@nuxodin/item 0.6.14 → 0.6.16

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@nuxodin/item",
-  "version": "0.6.14",
+  "version": "0.6.16",
   "description": "Reactive state library",
   "type": "module",
   "license": "MIT",

package/tools/schema/db/mysql/from-field.js

@@ -52,7 +52,8 @@ export function schemaFromField(row) {
     prop.maxLength = len;
   } else if (type in textBytes) {
     prop.type = "string";
-    if (type !== "text") prop.maxLength = Math.floor(textBytes[type] / B);
+    // also for TEXT: without its bound, a wider declaration would never widen it
+    prop.maxLength = Math.floor(textBytes[type] / B);
   } else if (type === "binary") {
     prop.type = "string";
     prop.contentEncoding = "base64";
@@ -78,6 +79,11 @@ export function schemaFromField(row) {
     prop.type = "integer";
     prop.minimum = 1901;
     prop.maximum = 2155;
+  } else if (type === "vector") {
+    // the length belongs to the application, not the schema — see to-field.js
+    prop.type = "array";
+    prop.items = { type: "number" };
+    prop["x-vector"] = true;
   } else if (type === "json") prop.type = "object";
   else if (type === "enum") {
     prop.type = "string";

package/tools/schema/db/mysql/to-db.js

@@ -34,7 +34,7 @@ function setEqual(a, b) {
 function secondaryIndexKind(prop) {
   if (prop["x-index"] === "unique") return "UNIQUE KEY";
   if (prop["x-index"] === "fulltext") return "FULLTEXT KEY";
-  if (prop["x-index"] === true) return "KEY";
+  if (prop["x-index"] === true) return prop["x-vector"] ? "VECTOR INDEX" : "KEY";
   return null;
 }
 
@@ -233,6 +233,8 @@ async function indexStatements(query, table, nextFields, currFields, { patch = f
     const currentKind = current
       ? current.indexType === "FULLTEXT"
         ? "FULLTEXT KEY"
+        : current.indexType === "VECTOR"
+        ? "VECTOR INDEX"
         : current.nonUnique
         ? "KEY"
         : "UNIQUE KEY"

package/tools/schema/db/mysql/to-field.js

@@ -15,6 +15,9 @@ export function toFieldDef(name, prop, { required = false } = {}) {
   if (schemaType === "boolean") type = "TINYINT(1)";
   else if (schemaType === "integer") ({ type, unsigned } = integerType(prop));
   else if (schemaType === "number") type = prop.multipleOf ? "DECIMAL" : "DOUBLE";
+  // `x-vector: true` leaves the length to the application (MariaDB needs one: it starts at 1 and
+  // widens with MODIFY, padding existing vectors with zeros); a number fixes it.
+  else if (prop["x-vector"]) type = `VECTOR(${prop["x-vector"] === true ? 1 : prop["x-vector"]})`;
   else if (schemaType === "object" || schemaType === "array") type = "JSON";
   else type = stringType(prop);
 

package/tools/schema/db/pg/from-field.js

@@ -43,6 +43,10 @@ export function schemaFromField(row) {
     } else if (type.startsWith('time') || dataType.startsWith('time')) {
         prop.type = 'string'
         prop.format = 'time'
+    } else if (type === 'vector') {
+        prop.type = 'array'
+        prop.items = { type: 'number' }
+        prop['x-vector'] = true
     } else if (type === 'json' || type === 'jsonb') prop.type = 'object'
     else prop.type = 'string'
 

package/tools/schema/db/pg/to-db.js

@@ -33,6 +33,8 @@ function preserveDefault(prop, currProp) {
 }
 
 function secondaryIndexKind(table, name, prop) {
+    // hnsw needs a fixed length and an operator class; the application builds vector indexes
+    if (prop['x-vector']) return null
     if (prop['x-index'] === 'unique') return 'UNIQUE INDEX'
     if (prop['x-index'] === true) return 'INDEX'
     if (prop['x-index'] === 'fulltext')
@@ -171,6 +173,10 @@ export async function schemaToDb(schema, query, { force = false, patch = false }
         }
     }
 
+    // x-vector columns are pgvector's type; the extension has to exist before they do
+    const vectors = nextTables.some(t => tableData(schema.properties[t]).fields.some(([, f]) => f['x-vector']))
+    if (vectors && stmts.length) stmts.unshift('CREATE EXTENSION IF NOT EXISTS vector;')
+
     for (const stmt of stmts) {
         console.log(stmt)
         await query(stmt)

package/tools/schema/db/pg/to-field.js

@@ -17,6 +17,7 @@ export function typeSql(prop) {
     if (t === 'boolean') return 'BOOLEAN'
     if (t === 'integer') return integerType(prop)
     if (t === 'number') return prop.multipleOf ? 'NUMERIC' : 'DOUBLE PRECISION'
+    if (prop['x-vector']) return prop['x-vector'] === true ? 'vector' : `vector(${prop['x-vector']})` // pgvector
     if (t === 'object' || t === 'array') return 'JSONB'
     if (prop.format in formats) return formats[prop.format]
     if (prop.contentEncoding === 'base64') return 'BYTEA'

package/tools/schema/db/sqlite/from-field.js

@@ -32,6 +32,7 @@ export function schemaFromField(row) {
     } else if (type === 'date')      { prop.type = 'string'; prop.format = 'date'
     } else if (type === 'time')      { prop.type = 'string'; prop.format = 'time'
     } else if (type === 'datetime' || type === 'timestamp') { prop.type = 'string'; prop.format = 'date-time'
+    } else if (type === 'vector')    { prop.type = 'array'; prop.items = { type: 'number' }; prop['x-vector'] = true
     } else if (type === 'json')      { prop.type = 'object'
     } else {
         prop.type = 'string'

package/tools/schema/db/sqlite/to-db.js

@@ -47,6 +47,7 @@ function createBody(table, fields, primaries) {
 }
 
 function indexKind(table, name, prop) {
+    if (prop['x-vector']) return null // sqlite-vec scans; a B-tree over vectors helps nothing
     if (prop['x-index'] === 'unique') return 'UNIQUE INDEX'
     if (prop['x-index'] === true) return 'INDEX'
     if (prop['x-index'] === 'fulltext')

package/tools/schema/db/sqlite/to-field.js

@@ -15,6 +15,7 @@ export function toFieldDef(name, prop) {
     if      (t === 'boolean')             type = 'BOOLEAN'
     else if (t === 'integer')             type = 'INTEGER'
     else if (t === 'number')              type = 'REAL'
+    else if (prop['x-vector'])            type = 'VECTOR'  // blobs keep their bytes; the name round-trips
     else if (t === 'object' || t === 'array') type = 'TEXT'  // JSON as text
     else {
         // Same reasoning as BOOLEAN: the declared name is the only carrier of the format, and