模式匹配

在 PostgreSQL 数据库中清理数据

Darryl Reeves, Ph.D.

Industry Assistant Professor, New York University

识别模式：示例

 camis    |              name               | inspection_date | score | nta  | ... 
 ---------+---------------------------------+-----------------+-------+------+-----
 ...      | ...                             | ...             | ...   | ...  | ...
 41659848 | LA BRISA DEL CIBAO              | 01/30/2018      | 20    | QN26 | ...
 40961447 | MESON SEVILLA RESTAURANT        | 03/19/2019      | 50    | MN15 | ...
 50063071 | WA BAR                          | 05/23/2018      | 15    | MN17 | ...
 50034992 | EMPANADAS MONUMENTAL            | 06/21/2019      | 17    | MN35 | ...
 50095871 | ALPHONSO'S PIZZERIA & TRATTORIA | 01/16/2020      | 10    | MN28 | ...
 41104041 | THE SPARROW TAVERN              | 09/17/2019      | 13    | QN72 | ...
 50016937 | BURGER KING                     | 09/14/2018      | 12    | QN55 | ...
 50066469 | DARBAR'S CHICKEN & RIBS         | 08/07/2017      | 11    | QN55 | ...
 41195691 | F & J PINE RESTAURANT           | 05/02/2019      | 26    | BX49 | ...
 50015706 | EL RINCONCITO DE LOS SABORES    | 12/18/2019      | 20    | QN35 | ...
 ...      | ...                             | ...             | ..    | ...  | ...

¹ https://www1.nyc.gov/site/planning/data-maps/open-data/dwn-nynta.page

识别模式：示例

NTA 代码的有效前缀

MN - 曼哈顿
BK - 布鲁克林
BX - 布朗克斯
QN - 皇后区
SI - 斯塔滕岛

识别模式：示例

用红框标出无效 NTA 代码的表格结果

LIKE 运算符

SELECT * FROM restaurant_inspection WHERE nta LIKE 'QN544';

SELECT * FROM restaurant_inspection WHERE nta = 'QN544';

LIKE 运算符

字符级模式匹配

% - 匹配任意长度（含0）的字符序列
_（下划线）- 匹配单个字符

SELECT 
* 
FROM 
restaurant_inspection 
WHERE nta LIKE 'QN%';

SELECT 
* 
FROM 
restaurant_inspection 
WHERE nta LIKE 'QN%' 
AND nta NOT LIKE 'QN__';

正则表达式（RE）

使用 LIKE 的模式匹配有局限
更具体的模式更有用
正则表达式（RE）提供更强的匹配能力

SIMILAR TO 运算符

提供额外的模式匹配功能

SELECT 
  camis, name, inspection_date, score, nta 
FROM 
  restaurant_inspection 
WHERE nta SIMILAR TO 'QN%' AND nta NOT SIMILAR TO 'QN__';

RE 基础

Metacharacter	Usage	Example RE	Example Match
`\d`	匹配数字 (0-9)	`\d\d\d`	'345'
`?`	匹配前一个字符出现0或1次	`x\d?`	'x5'
`+`	匹配前一个字符出现1次或多次	`\d+`	'10'
`*`	匹配前一个字符出现0次或多次	`\d*`	'3081'
`[]`	匹配方括号内任一字符	`[a-z]`	'f'

将 RE 与 SIMILAR TO 配合使用

SELECT 
    camis, name, inspection_date, score, nta 
FROM 
    restaurant_inspection 
WHERE nta SIMILAR TO 'QN%' AND nta NOT SIMILAR TO 'QN__';

SELECT 
    camis, name, inspection_date, score, nta 
FROM 
    restaurant_inspection 
WHERE 
    nta NOT SIMILAR TO '[A-Z][A-Z]\d\d';

 camis    |        name        | inspection_date | score | nta  
 ---------+--------------------+-----------------+-------+------
 41659848 | LA BRISA DEL CIBAO | 01/30/2018      | 20    | Q26
 41104041 | THE SPARROW TAVERN | 09/17/2019      | 13    | QN723

Passons à la pratique !

在 PostgreSQL 数据库中清理数据