From 10f23492e60e384f537ef7b6aeaf7a30ec778784 Mon Sep 17 00:00:00 2001
From: Kenny Daniel <platypii@gmail.com>
Date: Fri, 11 Apr 2025 03:13:35 -0700
Subject: [PATCH] Unconvert decimal type

---
 src/schema.js          | 13 ++++++-----
 src/unconvert.js       | 42 ++++++++++++++++++++++++++++++++++-
 test/unconvert.test.js | 50 +++++++++++++++++++++++++++++++++++++++++-
 3 files changed, 98 insertions(+), 7 deletions(-)

diff --git a/src/schema.js b/src/schema.js
index bae468f..bff23e3 100644
--- a/src/schema.js
+++ b/src/schema.js
@@ -19,11 +19,14 @@ export function schemaFromColumnData(columnData) {
       throw new Error('columns must have the same length')
     }
 
-    // auto-detect type if not provided
-    /** @type {SchemaElement} */
-    const schemaElement = column.type ? column : autoSchemaElement(column.name, column.data)
-    if (!schemaElement.type) throw new Error(`column ${column.name} cannot determine type`)
-    schema.push(schemaElement)
+    const { data, ...schemaElement } = column
+    if (column.type) {
+      // use provided type
+      schema.push(schemaElement)
+    } else {
+      // auto-detect type
+      schema.push(autoSchemaElement(column.name, data))
+    }
   }
 
   return schema
diff --git a/src/unconvert.js b/src/unconvert.js
index 0a772ea..6718439 100644
--- a/src/unconvert.js
+++ b/src/unconvert.js
@@ -1,3 +1,4 @@
+const dayMillis = 86400000 // 1 day in milliseconds
 
 /**
  * Convert from rich to primitive types.
@@ -9,7 +10,15 @@
  */
 export function unconvert(schemaElement, values) {
   const ctype = schemaElement.converted_type
-  // TODO: DECIMAL
+  if (ctype === 'DECIMAL') {
+    const scale = schemaElement.scale || 0
+    const factor = 10 ** scale
+    return values.map(v => {
+      if (v === null || v === undefined) return v
+      if (typeof v !== 'number') throw new Error('DECIMAL must be a number')
+      return unconvertDecimal(BigInt(Math.round(v * factor))) // to byte array
+    })
+  }
   if (ctype === 'DATE') {
     return values.map(v => v.getTime())
   }
@@ -68,6 +77,11 @@ export function unconvertMetadata(value, schema) {
     new DataView(buffer).setBigInt64(0, value, true)
     return new Uint8Array(buffer)
   }
+  if (type === 'INT32' && converted_type === 'DATE' && value instanceof Date) {
+    const buffer = new ArrayBuffer(8)
+    new DataView(buffer).setInt32(0, Math.floor(value.getTime() / dayMillis), true)
+    return new Uint8Array(buffer)
+  }
   if (type === 'INT64' && converted_type === 'TIMESTAMP_MILLIS' && value instanceof Date) {
     const buffer = new ArrayBuffer(8)
     new DataView(buffer).setBigInt64(0, BigInt(value.getTime()), true)
@@ -75,3 +89,29 @@ export function unconvertMetadata(value, schema) {
   }
   throw new Error(`unsupported type for statistics: ${type} with value ${value}`)
 }
+
+/**
+ * @param {bigint} value
+ * @returns {Uint8Array}
+ */
+export function unconvertDecimal(value) {
+  if (value === 0n) return new Uint8Array([])
+  const bytes = []
+  let current = value
+
+  while (true) {
+    // extract the lowest 8 bits
+    const byte = Number(current & 0xffn)
+    bytes.unshift(byte)
+    current >>= 8n
+
+    // for nonnegative: stop when top byte has signBit = 0 AND shifted value == 0n
+    // for negative: stop when top byte has signBit = 1 AND shifted value == -1n
+    const signBit = byte & 0x80
+    if (!signBit && current === 0n || signBit && current === -1n) {
+      break
+    }
+  }
+
+  return new Uint8Array(bytes)
+}
diff --git a/test/unconvert.test.js b/test/unconvert.test.js
index 4c4741d..581b4e9 100644
--- a/test/unconvert.test.js
+++ b/test/unconvert.test.js
@@ -1,5 +1,5 @@
 import { describe, expect, it } from 'vitest'
-import { unconvert, unconvertMetadata } from '../src/unconvert.js'
+import { unconvert, unconvertDecimal, unconvertMetadata } from '../src/unconvert.js'
 import { convertMetadata } from 'hyparquet/src/metadata.js'
 
 /**
@@ -155,3 +155,51 @@ describe('unconvertMetadata', () => {
       .toThrow('unsupported type for statistics: INT64 with value 123')
   })
 })
+
+describe('unconvertDecimal', () => {
+  const examples = [
+    { input: 0n, expected: new Uint8Array([]) },
+    { input: 1n, expected: new Uint8Array([0x01]) },
+    { input: -1n, expected: new Uint8Array([0xff]) },
+    { input: 1234n, expected: new Uint8Array([0x04, 0xd2]) },
+    { input: -1234n, expected: new Uint8Array([0xfb, 0x2e]) },
+    { input: 1234567890123456789n, expected: new Uint8Array([0x11, 0x22, 0x10, 0xf4, 0x7d, 0xe9, 0x81, 0x15]) },
+    { input: -1234567890123456789n, expected: new Uint8Array([0xee, 0xdd, 0xef, 0x0b, 0x82, 0x16, 0x7e, 0xeb]) },
+  ]
+
+  it.for(examples)('should convert %p', ({ input, expected }) => {
+    expect(parseDecimal(expected)).toEqual(input)
+  })
+
+  it.for(examples)('should unconvert %p', ({ input, expected }) => {
+    expect(unconvertDecimal(input)).toEqual(expected)
+  })
+
+  it.for(examples)('should roundtrip %p', ({ input }) => {
+    expect(parseDecimal(unconvertDecimal(input))).toEqual(input)
+  })
+
+  it.for(examples)('should reverse roundtrip %p', ({ expected }) => {
+    expect(unconvertDecimal(parseDecimal(expected))).toEqual(expected)
+  })
+})
+
+/**
+ * BigInt parseDecimal
+ * @param {Uint8Array} bytes
+ * @returns {bigint}
+ */
+function parseDecimal(bytes) {
+  let value = 0n
+  for (const byte of bytes) {
+    value = value * 256n + BigInt(byte)
+  }
+
+  // handle signed
+  const bits = BigInt(bytes.length) * 8n
+  if (bits && value >= 2n ** (bits - 1n)) {
+    value -= 2n ** bits
+  }
+
+  return value
+}