@aglyn/plugins-data 1.0.0-beta.221 → 1.0.0-beta.223

package/src/lib/artifact/dataset-schema.d.ts

@@ -0,0 +1,136 @@
+/**
+ * @license
+ * Copyright 2026 Aglyn LLC
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+/**
+ * A dataset's schema as it travels between workspaces (AGL-657): what a
+ * publisher's dataset is reduced to, what an installing workspace's new
+ * dataset is made from, and what an update compares. The data plugin's,
+ * because only it knows a dataset's model (AGL-3080); the installer stores and
+ * diffs it as content it does not read.
+ *
+ * Pure and context-free, so the owner's server answers
+ * (`dataset-schema-artifact.server.ts`) and their specs share one shape.
+ */
+/** Field cap on a published schema, mirroring the console's create limit. */
+export declare const DATASET_SCHEMA_MAX_FIELDS = 100;
+/** A published dataset schema: the model, with no records. */
+export interface PublishedDatasetSchema {
+    fields: Record<string, PublishedDatasetField>;
+    order: string[];
+}
+export interface PublishedDatasetField {
+    name: string;
+    type: string;
+    customType?: string;
+    description?: string;
+    required?: boolean;
+    default?: unknown;
+    validation?: Record<string, unknown>;
+    /**
+     * Kept so an install into an org that HAS the referenced dataset can relink
+     * it. `datasetId` is the publisher's id and is meaningless in another org,
+     * so `datasetLabel` carries the human name the installer matches on.
+     */
+    reference?: {
+        datasetId?: string;
+        datasetLabel?: string;
+        displayFieldId?: string;
+        multiple?: boolean;
+        onDelete?: string;
+    };
+}
+/**
+ * Validates and strips a `DatasetModel` for publishing as a dataset schema
+ * (AGL-657).
+ *
+ * Publishes STRUCTURE ONLY: records never travel. That is the whole safety
+ * story for this artifact type: a dataset's rows are the org's customer data,
+ * so the publish path reads the model and nothing else, and there is no code
+ * path here that can reach the `records` subcollection.
+ *
+ * Two cross-org hazards the issue called out, handled by carrying enough
+ * context for the INSTALLER to decide rather than by rejecting at publish:
+ * - `reference` FKs point at a dataset id that only exists in the publisher's
+ *   org. The id is kept alongside the referenced dataset's label so the
+ *   installer can relink by name and degrade the field when it can't (see
+ *   `resolveInstalledDatasetSchema`).
+ * - `customType` names a plugin-declared field type (AGL-434) whose plugin the
+ *   installing org may not have. Unknown custom types already degrade to their
+ *   base type at render, so the name rides along untouched.
+ */
+export declare function sanitizeDatasetSchema(model: {
+    fields?: Record<string, any>;
+    order?: unknown;
+}): {
+    ok: true;
+    schema: PublishedDatasetSchema;
+} | {
+    ok: false;
+    error: string;
+};
+/**
+ * Rewrites a published schema for the installing org (AGL-657).
+ *
+ * `reference` fields are the only part of a schema that can't cross an org
+ * boundary as-is: they name a dataset id in the PUBLISHER's org. Relink by the
+ * referenced dataset's label when the installing org has one by that name;
+ * otherwise degrade the field to plain `text` and report it, because a
+ * reference field pointing at a dataset that doesn't exist renders as a broken
+ * picker. Silently keeping the dead id would install something visibly
+ * broken and blame the installer for it.
+ *
+ * `existingDatasets` maps lowercased display name → dataset id in the target.
+ */
+export declare function resolveInstalledDatasetSchema(schema: PublishedDatasetSchema, existingDatasets: Readonly<Record<string, string>>): {
+    schema: PublishedDatasetSchema;
+    degradedFieldIds: string[];
+};
+/**
+ * Dataset schemas, classified by what an update would do to EXISTING RECORDS
+ * (AGL-1018).
+ *
+ * The sharp edge of updating an installed copy: adding a field is inert, but
+ * removing or retyping one reinterprets data that is already there. So the
+ * schema plan is not "safe vs conflict" but "additive vs destructive", and
+ * destructive changes are never taken without a count of the records at stake.
+ */
+export interface SchemaChangeSummary {
+    /** Field ids the new version adds. Applying these cannot harm a record. */
+    added: string[];
+    /** Field ids it removes: the values in existing records become orphaned. */
+    removed: string[];
+    /** Field ids whose TYPE changed: existing values may not survive a re-read. */
+    retyped: string[];
+    /** Anything else (labels, options, ordering). */
+    edited: string[];
+    /** True when nothing removes or retypes a field. */
+    additiveOnly: boolean;
+}
+interface SchemaShape {
+    order?: unknown;
+    fields?: Record<string, {
+        type?: unknown;
+    } | undefined>;
+}
+/**
+ * Compares two dataset schemas field by field.
+ *
+ * Deliberately ignores the base: for a schema the question is not "who edited
+ * this" but "what happens to the rows", and that is answered by the current
+ * schema against the incoming one regardless of who moved which field.
+ */
+export declare function summarizeSchemaChange(current: SchemaShape | null | undefined, incoming: SchemaShape | null | undefined): SchemaChangeSummary;
+export {};

package/src/lib/artifact/dataset-schema.js

@@ -0,0 +1,232 @@
+import { _ as _extends } from "@swc/helpers/_/_extends";
+import { _ as _object_without_properties_loose } from "@swc/helpers/_/_object_without_properties_loose";
+/**
+ * @license
+ * Copyright 2026 Aglyn LLC
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */ import { stableStringify } from "@aglyn/aglyn/app-utils/artifact-provenance";
+import { PORTABLE_DATASET_FIELD_TYPES, PORTABLE_DEFINITION_MAX_BYTES } from "@aglyn/aglyn/app-utils/node-definition-sanitizer";
+/**
+ * A dataset's schema as it travels between workspaces (AGL-657): what a
+ * publisher's dataset is reduced to, what an installing workspace's new
+ * dataset is made from, and what an update compares. The data plugin's,
+ * because only it knows a dataset's model (AGL-3080); the installer stores and
+ * diffs it as content it does not read.
+ *
+ * Pure and context-free, so the owner's server answers
+ * (`dataset-schema-artifact.server.ts`) and their specs share one shape.
+ */ /** Field cap on a published schema, mirroring the console's create limit. */ export const DATASET_SCHEMA_MAX_FIELDS = 100;
+const KEPT_VALIDATION_KEYS = [
+    'required',
+    'regex',
+    'min',
+    'max',
+    'options'
+];
+/**
+ * Validates and strips a `DatasetModel` for publishing as a dataset schema
+ * (AGL-657).
+ *
+ * Publishes STRUCTURE ONLY: records never travel. That is the whole safety
+ * story for this artifact type: a dataset's rows are the org's customer data,
+ * so the publish path reads the model and nothing else, and there is no code
+ * path here that can reach the `records` subcollection.
+ *
+ * Two cross-org hazards the issue called out, handled by carrying enough
+ * context for the INSTALLER to decide rather than by rejecting at publish:
+ * - `reference` FKs point at a dataset id that only exists in the publisher's
+ *   org. The id is kept alongside the referenced dataset's label so the
+ *   installer can relink by name and degrade the field when it can't (see
+ *   `resolveInstalledDatasetSchema`).
+ * - `customType` names a plugin-declared field type (AGL-434) whose plugin the
+ *   installing org may not have. Unknown custom types already degrade to their
+ *   base type at render, so the name rides along untouched.
+ */ export function sanitizeDatasetSchema(model) {
+    const fields = model == null ? void 0 : model.fields;
+    if (!fields || typeof fields !== 'object') {
+        return {
+            ok: false,
+            error: 'Dataset has no field model to publish'
+        };
+    }
+    // Order is the display contract; fall back to key order for models written
+    // before `order` was required so older datasets stay publishable.
+    const order = Array.isArray(model.order) ? model.order.map(String) : Object.keys(fields);
+    const kept = order.filter((id)=>fields[id]);
+    if (!kept.length) {
+        return {
+            ok: false,
+            error: 'Dataset has no fields to publish'
+        };
+    }
+    if (kept.length > DATASET_SCHEMA_MAX_FIELDS) {
+        return {
+            ok: false,
+            error: `Dataset schemas are limited to ${DATASET_SCHEMA_MAX_FIELDS} fields`
+        };
+    }
+    const schema = {
+        fields: {},
+        order: kept
+    };
+    for (const id of kept){
+        var _fields_id, _field_type, _field_name;
+        const field = (_fields_id = fields[id]) != null ? _fields_id : {};
+        const type = String((_field_type = field.type) != null ? _field_type : 'text');
+        if (!PORTABLE_DATASET_FIELD_TYPES.includes(type)) {
+            return {
+                ok: false,
+                error: `Field "${id}" has an unsupported type`
+            };
+        }
+        const safe = {
+            name: String((_field_name = field.name) != null ? _field_name : id).slice(0, 120),
+            type
+        };
+        if (field.customType) safe.customType = String(field.customType).slice(0, 60);
+        if (field.description) {
+            safe.description = String(field.description).slice(0, 500);
+        }
+        if (field.required === true) safe.required = true;
+        // Defaults are publisher-authored values, not data. Keep only primitives
+        // so a default can't smuggle a nested object into the installing org.
+        if (field.default !== undefined && (typeof field.default === 'string' || typeof field.default === 'number' || typeof field.default === 'boolean')) {
+            safe.default = field.default;
+        }
+        if (field.validation && typeof field.validation === 'object') {
+            const validation = {};
+            for (const key of KEPT_VALIDATION_KEYS){
+                if (field.validation[key] !== undefined) {
+                    validation[key] = field.validation[key];
+                }
+            }
+            if (Object.keys(validation).length) safe.validation = validation;
+        }
+        if (field.reference && typeof field.reference === 'object') {
+            safe.reference = _extends({}, field.reference.datasetId && {
+                datasetId: String(field.reference.datasetId)
+            }, field.reference.datasetLabel && {
+                datasetLabel: String(field.reference.datasetLabel).slice(0, 120)
+            }, field.reference.displayFieldId && {
+                displayFieldId: String(field.reference.displayFieldId)
+            }, field.reference.multiple === true && {
+                multiple: true
+            }, field.reference.onDelete && {
+                onDelete: String(field.reference.onDelete)
+            });
+        }
+        schema.fields[id] = safe;
+    }
+    const serialized = JSON.stringify(schema);
+    if (serialized.length > PORTABLE_DEFINITION_MAX_BYTES) {
+        return {
+            ok: false,
+            error: 'Dataset schema is too large to publish'
+        };
+    }
+    return {
+        ok: true,
+        schema
+    };
+}
+/**
+ * Rewrites a published schema for the installing org (AGL-657).
+ *
+ * `reference` fields are the only part of a schema that can't cross an org
+ * boundary as-is: they name a dataset id in the PUBLISHER's org. Relink by the
+ * referenced dataset's label when the installing org has one by that name;
+ * otherwise degrade the field to plain `text` and report it, because a
+ * reference field pointing at a dataset that doesn't exist renders as a broken
+ * picker. Silently keeping the dead id would install something visibly
+ * broken and blame the installer for it.
+ *
+ * `existingDatasets` maps lowercased display name → dataset id in the target.
+ */ export function resolveInstalledDatasetSchema(schema, existingDatasets) {
+    const degradedFieldIds = [];
+    const fields = {};
+    for (const id of schema.order){
+        var _ref;
+        var _field_reference;
+        const field = schema.fields[id];
+        if (!field) continue;
+        if (field.type !== 'reference' && !field.reference) {
+            fields[id] = field;
+            continue;
+        }
+        const label = String((_ref = (_field_reference = field.reference) == null ? void 0 : _field_reference.datasetLabel) != null ? _ref : '').toLowerCase();
+        const relinked = label ? existingDatasets[label] : undefined;
+        if (relinked) {
+            fields[id] = _extends({}, field, {
+                reference: _extends({}, field.reference, {
+                    datasetId: relinked
+                })
+            });
+            continue;
+        }
+        const { reference: _dropped } = field, rest = _object_without_properties_loose(field, [
+            "reference"
+        ]);
+        fields[id] = _extends({}, rest, {
+            type: 'text'
+        });
+        degradedFieldIds.push(id);
+    }
+    return {
+        schema: {
+            fields,
+            order: schema.order
+        },
+        degradedFieldIds
+    };
+}
+const same = (a, b)=>stableStringify(a) === stableStringify(b);
+/**
+ * Compares two dataset schemas field by field.
+ *
+ * Deliberately ignores the base: for a schema the question is not "who edited
+ * this" but "what happens to the rows", and that is answered by the current
+ * schema against the incoming one regardless of who moved which field.
+ */ export function summarizeSchemaChange(current, incoming) {
+    var _ref, _ref1;
+    const currentFields = (_ref = current == null ? void 0 : current.fields) != null ? _ref : {};
+    const incomingFields = (_ref1 = incoming == null ? void 0 : incoming.fields) != null ? _ref1 : {};
+    const added = [];
+    const removed = [];
+    const retyped = [];
+    const edited = [];
+    for (const id of Object.keys(incomingFields)){
+        var _currentFields_id, _incomingFields_id;
+        if (!(id in currentFields)) {
+            added.push(id);
+            continue;
+        }
+        if (String((_currentFields_id = currentFields[id]) == null ? void 0 : _currentFields_id.type) !== String((_incomingFields_id = incomingFields[id]) == null ? void 0 : _incomingFields_id.type)) {
+            retyped.push(id);
+        } else if (!same(currentFields[id], incomingFields[id])) {
+            edited.push(id);
+        }
+    }
+    for (const id of Object.keys(currentFields)){
+        if (!(id in incomingFields)) removed.push(id);
+    }
+    return {
+        added,
+        removed,
+        retyped,
+        edited,
+        additiveOnly: !removed.length && !retyped.length
+    };
+}
+
+//# sourceMappingURL=dataset-schema.js.map

package/src/lib/artifact/dataset-schema.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../../../../../../libs/plugins/data/src/lib/artifact/dataset-schema.ts"],"sourcesContent":["/**\n * @license\n * Copyright 2026 Aglyn LLC\n *\n * Licensed under the Apache License, Version 2.0 (the \"License\");\n * you may not use this file except in compliance with the License.\n * You may obtain a copy of the License at\n *\n *   http://www.apache.org/licenses/LICENSE-2.0\n *\n * Unless required by applicable law or agreed to in writing, software\n * distributed under the License is distributed on an \"AS IS\" BASIS,\n * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n * See the License for the specific language governing permissions and\n * limitations under the License.\n */\n\nimport { stableStringify } from '@aglyn/aglyn/app-utils/artifact-provenance'\nimport {\n  PORTABLE_DATASET_FIELD_TYPES,\n  PORTABLE_DEFINITION_MAX_BYTES,\n} from '@aglyn/aglyn/app-utils/node-definition-sanitizer'\n\n/**\n * A dataset's schema as it travels between workspaces (AGL-657): what a\n * publisher's dataset is reduced to, what an installing workspace's new\n * dataset is made from, and what an update compares. The data plugin's,\n * because only it knows a dataset's model (AGL-3080); the installer stores and\n * diffs it as content it does not read.\n *\n * Pure and context-free, so the owner's server answers\n * (`dataset-schema-artifact.server.ts`) and their specs share one shape.\n */\n\n/** Field cap on a published schema, mirroring the console's create limit. */\nexport const DATASET_SCHEMA_MAX_FIELDS = 100\n\n/** A published dataset schema: the model, with no records. */\nexport interface PublishedDatasetSchema {\n  fields: Record<string, PublishedDatasetField>\n  order: string[]\n}\n\nexport interface PublishedDatasetField {\n  name: string\n  type: string\n  customType?: string\n  description?: string\n  required?: boolean\n  default?: unknown\n  validation?: Record<string, unknown>\n  /**\n   * Kept so an install into an org that HAS the referenced dataset can relink\n   * it. `datasetId` is the publisher's id and is meaningless in another org,\n   * so `datasetLabel` carries the human name the installer matches on.\n   */\n  reference?: {\n    datasetId?: string\n    datasetLabel?: string\n    displayFieldId?: string\n    multiple?: boolean\n    onDelete?: string\n  }\n}\n\nconst KEPT_VALIDATION_KEYS = ['required', 'regex', 'min', 'max', 'options']\n\n/**\n * Validates and strips a `DatasetModel` for publishing as a dataset schema\n * (AGL-657).\n *\n * Publishes STRUCTURE ONLY: records never travel. That is the whole safety\n * story for this artifact type: a dataset's rows are the org's customer data,\n * so the publish path reads the model and nothing else, and there is no code\n * path here that can reach the `records` subcollection.\n *\n * Two cross-org hazards the issue called out, handled by carrying enough\n * context for the INSTALLER to decide rather than by rejecting at publish:\n * - `reference` FKs point at a dataset id that only exists in the publisher's\n *   org. The id is kept alongside the referenced dataset's label so the\n *   installer can relink by name and degrade the field when it can't (see\n *   `resolveInstalledDatasetSchema`).\n * - `customType` names a plugin-declared field type (AGL-434) whose plugin the\n *   installing org may not have. Unknown custom types already degrade to their\n *   base type at render, so the name rides along untouched.\n */\nexport function sanitizeDatasetSchema(model: {\n  fields?: Record<string, any>\n  order?: unknown\n}):\n  | { ok: true; schema: PublishedDatasetSchema }\n  | { ok: false; error: string } {\n  const fields = model?.fields\n  if (!fields || typeof fields !== 'object') {\n    return { ok: false, error: 'Dataset has no field model to publish' }\n  }\n  // Order is the display contract; fall back to key order for models written\n  // before `order` was required so older datasets stay publishable.\n  const order = Array.isArray(model.order)\n    ? model.order.map(String)\n    : Object.keys(fields)\n  const kept = order.filter((id) => fields[id])\n  if (!kept.length) {\n    return { ok: false, error: 'Dataset has no fields to publish' }\n  }\n  if (kept.length > DATASET_SCHEMA_MAX_FIELDS) {\n    return {\n      ok: false,\n      error: `Dataset schemas are limited to ${DATASET_SCHEMA_MAX_FIELDS} fields`,\n    }\n  }\n  const schema: PublishedDatasetSchema = { fields: {}, order: kept }\n  for (const id of kept) {\n    const field = fields[id] ?? {}\n    const type = String(field.type ?? 'text')\n    if (!PORTABLE_DATASET_FIELD_TYPES.includes(type)) {\n      return { ok: false, error: `Field \"${id}\" has an unsupported type` }\n    }\n    const safe: PublishedDatasetField = {\n      name: String(field.name ?? id).slice(0, 120),\n      type,\n    }\n    if (field.customType) safe.customType = String(field.customType).slice(0, 60)\n    if (field.description) {\n      safe.description = String(field.description).slice(0, 500)\n    }\n    if (field.required === true) safe.required = true\n    // Defaults are publisher-authored values, not data. Keep only primitives\n    // so a default can't smuggle a nested object into the installing org.\n    if (\n      field.default !== undefined &&\n      (typeof field.default === 'string' ||\n        typeof field.default === 'number' ||\n        typeof field.default === 'boolean')\n    ) {\n      safe.default = field.default\n    }\n    if (field.validation && typeof field.validation === 'object') {\n      const validation: Record<string, unknown> = {}\n      for (const key of KEPT_VALIDATION_KEYS) {\n        if (field.validation[key] !== undefined) {\n          validation[key] = field.validation[key]\n        }\n      }\n      if (Object.keys(validation).length) safe.validation = validation\n    }\n    if (field.reference && typeof field.reference === 'object') {\n      safe.reference = {\n        ...(field.reference.datasetId && {\n          datasetId: String(field.reference.datasetId),\n        }),\n        ...(field.reference.datasetLabel && {\n          datasetLabel: String(field.reference.datasetLabel).slice(0, 120),\n        }),\n        ...(field.reference.displayFieldId && {\n          displayFieldId: String(field.reference.displayFieldId),\n        }),\n        ...(field.reference.multiple === true && { multiple: true }),\n        ...(field.reference.onDelete && {\n          onDelete: String(field.reference.onDelete),\n        }),\n      }\n    }\n    schema.fields[id] = safe\n  }\n  const serialized = JSON.stringify(schema)\n  if (serialized.length > PORTABLE_DEFINITION_MAX_BYTES) {\n    return { ok: false, error: 'Dataset schema is too large to publish' }\n  }\n  return { ok: true, schema }\n}\n\n/**\n * Rewrites a published schema for the installing org (AGL-657).\n *\n * `reference` fields are the only part of a schema that can't cross an org\n * boundary as-is: they name a dataset id in the PUBLISHER's org. Relink by the\n * referenced dataset's label when the installing org has one by that name;\n * otherwise degrade the field to plain `text` and report it, because a\n * reference field pointing at a dataset that doesn't exist renders as a broken\n * picker. Silently keeping the dead id would install something visibly\n * broken and blame the installer for it.\n *\n * `existingDatasets` maps lowercased display name → dataset id in the target.\n */\nexport function resolveInstalledDatasetSchema(\n  schema: PublishedDatasetSchema,\n  existingDatasets: Readonly<Record<string, string>>,\n): { schema: PublishedDatasetSchema; degradedFieldIds: string[] } {\n  const degradedFieldIds: string[] = []\n  const fields: Record<string, PublishedDatasetField> = {}\n  for (const id of schema.order) {\n    const field = schema.fields[id]\n    if (!field) continue\n    if (field.type !== 'reference' && !field.reference) {\n      fields[id] = field\n      continue\n    }\n    const label = String(field.reference?.datasetLabel ?? '').toLowerCase()\n    const relinked = label ? existingDatasets[label] : undefined\n    if (relinked) {\n      fields[id] = {\n        ...field,\n        reference: { ...field.reference, datasetId: relinked },\n      }\n      continue\n    }\n    const { reference: _dropped, ...rest } = field\n    fields[id] = { ...rest, type: 'text' }\n    degradedFieldIds.push(id)\n  }\n  return { schema: { fields, order: schema.order }, degradedFieldIds }\n}\n\n/**\n * Dataset schemas, classified by what an update would do to EXISTING RECORDS\n * (AGL-1018).\n *\n * The sharp edge of updating an installed copy: adding a field is inert, but\n * removing or retyping one reinterprets data that is already there. So the\n * schema plan is not \"safe vs conflict\" but \"additive vs destructive\", and\n * destructive changes are never taken without a count of the records at stake.\n */\nexport interface SchemaChangeSummary {\n  /** Field ids the new version adds. Applying these cannot harm a record. */\n  added: string[]\n  /** Field ids it removes: the values in existing records become orphaned. */\n  removed: string[]\n  /** Field ids whose TYPE changed: existing values may not survive a re-read. */\n  retyped: string[]\n  /** Anything else (labels, options, ordering). */\n  edited: string[]\n  /** True when nothing removes or retypes a field. */\n  additiveOnly: boolean\n}\n\ninterface SchemaShape {\n  order?: unknown\n  fields?: Record<string, { type?: unknown } | undefined>\n}\n\nconst same = (a: unknown, b: unknown): boolean =>\n  stableStringify(a) === stableStringify(b)\n\n/**\n * Compares two dataset schemas field by field.\n *\n * Deliberately ignores the base: for a schema the question is not \"who edited\n * this\" but \"what happens to the rows\", and that is answered by the current\n * schema against the incoming one regardless of who moved which field.\n */\nexport function summarizeSchemaChange(\n  current: SchemaShape | null | undefined,\n  incoming: SchemaShape | null | undefined,\n): SchemaChangeSummary {\n  const currentFields = (current?.fields ?? {}) as Record<string, { type?: unknown }>\n  const incomingFields = (incoming?.fields ?? {}) as Record<string, { type?: unknown }>\n  const added: string[] = []\n  const removed: string[] = []\n  const retyped: string[] = []\n  const edited: string[] = []\n  for (const id of Object.keys(incomingFields)) {\n    if (!(id in currentFields)) {\n      added.push(id)\n      continue\n    }\n    if (String(currentFields[id]?.type) !== String(incomingFields[id]?.type)) {\n      retyped.push(id)\n    } else if (!same(currentFields[id], incomingFields[id])) {\n      edited.push(id)\n    }\n  }\n  for (const id of Object.keys(currentFields)) {\n    if (!(id in incomingFields)) removed.push(id)\n  }\n  return {\n    added,\n    removed,\n    retyped,\n    edited,\n    additiveOnly: !removed.length && !retyped.length,\n  }\n}\n"],"names":["stableStringify","PORTABLE_DATASET_FIELD_TYPES","PORTABLE_DEFINITION_MAX_BYTES","DATASET_SCHEMA_MAX_FIELDS","KEPT_VALIDATION_KEYS","sanitizeDatasetSchema","model","fields","ok","error","order","Array","isArray","map","String","Object","keys","kept","filter","id","length","schema","field","type","includes","safe","name","slice","customType","description","required","default","undefined","validation","key","reference","datasetId","datasetLabel","displayFieldId","multiple","onDelete","serialized","JSON","stringify","resolveInstalledDatasetSchema","existingDatasets","degradedFieldIds","label","toLowerCase","relinked","_dropped","rest","push","same","a","b","summarizeSchemaChange","current","incoming","currentFields","incomingFields","added","removed","retyped","edited","additiveOnly"],"mappings":";;AAAA;;;;;;;;;;;;;;;CAeC,GAED,SAASA,eAAe,QAAQ,6CAA4C;AAC5E,SACEC,4BAA4B,EAC5BC,6BAA6B,QACxB,mDAAkD;AAEzD;;;;;;;;;CASC,GAED,2EAA2E,GAC3E,OAAO,MAAMC,4BAA4B,IAAG;AA8B5C,MAAMC,uBAAuB;IAAC;IAAY;IAAS;IAAO;IAAO;CAAU;AAE3E;;;;;;;;;;;;;;;;;;CAkBC,GACD,OAAO,SAASC,sBAAsBC,KAGrC;IAGC,MAAMC,SAASD,yBAAAA,MAAOC,MAAM;IAC5B,IAAI,CAACA,UAAU,OAAOA,WAAW,UAAU;QACzC,OAAO;YAAEC,IAAI;YAAOC,OAAO;QAAwC;IACrE;IACA,2EAA2E;IAC3E,kEAAkE;IAClE,MAAMC,QAAQC,MAAMC,OAAO,CAACN,MAAMI,KAAK,IACnCJ,MAAMI,KAAK,CAACG,GAAG,CAACC,UAChBC,OAAOC,IAAI,CAACT;IAChB,MAAMU,OAAOP,MAAMQ,MAAM,CAAC,CAACC,KAAOZ,MAAM,CAACY,GAAG;IAC5C,IAAI,CAACF,KAAKG,MAAM,EAAE;QAChB,OAAO;YAAEZ,IAAI;YAAOC,OAAO;QAAmC;IAChE;IACA,IAAIQ,KAAKG,MAAM,GAAGjB,2BAA2B;QAC3C,OAAO;YACLK,IAAI;YACJC,OAAO,CAAC,+BAA+B,EAAEN,0BAA0B,OAAO,CAAC;QAC7E;IACF;IACA,MAAMkB,SAAiC;QAAEd,QAAQ,CAAC;QAAGG,OAAOO;IAAK;IACjE,KAAK,MAAME,MAAMF,KAAM;YACPV,YACMe,aAKLA;QANf,MAAMA,SAAQf,aAAAA,MAAM,CAACY,GAAG,YAAVZ,aAAc,CAAC;QAC7B,MAAMgB,OAAOT,QAAOQ,cAAAA,MAAMC,IAAI,YAAVD,cAAc;QAClC,IAAI,CAACrB,6BAA6BuB,QAAQ,CAACD,OAAO;YAChD,OAAO;gBAAEf,IAAI;gBAAOC,OAAO,CAAC,OAAO,EAAEU,GAAG,yBAAyB,CAAC;YAAC;QACrE;QACA,MAAMM,OAA8B;YAClCC,MAAMZ,QAAOQ,cAAAA,MAAMI,IAAI,YAAVJ,cAAcH,IAAIQ,KAAK,CAAC,GAAG;YACxCJ;QACF;QACA,IAAID,MAAMM,UAAU,EAAEH,KAAKG,UAAU,GAAGd,OAAOQ,MAAMM,UAAU,EAAED,KAAK,CAAC,GAAG;QAC1E,IAAIL,MAAMO,WAAW,EAAE;YACrBJ,KAAKI,WAAW,GAAGf,OAAOQ,MAAMO,WAAW,EAAEF,KAAK,CAAC,GAAG;QACxD;QACA,IAAIL,MAAMQ,QAAQ,KAAK,MAAML,KAAKK,QAAQ,GAAG;QAC7C,yEAAyE;QACzE,sEAAsE;QACtE,IACER,MAAMS,OAAO,KAAKC,aACjB,CAAA,OAAOV,MAAMS,OAAO,KAAK,YACxB,OAAOT,MAAMS,OAAO,KAAK,YACzB,OAAOT,MAAMS,OAAO,KAAK,SAAQ,GACnC;YACAN,KAAKM,OAAO,GAAGT,MAAMS,OAAO;QAC9B;QACA,IAAIT,MAAMW,UAAU,IAAI,OAAOX,MAAMW,UAAU,KAAK,UAAU;YAC5D,MAAMA,aAAsC,CAAC;YAC7C,KAAK,MAAMC,OAAO9B,qBAAsB;gBACtC,IAAIkB,MAAMW,UAAU,CAACC,IAAI,KAAKF,WAAW;oBACvCC,UAAU,CAACC,IAAI,GAAGZ,MAAMW,UAAU,CAACC,IAAI;gBACzC;YACF;YACA,IAAInB,OAAOC,IAAI,CAACiB,YAAYb,MAAM,EAAEK,KAAKQ,UAAU,GAAGA;QACxD;QACA,IAAIX,MAAMa,SAAS,IAAI,OAAOb,MAAMa,SAAS,KAAK,UAAU;YAC1DV,KAAKU,SAAS,GAAG,aACXb,MAAMa,SAAS,CAACC,SAAS,IAAI;gBAC/BA,WAAWtB,OAAOQ,MAAMa,SAAS,CAACC,SAAS;YAC7C,GACId,MAAMa,SAAS,CAACE,YAAY,IAAI;gBAClCA,cAAcvB,OAAOQ,MAAMa,SAAS,CAACE,YAAY,EAAEV,KAAK,CAAC,GAAG;YAC9D,GACIL,MAAMa,SAAS,CAACG,cAAc,IAAI;gBACpCA,gBAAgBxB,OAAOQ,MAAMa,SAAS,CAACG,cAAc;YACvD,GACIhB,MAAMa,SAAS,CAACI,QAAQ,KAAK,QAAQ;gBAAEA,UAAU;YAAK,GACtDjB,MAAMa,SAAS,CAACK,QAAQ,IAAI;gBAC9BA,UAAU1B,OAAOQ,MAAMa,SAAS,CAACK,QAAQ;YAC3C;QAEJ;QACAnB,OAAOd,MAAM,CAACY,GAAG,GAAGM;IACtB;IACA,MAAMgB,aAAaC,KAAKC,SAAS,CAACtB;IAClC,IAAIoB,WAAWrB,MAAM,GAAGlB,+BAA+B;QACrD,OAAO;YAAEM,IAAI;YAAOC,OAAO;QAAyC;IACtE;IACA,OAAO;QAAED,IAAI;QAAMa;IAAO;AAC5B;AAEA;;;;;;;;;;;;CAYC,GACD,OAAO,SAASuB,8BACdvB,MAA8B,EAC9BwB,gBAAkD;IAElD,MAAMC,mBAA6B,EAAE;IACrC,MAAMvC,SAAgD,CAAC;IACvD,KAAK,MAAMY,MAAME,OAAOX,KAAK,CAAE;;YAORY;QANrB,MAAMA,QAAQD,OAAOd,MAAM,CAACY,GAAG;QAC/B,IAAI,CAACG,OAAO;QACZ,IAAIA,MAAMC,IAAI,KAAK,eAAe,CAACD,MAAMa,SAAS,EAAE;YAClD5B,MAAM,CAACY,GAAG,GAAGG;YACb;QACF;QACA,MAAMyB,QAAQjC,gBAAOQ,mBAAAA,MAAMa,SAAS,qBAAfb,iBAAiBe,YAAY,mBAAI,IAAIW,WAAW;QACrE,MAAMC,WAAWF,QAAQF,gBAAgB,CAACE,MAAM,GAAGf;QACnD,IAAIiB,UAAU;YACZ1C,MAAM,CAACY,GAAG,GAAG,aACRG;gBACHa,WAAW,aAAKb,MAAMa,SAAS;oBAAEC,WAAWa;;;YAE9C;QACF;QACA,MAAM,EAAEd,WAAWe,QAAQ,EAAW,GAAG5B,OAAT6B,wCAAS7B;;;QACzCf,MAAM,CAACY,GAAG,GAAG,aAAKgC;YAAM5B,MAAM;;QAC9BuB,iBAAiBM,IAAI,CAACjC;IACxB;IACA,OAAO;QAAEE,QAAQ;YAAEd;YAAQG,OAAOW,OAAOX,KAAK;QAAC;QAAGoC;IAAiB;AACrE;AA6BA,MAAMO,OAAO,CAACC,GAAYC,IACxBvD,gBAAgBsD,OAAOtD,gBAAgBuD;AAEzC;;;;;;CAMC,GACD,OAAO,SAASC,sBACdC,OAAuC,EACvCC,QAAwC;;IAExC,MAAMC,wBAAiBF,2BAAAA,QAASlD,MAAM,mBAAI,CAAC;IAC3C,MAAMqD,0BAAkBF,4BAAAA,SAAUnD,MAAM,oBAAI,CAAC;IAC7C,MAAMsD,QAAkB,EAAE;IAC1B,MAAMC,UAAoB,EAAE;IAC5B,MAAMC,UAAoB,EAAE;IAC5B,MAAMC,SAAmB,EAAE;IAC3B,KAAK,MAAM7C,MAAMJ,OAAOC,IAAI,CAAC4C,gBAAiB;YAKjCD,mBAAoCC;QAJ/C,IAAI,CAAEzC,CAAAA,MAAMwC,aAAY,GAAI;YAC1BE,MAAMT,IAAI,CAACjC;YACX;QACF;QACA,IAAIL,QAAO6C,oBAAAA,aAAa,CAACxC,GAAG,qBAAjBwC,kBAAmBpC,IAAI,MAAMT,QAAO8C,qBAAAA,cAAc,CAACzC,GAAG,qBAAlByC,mBAAoBrC,IAAI,GAAG;YACxEwC,QAAQX,IAAI,CAACjC;QACf,OAAO,IAAI,CAACkC,KAAKM,aAAa,CAACxC,GAAG,EAAEyC,cAAc,CAACzC,GAAG,GAAG;YACvD6C,OAAOZ,IAAI,CAACjC;QACd;IACF;IACA,KAAK,MAAMA,MAAMJ,OAAOC,IAAI,CAAC2C,eAAgB;QAC3C,IAAI,CAAExC,CAAAA,MAAMyC,cAAa,GAAIE,QAAQV,IAAI,CAACjC;IAC5C;IACA,OAAO;QACL0C;QACAC;QACAC;QACAC;QACAC,cAAc,CAACH,QAAQ1C,MAAM,IAAI,CAAC2C,QAAQ3C,MAAM;IAClD;AACF"}

package/src/lib/components/dataset-record-dialog.component.d.ts

@@ -1,5 +1,5 @@
-import { type DatasetModel } from '@aglyn/aglyn';
 import { type DatasetReferenceResolver } from '../model/dataset-record-view';
+import { type DatasetModel } from '../model/dataset-models';
 /**
  * Looking at a record, as its own gesture.
  *

package/src/lib/components/dataset-record-dialog.component.js

@@ -16,10 +16,10 @@
  */ 'use client';
 import { _ as _extends } from "@swc/helpers/_/_extends";
 import { jsx as _jsx, jsxs as _jsxs } from "react/jsx-runtime";
-import { validateDocument } from "@aglyn/aglyn";
 import { datasetRecordFields } from "../model/dataset-record-view.js";
 import { Alert, Box, Button, Dialog, DialogActions, DialogContent, DialogTitle, Stack, Tooltip, Typography } from "@mui/material";
 import { useMemo } from "react";
+import { validateDocument } from "../model/dataset-models.js";
 /** What each "no value" state is called on screen. */ const PLACEHOLDER = {
     absent: 'Not set',
     null: 'Null',

package/src/lib/components/dataset-record-dialog.component.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../../../../../../../libs/plugins/data/src/lib/components/dataset-record-dialog.component.tsx"],"sourcesContent":["/**\n * @license\n * Copyright 2026 Aglyn LLC\n *\n * Licensed under the Apache License, Version 2.0 (the \"License\");\n * you may not use this file except in compliance with the License.\n * You may obtain a copy of the License at\n *\n *   http://www.apache.org/licenses/LICENSE-2.0\n *\n * Unless required by applicable law or agreed to in writing, software\n * distributed under the License is distributed on an \"AS IS\" BASIS,\n * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n * See the License for the specific language governing permissions and\n * limitations under the License.\n */\n'use client'\n\nimport { type DatasetModel, validateDocument } from '@aglyn/aglyn'\nimport {\n  type DatasetRecordField,\n  type DatasetReferenceResolver,\n  datasetRecordFields,\n} from '../model/dataset-record-view'\nimport {\n  Alert,\n  Box,\n  Button,\n  Dialog,\n  DialogActions,\n  DialogContent,\n  DialogTitle,\n  Stack,\n  Tooltip,\n  Typography,\n} from '@mui/material'\nimport { useMemo } from 'react'\n\n/**\n * Looking at a record, as its own gesture.\n *\n * The records table has room for the model's columns and one line per value,\n * so the only way to see what a record actually holds was to open it in the\n * EDITOR — which made reading a record and changing one the same click, on a\n * dialog whose primary button writes. This is the other half: the whole\n * record, nothing writable.\n *\n * ## It reads nothing\n *\n * Every value on screen arrives as a prop, out of the page the table already\n * listens to. There is no query, no `getDoc`, and no effect — opening a record\n * costs zero documents, and rendering the table does not open one dialog per\n * row to find out. That is not an optimization detail: a viewer that fetched\n * the record it was handed would bill a read per glance, and one that mounted\n * per row would bill the whole page on arrival.\n *\n * ## Read-only means read-only\n *\n * Nothing here takes input and nothing here writes. `Edit record` is a\n * separate, explicitly labelled action that closes this and opens the editor —\n * the dialog that has always owned writing — so the two gestures stay\n * distinct rather than one being a slip away from the other.\n */\nexport interface DatasetRecordDialogProps {\n  /**\n   * The record row itself, straight from the table's page. `null` closes the\n   * dialog — including when the row leaves the page under it, which is the\n   * honest response to a record this reader can no longer see.\n   */\n  record: { $id: string; values?: Record<string, unknown> } | null\n  model: DatasetModel\n  /** Reference target ID → label, or `null` when it resolves to nothing. */\n  resolveReference?: DatasetReferenceResolver\n  onClose: () => void\n  /**\n   * Opens the record in the editor. Omitted for a reader who cannot write, so\n   * the control is absent rather than present-and-refusing.\n   */\n  onEdit?: () => void\n}\n\n/** What each \"no value\" state is called on screen. */\nconst PLACEHOLDER: Record<string, string> = {\n  absent: 'Not set',\n  null: 'Null',\n  'empty-text': 'Empty text',\n  'empty-list': 'Empty list',\n  'empty-map': 'Empty map',\n}\n\n/**\n * One field's value.\n *\n * The four empty states get four different words, deliberately. A table cell\n * prints `--` for all of them because it has one line to work with; here there\n * is room to say which, and the difference decides what a reader does next —\n * a field that was never written and a field holding a real empty string are\n * not the same finding.\n */\nfunction RecordFieldValue(props: { field: DatasetRecordField }) {\n  const { value } = props.field\n  if (value.kind === 'opaque') {\n    return (\n      <Typography\n        variant=\"body2\"\n        color=\"text.secondary\"\n        sx={{ fontStyle: 'italic' }}\n      >\n        {value.text}\n      </Typography>\n    )\n  }\n  if (value.kind !== 'value') {\n    return (\n      <Typography\n        variant=\"body2\"\n        color=\"text.disabled\"\n        sx={{ fontStyle: 'italic' }}\n      >\n        {PLACEHOLDER[value.kind] ?? 'Not set'}\n      </Typography>\n    )\n  }\n  if (value.references) {\n    return (\n      <Stack direction=\"row\" spacing={1} sx={{ flexWrap: 'wrap', rowGap: 0.5 }}>\n        {value.references.map((reference) => (\n          <Typography\n            key={reference.id}\n            variant=\"body2\"\n            // An unresolved target is a fact about the data, not a rendering\n            // failure to paper over: the ID is shown either way, and only the\n            // resolved one is allowed to look like a working link.\n            color={reference.label ? 'text.primary' : 'warning.main'}\n            sx={{ overflowWrap: 'anywhere' }}\n          >\n            {reference.label ?? `${reference.id} · unresolved`}\n          </Typography>\n        ))}\n      </Stack>\n    )\n  }\n  if (value.items) {\n    return (\n      // One line per item. The comma-joined form a cell shows cannot say\n      // whether `a, b` is one item or two; a list of lines can.\n      <Stack component=\"ul\" spacing={0.25} sx={{ m: 0, pl: 2.5 }}>\n        {value.items.map((item, index) => (\n          <Typography\n            // Items are values, not identities — two equal strings are two\n            // real entries and the position is what tells them apart.\n            key={`${index}:${item}`}\n            component=\"li\"\n            variant=\"body2\"\n            sx={{ overflowWrap: 'anywhere' }}\n          >\n            {item}\n          </Typography>\n        ))}\n      </Stack>\n    )\n  }\n  return (\n    <Typography\n      variant=\"body2\"\n      // Pretty-printed JSON is only readable where its whitespace survives.\n      {...(value.block\n        ? {\n            component: 'pre',\n            sx: {\n              m: 0,\n              fontFamily: 'monospace',\n              whiteSpace: 'pre-wrap',\n              overflowWrap: 'anywhere',\n            },\n          }\n        : { sx: { overflowWrap: 'anywhere' } })}\n    >\n      {value.text}\n    </Typography>\n  )\n}\nRecordFieldValue.displayName = 'RecordFieldValue'\n\nexport function DatasetRecordDialog(props: DatasetRecordDialogProps) {\n  const { record, model, resolveReference, onClose, onEdit } = props\n  const values = record?.values\n  const fields = useMemo(\n    () => (record ? datasetRecordFields(model, values, resolveReference) : []),\n    [record, model, values, resolveReference],\n  )\n  /**\n   * The same mismatch report the editor shows, for the same reason: a type\n   * change never rewrites stored documents, so a record may legitimately hold\n   * a value its model no longer describes. Reporting it beside the value is\n   * what stops the view claiming the record conforms.\n   */\n  const mismatches = useMemo(\n    () => (record ? validateDocument(model, values ?? {}) : {}),\n    [record, model, values],\n  )\n\n  return (\n    <Dialog\n      open={Boolean(record)}\n      onClose={onClose}\n      maxWidth=\"sm\"\n      fullWidth\n      aria-labelledby=\"dataset-record-view-title\"\n    >\n      <DialogTitle id=\"dataset-record-view-title\">{'View record'}</DialogTitle>\n      <DialogContent dividers>\n        <Stack spacing={2}>\n          <Box>\n            <Typography\n              variant=\"overline\"\n              color=\"text.secondary\"\n              component=\"div\"\n            >\n              {'Record ID'}\n            </Typography>\n            <Typography variant=\"body2\" sx={{ overflowWrap: 'anywhere' }}>\n              {record?.$id ?? ''}\n            </Typography>\n          </Box>\n          {fields.map((field) => (\n            <Box key={field.fieldId}>\n              <Stack\n                direction=\"row\"\n                spacing={1}\n                sx={{ alignItems: 'baseline', flexWrap: 'wrap' }}\n              >\n                <Typography\n                  variant=\"overline\"\n                  color=\"text.secondary\"\n                  component=\"div\"\n                >\n                  {field.label}\n                </Typography>\n                {field.source === 'extra' ? (\n                  <Tooltip\n                    title={\n                      'Stored on this record with no field in the ' +\n                      'collection’s schema. The table cannot show it, and ' +\n                      'the next save through the editor removes it.'\n                    }\n                  >\n                    <Typography\n                      variant=\"caption\"\n                      color=\"warning.main\"\n                      sx={{ cursor: 'help' }}\n                    >\n                      {'not in schema'}\n                    </Typography>\n                  </Tooltip>\n                ) : null}\n              </Stack>\n              <RecordFieldValue field={field} />\n              {mismatches[field.fieldId] ? (\n                <Typography variant=\"caption\" color=\"warning.main\">\n                  {mismatches[field.fieldId]}\n                </Typography>\n              ) : null}\n              {field.description ? (\n                <Typography\n                  variant=\"caption\"\n                  color=\"text.secondary\"\n                  component=\"div\"\n                >\n                  {field.description}\n                </Typography>\n              ) : null}\n            </Box>\n          ))}\n          {fields.length === 0 ? (\n            <Alert severity=\"info\">{'This record holds no fields.'}</Alert>\n          ) : null}\n        </Stack>\n      </DialogContent>\n      <DialogActions>\n        {onEdit ? <Button onClick={onEdit}>{'Edit record'}</Button> : null}\n        <Button variant=\"contained\" color=\"primary\" onClick={onClose}>\n          {'Close'}\n        </Button>\n      </DialogActions>\n    </Dialog>\n  )\n}\nDatasetRecordDialog.displayName = 'DatasetRecordDialog'\n\nexport default DatasetRecordDialog\n"],"names":["validateDocument","datasetRecordFields","Alert","Box","Button","Dialog","DialogActions","DialogContent","DialogTitle","Stack","Tooltip","Typography","useMemo","PLACEHOLDER","absent","null","RecordFieldValue","props","value","field","kind","variant","color","sx","fontStyle","text","references","direction","spacing","flexWrap","rowGap","map","reference","label","overflowWrap","id","items","component","m","pl","item","index","block","fontFamily","whiteSpace","displayName","DatasetRecordDialog","record","model","resolveReference","onClose","onEdit","values","fields","mismatches","open","Boolean","maxWidth","fullWidth","aria-labelledby","dividers","$id","alignItems","source","title","cursor","fieldId","description","length","severity","onClick"],"mappings":"AAAA;;;;;;;;;;;;;;;CAeC,GACD;;;AAEA,SAA4BA,gBAAgB,QAAQ,eAAc;AAClE,SAGEC,mBAAmB,QACd,kCAA8B;AACrC,SACEC,KAAK,EACLC,GAAG,EACHC,MAAM,EACNC,MAAM,EACNC,aAAa,EACbC,aAAa,EACbC,WAAW,EACXC,KAAK,EACLC,OAAO,EACPC,UAAU,QACL,gBAAe;AACtB,SAASC,OAAO,QAAQ,QAAO;AA6C/B,oDAAoD,GACpD,MAAMC,cAAsC;IAC1CC,QAAQ;IACRC,MAAM;IACN,cAAc;IACd,cAAc;IACd,aAAa;AACf;AAEA;;;;;;;;CAQC,GACD,SAASC,iBAAiBC,KAAoC;IAC5D,MAAM,EAAEC,KAAK,EAAE,GAAGD,MAAME,KAAK;IAC7B,IAAID,MAAME,IAAI,KAAK,UAAU;QAC3B,qBACE,KAACT;YACCU,SAAQ;YACRC,OAAM;YACNC,IAAI;gBAAEC,WAAW;YAAS;sBAEzBN,MAAMO,IAAI;;IAGjB;IACA,IAAIP,MAAME,IAAI,KAAK,SAAS;YAOrBP;QANL,qBACE,KAACF;YACCU,SAAQ;YACRC,OAAM;YACNC,IAAI;gBAAEC,WAAW;YAAS;uBAEzBX,0BAAAA,WAAW,CAACK,MAAME,IAAI,CAAC,YAAvBP,0BAA2B;;IAGlC;IACA,IAAIK,MAAMQ,UAAU,EAAE;QACpB,qBACE,KAACjB;YAAMkB,WAAU;YAAMC,SAAS;YAAGL,IAAI;gBAAEM,UAAU;gBAAQC,QAAQ;YAAI;sBACpEZ,MAAMQ,UAAU,CAACK,GAAG,CAAC,CAACC;oBAUlBA;qCATH,KAACrB;oBAECU,SAAQ;oBACR,iEAAiE;oBACjE,kEAAkE;oBAClE,uDAAuD;oBACvDC,OAAOU,UAAUC,KAAK,GAAG,iBAAiB;oBAC1CV,IAAI;wBAAEW,cAAc;oBAAW;+BAE9BF,mBAAAA,UAAUC,KAAK,YAAfD,mBAAmB,GAAGA,UAAUG,EAAE,CAAC,aAAa,CAAC;mBAR7CH,UAAUG,EAAE;;;IAa3B;IACA,IAAIjB,MAAMkB,KAAK,EAAE;QACf,OACE,mEAAmE;QACnE,0DAA0D;sBAC1D,KAAC3B;YAAM4B,WAAU;YAAKT,SAAS;YAAML,IAAI;gBAAEe,GAAG;gBAAGC,IAAI;YAAI;sBACtDrB,MAAMkB,KAAK,CAACL,GAAG,CAAC,CAACS,MAAMC,sBACtB,KAAC9B;oBAIC0B,WAAU;oBACVhB,SAAQ;oBACRE,IAAI;wBAAEW,cAAc;oBAAW;8BAE9BM;mBALI,GAAGC,MAAM,CAAC,EAAED,MAAM;;IAUjC;IACA,qBACE,KAAC7B;QACCU,SAAQ;OAEHH,MAAMwB,KAAK,GACZ;QACEL,WAAW;QACXd,IAAI;YACFe,GAAG;YACHK,YAAY;YACZC,YAAY;YACZV,cAAc;QAChB;IACF,IACA;QAAEX,IAAI;YAAEW,cAAc;QAAW;IAAE;kBAEtChB,MAAMO,IAAI;;AAGjB;AACAT,iBAAiB6B,WAAW,GAAG;AAE/B,OAAO,SAASC,oBAAoB7B,KAA+B;;IACjE,MAAM,EAAE8B,MAAM,EAAEC,KAAK,EAAEC,gBAAgB,EAAEC,OAAO,EAAEC,MAAM,EAAE,GAAGlC;IAC7D,MAAMmC,SAASL,0BAAAA,OAAQK,MAAM;IAC7B,MAAMC,SAASzC,QACb,IAAOmC,SAAS9C,oBAAoB+C,OAAOI,QAAQH,oBAAoB,EAAE,EACzE;QAACF;QAAQC;QAAOI;QAAQH;KAAiB;IAE3C;;;;;GAKC,GACD,MAAMK,aAAa1C,QACjB,IAAOmC,SAAS/C,iBAAiBgD,OAAOI,iBAAAA,SAAU,CAAC,KAAK,CAAC,GACzD;QAACL;QAAQC;QAAOI;KAAO;IAGzB,qBACE,MAAC/C;QACCkD,MAAMC,QAAQT;QACdG,SAASA;QACTO,UAAS;QACTC,SAAS;QACTC,mBAAgB;;0BAEhB,KAACnD;gBAAY2B,IAAG;0BAA6B;;0BAC7C,KAAC5B;gBAAcqD,QAAQ;0BACrB,cAAA,MAACnD;oBAAMmB,SAAS;;sCACd,MAACzB;;8CACC,KAACQ;oCACCU,SAAQ;oCACRC,OAAM;oCACNe,WAAU;8CAET;;8CAEH,KAAC1B;oCAAWU,SAAQ;oCAAQE,IAAI;wCAAEW,cAAc;oCAAW;sDACxDa,0BAAAA,OAAQc,GAAG,mBAAI;;;;wBAGnBR,OAAOtB,GAAG,CAAC,CAACZ,sBACX,MAAChB;;kDACC,MAACM;wCACCkB,WAAU;wCACVC,SAAS;wCACTL,IAAI;4CAAEuC,YAAY;4CAAYjC,UAAU;wCAAO;;0DAE/C,KAAClB;gDACCU,SAAQ;gDACRC,OAAM;gDACNe,WAAU;0DAETlB,MAAMc,KAAK;;4CAEbd,MAAM4C,MAAM,KAAK,wBAChB,KAACrD;gDACCsD,OACE,gDACA,wDACA;0DAGF,cAAA,KAACrD;oDACCU,SAAQ;oDACRC,OAAM;oDACNC,IAAI;wDAAE0C,QAAQ;oDAAO;8DAEpB;;iDAGH;;;kDAEN,KAACjD;wCAAiBG,OAAOA;;oCACxBmC,UAAU,CAACnC,MAAM+C,OAAO,CAAC,iBACxB,KAACvD;wCAAWU,SAAQ;wCAAUC,OAAM;kDACjCgC,UAAU,CAACnC,MAAM+C,OAAO,CAAC;yCAE1B;oCACH/C,MAAMgD,WAAW,iBAChB,KAACxD;wCACCU,SAAQ;wCACRC,OAAM;wCACNe,WAAU;kDAETlB,MAAMgD,WAAW;yCAElB;;+BA7CIhD,MAAM+C,OAAO;wBAgDxBb,OAAOe,MAAM,KAAK,kBACjB,KAAClE;4BAAMmE,UAAS;sCAAQ;6BACtB;;;;0BAGR,MAAC/D;;oBACE6C,uBAAS,KAAC/C;wBAAOkE,SAASnB;kCAAS;yBAA0B;kCAC9D,KAAC/C;wBAAOiB,SAAQ;wBAAYC,OAAM;wBAAUgD,SAASpB;kCAClD;;;;;;AAKX;AACAJ,oBAAoBD,WAAW,GAAG;AAElC,eAAeC,oBAAmB"}
+{"version":3,"sources":["../../../../../../../libs/plugins/data/src/lib/components/dataset-record-dialog.component.tsx"],"sourcesContent":["/**\n * @license\n * Copyright 2026 Aglyn LLC\n *\n * Licensed under the Apache License, Version 2.0 (the \"License\");\n * you may not use this file except in compliance with the License.\n * You may obtain a copy of the License at\n *\n *   http://www.apache.org/licenses/LICENSE-2.0\n *\n * Unless required by applicable law or agreed to in writing, software\n * distributed under the License is distributed on an \"AS IS\" BASIS,\n * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n * See the License for the specific language governing permissions and\n * limitations under the License.\n */\n'use client'\n\nimport {\n  type DatasetRecordField,\n  type DatasetReferenceResolver,\n  datasetRecordFields,\n} from '../model/dataset-record-view'\nimport {\n  Alert,\n  Box,\n  Button,\n  Dialog,\n  DialogActions,\n  DialogContent,\n  DialogTitle,\n  Stack,\n  Tooltip,\n  Typography,\n} from '@mui/material'\nimport { useMemo } from 'react'\nimport { type DatasetModel, validateDocument } from '../model/dataset-models'\n\n/**\n * Looking at a record, as its own gesture.\n *\n * The records table has room for the model's columns and one line per value,\n * so the only way to see what a record actually holds was to open it in the\n * EDITOR — which made reading a record and changing one the same click, on a\n * dialog whose primary button writes. This is the other half: the whole\n * record, nothing writable.\n *\n * ## It reads nothing\n *\n * Every value on screen arrives as a prop, out of the page the table already\n * listens to. There is no query, no `getDoc`, and no effect — opening a record\n * costs zero documents, and rendering the table does not open one dialog per\n * row to find out. That is not an optimization detail: a viewer that fetched\n * the record it was handed would bill a read per glance, and one that mounted\n * per row would bill the whole page on arrival.\n *\n * ## Read-only means read-only\n *\n * Nothing here takes input and nothing here writes. `Edit record` is a\n * separate, explicitly labelled action that closes this and opens the editor —\n * the dialog that has always owned writing — so the two gestures stay\n * distinct rather than one being a slip away from the other.\n */\nexport interface DatasetRecordDialogProps {\n  /**\n   * The record row itself, straight from the table's page. `null` closes the\n   * dialog — including when the row leaves the page under it, which is the\n   * honest response to a record this reader can no longer see.\n   */\n  record: { $id: string; values?: Record<string, unknown> } | null\n  model: DatasetModel\n  /** Reference target ID → label, or `null` when it resolves to nothing. */\n  resolveReference?: DatasetReferenceResolver\n  onClose: () => void\n  /**\n   * Opens the record in the editor. Omitted for a reader who cannot write, so\n   * the control is absent rather than present-and-refusing.\n   */\n  onEdit?: () => void\n}\n\n/** What each \"no value\" state is called on screen. */\nconst PLACEHOLDER: Record<string, string> = {\n  absent: 'Not set',\n  null: 'Null',\n  'empty-text': 'Empty text',\n  'empty-list': 'Empty list',\n  'empty-map': 'Empty map',\n}\n\n/**\n * One field's value.\n *\n * The four empty states get four different words, deliberately. A table cell\n * prints `--` for all of them because it has one line to work with; here there\n * is room to say which, and the difference decides what a reader does next —\n * a field that was never written and a field holding a real empty string are\n * not the same finding.\n */\nfunction RecordFieldValue(props: { field: DatasetRecordField }) {\n  const { value } = props.field\n  if (value.kind === 'opaque') {\n    return (\n      <Typography\n        variant=\"body2\"\n        color=\"text.secondary\"\n        sx={{ fontStyle: 'italic' }}\n      >\n        {value.text}\n      </Typography>\n    )\n  }\n  if (value.kind !== 'value') {\n    return (\n      <Typography\n        variant=\"body2\"\n        color=\"text.disabled\"\n        sx={{ fontStyle: 'italic' }}\n      >\n        {PLACEHOLDER[value.kind] ?? 'Not set'}\n      </Typography>\n    )\n  }\n  if (value.references) {\n    return (\n      <Stack direction=\"row\" spacing={1} sx={{ flexWrap: 'wrap', rowGap: 0.5 }}>\n        {value.references.map((reference) => (\n          <Typography\n            key={reference.id}\n            variant=\"body2\"\n            // An unresolved target is a fact about the data, not a rendering\n            // failure to paper over: the ID is shown either way, and only the\n            // resolved one is allowed to look like a working link.\n            color={reference.label ? 'text.primary' : 'warning.main'}\n            sx={{ overflowWrap: 'anywhere' }}\n          >\n            {reference.label ?? `${reference.id} · unresolved`}\n          </Typography>\n        ))}\n      </Stack>\n    )\n  }\n  if (value.items) {\n    return (\n      // One line per item. The comma-joined form a cell shows cannot say\n      // whether `a, b` is one item or two; a list of lines can.\n      <Stack component=\"ul\" spacing={0.25} sx={{ m: 0, pl: 2.5 }}>\n        {value.items.map((item, index) => (\n          <Typography\n            // Items are values, not identities — two equal strings are two\n            // real entries and the position is what tells them apart.\n            key={`${index}:${item}`}\n            component=\"li\"\n            variant=\"body2\"\n            sx={{ overflowWrap: 'anywhere' }}\n          >\n            {item}\n          </Typography>\n        ))}\n      </Stack>\n    )\n  }\n  return (\n    <Typography\n      variant=\"body2\"\n      // Pretty-printed JSON is only readable where its whitespace survives.\n      {...(value.block\n        ? {\n            component: 'pre',\n            sx: {\n              m: 0,\n              fontFamily: 'monospace',\n              whiteSpace: 'pre-wrap',\n              overflowWrap: 'anywhere',\n            },\n          }\n        : { sx: { overflowWrap: 'anywhere' } })}\n    >\n      {value.text}\n    </Typography>\n  )\n}\nRecordFieldValue.displayName = 'RecordFieldValue'\n\nexport function DatasetRecordDialog(props: DatasetRecordDialogProps) {\n  const { record, model, resolveReference, onClose, onEdit } = props\n  const values = record?.values\n  const fields = useMemo(\n    () => (record ? datasetRecordFields(model, values, resolveReference) : []),\n    [record, model, values, resolveReference],\n  )\n  /**\n   * The same mismatch report the editor shows, for the same reason: a type\n   * change never rewrites stored documents, so a record may legitimately hold\n   * a value its model no longer describes. Reporting it beside the value is\n   * what stops the view claiming the record conforms.\n   */\n  const mismatches = useMemo(\n    () => (record ? validateDocument(model, values ?? {}) : {}),\n    [record, model, values],\n  )\n\n  return (\n    <Dialog\n      open={Boolean(record)}\n      onClose={onClose}\n      maxWidth=\"sm\"\n      fullWidth\n      aria-labelledby=\"dataset-record-view-title\"\n    >\n      <DialogTitle id=\"dataset-record-view-title\">{'View record'}</DialogTitle>\n      <DialogContent dividers>\n        <Stack spacing={2}>\n          <Box>\n            <Typography\n              variant=\"overline\"\n              color=\"text.secondary\"\n              component=\"div\"\n            >\n              {'Record ID'}\n            </Typography>\n            <Typography variant=\"body2\" sx={{ overflowWrap: 'anywhere' }}>\n              {record?.$id ?? ''}\n            </Typography>\n          </Box>\n          {fields.map((field) => (\n            <Box key={field.fieldId}>\n              <Stack\n                direction=\"row\"\n                spacing={1}\n                sx={{ alignItems: 'baseline', flexWrap: 'wrap' }}\n              >\n                <Typography\n                  variant=\"overline\"\n                  color=\"text.secondary\"\n                  component=\"div\"\n                >\n                  {field.label}\n                </Typography>\n                {field.source === 'extra' ? (\n                  <Tooltip\n                    title={\n                      'Stored on this record with no field in the ' +\n                      'collection’s schema. The table cannot show it, and ' +\n                      'the next save through the editor removes it.'\n                    }\n                  >\n                    <Typography\n                      variant=\"caption\"\n                      color=\"warning.main\"\n                      sx={{ cursor: 'help' }}\n                    >\n                      {'not in schema'}\n                    </Typography>\n                  </Tooltip>\n                ) : null}\n              </Stack>\n              <RecordFieldValue field={field} />\n              {mismatches[field.fieldId] ? (\n                <Typography variant=\"caption\" color=\"warning.main\">\n                  {mismatches[field.fieldId]}\n                </Typography>\n              ) : null}\n              {field.description ? (\n                <Typography\n                  variant=\"caption\"\n                  color=\"text.secondary\"\n                  component=\"div\"\n                >\n                  {field.description}\n                </Typography>\n              ) : null}\n            </Box>\n          ))}\n          {fields.length === 0 ? (\n            <Alert severity=\"info\">{'This record holds no fields.'}</Alert>\n          ) : null}\n        </Stack>\n      </DialogContent>\n      <DialogActions>\n        {onEdit ? <Button onClick={onEdit}>{'Edit record'}</Button> : null}\n        <Button variant=\"contained\" color=\"primary\" onClick={onClose}>\n          {'Close'}\n        </Button>\n      </DialogActions>\n    </Dialog>\n  )\n}\nDatasetRecordDialog.displayName = 'DatasetRecordDialog'\n\nexport default DatasetRecordDialog\n"],"names":["datasetRecordFields","Alert","Box","Button","Dialog","DialogActions","DialogContent","DialogTitle","Stack","Tooltip","Typography","useMemo","validateDocument","PLACEHOLDER","absent","null","RecordFieldValue","props","value","field","kind","variant","color","sx","fontStyle","text","references","direction","spacing","flexWrap","rowGap","map","reference","label","overflowWrap","id","items","component","m","pl","item","index","block","fontFamily","whiteSpace","displayName","DatasetRecordDialog","record","model","resolveReference","onClose","onEdit","values","fields","mismatches","open","Boolean","maxWidth","fullWidth","aria-labelledby","dividers","$id","alignItems","source","title","cursor","fieldId","description","length","severity","onClick"],"mappings":"AAAA;;;;;;;;;;;;;;;CAeC,GACD;;;AAEA,SAGEA,mBAAmB,QACd,kCAA8B;AACrC,SACEC,KAAK,EACLC,GAAG,EACHC,MAAM,EACNC,MAAM,EACNC,aAAa,EACbC,aAAa,EACbC,WAAW,EACXC,KAAK,EACLC,OAAO,EACPC,UAAU,QACL,gBAAe;AACtB,SAASC,OAAO,QAAQ,QAAO;AAC/B,SAA4BC,gBAAgB,QAAQ,6BAAyB;AA6C7E,oDAAoD,GACpD,MAAMC,cAAsC;IAC1CC,QAAQ;IACRC,MAAM;IACN,cAAc;IACd,cAAc;IACd,aAAa;AACf;AAEA;;;;;;;;CAQC,GACD,SAASC,iBAAiBC,KAAoC;IAC5D,MAAM,EAAEC,KAAK,EAAE,GAAGD,MAAME,KAAK;IAC7B,IAAID,MAAME,IAAI,KAAK,UAAU;QAC3B,qBACE,KAACV;YACCW,SAAQ;YACRC,OAAM;YACNC,IAAI;gBAAEC,WAAW;YAAS;sBAEzBN,MAAMO,IAAI;;IAGjB;IACA,IAAIP,MAAME,IAAI,KAAK,SAAS;YAOrBP;QANL,qBACE,KAACH;YACCW,SAAQ;YACRC,OAAM;YACNC,IAAI;gBAAEC,WAAW;YAAS;uBAEzBX,0BAAAA,WAAW,CAACK,MAAME,IAAI,CAAC,YAAvBP,0BAA2B;;IAGlC;IACA,IAAIK,MAAMQ,UAAU,EAAE;QACpB,qBACE,KAAClB;YAAMmB,WAAU;YAAMC,SAAS;YAAGL,IAAI;gBAAEM,UAAU;gBAAQC,QAAQ;YAAI;sBACpEZ,MAAMQ,UAAU,CAACK,GAAG,CAAC,CAACC;oBAUlBA;qCATH,KAACtB;oBAECW,SAAQ;oBACR,iEAAiE;oBACjE,kEAAkE;oBAClE,uDAAuD;oBACvDC,OAAOU,UAAUC,KAAK,GAAG,iBAAiB;oBAC1CV,IAAI;wBAAEW,cAAc;oBAAW;+BAE9BF,mBAAAA,UAAUC,KAAK,YAAfD,mBAAmB,GAAGA,UAAUG,EAAE,CAAC,aAAa,CAAC;mBAR7CH,UAAUG,EAAE;;;IAa3B;IACA,IAAIjB,MAAMkB,KAAK,EAAE;QACf,OACE,mEAAmE;QACnE,0DAA0D;sBAC1D,KAAC5B;YAAM6B,WAAU;YAAKT,SAAS;YAAML,IAAI;gBAAEe,GAAG;gBAAGC,IAAI;YAAI;sBACtDrB,MAAMkB,KAAK,CAACL,GAAG,CAAC,CAACS,MAAMC,sBACtB,KAAC/B;oBAIC2B,WAAU;oBACVhB,SAAQ;oBACRE,IAAI;wBAAEW,cAAc;oBAAW;8BAE9BM;mBALI,GAAGC,MAAM,CAAC,EAAED,MAAM;;IAUjC;IACA,qBACE,KAAC9B;QACCW,SAAQ;OAEHH,MAAMwB,KAAK,GACZ;QACEL,WAAW;QACXd,IAAI;YACFe,GAAG;YACHK,YAAY;YACZC,YAAY;YACZV,cAAc;QAChB;IACF,IACA;QAAEX,IAAI;YAAEW,cAAc;QAAW;IAAE;kBAEtChB,MAAMO,IAAI;;AAGjB;AACAT,iBAAiB6B,WAAW,GAAG;AAE/B,OAAO,SAASC,oBAAoB7B,KAA+B;;IACjE,MAAM,EAAE8B,MAAM,EAAEC,KAAK,EAAEC,gBAAgB,EAAEC,OAAO,EAAEC,MAAM,EAAE,GAAGlC;IAC7D,MAAMmC,SAASL,0BAAAA,OAAQK,MAAM;IAC7B,MAAMC,SAAS1C,QACb,IAAOoC,SAAS/C,oBAAoBgD,OAAOI,QAAQH,oBAAoB,EAAE,EACzE;QAACF;QAAQC;QAAOI;QAAQH;KAAiB;IAE3C;;;;;GAKC,GACD,MAAMK,aAAa3C,QACjB,IAAOoC,SAASnC,iBAAiBoC,OAAOI,iBAAAA,SAAU,CAAC,KAAK,CAAC,GACzD;QAACL;QAAQC;QAAOI;KAAO;IAGzB,qBACE,MAAChD;QACCmD,MAAMC,QAAQT;QACdG,SAASA;QACTO,UAAS;QACTC,SAAS;QACTC,mBAAgB;;0BAEhB,KAACpD;gBAAY4B,IAAG;0BAA6B;;0BAC7C,KAAC7B;gBAAcsD,QAAQ;0BACrB,cAAA,MAACpD;oBAAMoB,SAAS;;sCACd,MAAC1B;;8CACC,KAACQ;oCACCW,SAAQ;oCACRC,OAAM;oCACNe,WAAU;8CAET;;8CAEH,KAAC3B;oCAAWW,SAAQ;oCAAQE,IAAI;wCAAEW,cAAc;oCAAW;sDACxDa,0BAAAA,OAAQc,GAAG,mBAAI;;;;wBAGnBR,OAAOtB,GAAG,CAAC,CAACZ,sBACX,MAACjB;;kDACC,MAACM;wCACCmB,WAAU;wCACVC,SAAS;wCACTL,IAAI;4CAAEuC,YAAY;4CAAYjC,UAAU;wCAAO;;0DAE/C,KAACnB;gDACCW,SAAQ;gDACRC,OAAM;gDACNe,WAAU;0DAETlB,MAAMc,KAAK;;4CAEbd,MAAM4C,MAAM,KAAK,wBAChB,KAACtD;gDACCuD,OACE,gDACA,wDACA;0DAGF,cAAA,KAACtD;oDACCW,SAAQ;oDACRC,OAAM;oDACNC,IAAI;wDAAE0C,QAAQ;oDAAO;8DAEpB;;iDAGH;;;kDAEN,KAACjD;wCAAiBG,OAAOA;;oCACxBmC,UAAU,CAACnC,MAAM+C,OAAO,CAAC,iBACxB,KAACxD;wCAAWW,SAAQ;wCAAUC,OAAM;kDACjCgC,UAAU,CAACnC,MAAM+C,OAAO,CAAC;yCAE1B;oCACH/C,MAAMgD,WAAW,iBAChB,KAACzD;wCACCW,SAAQ;wCACRC,OAAM;wCACNe,WAAU;kDAETlB,MAAMgD,WAAW;yCAElB;;+BA7CIhD,MAAM+C,OAAO;wBAgDxBb,OAAOe,MAAM,KAAK,kBACjB,KAACnE;4BAAMoE,UAAS;sCAAQ;6BACtB;;;;0BAGR,MAAChE;;oBACE8C,uBAAS,KAAChD;wBAAOmE,SAASnB;kCAAS;yBAA0B;kCAC9D,KAAChD;wBAAOkB,SAAQ;wBAAYC,OAAM;wBAAUgD,SAASpB;kCAClD;;;;;;AAKX;AACAJ,oBAAoBD,WAAW,GAAG;AAElC,eAAeC,oBAAmB"}

package/src/lib/components/dataset-record-filter.d.ts

@@ -14,10 +14,10 @@
  * See the License for the specific language governing permissions and
  * limitations under the License.
  */
-import { type DatasetModel } from '@aglyn/aglyn';
 import type { ListFilterField } from '@aglyn/shared-ui-jsx/const/list-filter';
 import type { ListFilterClause, ListFilterOption } from '@aglyn/shared-ui-jsx/const/list-grid-filter';
 import { type ListQueryDeclaration, type ListQueryNormalizers, type ListQueryPlan } from '@aglyn/shared-ui-jsx/const/list-query-plan';
+import { type DatasetModel } from '../model/dataset-models';
 /**
  * The grid column a dataset field is shown in. Prefixed, because field ids
  * are the dataset author's and may be any name, `actions` included.

package/src/lib/components/dataset-record-filter.js

@@ -1,4 +1,3 @@
-import { _ as _extends } from "@swc/helpers/_/_extends";
 /**
  * @license
  * Copyright 2026 Aglyn LLC
@@ -14,8 +13,9 @@ import { _ as _extends } from "@swc/helpers/_/_extends";
  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  * See the License for the specific language governing permissions and
  * limitations under the License.
- */ import { DATASET_FILTER_PREFIX_MAX, datasetFilterTextKey, datasetFilterToken, datasetFilterValuePath, datasetFilterWords, datasetSearchToken } from "@aglyn/aglyn";
+ */ import { _ as _extends } from "@swc/helpers/_/_extends";
 import { LIST_QUERY_ID_PATH, planListQuery } from "@aglyn/shared-ui-jsx/const/list-query-plan";
+import { DATASET_FILTER_PREFIX_MAX, datasetFilterTextKey, datasetFilterToken, datasetFilterValuePath, datasetFilterWords, datasetSearchToken } from "../model/dataset-models.js";
 /*
  * THE RECORDS TABLE IS ONE QUERY (AGL-3321).
  *

package/src/lib/components/dataset-record-filter.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../../../../../../../libs/plugins/data/src/lib/components/dataset-record-filter.ts"],"sourcesContent":["/**\n * @license\n * Copyright 2026 Aglyn LLC\n *\n * Licensed under the Apache License, Version 2.0 (the \"License\");\n * you may not use this file except in compliance with the License.\n * You may obtain a copy of the License at\n *\n *   http://www.apache.org/licenses/LICENSE-2.0\n *\n * Unless required by applicable law or agreed to in writing, software\n * distributed under the License is distributed on an \"AS IS\" BASIS,\n * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n * See the License for the specific language governing permissions and\n * limitations under the License.\n */\n\nimport {\n  DATASET_FILTER_PREFIX_MAX,\n  type DatasetModel,\n  datasetFilterTextKey,\n  datasetFilterToken,\n  datasetFilterValuePath,\n  datasetFilterWords,\n  datasetSearchToken,\n} from '@aglyn/aglyn'\nimport type { ListFilterField, ListFilterRequest } from '@aglyn/shared-ui-jsx/const/list-filter'\nimport type {\n  ListFilterClause,\n  ListFilterOption,\n} from '@aglyn/shared-ui-jsx/const/list-grid-filter'\nimport {\n  LIST_QUERY_ID_PATH,\n  type ListQueryDeclaration,\n  type ListQueryNormalizers,\n  type ListQueryPlan,\n  planListQuery,\n} from '@aglyn/shared-ui-jsx/const/list-query-plan'\n\n/*\n * THE RECORDS TABLE IS ONE QUERY (AGL-3321).\n *\n * Every clause in the Filters panel and the quick-search word are\n * predicates on the records query, planned by the shared `planListQuery`\n * and answered by Firestore; nothing is matched over rows already loaded.\n *\n * A record's fields live under `values`, which is exempt from indexing (see\n * `datasetFilterKeys`), so the query reads the two fields every writer\n * stamps beside it:\n *\n *   `filterValues.<fieldId> ==`       an equality — a picked option, a\n *                                     boolean, a number, plain text (by its\n *                                     lower-cased key);\n *   `filterKeys array-contains`       ONE word-level clause — a text\n *                                     `contains` (`f:<id>^<prefix>`), a list\n *                                     member (`f:<id>=<member>`), or the\n *                                     quick-search word (`s:<prefix>`).\n *\n * Firestore's automatic single-field indexes compose any number of those\n * equalities and one `array-contains` under `orderBy(documentId())` by index\n * merging, so the table needs NO composite index — which it could not have:\n * every dataset's fields are its author's, and a composite per field per\n * dataset is not something an index file can list.\n *\n * What one such query cannot hold is not offered, or refused by name:\n *\n *   - no ranges. `>` on a number or a date on a timestamp is one inequality,\n *     which must then lead the order, and beside any equality it needs a\n *     composite per dataset. Numbers offer `=`, timestamps nothing.\n *   - no negations and no empty checks: Firestore has neither for a field a\n *     writer omits.\n *   - one array clause: a `contains` beside the search, or a second\n *     `contains`, is refused through `ListQueryNotices`, never half-applied.\n *   - one word of a `contains` or a search: the first word is asked, and the\n *     table says so for the rest.\n */\n\n/** The array the word-level clauses and the quick search ask. */\nconst TOKENS_PATH = 'filterKeys'\n\n/**\n * The grid column a dataset field is shown in. Prefixed, because field ids\n * are the dataset author's and may be any name, `actions` included.\n */\nexport const recordColumn = (fieldId: string) => `values.${fieldId}`\nconst COLUMN_PREFIX = recordColumn('')\nconst fieldIdOf = (column: string) =>\n  column.startsWith(COLUMN_PREFIX) ? column.slice(COLUMN_PREFIX.length) : column\n\n/**\n * How the records' filter fields were normalized (`dataset-models.ts`): a\n * typed text value becomes its `filterValues` key, and a search word its\n * `s:` token — so the query asks for exactly what the writers stored.\n */\nexport const datasetRecordNormalizers: ListQueryNormalizers = {\n  key: datasetFilterTextKey,\n  token: (value) => datasetSearchToken(value) ?? '',\n  // No field offers `endsWith`, so nothing is ever reversed.\n  reversed: (value) => value,\n  maxPrefix: DATASET_FILTER_PREFIX_MAX,\n}\n\n/** What the records grid's Filters panel offers for one dataset model. */\nexport interface DatasetRecordFilter {\n  /** The query declaration: every field below, the id order, the search. */\n  declaration: ListQueryDeclaration\n  fields: ListFilterField[]\n  options: Record<string, ListFilterOption[]>\n  headers: Record<string, string>\n  selectFields: string[]\n}\n\nconst BOOLEAN_OPTIONS: ListFilterOption[] = [\n  { value: 'true', label: 'True' },\n  { value: 'false', label: 'False' },\n]\n\n/**\n * The model's fields as query fields, each shown in column\n * `values.<fieldId>`:\n *\n *   enum text     picked: is, is any of          `filterValues.<id>`\n *   plain text    contains (one word), equals,   `filterKeys`,\n *                 is any of (ignoring case)      `filterValues.<id>`\n *   bool          picked: is                     `filterValues.<id>`\n *   numbers       =                              `filterValues.<id>`\n *   list          contains (a whole member)      `filterKeys`\n *\n * Timestamps, references, maps, bytes, coordinates and nil fields offer no\n * filter. A field whose id no query path can name (a v1 column with a `.`\n * in it — `datasetFilterValuePath`) offers only its `contains`.\n */\nexport function datasetRecordFilter(model: DatasetModel): DatasetRecordFilter {\n  const fields: ListFilterField[] = []\n  const options: Record<string, ListFilterOption[]> = {}\n  const headers: Record<string, string> = {}\n  const selectFields: string[] = []\n  for (const fieldId of model.order ?? []) {\n    const field = model.fields?.[fieldId]\n    if (!field) continue\n    const column = recordColumn(fieldId)\n    const valuePath = datasetFilterValuePath(fieldId)\n    const choices = field.type === 'text' ? (field.validation?.options ?? []) : []\n    let declared: ListFilterField | null = null\n    if (field.type === 'text' && choices.length) {\n      if (valuePath) {\n        declared = { column, kind: 'exact', path: valuePath, operators: ['equals', 'isAnyOf'] }\n        options[column] = choices.map((choice) => ({ value: choice, label: choice }))\n      }\n    } else if (field.type === 'text') {\n      // A `contains` asks the token the clause's value is turned into\n      // (`recordQueryClause`), matched as given.\n      declared = {\n        column,\n        kind: 'text',\n        path: valuePath ?? column,\n        tokensPath: TOKENS_PATH,\n        verbatimTokens: true,\n        ...(valuePath ? { lowerPath: valuePath } : {}),\n        operators: valuePath ? ['contains', 'equals', 'isAnyOf'] : ['contains'],\n      }\n    } else if (field.type === 'bool') {\n      if (valuePath) {\n        declared = { column, kind: 'boolean', path: valuePath, operators: ['equals'] }\n        options[column] = BOOLEAN_OPTIONS\n      }\n    } else if (field.type === 'int32' || field.type === 'int64' || field.type === 'float') {\n      if (valuePath) declared = { column, kind: 'number', path: valuePath, operators: ['='] }\n    } else if (field.type === 'sorted') {\n      declared = {\n        column,\n        kind: 'text',\n        path: column,\n        tokensPath: TOKENS_PATH,\n        verbatimTokens: true,\n        operators: ['contains'],\n      }\n    }\n    if (!declared) continue\n    fields.push(declared)\n    headers[column] = field.name || fieldId\n    if (options[column]) selectFields.push(column)\n  }\n  return {\n    declaration: {\n      fields,\n      // The document name: the one order no record can be missing (see the\n      // card), and the one every automatic index already ends in.\n      sorts: [{ path: LIST_QUERY_ID_PATH, direction: 'asc' }],\n      search: { tokensPath: TOKENS_PATH },\n    },\n    fields,\n    options,\n    headers,\n    selectFields,\n  }\n}\n\n/**\n * A clause as the query asks it: a `contains` becomes the `filterKeys`\n * token of its first word (text) or of its member (list). Everything else\n * goes as typed — the planner applies the text key, the number, the boolean.\n */\nfunction recordQueryClause(model: DatasetModel, clause: ListFilterClause): ListFilterRequest {\n  if (clause.op !== 'contains') return { field: clause.field, op: clause.op, value: clause.value }\n  const token = datasetFilterToken(model, {\n    field: fieldIdOf(clause.field),\n    op: 'contains',\n    value: clause.value,\n  })\n  return { field: clause.field, op: clause.op, value: token ?? '' }\n}\n\nconst clipWord = (word: string) =>\n  Array.from(word).slice(0, DATASET_FILTER_PREFIX_MAX).join('')\n\n/** The records query, and what it says to the reader. */\nexport interface DatasetRecordPlan {\n  plan: ListQueryPlan\n  /** What was not put on the query, as the reader asked it, and why. */\n  refused: Array<{ clause: ListFilterClause | 'search'; reason: string }>\n  /** What WAS served, where it differs from what was typed. */\n  notices: string[]\n}\n\n/**\n * The clauses and search words as ONE records query (`planListQuery`), with\n * the refusals mapped back to the clauses as the reader typed them, and the\n * notices said in the reader's words rather than as stored tokens.\n */\nexport function planRecordQuery(\n  model: DatasetModel,\n  filter: DatasetRecordFilter,\n  clauses: readonly ListFilterClause[],\n  searchWords: readonly string[],\n): DatasetRecordPlan {\n  const asked = new Map<ListFilterRequest, ListFilterClause>()\n  const queryClauses = clauses.map((clause) => {\n    const queryClause = recordQueryClause(model, clause)\n    asked.set(queryClause, clause)\n    return queryClause\n  })\n  const plan = planListQuery(\n    filter.declaration,\n    { clauses: queryClauses, search: searchWords },\n    datasetRecordNormalizers,\n  )\n  const refused = plan.refused.map((entry) => ({\n    clause: entry.clause === 'search' ? ('search' as const) : (asked.get(entry.clause) ?? entry.clause),\n    reason: entry.reason,\n  }))\n\n  const notices: string[] = []\n  const oneWord = (label: string, text: string) => {\n    const words = datasetFilterWords(text)\n    if (words.length > 1) {\n      notices.push(\n        `${label} matches one word at a time: showing records with a word starting \"${clipWord(words[0])}\".`,\n      )\n    }\n    if (words.some((word) => Array.from(word).length > DATASET_FILTER_PREFIX_MAX)) {\n      notices.push(`${label} reads the first ${DATASET_FILTER_PREFIX_MAX} letters of a word.`)\n    }\n  }\n  if (plan.searched) oneWord('Search', searchWords.join(' '))\n  for (const served of plan.served) {\n    const clause = asked.get(served)\n    // A list's `contains` is a whole member; only text is asked by word.\n    if (clause?.op !== 'contains') continue\n    if (model.fields?.[fieldIdOf(clause.field)]?.type !== 'text') continue\n    oneWord(`${filter.headers[clause.field] ?? clause.field} contains`, clause.value)\n  }\n  return { plan, refused, notices }\n}\n"],"names":["DATASET_FILTER_PREFIX_MAX","datasetFilterTextKey","datasetFilterToken","datasetFilterValuePath","datasetFilterWords","datasetSearchToken","LIST_QUERY_ID_PATH","planListQuery","TOKENS_PATH","recordColumn","fieldId","COLUMN_PREFIX","fieldIdOf","column","startsWith","slice","length","datasetRecordNormalizers","key","token","value","reversed","maxPrefix","BOOLEAN_OPTIONS","label","datasetRecordFilter","model","fields","options","headers","selectFields","order","field","valuePath","choices","type","validation","declared","kind","path","operators","map","choice","tokensPath","verbatimTokens","lowerPath","push","name","declaration","sorts","direction","search","recordQueryClause","clause","op","clipWord","word","Array","from","join","planRecordQuery","filter","clauses","searchWords","asked","Map","queryClauses","queryClause","set","plan","refused","entry","get","reason","notices","oneWord","text","words","some","searched","served"],"mappings":";AAAA;;;;;;;;;;;;;;;CAeC,GAED,SACEA,yBAAyB,EAEzBC,oBAAoB,EACpBC,kBAAkB,EAClBC,sBAAsB,EACtBC,kBAAkB,EAClBC,kBAAkB,QACb,eAAc;AAMrB,SACEC,kBAAkB,EAIlBC,aAAa,QACR,6CAA4C;AAEnD;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;CAoCC,GAED,+DAA+D,GAC/D,MAAMC,cAAc;AAEpB;;;CAGC,GACD,OAAO,MAAMC,eAAe,CAACC,UAAoB,CAAC,OAAO,EAAEA,SAAS,CAAA;AACpE,MAAMC,gBAAgBF,aAAa;AACnC,MAAMG,YAAY,CAACC,SACjBA,OAAOC,UAAU,CAACH,iBAAiBE,OAAOE,KAAK,CAACJ,cAAcK,MAAM,IAAIH;AAE1E;;;;CAIC,GACD,OAAO,MAAMI,2BAAiD;IAC5DC,KAAKjB;IACLkB,OAAO,CAACC;YAAUf;gBAAAA,sBAAAA,mBAAmBe,kBAAnBf,sBAA6B;;IAC/C,2DAA2D;IAC3DgB,UAAU,CAACD,QAAUA;IACrBE,WAAWtB;AACb,EAAC;AAYD,MAAMuB,kBAAsC;IAC1C;QAAEH,OAAO;QAAQI,OAAO;IAAO;IAC/B;QAAEJ,OAAO;QAASI,OAAO;IAAQ;CAClC;AAED;;;;;;;;;;;;;;CAcC,GACD,OAAO,SAASC,oBAAoBC,KAAmB;QAK/BA;IAJtB,MAAMC,SAA4B,EAAE;IACpC,MAAMC,UAA8C,CAAC;IACrD,MAAMC,UAAkC,CAAC;IACzC,MAAMC,eAAyB,EAAE;IACjC,KAAK,MAAMpB,YAAWgB,eAAAA,MAAMK,KAAK,YAAXL,eAAe,EAAE,CAAE;;YACzBA,eAI2BM;QAJzC,MAAMA,SAAQN,gBAAAA,MAAMC,MAAM,qBAAZD,aAAc,CAAChB,QAAQ;QACrC,IAAI,CAACsB,OAAO;QACZ,MAAMnB,SAASJ,aAAaC;QAC5B,MAAMuB,YAAY9B,uBAAuBO;QACzC,MAAMwB,UAAUF,MAAMG,IAAI,KAAK,kBAAUH,oBAAAA,MAAMI,UAAU,qBAAhBJ,kBAAkBJ,OAAO,mBAAI,EAAE,GAAI,EAAE;QAC9E,IAAIS,WAAmC;QACvC,IAAIL,MAAMG,IAAI,KAAK,UAAUD,QAAQlB,MAAM,EAAE;YAC3C,IAAIiB,WAAW;gBACbI,WAAW;oBAAExB;oBAAQyB,MAAM;oBAASC,MAAMN;oBAAWO,WAAW;wBAAC;wBAAU;qBAAU;gBAAC;gBACtFZ,OAAO,CAACf,OAAO,GAAGqB,QAAQO,GAAG,CAAC,CAACC,SAAY,CAAA;wBAAEtB,OAAOsB;wBAAQlB,OAAOkB;oBAAO,CAAA;YAC5E;QACF,OAAO,IAAIV,MAAMG,IAAI,KAAK,QAAQ;YAChC,gEAAgE;YAChE,2CAA2C;YAC3CE,WAAW;gBACTxB;gBACAyB,MAAM;gBACNC,IAAI,EAAEN,oBAAAA,YAAapB;gBACnB8B,YAAYnC;gBACZoC,gBAAgB;eACZX,YAAY;gBAAEY,WAAWZ;YAAU,IAAI,CAAC;gBAC5CO,WAAWP,YAAY;oBAAC;oBAAY;oBAAU;iBAAU,GAAG;oBAAC;iBAAW;;QAE3E,OAAO,IAAID,MAAMG,IAAI,KAAK,QAAQ;YAChC,IAAIF,WAAW;gBACbI,WAAW;oBAAExB;oBAAQyB,MAAM;oBAAWC,MAAMN;oBAAWO,WAAW;wBAAC;qBAAS;gBAAC;gBAC7EZ,OAAO,CAACf,OAAO,GAAGU;YACpB;QACF,OAAO,IAAIS,MAAMG,IAAI,KAAK,WAAWH,MAAMG,IAAI,KAAK,WAAWH,MAAMG,IAAI,KAAK,SAAS;YACrF,IAAIF,WAAWI,WAAW;gBAAExB;gBAAQyB,MAAM;gBAAUC,MAAMN;gBAAWO,WAAW;oBAAC;iBAAI;YAAC;QACxF,OAAO,IAAIR,MAAMG,IAAI,KAAK,UAAU;YAClCE,WAAW;gBACTxB;gBACAyB,MAAM;gBACNC,MAAM1B;gBACN8B,YAAYnC;gBACZoC,gBAAgB;gBAChBJ,WAAW;oBAAC;iBAAW;YACzB;QACF;QACA,IAAI,CAACH,UAAU;QACfV,OAAOmB,IAAI,CAACT;QACZR,OAAO,CAAChB,OAAO,GAAGmB,MAAMe,IAAI,IAAIrC;QAChC,IAAIkB,OAAO,CAACf,OAAO,EAAEiB,aAAagB,IAAI,CAACjC;IACzC;IACA,OAAO;QACLmC,aAAa;YACXrB;YACA,qEAAqE;YACrE,4DAA4D;YAC5DsB,OAAO;gBAAC;oBAAEV,MAAMjC;oBAAoB4C,WAAW;gBAAM;aAAE;YACvDC,QAAQ;gBAAER,YAAYnC;YAAY;QACpC;QACAmB;QACAC;QACAC;QACAC;IACF;AACF;AAEA;;;;CAIC,GACD,SAASsB,kBAAkB1B,KAAmB,EAAE2B,MAAwB;IACtE,IAAIA,OAAOC,EAAE,KAAK,YAAY,OAAO;QAAEtB,OAAOqB,OAAOrB,KAAK;QAAEsB,IAAID,OAAOC,EAAE;QAAElC,OAAOiC,OAAOjC,KAAK;IAAC;IAC/F,MAAMD,QAAQjB,mBAAmBwB,OAAO;QACtCM,OAAOpB,UAAUyC,OAAOrB,KAAK;QAC7BsB,IAAI;QACJlC,OAAOiC,OAAOjC,KAAK;IACrB;IACA,OAAO;QAAEY,OAAOqB,OAAOrB,KAAK;QAAEsB,IAAID,OAAOC,EAAE;QAAElC,KAAK,EAAED,gBAAAA,QAAS;IAAG;AAClE;AAEA,MAAMoC,WAAW,CAACC,OAChBC,MAAMC,IAAI,CAACF,MAAMzC,KAAK,CAAC,GAAGf,2BAA2B2D,IAAI,CAAC;AAW5D;;;;CAIC,GACD,OAAO,SAASC,gBACdlC,KAAmB,EACnBmC,MAA2B,EAC3BC,OAAoC,EACpCC,WAA8B;IAE9B,MAAMC,QAAQ,IAAIC;IAClB,MAAMC,eAAeJ,QAAQrB,GAAG,CAAC,CAACY;QAChC,MAAMc,cAAcf,kBAAkB1B,OAAO2B;QAC7CW,MAAMI,GAAG,CAACD,aAAad;QACvB,OAAOc;IACT;IACA,MAAME,OAAO9D,cACXsD,OAAOb,WAAW,EAClB;QAAEc,SAASI;QAAcf,QAAQY;IAAY,GAC7C9C;IAEF,MAAMqD,UAAUD,KAAKC,OAAO,CAAC7B,GAAG,CAAC,CAAC8B;YAC2BP;eADhB;YAC3CX,QAAQkB,MAAMlB,MAAM,KAAK,WAAY,YAAsBW,aAAAA,MAAMQ,GAAG,CAACD,MAAMlB,MAAM,aAAtBW,aAA2BO,MAAMlB,MAAM;YAClGoB,QAAQF,MAAME,MAAM;QACtB;;IAEA,MAAMC,UAAoB,EAAE;IAC5B,MAAMC,UAAU,CAACnD,OAAeoD;QAC9B,MAAMC,QAAQzE,mBAAmBwE;QACjC,IAAIC,MAAM7D,MAAM,GAAG,GAAG;YACpB0D,QAAQ5B,IAAI,CACV,GAAGtB,MAAM,mEAAmE,EAAE+B,SAASsB,KAAK,CAAC,EAAE,EAAE,EAAE,CAAC;QAExG;QACA,IAAIA,MAAMC,IAAI,CAAC,CAACtB,OAASC,MAAMC,IAAI,CAACF,MAAMxC,MAAM,GAAGhB,4BAA4B;YAC7E0E,QAAQ5B,IAAI,CAAC,GAAGtB,MAAM,iBAAiB,EAAExB,0BAA0B,mBAAmB,CAAC;QACzF;IACF;IACA,IAAIqE,KAAKU,QAAQ,EAAEJ,QAAQ,UAAUZ,YAAYJ,IAAI,CAAC;IACtD,KAAK,MAAMqB,UAAUX,KAAKW,MAAM,CAAE;YAKrBnB;YADPnC,yBAAAA;QAHJ,MAAM2B,SAASW,MAAMQ,GAAG,CAACQ;QACzB,qEAAqE;QACrE,IAAI3B,CAAAA,0BAAAA,OAAQC,EAAE,MAAK,YAAY;QAC/B,IAAI5B,EAAAA,gBAAAA,MAAMC,MAAM,sBAAZD,0BAAAA,aAAc,CAACd,UAAUyC,OAAOrB,KAAK,EAAE,qBAAvCN,wBAAyCS,IAAI,MAAK,QAAQ;QAC9DwC,QAAQ,IAAGd,+BAAAA,OAAOhC,OAAO,CAACwB,OAAOrB,KAAK,CAAC,YAA5B6B,+BAAgCR,OAAOrB,KAAK,CAAC,SAAS,CAAC,EAAEqB,OAAOjC,KAAK;IAClF;IACA,OAAO;QAAEiD;QAAMC;QAASI;IAAQ;AAClC"}
+{"version":3,"sources":["../../../../../../../libs/plugins/data/src/lib/components/dataset-record-filter.ts"],"sourcesContent":["/**\n * @license\n * Copyright 2026 Aglyn LLC\n *\n * Licensed under the Apache License, Version 2.0 (the \"License\");\n * you may not use this file except in compliance with the License.\n * You may obtain a copy of the License at\n *\n *   http://www.apache.org/licenses/LICENSE-2.0\n *\n * Unless required by applicable law or agreed to in writing, software\n * distributed under the License is distributed on an \"AS IS\" BASIS,\n * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n * See the License for the specific language governing permissions and\n * limitations under the License.\n */\n\nimport type { ListFilterField, ListFilterRequest } from '@aglyn/shared-ui-jsx/const/list-filter'\nimport type {\n  ListFilterClause,\n  ListFilterOption,\n} from '@aglyn/shared-ui-jsx/const/list-grid-filter'\nimport {\n  LIST_QUERY_ID_PATH,\n  type ListQueryDeclaration,\n  type ListQueryNormalizers,\n  type ListQueryPlan,\n  planListQuery,\n} from '@aglyn/shared-ui-jsx/const/list-query-plan'\nimport { DATASET_FILTER_PREFIX_MAX, type DatasetModel, datasetFilterTextKey, datasetFilterToken, datasetFilterValuePath, datasetFilterWords, datasetSearchToken } from '../model/dataset-models'\n\n/*\n * THE RECORDS TABLE IS ONE QUERY (AGL-3321).\n *\n * Every clause in the Filters panel and the quick-search word are\n * predicates on the records query, planned by the shared `planListQuery`\n * and answered by Firestore; nothing is matched over rows already loaded.\n *\n * A record's fields live under `values`, which is exempt from indexing (see\n * `datasetFilterKeys`), so the query reads the two fields every writer\n * stamps beside it:\n *\n *   `filterValues.<fieldId> ==`       an equality — a picked option, a\n *                                     boolean, a number, plain text (by its\n *                                     lower-cased key);\n *   `filterKeys array-contains`       ONE word-level clause — a text\n *                                     `contains` (`f:<id>^<prefix>`), a list\n *                                     member (`f:<id>=<member>`), or the\n *                                     quick-search word (`s:<prefix>`).\n *\n * Firestore's automatic single-field indexes compose any number of those\n * equalities and one `array-contains` under `orderBy(documentId())` by index\n * merging, so the table needs NO composite index — which it could not have:\n * every dataset's fields are its author's, and a composite per field per\n * dataset is not something an index file can list.\n *\n * What one such query cannot hold is not offered, or refused by name:\n *\n *   - no ranges. `>` on a number or a date on a timestamp is one inequality,\n *     which must then lead the order, and beside any equality it needs a\n *     composite per dataset. Numbers offer `=`, timestamps nothing.\n *   - no negations and no empty checks: Firestore has neither for a field a\n *     writer omits.\n *   - one array clause: a `contains` beside the search, or a second\n *     `contains`, is refused through `ListQueryNotices`, never half-applied.\n *   - one word of a `contains` or a search: the first word is asked, and the\n *     table says so for the rest.\n */\n\n/** The array the word-level clauses and the quick search ask. */\nconst TOKENS_PATH = 'filterKeys'\n\n/**\n * The grid column a dataset field is shown in. Prefixed, because field ids\n * are the dataset author's and may be any name, `actions` included.\n */\nexport const recordColumn = (fieldId: string) => `values.${fieldId}`\nconst COLUMN_PREFIX = recordColumn('')\nconst fieldIdOf = (column: string) =>\n  column.startsWith(COLUMN_PREFIX) ? column.slice(COLUMN_PREFIX.length) : column\n\n/**\n * How the records' filter fields were normalized (`dataset-models.ts`): a\n * typed text value becomes its `filterValues` key, and a search word its\n * `s:` token — so the query asks for exactly what the writers stored.\n */\nexport const datasetRecordNormalizers: ListQueryNormalizers = {\n  key: datasetFilterTextKey,\n  token: (value) => datasetSearchToken(value) ?? '',\n  // No field offers `endsWith`, so nothing is ever reversed.\n  reversed: (value) => value,\n  maxPrefix: DATASET_FILTER_PREFIX_MAX,\n}\n\n/** What the records grid's Filters panel offers for one dataset model. */\nexport interface DatasetRecordFilter {\n  /** The query declaration: every field below, the id order, the search. */\n  declaration: ListQueryDeclaration\n  fields: ListFilterField[]\n  options: Record<string, ListFilterOption[]>\n  headers: Record<string, string>\n  selectFields: string[]\n}\n\nconst BOOLEAN_OPTIONS: ListFilterOption[] = [\n  { value: 'true', label: 'True' },\n  { value: 'false', label: 'False' },\n]\n\n/**\n * The model's fields as query fields, each shown in column\n * `values.<fieldId>`:\n *\n *   enum text     picked: is, is any of          `filterValues.<id>`\n *   plain text    contains (one word), equals,   `filterKeys`,\n *                 is any of (ignoring case)      `filterValues.<id>`\n *   bool          picked: is                     `filterValues.<id>`\n *   numbers       =                              `filterValues.<id>`\n *   list          contains (a whole member)      `filterKeys`\n *\n * Timestamps, references, maps, bytes, coordinates and nil fields offer no\n * filter. A field whose id no query path can name (a v1 column with a `.`\n * in it — `datasetFilterValuePath`) offers only its `contains`.\n */\nexport function datasetRecordFilter(model: DatasetModel): DatasetRecordFilter {\n  const fields: ListFilterField[] = []\n  const options: Record<string, ListFilterOption[]> = {}\n  const headers: Record<string, string> = {}\n  const selectFields: string[] = []\n  for (const fieldId of model.order ?? []) {\n    const field = model.fields?.[fieldId]\n    if (!field) continue\n    const column = recordColumn(fieldId)\n    const valuePath = datasetFilterValuePath(fieldId)\n    const choices = field.type === 'text' ? (field.validation?.options ?? []) : []\n    let declared: ListFilterField | null = null\n    if (field.type === 'text' && choices.length) {\n      if (valuePath) {\n        declared = { column, kind: 'exact', path: valuePath, operators: ['equals', 'isAnyOf'] }\n        options[column] = choices.map((choice) => ({ value: choice, label: choice }))\n      }\n    } else if (field.type === 'text') {\n      // A `contains` asks the token the clause's value is turned into\n      // (`recordQueryClause`), matched as given.\n      declared = {\n        column,\n        kind: 'text',\n        path: valuePath ?? column,\n        tokensPath: TOKENS_PATH,\n        verbatimTokens: true,\n        ...(valuePath ? { lowerPath: valuePath } : {}),\n        operators: valuePath ? ['contains', 'equals', 'isAnyOf'] : ['contains'],\n      }\n    } else if (field.type === 'bool') {\n      if (valuePath) {\n        declared = { column, kind: 'boolean', path: valuePath, operators: ['equals'] }\n        options[column] = BOOLEAN_OPTIONS\n      }\n    } else if (field.type === 'int32' || field.type === 'int64' || field.type === 'float') {\n      if (valuePath) declared = { column, kind: 'number', path: valuePath, operators: ['='] }\n    } else if (field.type === 'sorted') {\n      declared = {\n        column,\n        kind: 'text',\n        path: column,\n        tokensPath: TOKENS_PATH,\n        verbatimTokens: true,\n        operators: ['contains'],\n      }\n    }\n    if (!declared) continue\n    fields.push(declared)\n    headers[column] = field.name || fieldId\n    if (options[column]) selectFields.push(column)\n  }\n  return {\n    declaration: {\n      fields,\n      // The document name: the one order no record can be missing (see the\n      // card), and the one every automatic index already ends in.\n      sorts: [{ path: LIST_QUERY_ID_PATH, direction: 'asc' }],\n      search: { tokensPath: TOKENS_PATH },\n    },\n    fields,\n    options,\n    headers,\n    selectFields,\n  }\n}\n\n/**\n * A clause as the query asks it: a `contains` becomes the `filterKeys`\n * token of its first word (text) or of its member (list). Everything else\n * goes as typed — the planner applies the text key, the number, the boolean.\n */\nfunction recordQueryClause(model: DatasetModel, clause: ListFilterClause): ListFilterRequest {\n  if (clause.op !== 'contains') return { field: clause.field, op: clause.op, value: clause.value }\n  const token = datasetFilterToken(model, {\n    field: fieldIdOf(clause.field),\n    op: 'contains',\n    value: clause.value,\n  })\n  return { field: clause.field, op: clause.op, value: token ?? '' }\n}\n\nconst clipWord = (word: string) =>\n  Array.from(word).slice(0, DATASET_FILTER_PREFIX_MAX).join('')\n\n/** The records query, and what it says to the reader. */\nexport interface DatasetRecordPlan {\n  plan: ListQueryPlan\n  /** What was not put on the query, as the reader asked it, and why. */\n  refused: Array<{ clause: ListFilterClause | 'search'; reason: string }>\n  /** What WAS served, where it differs from what was typed. */\n  notices: string[]\n}\n\n/**\n * The clauses and search words as ONE records query (`planListQuery`), with\n * the refusals mapped back to the clauses as the reader typed them, and the\n * notices said in the reader's words rather than as stored tokens.\n */\nexport function planRecordQuery(\n  model: DatasetModel,\n  filter: DatasetRecordFilter,\n  clauses: readonly ListFilterClause[],\n  searchWords: readonly string[],\n): DatasetRecordPlan {\n  const asked = new Map<ListFilterRequest, ListFilterClause>()\n  const queryClauses = clauses.map((clause) => {\n    const queryClause = recordQueryClause(model, clause)\n    asked.set(queryClause, clause)\n    return queryClause\n  })\n  const plan = planListQuery(\n    filter.declaration,\n    { clauses: queryClauses, search: searchWords },\n    datasetRecordNormalizers,\n  )\n  const refused = plan.refused.map((entry) => ({\n    clause: entry.clause === 'search' ? ('search' as const) : (asked.get(entry.clause) ?? entry.clause),\n    reason: entry.reason,\n  }))\n\n  const notices: string[] = []\n  const oneWord = (label: string, text: string) => {\n    const words = datasetFilterWords(text)\n    if (words.length > 1) {\n      notices.push(\n        `${label} matches one word at a time: showing records with a word starting \"${clipWord(words[0])}\".`,\n      )\n    }\n    if (words.some((word) => Array.from(word).length > DATASET_FILTER_PREFIX_MAX)) {\n      notices.push(`${label} reads the first ${DATASET_FILTER_PREFIX_MAX} letters of a word.`)\n    }\n  }\n  if (plan.searched) oneWord('Search', searchWords.join(' '))\n  for (const served of plan.served) {\n    const clause = asked.get(served)\n    // A list's `contains` is a whole member; only text is asked by word.\n    if (clause?.op !== 'contains') continue\n    if (model.fields?.[fieldIdOf(clause.field)]?.type !== 'text') continue\n    oneWord(`${filter.headers[clause.field] ?? clause.field} contains`, clause.value)\n  }\n  return { plan, refused, notices }\n}\n"],"names":["LIST_QUERY_ID_PATH","planListQuery","DATASET_FILTER_PREFIX_MAX","datasetFilterTextKey","datasetFilterToken","datasetFilterValuePath","datasetFilterWords","datasetSearchToken","TOKENS_PATH","recordColumn","fieldId","COLUMN_PREFIX","fieldIdOf","column","startsWith","slice","length","datasetRecordNormalizers","key","token","value","reversed","maxPrefix","BOOLEAN_OPTIONS","label","datasetRecordFilter","model","fields","options","headers","selectFields","order","field","valuePath","choices","type","validation","declared","kind","path","operators","map","choice","tokensPath","verbatimTokens","lowerPath","push","name","declaration","sorts","direction","search","recordQueryClause","clause","op","clipWord","word","Array","from","join","planRecordQuery","filter","clauses","searchWords","asked","Map","queryClauses","queryClause","set","plan","refused","entry","get","reason","notices","oneWord","text","words","some","searched","served"],"mappings":"AAAA;;;;;;;;;;;;;;;CAeC;AAOD,SACEA,kBAAkB,EAIlBC,aAAa,QACR,6CAA4C;AACnD,SAASC,yBAAyB,EAAqBC,oBAAoB,EAAEC,kBAAkB,EAAEC,sBAAsB,EAAEC,kBAAkB,EAAEC,kBAAkB,QAAQ,6BAAyB;AAEhM;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;CAoCC,GAED,+DAA+D,GAC/D,MAAMC,cAAc;AAEpB;;;CAGC,GACD,OAAO,MAAMC,eAAe,CAACC,UAAoB,CAAC,OAAO,EAAEA,SAAS,CAAA;AACpE,MAAMC,gBAAgBF,aAAa;AACnC,MAAMG,YAAY,CAACC,SACjBA,OAAOC,UAAU,CAACH,iBAAiBE,OAAOE,KAAK,CAACJ,cAAcK,MAAM,IAAIH;AAE1E;;;;CAIC,GACD,OAAO,MAAMI,2BAAiD;IAC5DC,KAAKf;IACLgB,OAAO,CAACC;YAAUb;gBAAAA,sBAAAA,mBAAmBa,kBAAnBb,sBAA6B;;IAC/C,2DAA2D;IAC3Dc,UAAU,CAACD,QAAUA;IACrBE,WAAWpB;AACb,EAAC;AAYD,MAAMqB,kBAAsC;IAC1C;QAAEH,OAAO;QAAQI,OAAO;IAAO;IAC/B;QAAEJ,OAAO;QAASI,OAAO;IAAQ;CAClC;AAED;;;;;;;;;;;;;;CAcC,GACD,OAAO,SAASC,oBAAoBC,KAAmB;QAK/BA;IAJtB,MAAMC,SAA4B,EAAE;IACpC,MAAMC,UAA8C,CAAC;IACrD,MAAMC,UAAkC,CAAC;IACzC,MAAMC,eAAyB,EAAE;IACjC,KAAK,MAAMpB,YAAWgB,eAAAA,MAAMK,KAAK,YAAXL,eAAe,EAAE,CAAE;;YACzBA,eAI2BM;QAJzC,MAAMA,SAAQN,gBAAAA,MAAMC,MAAM,qBAAZD,aAAc,CAAChB,QAAQ;QACrC,IAAI,CAACsB,OAAO;QACZ,MAAMnB,SAASJ,aAAaC;QAC5B,MAAMuB,YAAY5B,uBAAuBK;QACzC,MAAMwB,UAAUF,MAAMG,IAAI,KAAK,kBAAUH,oBAAAA,MAAMI,UAAU,qBAAhBJ,kBAAkBJ,OAAO,mBAAI,EAAE,GAAI,EAAE;QAC9E,IAAIS,WAAmC;QACvC,IAAIL,MAAMG,IAAI,KAAK,UAAUD,QAAQlB,MAAM,EAAE;YAC3C,IAAIiB,WAAW;gBACbI,WAAW;oBAAExB;oBAAQyB,MAAM;oBAASC,MAAMN;oBAAWO,WAAW;wBAAC;wBAAU;qBAAU;gBAAC;gBACtFZ,OAAO,CAACf,OAAO,GAAGqB,QAAQO,GAAG,CAAC,CAACC,SAAY,CAAA;wBAAEtB,OAAOsB;wBAAQlB,OAAOkB;oBAAO,CAAA;YAC5E;QACF,OAAO,IAAIV,MAAMG,IAAI,KAAK,QAAQ;YAChC,gEAAgE;YAChE,2CAA2C;YAC3CE,WAAW;gBACTxB;gBACAyB,MAAM;gBACNC,IAAI,EAAEN,oBAAAA,YAAapB;gBACnB8B,YAAYnC;gBACZoC,gBAAgB;eACZX,YAAY;gBAAEY,WAAWZ;YAAU,IAAI,CAAC;gBAC5CO,WAAWP,YAAY;oBAAC;oBAAY;oBAAU;iBAAU,GAAG;oBAAC;iBAAW;;QAE3E,OAAO,IAAID,MAAMG,IAAI,KAAK,QAAQ;YAChC,IAAIF,WAAW;gBACbI,WAAW;oBAAExB;oBAAQyB,MAAM;oBAAWC,MAAMN;oBAAWO,WAAW;wBAAC;qBAAS;gBAAC;gBAC7EZ,OAAO,CAACf,OAAO,GAAGU;YACpB;QACF,OAAO,IAAIS,MAAMG,IAAI,KAAK,WAAWH,MAAMG,IAAI,KAAK,WAAWH,MAAMG,IAAI,KAAK,SAAS;YACrF,IAAIF,WAAWI,WAAW;gBAAExB;gBAAQyB,MAAM;gBAAUC,MAAMN;gBAAWO,WAAW;oBAAC;iBAAI;YAAC;QACxF,OAAO,IAAIR,MAAMG,IAAI,KAAK,UAAU;YAClCE,WAAW;gBACTxB;gBACAyB,MAAM;gBACNC,MAAM1B;gBACN8B,YAAYnC;gBACZoC,gBAAgB;gBAChBJ,WAAW;oBAAC;iBAAW;YACzB;QACF;QACA,IAAI,CAACH,UAAU;QACfV,OAAOmB,IAAI,CAACT;QACZR,OAAO,CAAChB,OAAO,GAAGmB,MAAMe,IAAI,IAAIrC;QAChC,IAAIkB,OAAO,CAACf,OAAO,EAAEiB,aAAagB,IAAI,CAACjC;IACzC;IACA,OAAO;QACLmC,aAAa;YACXrB;YACA,qEAAqE;YACrE,4DAA4D;YAC5DsB,OAAO;gBAAC;oBAAEV,MAAMvC;oBAAoBkD,WAAW;gBAAM;aAAE;YACvDC,QAAQ;gBAAER,YAAYnC;YAAY;QACpC;QACAmB;QACAC;QACAC;QACAC;IACF;AACF;AAEA;;;;CAIC,GACD,SAASsB,kBAAkB1B,KAAmB,EAAE2B,MAAwB;IACtE,IAAIA,OAAOC,EAAE,KAAK,YAAY,OAAO;QAAEtB,OAAOqB,OAAOrB,KAAK;QAAEsB,IAAID,OAAOC,EAAE;QAAElC,OAAOiC,OAAOjC,KAAK;IAAC;IAC/F,MAAMD,QAAQf,mBAAmBsB,OAAO;QACtCM,OAAOpB,UAAUyC,OAAOrB,KAAK;QAC7BsB,IAAI;QACJlC,OAAOiC,OAAOjC,KAAK;IACrB;IACA,OAAO;QAAEY,OAAOqB,OAAOrB,KAAK;QAAEsB,IAAID,OAAOC,EAAE;QAAElC,KAAK,EAAED,gBAAAA,QAAS;IAAG;AAClE;AAEA,MAAMoC,WAAW,CAACC,OAChBC,MAAMC,IAAI,CAACF,MAAMzC,KAAK,CAAC,GAAGb,2BAA2ByD,IAAI,CAAC;AAW5D;;;;CAIC,GACD,OAAO,SAASC,gBACdlC,KAAmB,EACnBmC,MAA2B,EAC3BC,OAAoC,EACpCC,WAA8B;IAE9B,MAAMC,QAAQ,IAAIC;IAClB,MAAMC,eAAeJ,QAAQrB,GAAG,CAAC,CAACY;QAChC,MAAMc,cAAcf,kBAAkB1B,OAAO2B;QAC7CW,MAAMI,GAAG,CAACD,aAAad;QACvB,OAAOc;IACT;IACA,MAAME,OAAOpE,cACX4D,OAAOb,WAAW,EAClB;QAAEc,SAASI;QAAcf,QAAQY;IAAY,GAC7C9C;IAEF,MAAMqD,UAAUD,KAAKC,OAAO,CAAC7B,GAAG,CAAC,CAAC8B;YAC2BP;eADhB;YAC3CX,QAAQkB,MAAMlB,MAAM,KAAK,WAAY,YAAsBW,aAAAA,MAAMQ,GAAG,CAACD,MAAMlB,MAAM,aAAtBW,aAA2BO,MAAMlB,MAAM;YAClGoB,QAAQF,MAAME,MAAM;QACtB;;IAEA,MAAMC,UAAoB,EAAE;IAC5B,MAAMC,UAAU,CAACnD,OAAeoD;QAC9B,MAAMC,QAAQvE,mBAAmBsE;QACjC,IAAIC,MAAM7D,MAAM,GAAG,GAAG;YACpB0D,QAAQ5B,IAAI,CACV,GAAGtB,MAAM,mEAAmE,EAAE+B,SAASsB,KAAK,CAAC,EAAE,EAAE,EAAE,CAAC;QAExG;QACA,IAAIA,MAAMC,IAAI,CAAC,CAACtB,OAASC,MAAMC,IAAI,CAACF,MAAMxC,MAAM,GAAGd,4BAA4B;YAC7EwE,QAAQ5B,IAAI,CAAC,GAAGtB,MAAM,iBAAiB,EAAEtB,0BAA0B,mBAAmB,CAAC;QACzF;IACF;IACA,IAAImE,KAAKU,QAAQ,EAAEJ,QAAQ,UAAUZ,YAAYJ,IAAI,CAAC;IACtD,KAAK,MAAMqB,UAAUX,KAAKW,MAAM,CAAE;YAKrBnB;YADPnC,yBAAAA;QAHJ,MAAM2B,SAASW,MAAMQ,GAAG,CAACQ;QACzB,qEAAqE;QACrE,IAAI3B,CAAAA,0BAAAA,OAAQC,EAAE,MAAK,YAAY;QAC/B,IAAI5B,EAAAA,gBAAAA,MAAMC,MAAM,sBAAZD,0BAAAA,aAAc,CAACd,UAAUyC,OAAOrB,KAAK,EAAE,qBAAvCN,wBAAyCS,IAAI,MAAK,QAAQ;QAC9DwC,QAAQ,IAAGd,+BAAAA,OAAOhC,OAAO,CAACwB,OAAOrB,KAAK,CAAC,YAA5B6B,+BAAgCR,OAAOrB,KAAK,CAAC,SAAS,CAAC,EAAEqB,OAAOjC,KAAK;IAClF;IACA,OAAO;QAAEiD;QAAMC;QAASI;IAAQ;AAClC"}

package/src/lib/components/dataset-schema-dialog.component.d.ts

@@ -1,4 +1,4 @@
-import { type DatasetModel } from '@aglyn/aglyn';
+import { type DatasetModel } from '../model/dataset-models';
 export interface DatasetSchemaDialogProps {
     /** Host context; resolves the owning org for the data scope. */
     hostId?: string;