@lakeql/schema-generator 0.2.1 → 0.4.0

package/dist/endpoint-schema.d.mts

@@ -15,14 +15,59 @@ declare const metadataFieldPattern: RegExp;
  */
 declare const loadStrategies: readonly ["full_load", "full_load_append", "append"];
 type LoadStrategy = (typeof loadStrategies)[number];
+/**
+ * Supported storage adapter types.
+ */
+declare const storageTypes: readonly ["s3", "minio"];
+type StorageType = (typeof storageTypes)[number];
+/**
+ * Supported partitioning format granularities.
+ */
+declare const partitioningFormats: readonly ["year", "year/month", "year/month/day"];
+type PartitioningFormat = (typeof partitioningFormats)[number];
+/**
+ * Supported date components for custom partition format segments.
+ */
+declare const partitioningComponents: readonly ["year", "month", "day", "hour", "minute", "second"];
+type PartitioningComponent = (typeof partitioningComponents)[number];
+/**
+ * Validates a custom partitioning format string.
+ * - If the value contains no `/` and no `:`, it must match fieldNamePattern (legacy single-field mode).
+ * - If it contains `/` or `:`, each segment is validated:
+ *   - Plain segments must match fieldNamePattern.
+ *   - Colon segments must be `fieldname:component` where fieldname matches pattern
+ *     and component is one of the supported partitioning components.
+ *
+ * Returns null if valid, or an error message string if invalid.
+ */
+declare function validatePartitioningFormat(value: string): string | null;
+/**
+ * Partitioning configuration value:
+ * - true: partition by write timestamp (default)
+ * - false: no partitioning (flat layout)
+ * - string: partition by named field
+ */
+type PartitioningValue = boolean | string;
 /**
  * Mutation pipeline configuration for an endpoint.
  */
 interface MutationConfig {
   /** Load strategy for the write pipeline. */
   loadStrategy: LoadStrategy;
-  /** S3 base path for endpoint data. */
+  /** Storage adapter type. Defaults to "s3". */
+  type: StorageType;
+  /** S3/MinIO bucket name for endpoint data. */
+  bucket: string;
+  /** Base path for endpoint data. */
   basePath: string;
+  /** Optional region override (falls back to AWS_DEFAULT_REGION env var for S3). */
+  region?: string;
+  /** Optional custom endpoint. Required for MinIO, optional for S3. */
+  endpoint?: string;
+  /** Partitioning mode. Default: true */
+  partitioning?: PartitioningValue;
+  /** Partition path granularity. Default: "year/month/day" */
+  partitioningFormat?: string;
 }
 /**
  * Zod schema for the mutation configuration object.
@@ -33,7 +78,16 @@ declare const mutationConfigSchema: z.ZodObject<{
     full_load_append: "full_load_append";
     append: "append";
   }>;
+  type: z.ZodDefault<z.ZodEnum<{
+    s3: "s3";
+    minio: "minio";
+  }>>;
+  bucket: z.ZodString;
   basePath: z.ZodString;
+  region: z.ZodOptional<z.ZodString>;
+  endpoint: z.ZodOptional<z.ZodString>;
+  partitioning: z.ZodOptional<z.ZodUnion<readonly [z.ZodBoolean, z.ZodString]>>;
+  partitioningFormat: z.ZodOptional<z.ZodString>;
 }, z.core.$strip>;
 /**
  * Zod schema for the mutation field: either `false` (disabled) or a config object (enabled).
@@ -44,7 +98,16 @@ declare const mutationSchema: z.ZodUnion<readonly [z.ZodLiteral<false>, z.ZodObj
     full_load_append: "full_load_append";
     append: "append";
   }>;
+  type: z.ZodDefault<z.ZodEnum<{
+    s3: "s3";
+    minio: "minio";
+  }>>;
+  bucket: z.ZodString;
   basePath: z.ZodString;
+  region: z.ZodOptional<z.ZodString>;
+  endpoint: z.ZodOptional<z.ZodString>;
+  partitioning: z.ZodOptional<z.ZodUnion<readonly [z.ZodBoolean, z.ZodString]>>;
+  partitioningFormat: z.ZodOptional<z.ZodString>;
 }, z.core.$strip>]>;
 /**
  * Supported primitive types for field definitions.
@@ -118,7 +181,16 @@ declare const endpointDefinitionSchema: z.ZodObject<{
       full_load_append: "full_load_append";
       append: "append";
     }>;
+    type: z.ZodDefault<z.ZodEnum<{
+      s3: "s3";
+      minio: "minio";
+    }>>;
+    bucket: z.ZodString;
     basePath: z.ZodString;
+    region: z.ZodOptional<z.ZodString>;
+    endpoint: z.ZodOptional<z.ZodString>;
+    partitioning: z.ZodOptional<z.ZodUnion<readonly [z.ZodBoolean, z.ZodString]>>;
+    partitioningFormat: z.ZodOptional<z.ZodString>;
   }, z.core.$strip>]>>;
 }, z.core.$strip>;
 type EndpointDefinitionFormat = z.infer<typeof endpointDefinitionSchema>;
@@ -131,4 +203,4 @@ declare function findDuplicateFieldNames(fields: FieldDefinition[], path?: strin
   path: string[];
 }[];
 //#endregion
-export { ArrayItemDefinition, EndpointDefinitionFormat, FieldDefinition, FieldOptions, FieldValidation, LoadStrategy, MutationConfig, PrimitiveType, endpointDefinitionSchema, fieldDefinitionSchema, fieldNamePattern, fieldOptionsSchema, fieldValidationSchema, findDuplicateFieldNames, loadStrategies, metadataFieldPattern, mutationConfigSchema, mutationSchema, primitiveTypes };
+export { ArrayItemDefinition, EndpointDefinitionFormat, FieldDefinition, FieldOptions, FieldValidation, LoadStrategy, MutationConfig, PartitioningComponent, PartitioningFormat, PartitioningValue, PrimitiveType, StorageType, endpointDefinitionSchema, fieldDefinitionSchema, fieldNamePattern, fieldOptionsSchema, fieldValidationSchema, findDuplicateFieldNames, loadStrategies, metadataFieldPattern, mutationConfigSchema, mutationSchema, partitioningComponents, partitioningFormats, primitiveTypes, storageTypes, validatePartitioningFormat };

package/dist/endpoint-schema.mjs

@@ -1 +1 @@
-import{z as e}from"zod";const t=/^[a-zA-Z_][a-zA-Z0-9_]{0,63}$/u,n=/^[a-zA-Z_][a-zA-Z0-9_]{0,127}$/u,r=[`full_load`,`full_load_append`,`append`],i=e.object({loadStrategy:e.enum(r),basePath:e.string().min(1)}),a=e.union([e.literal(!1),i]),o=[`String`,`Integer`,`Float`,`Boolean`,`Date`,`DateTime`],s=e.union([e.object({type:e.literal(`email`)}),e.object({type:e.literal(`url`)}),e.object({type:e.literal(`uuid`)}),e.object({type:e.literal(`min`),value:e.number()}),e.object({type:e.literal(`max`),value:e.number()}),e.object({type:e.literal(`regex`),pattern:e.string()})]),c=e.object({required:e.boolean().optional(),validations:e.array(s).optional()}),l=e.lazy(()=>e.discriminatedUnion(`type`,[e.object({name:e.string().regex(t),type:e.literal(`String`),options:c.optional()}),e.object({name:e.string().regex(t),type:e.literal(`Integer`),options:c.optional()}),e.object({name:e.string().regex(t),type:e.literal(`Float`),options:c.optional()}),e.object({name:e.string().regex(t),type:e.literal(`Boolean`),options:c.optional()}),e.object({name:e.string().regex(t),type:e.literal(`Date`),options:c.optional()}),e.object({name:e.string().regex(t),type:e.literal(`DateTime`),options:c.optional()}),e.object({name:e.string().regex(t),type:e.literal(`Object`),fields:e.array(l).min(1),options:c.optional()}),e.object({name:e.string().regex(t),type:e.literal(`Array`),items:e.union([e.object({type:e.enum(o)}),e.object({type:e.literal(`Object`),fields:e.array(l).min(1)})]),options:c.optional()})])),u=e.object({version:e.literal(`1.0`),tableName:e.string().regex(n),catalog:e.string().regex(n),schema:e.string().regex(n),fields:e.array(l),mutation:a.optional()});function d(e,t=[]){let n=[],r=new Set;for(let i of e)r.has(i.name)?n.push({name:i.name,path:t}):r.add(i.name),i.type===`Object`&&i.fields&&n.push(...d(i.fields,[...t,i.name])),i.type===`Array`&&i.items?.type===`Object`&&i.items.fields&&n.push(...d(i.items.fields,[...t,i.name,`items`]));return n}export{u as endpointDefinitionSchema,l as fieldDefinitionSchema,t as fieldNamePattern,c as fieldOptionsSchema,s as fieldValidationSchema,d as findDuplicateFieldNames,r as loadStrategies,n as metadataFieldPattern,i as mutationConfigSchema,a as mutationSchema,o as primitiveTypes};
+import{z as e}from"zod";const t=/^[a-zA-Z_][a-zA-Z0-9_]{0,63}$/u,n=/^[a-zA-Z_][a-zA-Z0-9_]{0,127}$/u,r=[`full_load`,`full_load_append`,`append`],i=[`s3`,`minio`],a=[`year`,`year/month`,`year/month/day`],o=[`year`,`month`,`day`,`hour`,`minute`,`second`];function s(e){if(!(e.includes(`/`)||e.includes(`:`)))return t.test(e)?null:`Invalid partition field name "${e}". Must match pattern: letters/underscores, no leading digit, max 64 chars.`;let n=e.split(`/`);for(let r of n){if(r===``)return`Invalid partition format: empty segment found in "${e}".`;if(r.includes(`:`)){let e=r.split(`:`);if(e.length!==2)return`Invalid partition segment "${r}": expected format "fieldname:component".`;let[n,i]=e;if(!t.test(n))return`Invalid field name "${n}" in partition segment "${r}".`;if(!o.includes(i))return`Invalid date component "${i}" in partition segment "${r}". Expected one of: ${o.join(`, `)}.`}else if(!t.test(r))return`Invalid field name "${r}" in partition format.`}return null}const c=e.object({loadStrategy:e.enum(r),type:e.enum(i).default(`s3`),bucket:e.string().min(1),basePath:e.string().min(1),region:e.string().optional(),endpoint:e.string().optional(),partitioning:e.union([e.boolean(),e.string().min(1)]).optional(),partitioningFormat:e.string().optional()}).superRefine((e,t)=>{if(e.partitioning!==!1){if(typeof e.partitioning==`string`){let n=s(e.partitioning);if(n){t.addIssue({code:`custom`,path:[`partitioning`],message:n});return}if(e.partitioning.includes(`/`)||e.partitioning.includes(`:`))return}e.partitioningFormat!==void 0&&!a.includes(e.partitioningFormat)&&t.addIssue({code:`invalid_value`,values:[...a],path:[`partitioningFormat`],message:`Invalid partitioning format. Expected one of: ${a.join(`, `)}`})}}),l=e.union([e.literal(!1),c]),u=[`String`,`Integer`,`Float`,`Boolean`,`Date`,`DateTime`],d=e.union([e.object({type:e.literal(`email`)}),e.object({type:e.literal(`url`)}),e.object({type:e.literal(`uuid`)}),e.object({type:e.literal(`min`),value:e.number()}),e.object({type:e.literal(`max`),value:e.number()}),e.object({type:e.literal(`regex`),pattern:e.string()})]),f=e.object({required:e.boolean().optional(),validations:e.array(d).optional()}),p=e.lazy(()=>e.discriminatedUnion(`type`,[e.object({name:e.string().regex(t),type:e.literal(`String`),options:f.optional()}),e.object({name:e.string().regex(t),type:e.literal(`Integer`),options:f.optional()}),e.object({name:e.string().regex(t),type:e.literal(`Float`),options:f.optional()}),e.object({name:e.string().regex(t),type:e.literal(`Boolean`),options:f.optional()}),e.object({name:e.string().regex(t),type:e.literal(`Date`),options:f.optional()}),e.object({name:e.string().regex(t),type:e.literal(`DateTime`),options:f.optional()}),e.object({name:e.string().regex(t),type:e.literal(`Object`),fields:e.array(p).min(1),options:f.optional()}),e.object({name:e.string().regex(t),type:e.literal(`Array`),items:e.union([e.object({type:e.enum(u)}),e.object({type:e.literal(`Object`),fields:e.array(p).min(1)})]),options:f.optional()})])),m=e.object({version:e.literal(`1.0`),tableName:e.string().regex(n),catalog:e.string().regex(n),schema:e.string().regex(n),fields:e.array(p),mutation:l.optional()});function h(e,t=[]){let n=[],r=new Set;for(let i of e)r.has(i.name)?n.push({name:i.name,path:t}):r.add(i.name),i.type===`Object`&&i.fields&&n.push(...h(i.fields,[...t,i.name])),i.type===`Array`&&i.items?.type===`Object`&&i.items.fields&&n.push(...h(i.items.fields,[...t,i.name,`items`]));return n}export{m as endpointDefinitionSchema,p as fieldDefinitionSchema,t as fieldNamePattern,f as fieldOptionsSchema,d as fieldValidationSchema,h as findDuplicateFieldNames,r as loadStrategies,n as metadataFieldPattern,c as mutationConfigSchema,l as mutationSchema,o as partitioningComponents,a as partitioningFormats,u as primitiveTypes,i as storageTypes,s as validatePartitioningFormat};

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@lakeql/schema-generator",
-  "version": "0.2.1",
+  "version": "0.4.0",
   "description": "Generates GraphQL schemas, Hive table definitions, and JSON schemas from Trino metadata",
   "keywords": [
     "codegen",
@@ -55,7 +55,7 @@
     "lodash-es": "4.18.1",
     "magic-regexp": "0.11.0",
     "zod": "4.4.3",
-    "@lakeql/helpers": "0.1.2"
+    "@lakeql/helpers": "0.1.3"
   },
   "devDependencies": {
     "@types/json-schema": "7.0.15",
@@ -65,7 +65,7 @@
     "json-column-parser": "2.0.0",
     "tsdown": "0.22.2",
     "typescript": "6.0.3",
-    "vitest": "4.1.8",
+    "vitest": "4.1.9",
     "@lakeql/tsconfig": "0.1.0"
   },
   "scripts": {