canopycms 0.3.2 → 0.4.0

package/dist/cli/cli.js

@@ -4552,16 +4552,26 @@ function flattenGroupFields(fields) {
   return result;
 }
 function findBodyFieldName(fields) {
+  return findExplicitBodyFieldName(fields) ?? "body";
+}
+function findExplicitBodyFieldName(fields) {
   for (const field of flattenGroupFields(fields)) {
     if ("isBody" in field && field.isBody)
       return field.name;
   }
-  return "body";
+  return void 0;
+}
+function linkOnlyReferenceData(data, format, fields) {
+  const bodyFieldName = isDataOnlyFormat(format) && fields ? findExplicitBodyFieldName(fields) : void 0;
+  if (bodyFieldName === void 0 || !(bodyFieldName in data))
+    return data;
+  const { [bodyFieldName]: _body, ...rest } = data;
+  return rest;
 }
 function extractSeoFields(entryData, opts = {}) {
   const names = { ...DEFAULT_SEO_FIELD_NAMES, ...opts.fields };
-  const entry = asRecord(entryData);
-  const source = opts.group ? asRecord(entry?.[opts.group]) : entry;
+  const entry = asRecord2(entryData);
+  const source = opts.group ? asRecord2(entry?.[opts.group]) : entry;
   return {
     title: asText(source?.[names.title]) ?? asText(opts.fallbackTitle),
     description: asText(source?.[names.description]) ?? asText(opts.fallbackDescription),
@@ -6285,14 +6295,18 @@ async function generateAIContent(options) {
     const pending = pendingById.get(id);
     if (!pending || !exported.has(id))
       return void 0;
-    const { slug } = pending.entry;
-    const target = buildResolvedReference(structuredClone(pending.storedData), {
+    const { format } = pending;
+    const { slug, fields } = pending.entry;
+    const storedData = structuredClone(pending.storedData);
+    const target = buildResolvedReference(linkOnlyReferenceData(storedData, format, fields), {
       id,
       slug,
       collection: pending.collectionPath,
       urlPath: computeEntryUrl(pending.collectionPath, slug, contentRoot)
     });
-    return { target, body: pending.body };
+    const dataBodyField = isDataOnlyFormat(format) ? findExplicitBodyFieldName(fields) : void 0;
+    const dataBody = dataBodyField === void 0 ? void 0 : storedData[dataBodyField];
+    return { target, body: pending.body ?? (typeof dataBody === "string" ? dataBody : void 0) };
   };
   const linkFor = config?.referenceLink ? await decideReferenceLinks(config.referenceLink, allPending, subjectOf, entryLinkUrl) : void 0;
   const bodyLinkUrl = linkFor ? (location) => {
@@ -6434,6 +6448,7 @@ async function collectEntries(read, collection, cleanPath) {
         entry: aiEntry,
         body,
         storedData,
+        format: doc.format,
         linkedIds: [],
         contentId: extractIdFromFilename(path13.basename(listEntry.relativePath)),
         absolutePath: doc.absolutePath,
@@ -7770,7 +7785,7 @@ async function generateAIContentCLI(options) {
   console.log(`  Output: ${result.outputDir}
 `);
 }
-var __defProp3, __getOwnPropNames3, __esm3, __export3, init_error3, init_atomic_write2, init_field_traversal, ITEM_KEY, ITEM_KEY_PATTERN, isItemKey, BASE58, isListField, init_item_keys, active2, init_logger2, SENTINEL_BASE2, DOT_SEGMENT2, init_sanitize_href2, init_normalize, ROOT_COLLECTION_ID, init_types2, normalizePathValue, flattenSchema, init_flatten, scalarFieldTypes2, markdownFieldTypes2, primitiveFieldTypes2, fieldTypes2, init_types22, SAFE_HTML_TAGS2, COMPONENT_NAME2, MDX_BINDINGS2, FORBIDDEN_ATTRIBUTES2, init_mdx_allowlist2, fieldBaseSchema2, selectOptionSchema2, referenceOptionSchema2, primitiveFieldSchema2, mdxPropAllowSchema2, mdxAllowlistSchema2, markdownFieldSchema2, ASPECT_RATIO_RE2, imageFieldSchema2, selectFieldSchema2, referenceFieldSchema2, fieldHolder2, blockSchema2, blockFieldSchema2, objectFieldSchema2, inlineGroupFieldSchema2, customFieldSchema2, knownFieldSchema2, fieldSchema2, init_field2, relativePathSchema2, entryTypeSchema2, collectionSchema2, init_collection2, uploadTargetUrlSchema2, previewPrefixSchema2, init_url2, mediaSchema2, init_media2, defaultBranchAccessSchema2, defaultPathAccessLevelSchema2, defaultPathAccessLevelsSchema2, defaultPathAccessSchema2, defaultBaseBranchSchema2, defaultRemoteNameSchema2, defaultRemoteUrlSchema2, gitBotAuthorNameSchema2, gitBotAuthorEmailSchema2, githubTokenEnvVarSchema2, operatingModeSchema2, deployedAsSchema2, contentRootSchema2, sourceRootSchema2, basePathSchema2, deploymentNameSchema2, devContentSyncSchema2, devConfigSchema2, editorConfigSchema2, CanopyConfigSchema2, DEFAULT_PROD_WORKSPACE2, init_config3, SERVER_MODE_ENV_VAR, BROWSER_MODE_ENV_VAR, warned2, init_mode_env, forEachReferenceField, init_validation, ProdClientSafeStrategy2, DevClientSafeStrategy2, clientStrategyCache, init_client_safe_strategy, warned3, VALID_DEPLOYMENT_NAME2, isValidDeploymentName2, init_deployment_name, init_config22, ProdStrategy2, DevStrategy2, strategyCache2, init_client_unsafe_strategy, operating_mode_exports, init_operating_mode, task_queue_config_exports2, init_task_queue_config2, worker_status_exports, WORKER_STATUS_FILE, init_worker_status, BLOCK_DISCRIMINATOR_KEYS, BLOCK_STRUCTURAL_KEYS, envFlag2, LOG_LEVELS2, DebugLogger2, testLogger2, END, FRESH, LINE_WIDTH, MIN_LINE_WIDTH, SourceSplicer, log2, MAX_LCS_CELLS, Splicer, log22, EVIDENCE_MAX_DEPTH, log3, locks, PROVISIONING_LOCK_STALE_MS, sleep, META_DIR, CONTENT_WRITE_LOCK_NAME, DEFAULT_CONTENT_WRITE_LOCK_WAIT_MS, ContentWriteLockBusyError, sleep2, CONTENT_WRITE_LOCK_STALE_MS, heldRoots, DEFAULT_SEO_FIELD_NAMES, asRecord, asText, asFlag, asImageUrl, asOgType, asTwitterCard, RESTRICTED_REFERENCE_MARKER, SEO_GROUP_FIELDS, MAX_WARNED, warned4, SchemaUnavailableError, BASE58_PATTERN, CONTENT_ID_PATTERN, PHYSICAL_SEGMENT_PATTERN, isValidId, EMPTY_LOGICAL_PATH, ContentIdIndex, parseTypedFilename, CONTENT_EXTENSIONS, registry, finalization, log4, META_DIR2, entryTypeMetaSchema, collectionMetaSchema, rootCollectionMetaSchema, fingerprints, editDistance, closestName, forEachEntryType, collectEntryTypeNames, validateReferenceEntryTypes, isDeployedStatic, isBuildMode, readsFromCheckout, STATIC_DEPLOY_USER, storage, log5, BaseBranchUnresolvedError, NETWORK_SCHEME_PATTERN, TRANSPORT_HELPER_PATTERN, LEADING_OPTION_PATTERN, WINDOWS_DRIVE_PATTERN, COLON_BEFORE_SLASH_PATTERN, MISSING_REMOTE_REF_REASONS, CANOPY_META_DIR, SCHEMA_CACHE_VERSION, MTIME_CHECK_DEBOUNCE_MS, SCHEMA_CACHE_FILE, SCHEMA_GENERATION_RESOURCE, reportedIssues, BranchSchemaCache, RESOURCE, getFormatExtension, isDataOnlyFormat, asRecord2, RESERVED_SETTINGS_BRANCH_PREFIX, BranchPathError, resolveContentBranchesRoot, ContentStoreError, ContentConflictError, EntryGoneError, BranchSyncingError, DuplicateContentIdError, UrlPathConflictError, isEmptyFieldValue, DEFAULT_INDEX_FRESHNESS_INTERVAL_MS, FORCED_REFRESH_MIN_INTERVAL_MS, STALE_LOOKUP, isMissingReference, createReferenceResolveCache, ContentStore, ATTR_CONTENT, BLOCK_PREFIX, BLOCK_SUFFIX, BLOCK_RESTORE_RE, INLINE_PREFIX, INLINE_SUFFIX, INLINE_RESTORE_RE, DEFAULT_REFERENCE_RENDERING, FLAT_CELL_TYPES, log6, BASE58_CHAR, ENTRY_LINK_PATTERN, INLINE_LINK_TARGET, BranchMetadataCorruptError, BRANCH_META_DIR, BRANCH_META_FILE, branchMetadataFilePath, UNKNOWN_TIME, enumOf, branchMetadataShape, branchMetadataSchema, branchMetadataFileSchema, log7, REGISTRY_FILE, REGISTRY_STALE_FILE, REGISTRY_TEMP_FILE, REGISTRY_VERSION, RESOURCE2, GET_MISS_REFRESH_MIN_INTERVAL_MS, snapshotSchema, reportedUnreadable, BranchRegistry, log8, OccWriteConflictError, CURRENT_SCHEMA_VERSION, BranchMetadataConflictError, BranchMetadataFileManager, loadBranchContext, log9, GIT_ENV_PASSTHROUGH, FORCE_C_LOCALE, GIT_NETWORK_ENV_PASSTHROUGH, NO_AUTO_GC_CONFIG, REMOTE_GIT_CONFIG, REMOTE_GIT_MAX_LOOSE_OBJECTS, REMOTE_GIT_MAX_PACKS, PARALLEL_CHECKOUT_ARGS, remoteInitLocks, GITHUB_TRACKING_REF_PREFIX, GitConflictError, GitRemoteRefMissingError, RemoteNotReadyError, SettingsRemoteUnreadableError, SettingsBranchDivergedError, SETTINGS_BRANCH_FILES, SettingsBranchHasContentHistoryError, GitManager, PUBLISH_LOCK_WAIT_MS, CREATE_RESIDUE_MIN_QUIET_MS, STAGING_MAX_AGE_MS, REPAIR_MAX_AGE_MS, DELETING_MAX_AGE_MS, SETTINGS_TRASH_MAX_AGE_MS, BranchProvisioningBusyError, BranchDirOccupiedError, STAGING_PREFIX, REPAIR_PREFIX, TRASH_PREFIX, stagingDirName, NO_GIT, RENAME_BLOCKED, SPARSE_CONE_RECORD, defaultSink, sink, ProvisionLog, testHooks, provisionsInFlight, BranchWorkspaceManager, log10, normalizePath, readEntryData, createReferenceResolver, withoutUnavailableEntries, resolveCollectionItemReferences, looksLikeMalformedEntry, listCollectionEntries, MAX_INPUT_PIXELS, CROP_PRECISION, KNOWN_FIELD_TYPES, STRING_FIELD_TYPES, selectOptionValues, GENERATED_RECORD_FILENAME, SAFE_BUILD_ID, BUILD_ID_RULE, jiti;
+var __defProp3, __getOwnPropNames3, __esm3, __export3, init_error3, init_atomic_write2, init_field_traversal, ITEM_KEY, ITEM_KEY_PATTERN, isItemKey, BASE58, isListField, init_item_keys, active2, init_logger2, SENTINEL_BASE2, DOT_SEGMENT2, init_sanitize_href2, init_normalize, ROOT_COLLECTION_ID, init_types2, normalizePathValue, flattenSchema, init_flatten, scalarFieldTypes2, markdownFieldTypes2, primitiveFieldTypes2, fieldTypes2, init_types22, SAFE_HTML_TAGS2, COMPONENT_NAME2, MDX_BINDINGS2, FORBIDDEN_ATTRIBUTES2, init_mdx_allowlist2, fieldBaseSchema2, selectOptionSchema2, referenceOptionSchema2, primitiveFieldSchema2, mdxPropAllowSchema2, mdxAllowlistSchema2, markdownFieldSchema2, ASPECT_RATIO_RE2, imageFieldSchema2, selectFieldSchema2, referenceFieldSchema2, fieldHolder2, blockSchema2, blockFieldSchema2, objectFieldSchema2, inlineGroupFieldSchema2, customFieldSchema2, knownFieldSchema2, fieldSchema2, init_field2, relativePathSchema2, entryTypeSchema2, collectionSchema2, init_collection2, uploadTargetUrlSchema2, previewPrefixSchema2, init_url2, mediaSchema2, init_media2, defaultBranchAccessSchema2, defaultPathAccessLevelSchema2, defaultPathAccessLevelsSchema2, defaultPathAccessSchema2, defaultBaseBranchSchema2, defaultRemoteNameSchema2, defaultRemoteUrlSchema2, gitBotAuthorNameSchema2, gitBotAuthorEmailSchema2, githubTokenEnvVarSchema2, operatingModeSchema2, deployedAsSchema2, contentRootSchema2, sourceRootSchema2, basePathSchema2, deploymentNameSchema2, devContentSyncSchema2, devConfigSchema2, editorConfigSchema2, CanopyConfigSchema2, DEFAULT_PROD_WORKSPACE2, init_config3, SERVER_MODE_ENV_VAR, BROWSER_MODE_ENV_VAR, warned2, init_mode_env, forEachReferenceField, init_validation, ProdClientSafeStrategy2, DevClientSafeStrategy2, clientStrategyCache, init_client_safe_strategy, warned3, VALID_DEPLOYMENT_NAME2, isValidDeploymentName2, init_deployment_name, init_config22, ProdStrategy2, DevStrategy2, strategyCache2, init_client_unsafe_strategy, operating_mode_exports, init_operating_mode, task_queue_config_exports2, init_task_queue_config2, worker_status_exports, WORKER_STATUS_FILE, init_worker_status, BLOCK_DISCRIMINATOR_KEYS, BLOCK_STRUCTURAL_KEYS, envFlag2, LOG_LEVELS2, DebugLogger2, testLogger2, END, FRESH, LINE_WIDTH, MIN_LINE_WIDTH, SourceSplicer, log2, MAX_LCS_CELLS, Splicer, log22, EVIDENCE_MAX_DEPTH, log3, locks, PROVISIONING_LOCK_STALE_MS, sleep, META_DIR, CONTENT_WRITE_LOCK_NAME, DEFAULT_CONTENT_WRITE_LOCK_WAIT_MS, ContentWriteLockBusyError, sleep2, CONTENT_WRITE_LOCK_STALE_MS, heldRoots, getFormatExtension, isDataOnlyFormat, asRecord, DEFAULT_SEO_FIELD_NAMES, asRecord2, asText, asFlag, asImageUrl, asOgType, asTwitterCard, RESTRICTED_REFERENCE_MARKER, SEO_GROUP_FIELDS, MAX_WARNED, warned4, SchemaUnavailableError, BASE58_PATTERN, CONTENT_ID_PATTERN, PHYSICAL_SEGMENT_PATTERN, isValidId, EMPTY_LOGICAL_PATH, ContentIdIndex, parseTypedFilename, CONTENT_EXTENSIONS, registry, finalization, log4, META_DIR2, entryTypeMetaSchema, collectionMetaSchema, rootCollectionMetaSchema, fingerprints, editDistance, closestName, forEachEntryType, collectEntryTypeNames, validateReferenceEntryTypes, isDeployedStatic, isBuildMode, readsFromCheckout, STATIC_DEPLOY_USER, storage, log5, BaseBranchUnresolvedError, NETWORK_SCHEME_PATTERN, TRANSPORT_HELPER_PATTERN, LEADING_OPTION_PATTERN, WINDOWS_DRIVE_PATTERN, COLON_BEFORE_SLASH_PATTERN, MISSING_REMOTE_REF_REASONS, CANOPY_META_DIR, SCHEMA_CACHE_VERSION, MTIME_CHECK_DEBOUNCE_MS, SCHEMA_CACHE_FILE, SCHEMA_GENERATION_RESOURCE, reportedIssues, BranchSchemaCache, RESOURCE, RESERVED_SETTINGS_BRANCH_PREFIX, BranchPathError, resolveContentBranchesRoot, ContentStoreError, ContentConflictError, EntryGoneError, BranchSyncingError, DuplicateContentIdError, UrlPathConflictError, isEmptyFieldValue, DEFAULT_INDEX_FRESHNESS_INTERVAL_MS, FORCED_REFRESH_MIN_INTERVAL_MS, STALE_LOOKUP, isMissingReference, createReferenceResolveCache, ContentStore, ATTR_CONTENT, BLOCK_PREFIX, BLOCK_SUFFIX, BLOCK_RESTORE_RE, INLINE_PREFIX, INLINE_SUFFIX, INLINE_RESTORE_RE, DEFAULT_REFERENCE_RENDERING, FLAT_CELL_TYPES, log6, BASE58_CHAR, ENTRY_LINK_PATTERN, INLINE_LINK_TARGET, BranchMetadataCorruptError, BRANCH_META_DIR, BRANCH_META_FILE, branchMetadataFilePath, UNKNOWN_TIME, enumOf, branchMetadataShape, branchMetadataSchema, branchMetadataFileSchema, log7, REGISTRY_FILE, REGISTRY_STALE_FILE, REGISTRY_TEMP_FILE, REGISTRY_VERSION, RESOURCE2, GET_MISS_REFRESH_MIN_INTERVAL_MS, snapshotSchema, reportedUnreadable, BranchRegistry, log8, OccWriteConflictError, CURRENT_SCHEMA_VERSION, BranchMetadataConflictError, BranchMetadataFileManager, loadBranchContext, log9, GIT_ENV_PASSTHROUGH, FORCE_C_LOCALE, GIT_NETWORK_ENV_PASSTHROUGH, NO_AUTO_GC_CONFIG, REMOTE_GIT_CONFIG, REMOTE_GIT_MAX_LOOSE_OBJECTS, REMOTE_GIT_MAX_PACKS, PARALLEL_CHECKOUT_ARGS, remoteInitLocks, GITHUB_TRACKING_REF_PREFIX, GitConflictError, GitRemoteRefMissingError, RemoteNotReadyError, SettingsRemoteUnreadableError, SettingsBranchDivergedError, SETTINGS_BRANCH_FILES, SettingsBranchHasContentHistoryError, GitManager, PUBLISH_LOCK_WAIT_MS, CREATE_RESIDUE_MIN_QUIET_MS, STAGING_MAX_AGE_MS, REPAIR_MAX_AGE_MS, DELETING_MAX_AGE_MS, SETTINGS_TRASH_MAX_AGE_MS, BranchProvisioningBusyError, BranchDirOccupiedError, STAGING_PREFIX, REPAIR_PREFIX, TRASH_PREFIX, stagingDirName, NO_GIT, RENAME_BLOCKED, SPARSE_CONE_RECORD, defaultSink, sink, ProvisionLog, testHooks, provisionsInFlight, BranchWorkspaceManager, log10, normalizePath, readEntryData, createReferenceResolver, withoutUnavailableEntries, resolveCollectionItemReferences, looksLikeMalformedEntry, listCollectionEntries, MAX_INPUT_PIXELS, CROP_PRECISION, KNOWN_FIELD_TYPES, STRING_FIELD_TYPES, selectOptionValues, GENERATED_RECORD_FILENAME, SAFE_BUILD_ID, BUILD_ID_RULE, jiti;
 var init_generate_ai_content = __esm({
   "dist/cli/generate-ai-content.js"() {
     "use strict";
@@ -9216,6 +9231,17 @@ var init_generate_ai_content = __esm({
     });
     CONTENT_WRITE_LOCK_STALE_MS = 3e4;
     heldRoots = new AsyncLocalStorage();
+    getFormatExtension = (format) => {
+      if (format === "md")
+        return ".md";
+      if (format === "mdx")
+        return ".mdx";
+      if (format === "yaml")
+        return ".yaml";
+      return ".json";
+    };
+    isDataOnlyFormat = (format) => format === "json" || format === "yaml";
+    asRecord = (value) => value && typeof value === "object" && !Array.isArray(value) ? value : {};
     init_item_keys();
     DEFAULT_SEO_FIELD_NAMES = {
       title: "metaTitle",
@@ -9226,7 +9252,7 @@ var init_generate_ai_content = __esm({
       noindex: "noindex",
       twitterCard: "twitterCard"
     };
-    asRecord = (value) => typeof value === "object" && value !== null && !Array.isArray(value) ? value : void 0;
+    asRecord2 = (value) => typeof value === "object" && value !== null && !Array.isArray(value) ? value : void 0;
     asText = (value) => {
       if (typeof value !== "string")
         return void 0;
@@ -9238,7 +9264,7 @@ var init_generate_ai_content = __esm({
       const text3 = asText(value);
       if (text3)
         return text3;
-      return asText(asRecord(value)?.src);
+      return asText(asRecord2(value)?.src);
     };
     asOgType = (value) => {
       const text3 = asText(value);
@@ -9939,17 +9965,6 @@ var init_generate_ai_content = __esm({
     RESOURCE = "content-index";
     init_error3();
     init_error3();
-    getFormatExtension = (format) => {
-      if (format === "md")
-        return ".md";
-      if (format === "mdx")
-        return ".mdx";
-      if (format === "yaml")
-        return ".yaml";
-      return ".json";
-    };
-    isDataOnlyFormat = (format) => format === "json" || format === "yaml";
-    asRecord2 = (value) => value && typeof value === "object" && !Array.isArray(value) ? value : {};
     init_normalize();
     init_operating_mode();
     RESERVED_SETTINGS_BRANCH_PREFIX = "canopycms-settings-";
@@ -10448,7 +10463,7 @@ var init_generate_ai_content = __esm({
           fields = entryTypeConfig?.schema || [];
         }
         if (format === "json") {
-          const data = asRecord2(JSON.parse(raw));
+          const data = asRecord(JSON.parse(raw));
           doc = {
             collection: schemaItem.logicalPath,
             collectionName: schemaItem.name,
@@ -10458,7 +10473,7 @@ var init_generate_ai_content = __esm({
             absolutePath
           };
         } else if (format === "yaml") {
-          const data = asRecord2(yamlParse(raw));
+          const data = asRecord(yamlParse(raw));
           doc = {
             collection: schemaItem.logicalPath,
             collectionName: schemaItem.name,
@@ -11061,10 +11076,11 @@ var init_generate_ai_content = __esm({
        * Measured (2000 occurrences of one target, local disk, the friendliest case for NOT
        * caching): cloning a snippet-sized target is ~63x cheaper than re-reading and parsing it, a
        * 265KB JSON target ~0.8x -- a wash at the pathological end, never a blow-up, and on the
-       * EFS/NFS this targets the syscall the memo removes dominates both. An md/mdx target
-       * resolves to its FRONTMATTER unless the field sets `includeBody`, which is why
-       * `includeBody` is opt-in per field: an embedded body sits inside the memoized object and is
-       * cloned once per referencing entry.
+       * EFS/NFS this targets the syscall the memo removes dominates both. A target resolves
+       * without its body (an md/mdx target to its frontmatter, a json/yaml one without its `isBody`
+       * field) unless the field sets `includeBody`, which is why `includeBody` is opt-in per
+       * field: an embedded body sits inside the memoized object and is cloned once per referencing
+       * entry.
        */
       resolveSingleReference(id, idIndex, includeBody, cache, access) {
         if (!cache)
@@ -11159,7 +11175,8 @@ var init_generate_ai_content = __esm({
             }));
           }
           const bodyForEmbed = includeBody && "body" in doc ? { fieldName: doc.bodyFieldName, value: doc.body } : void 0;
-          return buildResolvedReference(doc.data, meta, bodyForEmbed);
+          const data = includeBody ? doc.data : linkOnlyReferenceData(doc.data, doc.format, this.entryTypeOf(location.collection, location.relativePath)?.schema);
+          return buildResolvedReference(data, meta, bodyForEmbed);
         } catch (error) {
           if (isNodeError3(error) && error.code === "ENOENT")
             return STALE_LOOKUP;
@@ -13182,10 +13199,10 @@ var init_generate_ai_content = __esm({
       try {
         const raw = await fs20.readFile(filePath, "utf8");
         if (format === "json") {
-          return asRecord2(JSON.parse(raw));
+          return asRecord(JSON.parse(raw));
         }
         if (format === "yaml") {
-          return asRecord2(yamlParse2(raw));
+          return asRecord(yamlParse2(raw));
         }
         const parsed = matter3(raw);
         const data = { ...parsed.data ?? {} };

package/dist/cli/generate-ai-content.js

@@ -2912,13 +2912,36 @@ function flattenGroupFields(fields) {
   return result;
 }
 
+// dist/utils/format.js
+var getFormatExtension = (format) => {
+  if (format === "md")
+    return ".md";
+  if (format === "mdx")
+    return ".mdx";
+  if (format === "yaml")
+    return ".yaml";
+  return ".json";
+};
+var isDataOnlyFormat = (format) => format === "json" || format === "yaml";
+var asRecord = (value) => value && typeof value === "object" && !Array.isArray(value) ? value : {};
+
 // dist/utils/body-field.js
 function findBodyFieldName(fields) {
+  return findExplicitBodyFieldName(fields) ?? "body";
+}
+function findExplicitBodyFieldName(fields) {
   for (const field of flattenGroupFields(fields)) {
     if ("isBody" in field && field.isBody)
       return field.name;
   }
-  return "body";
+  return void 0;
+}
+function linkOnlyReferenceData(data, format, fields) {
+  const bodyFieldName = isDataOnlyFormat(format) && fields ? findExplicitBodyFieldName(fields) : void 0;
+  if (bodyFieldName === void 0 || !(bodyFieldName in data))
+    return data;
+  const { [bodyFieldName]: _body, ...rest } = data;
+  return rest;
 }
 
 // dist/content-store.js
@@ -2934,7 +2957,7 @@ var DEFAULT_SEO_FIELD_NAMES = {
   noindex: "noindex",
   twitterCard: "twitterCard"
 };
-var asRecord = (value) => typeof value === "object" && value !== null && !Array.isArray(value) ? value : void 0;
+var asRecord2 = (value) => typeof value === "object" && value !== null && !Array.isArray(value) ? value : void 0;
 var asText = (value) => {
   if (typeof value !== "string")
     return void 0;
@@ -2946,7 +2969,7 @@ var asImageUrl = (value) => {
   const text = asText(value);
   if (text)
     return text;
-  return asText(asRecord(value)?.src);
+  return asText(asRecord2(value)?.src);
 };
 var asOgType = (value) => {
   const text = asText(value);
@@ -2958,8 +2981,8 @@ var asTwitterCard = (value) => {
 };
 function extractSeoFields(entryData, opts = {}) {
   const names = { ...DEFAULT_SEO_FIELD_NAMES, ...opts.fields };
-  const entry = asRecord(entryData);
-  const source = opts.group ? asRecord(entry?.[opts.group]) : entry;
+  const entry = asRecord2(entryData);
+  const source = opts.group ? asRecord2(entry?.[opts.group]) : entry;
   return {
     title: asText(source?.[names.title]) ?? asText(opts.fallbackTitle),
     description: asText(source?.[names.description]) ?? asText(opts.fallbackDescription),
@@ -4459,19 +4482,6 @@ async function readFileWithMtimeIfExists(filePath) {
   }
 }
 
-// dist/utils/format.js
-var getFormatExtension = (format) => {
-  if (format === "md")
-    return ".md";
-  if (format === "mdx")
-    return ".mdx";
-  if (format === "yaml")
-    return ".yaml";
-  return ".json";
-};
-var isDataOnlyFormat = (format) => format === "json" || format === "yaml";
-var asRecord2 = (value) => value && typeof value === "object" && !Array.isArray(value) ? value : {};
-
 // dist/paths/index.js
 init_normalize();
 
@@ -5085,7 +5095,7 @@ var ContentStore = class {
       fields = entryTypeConfig?.schema || [];
     }
     if (format === "json") {
-      const data = asRecord2(JSON.parse(raw));
+      const data = asRecord(JSON.parse(raw));
       doc = {
         collection: schemaItem.logicalPath,
         collectionName: schemaItem.name,
@@ -5095,7 +5105,7 @@ var ContentStore = class {
         absolutePath
       };
     } else if (format === "yaml") {
-      const data = asRecord2(yamlParse(raw));
+      const data = asRecord(yamlParse(raw));
       doc = {
         collection: schemaItem.logicalPath,
         collectionName: schemaItem.name,
@@ -5698,10 +5708,11 @@ var ContentStore = class {
    * Measured (2000 occurrences of one target, local disk, the friendliest case for NOT
    * caching): cloning a snippet-sized target is ~63x cheaper than re-reading and parsing it, a
    * 265KB JSON target ~0.8x -- a wash at the pathological end, never a blow-up, and on the
-   * EFS/NFS this targets the syscall the memo removes dominates both. An md/mdx target
-   * resolves to its FRONTMATTER unless the field sets `includeBody`, which is why
-   * `includeBody` is opt-in per field: an embedded body sits inside the memoized object and is
-   * cloned once per referencing entry.
+   * EFS/NFS this targets the syscall the memo removes dominates both. A target resolves
+   * without its body (an md/mdx target to its frontmatter, a json/yaml one without its `isBody`
+   * field) unless the field sets `includeBody`, which is why `includeBody` is opt-in per
+   * field: an embedded body sits inside the memoized object and is cloned once per referencing
+   * entry.
    */
   resolveSingleReference(id, idIndex, includeBody, cache, access) {
     if (!cache)
@@ -5796,7 +5807,8 @@ var ContentStore = class {
         }));
       }
       const bodyForEmbed = includeBody && "body" in doc ? { fieldName: doc.bodyFieldName, value: doc.body } : void 0;
-      return buildResolvedReference(doc.data, meta, bodyForEmbed);
+      const data = includeBody ? doc.data : linkOnlyReferenceData(doc.data, doc.format, this.entryTypeOf(location.collection, location.relativePath)?.schema);
+      return buildResolvedReference(data, meta, bodyForEmbed);
     } catch (error) {
       if (isNodeError(error) && error.code === "ENOENT")
         return STALE_LOOKUP;
@@ -6752,14 +6764,18 @@ async function generateAIContent(options) {
     const pending = pendingById.get(id);
     if (!pending || !exported.has(id))
       return void 0;
-    const { slug } = pending.entry;
-    const target = buildResolvedReference(structuredClone(pending.storedData), {
+    const { format } = pending;
+    const { slug, fields } = pending.entry;
+    const storedData = structuredClone(pending.storedData);
+    const target = buildResolvedReference(linkOnlyReferenceData(storedData, format, fields), {
       id,
       slug,
       collection: pending.collectionPath,
       urlPath: computeEntryUrl(pending.collectionPath, slug, contentRoot)
     });
-    return { target, body: pending.body };
+    const dataBodyField = isDataOnlyFormat(format) ? findExplicitBodyFieldName(fields) : void 0;
+    const dataBody = dataBodyField === void 0 ? void 0 : storedData[dataBodyField];
+    return { target, body: pending.body ?? (typeof dataBody === "string" ? dataBody : void 0) };
   };
   const linkFor = config?.referenceLink ? await decideReferenceLinks(config.referenceLink, allPending, subjectOf, entryLinkUrl) : void 0;
   const bodyLinkUrl = linkFor ? (location) => {
@@ -6901,6 +6917,7 @@ async function collectEntries(read, collection, cleanPath) {
         entry: aiEntry,
         body,
         storedData,
+        format: doc.format,
         linkedIds: [],
         contentId: extractIdFromFilename(path13.basename(listEntry.relativePath)),
         absolutePath: doc.absolutePath,
@@ -9794,10 +9811,10 @@ var readEntryData = async (filePath, format, bodyFieldName = "body") => {
   try {
     const raw = await fs20.readFile(filePath, "utf8");
     if (format === "json") {
-      return asRecord2(JSON.parse(raw));
+      return asRecord(JSON.parse(raw));
     }
     if (format === "yaml") {
-      return asRecord2(yamlParse2(raw));
+      return asRecord(yamlParse2(raw));
     }
     const parsed = matter3(raw);
     const data = { ...parsed.data ?? {} };

package/dist/config/types.d.ts

@@ -45,10 +45,11 @@ interface BaseFieldConfig {
     /** Mark this field as the display title for entries using this schema. At most one field per schema. */
     isTitle?: boolean;
     /**
-     * Mark this field as the body content for md/mdx entries.
-     * The field must be type 'markdown' or 'mdx'. At most one per schema.
-     * When present, the file's markdown content is mapped to this field,
-     * making TypeFromEntrySchema include it without manual `& { body: string }`.
+     * Mark this field as the entry's body. The field must be type 'markdown' or 'mdx'. At most one
+     * per schema. On an md/mdx entry, the file's markdown content is mapped to this field, making
+     * TypeFromEntrySchema include it without manual `& { body: string }`. On a json/yaml entry, it
+     * is the prose a link-only reference omits. Either way a reference carries it only with
+     * `includeBody: true`.
      */
     isBody?: boolean;
 }
@@ -117,7 +118,9 @@ export interface ReferenceFieldConfig extends BaseFieldConfig {
     options?: ReferenceOption[];
     /**
      * Include the referenced entry's body in the resolved value, under that entry type's own
-     * body field name (`isBody: true`, else `body`). Defaults to `false`.
+     * body field name (`isBody: true`, else `body`). Defaults to `false`. The body is an md/mdx
+     * target's markdown content, or a json/yaml target's `isBody` field; a json/yaml field merely
+     * named `body` is ordinary data and always resolves.
      *
      * This is the embed-vs-link distinction, declared where it is actually known. A reference
      * that **embeds** its target — a shared call-to-action block rendered inline — needs the
@@ -131,9 +134,6 @@ export interface ReferenceFieldConfig extends BaseFieldConfig {
      * contains both kinds: a page with a shared CTA *and* a related-posts list cannot be served
      * by a single call-level setting, and a call-level flag would mean the same field resolved
      * to different shapes in different places.
-     *
-     * Only affects formats that HAVE a separate body — md and mdx. A json/yaml target's whole
-     * document is already its data, so this is a no-op there.
      */
     includeBody?: boolean;
 }

package/dist/content-store.d.ts

@@ -557,10 +557,11 @@ export declare class ContentStore {
      * Measured (2000 occurrences of one target, local disk, the friendliest case for NOT
      * caching): cloning a snippet-sized target is ~63x cheaper than re-reading and parsing it, a
      * 265KB JSON target ~0.8x -- a wash at the pathological end, never a blow-up, and on the
-     * EFS/NFS this targets the syscall the memo removes dominates both. An md/mdx target
-     * resolves to its FRONTMATTER unless the field sets `includeBody`, which is why
-     * `includeBody` is opt-in per field: an embedded body sits inside the memoized object and is
-     * cloned once per referencing entry.
+     * EFS/NFS this targets the syscall the memo removes dominates both. A target resolves
+     * without its body (an md/mdx target to its frontmatter, a json/yaml one without its `isBody`
+     * field) unless the field sets `includeBody`, which is why `includeBody` is opt-in per
+     * field: an embedded body sits inside the memoized object and is cloned once per referencing
+     * entry.
      */
     private resolveSingleReference;
     /**

package/dist/content-store.js

@@ -20,7 +20,7 @@ import { serializeFrontmatter, serializeYaml } from './utils/content-serialize.j
 import { serializeJson } from './utils/json-source-splice.js';
 import { withLock } from './utils/async-mutex.js';
 import { assertContentWriteLockHeld, ContentWriteLockBusyError, DEFAULT_CONTENT_WRITE_LOCK_WAIT_MS, withContentWriteLock, } from './utils/content-write-lock.js';
-import { findBodyFieldName } from './utils/body-field.js';
+import { findBodyFieldName, linkOnlyReferenceData } from './utils/body-field.js';
 import { withItemKeys } from './item-keys.js';
 import { buildMissingReference, buildResolvedReference, buildRestrictedReference, } from './entry-schema.js';
 import { warnDanglingReference } from './dangling-reference-log.js';
@@ -1611,10 +1611,11 @@ export class ContentStore {
      * Measured (2000 occurrences of one target, local disk, the friendliest case for NOT
      * caching): cloning a snippet-sized target is ~63x cheaper than re-reading and parsing it, a
      * 265KB JSON target ~0.8x -- a wash at the pathological end, never a blow-up, and on the
-     * EFS/NFS this targets the syscall the memo removes dominates both. An md/mdx target
-     * resolves to its FRONTMATTER unless the field sets `includeBody`, which is why
-     * `includeBody` is opt-in per field: an embedded body sits inside the memoized object and is
-     * cloned once per referencing entry.
+     * EFS/NFS this targets the syscall the memo removes dominates both. A target resolves
+     * without its body (an md/mdx target to its frontmatter, a json/yaml one without its `isBody`
+     * field) unless the field sets `includeBody`, which is why `includeBody` is opt-in per
+     * field: an embedded body sits inside the memoized object and is cloned once per referencing
+     * entry.
      */
     resolveSingleReference(id, idIndex, includeBody, cache, access) {
         if (!cache)
@@ -1738,9 +1739,13 @@ export class ContentStore {
             //
             // `'body' in doc` narrows the ContentDocument union to its markdown variant, so
             // `doc.body`/`doc.bodyFieldName` are reachable at all -- a type guard, not a redundant
-            // runtime check. Only a field that asked to EMBED its target passes a body at all.
+            // runtime check. Only a field that asked to EMBED its target passes a body at all, and
+            // only such a field keeps a json/yaml target's `isBody` field.
             const bodyForEmbed = includeBody && 'body' in doc ? { fieldName: doc.bodyFieldName, value: doc.body } : undefined;
-            return buildResolvedReference(doc.data, meta, bodyForEmbed);
+            const data = includeBody
+                ? doc.data
+                : linkOnlyReferenceData(doc.data, doc.format, this.entryTypeOf(location.collection, location.relativePath)?.schema);
+            return buildResolvedReference(data, meta, bodyForEmbed);
         }
         catch (error) {
             // Index hit but the file is gone — the typical symptom of an external

package/dist/editor/fields/gfm-autolink-literal.d.ts

@@ -0,0 +1,11 @@
+/** GFM's autolink literals (bare `www.`, `http(s)://` and email addresses), by GFM's own parser. */
+import { gfmAutolinkLiteralFromMarkdown } from 'mdast-util-gfm-autolink-literal';
+import { gfmAutolinkLiteral } from 'micromark-extension-gfm-autolink-literal';
+export { gfmAutolinkLiteral, gfmAutolinkLiteralFromMarkdown };
+export interface AutolinkLiteral {
+    /** GFM's href: `www.` gains `http://`, an email `mailto:`. */
+    readonly url: string;
+    readonly length: number;
+}
+/** The literal GFM reads at `offset` in `text`, if any. */
+export declare function autolinkLiteralAt(text: string, offset: number): AutolinkLiteral | undefined;

package/dist/editor/fields/gfm-autolink-literal.js

@@ -0,0 +1,45 @@
+/** GFM's autolink literals (bare `www.`, `http(s)://` and email addresses), by GFM's own parser. */
+import { fromMarkdown } from 'mdast-util-from-markdown';
+import { gfmAutolinkLiteralFromMarkdown } from 'mdast-util-gfm-autolink-literal';
+import { gfmAutolinkLiteral } from 'micromark-extension-gfm-autolink-literal';
+export { gfmAutolinkLiteral, gfmAutolinkLiteralFromMarkdown };
+/** An export asks about a paragraph once per link in it, so recent parses are kept. */
+const MAX_CACHED = 16;
+const cache = new Map();
+/** `text`'s literals by offset. `[a](b)` and `<b>` are none, and lexical text holds neither. */
+function literalsIn(text) {
+    const cached = cache.get(text);
+    if (cached)
+        return cached;
+    const tree = fromMarkdown(text, {
+        extensions: [gfmAutolinkLiteral()],
+        mdastExtensions: [gfmAutolinkLiteralFromMarkdown()],
+    });
+    const literals = new Map();
+    const pending = [tree];
+    for (let node = pending.pop(); node !== undefined; node = pending.pop()) {
+        const start = node.position?.start.offset;
+        const end = node.position?.end.offset;
+        const [only, ...rest] = node.children ?? [];
+        if (node.type === 'link' &&
+            node.url !== undefined &&
+            start !== undefined &&
+            end !== undefined &&
+            rest.length === 0 &&
+            only?.type === 'text' &&
+            only.value === text.slice(start, end)) {
+            literals.set(start, { url: node.url, length: end - start });
+        }
+        else {
+            pending.push(...(node.children ?? []));
+        }
+    }
+    if (cache.size >= MAX_CACHED)
+        cache.delete(cache.keys().next().value ?? '');
+    cache.set(text, literals);
+    return literals;
+}
+/** The literal GFM reads at `offset` in `text`, if any. */
+export function autolinkLiteralAt(text, offset) {
+    return literalsIn(text).get(offset);
+}

package/dist/editor/fields/markdown-fidelity-visitors.d.ts

@@ -1,8 +1,7 @@
 /**
  * MarkdownField's MDXEditor visitors for markdown its own import and export would change the
- * meaning of: an ordered list's `start`, a link inside a formatted span, an auto-link inside code,
- * and the code spans either side of one. MDXEditor is passed in, as `mdx-jsx-support.tsx`
- * explains.
+ * meaning of: an ordered list's `start`, a bare URL, a link inside a formatted span, an auto-link
+ * inside code, and the code spans either side of one. MDXEditor is passed in (`mdx-jsx-support`).
  */
 import type * as MdxEditor from '@mdxeditor/editor';
 type MdxEditorModule = typeof MdxEditor;