获取JOINed数据并使其成为一行而不是列

Question

我正在处理一个使用连接来提取某些信息的查询。

import tensorflow as tf
import numpy as np

# Dummy values
indices = np.array([[1, 2], [2, 3]])
values = np.array([100, 200])

# Placeholders
indices_ = tf.placeholder(tf.int32, shape=(2, 2))
values_ = tf.placeholder(tf.float32, shape=(2))

# Use the Cantor tuple to create a one-to-one correspondence between the coordinates
# and a single value
x = tf.cast(indices_[:, 0], tf.float32)
y = tf.cast(indices_[:, 1], tf.float32)
z = (x + y) * (x + y + 1) / 2 + y  # shape = (2)

# Collect unique indices, treated as single values
# Drop the indices position into z because are useless
unique_cantor, _ = tf.unique(z)

# Go back from cantor numbers to pairs of values
w = tf.floor((tf.sqrt(8 * unique_cantor + 1) - 1) / 2)
t = (tf.pow(w, 2) + w) / 2
y = z - t
x = w - y

# Recreate a batch of coordinates that are uniques
unique_indices = tf.cast(tf.stack([x, y], axis=1), tf.int32)

# Update without accumulator
go = tf.scatter_nd(unique_indices, values_, [4, 5])

with tf.Session() as sess:
    print(sess.run(go, feed_dict={indices_: indices, values_: values}))

加入的数据作为该行的另一列进入。我希望这个专栏成为另一行：

12,0，PAR PHARMA，current_timestamp，''

我该怎么做？

Answer 1

我认为您需要Union all将这些记录作为行：

    SELECT ca.item_id
        ,ca.FIELD_ID
        ,ca.attr_val
        ,ca.upd_dtt
        ,ca.upd_usr
    FROM contract_attr ca
    UNION ALL
    SELECT ... cols along with, --If no columns available in mfr then provide nulls accordingly before providing column name Item_Name
      ,mf.[ITEM_NAME]
    FROM mfr mf

列列表必须是相同的顺序

用JOIN完成问题：

  SELECT ca.item_id
        ,ca.FIELD_ID
        ,ca.attr_val
        ,ca.upd_dtt
        ,ca.upd_usr
    FROM contract_attr ca
    UNION ALL
    SELECT ca.item_id,9999,mf.[ITEM_NAME],'',''
 FROM mfr mf
 JOIN contract_attr ca on ca.attr_val = mf.[ITEM_PK]
 Order by ca.item_id