Python的标准库有一个处理CSV文件的模块。它的DictReader类假设输入文件的第一行是列名（除非你提供fieldnames参数）。使用它，你只需要对字段名'tests'做一些特殊的事情。不幸的是，它还不能处理pathlib.Path()示例，所以你必须自己打开文件。
您应该使用ruamel.yaml转储生成的数据结构，您必须将其安装在virtualenv中，例如使用python -m pip install ruamel.yaml。它是YAML 1.2（免责声明：我是那个软件包的作者）。

import csv
from pathlib import Path
import ruamel.yaml

input = Path('input.csv')
output = Path('output.yaml')

data = []
reader = csv.DictReader(input.open(newline=''))
for row in reader:
    d = {}
    data.append(d)
    for field in reader.fieldnames:
        d[field] = row[field].split('|') if field == 'tests' else row[field]

yaml = ruamel.yaml.YAML()
yaml.dump(data, output)

print(output.read_text())

其给出：

- name: product_id
  description: ID of the product
  tests:
  - not null
  - unique
- name: product_name
  description: Name of the product
  tests:
  - not null

在我看来，一个更像Python的解决方案是：

import yaml
import csv

with open('sample_data.csv', 'r') as f:
    reader = csv.DictReader(f)
    data = []
    for row in reader:
        item = {}
        item['name'] = row['name']
        item['description'] = row['description']
        item['tests'] = row['tests'].split('|')
        data.append(item)

with open('output.yaml', 'w') as f:
    yaml.dump(data, f)

输入：

name,description,tests
product_id,ID of the product,not null|unique
product_name,Name of the product,not null

输出：

- description: ID of the product
  name: product_id
  tests:
  - not null
  - unique
- description: Name of the product
  name: product_name
  tests:
  - not null