into the contract.
         const baseUrl = activeCaptures.length === 0
-            ? "https://example.com"
+            ? ""
             : deriveBaseUrl(activeCaptures, ownBackendHostnames, submitPatterns);
         // Gate on the flow's declared own-backend hosts (or the registrable-domain
         // fallback of baseUrl) via the same predicate recon-http.ts applies at
@@ -14497,7 +14565,10 @@ async function main() {
     logger.info(`wrote ${outDir}/contract.ts`);
     // Same opts fed to emitContractTs above, so this can never drift from what
     // the header used to embed.
-    const checklist = buildContractChecklist(final.contractOpts);
+    const checklist = [
+        ...buildContractChecklist(final.contractOpts),
+        ...unreadPayloadFieldChecklistLines(unreadDeclaredPayloadFields(final.contractCode, final.browserFlow.code, flowSteps)),
+    ];
     logger.info(`review checklist for ${outDir}/contract.ts:\n${checklist.map((item) => `  [ ] ${item}`).join("\n")}`);
     (0, node_fs_1.writeFileSync)(`${outDir}/flows/browser-flow.ts`, final.browserFlow.code);
     logger.info(`wrote ${outDir}/flows/browser-flow.ts`);