Remember that a development is encountered in the pit of origin string sorting

        Briefly describe the situation , In the stored procedure, a field of string type is used as the sorting number of patient visits , The result is inexplicably found that the call is over 1 Called after the th 11、12 Wait for patients . User feedback may not be accurate , I added my own log , Copy the log quickly to observe . It turns out that the actual situation is like this , It's wrong to say , Start looking at the code , Fields used for analysis and sorting , After analysis, it is found that since this field is a string, the collation is a character by character comparison , Write a separate statement to test , Right enough .

WITH T AS
(SELECT '11' STR
    FROM DUAL
  UNION ALL
  SELECT '2' STR FROM DUAL
  UNION ALL
  SELECT '3' STR FROM DUAL
  UNION ALL
  SELECT '21' STR FROM DUAL
  UNION ALL
  SELECT '30' STR FROM DUAL
  UNION ALL
  SELECT '1' STR FROM DUAL)
SELECT * FROM T ORDER BY STR;

The output is as follows ：

1    1
2    11
3    2
4    21
5    3
6    30

 

Because this field will store Chinese characters , In the end use state The numeric part of the field and the fetch string is converted to a numeric value , And then sort it . as for

Take the numerical part , Using regular expressions

to_number(regexp_substr(nvl( Field a,0),'[0-9]+'))

Record the pit you encountered , It could be very simple , But easy to ignore . Be diligent in practice 、 Willing to record , Love to share .