Python REST API数据清洗：利用模糊匹配识别姓名拼写变体与错别字

import requests
import json

# 假设API参数只支持精确匹配
Payee_Parameter = {
    "contribution_payee": "John Smith", # 只能匹配精确的“John Smith”
    "dt_posted": "ascending",
    "key": "YOUR_API_KEY" # 替换为你的API密钥
}

ContributionsLink = "https://lda.senate.gov/api/v1/contributions/"
response = requests.get(ContributionsLink, params=Payee_Parameter)
data = response.json()
# ... 后续处理 ...

pip install fuzzywuzzy

pip install python-Levenshtein

from fuzzywuzzy import fuzz
from fuzzywuzzy import process

print(fuzz.ratio("John Doe", "Joe Dow"))        # 输出: 67
print(fuzz.ratio("John Doe", "John M. Doe"))    # 输出: 84
print(fuzz.ratio("John Doe", "Billy Jean"))     # 输出: 22

print(fuzz.partial_ratio("apple pie", "apple")) # 输出: 100 (因为"apple"是"apple pie"的一部分)
print(fuzz.token_sort_ratio("fuzzy wuzzy was a bear", "wuzzy fuzzy bear was a")) # 输出: 100 (忽略词序)

import requests
import json
from fuzzywuzzy import fuzz
from fuzzywuzzy import process

# 目标匹配的姓名
target_candidate_name = "John Smith"
# 相似度阈值：只有相似度达到或超过这个值才被认为是匹配
similarity_threshold = 80

# 模拟从API获取的数据。
# 实际应用中，这里会是一个API调用，可能通过宽泛的查询参数获取大量数据。
# 例如：
# Payee_Parameter = {
#     "contribution_payee": "Smith", # 宽泛查询，例如只用姓氏
#     "dt_posted": "ascending",
#     "key": "YOUR_API_KEY"
# }
# ContributionsLink = "https://lda.senate.gov/api/v1/contributions/"
# response = requests.get(ContributionsLink, params=Payee_Parameter)
# api_data = response.json()
# api_data_results = api_data.get("results", []) # 假设数据在"results"键下

# 为了演示，我们使用一个硬编码的列表来模拟API返回的数据
api_data_results = [
    {"payee_name": "John Smith", "amount": 100, "id": "rec1"},
    {"payee_name": "Jonathan Smith", "amount": 150, "id": "rec2"},
    {"payee_name": "Jon Smith", "amount": 200, "id": "rec3"},
    {"payee_name": "Johnathon Smith", "amount": 50, "id": "rec4"},
    {"payee_name": "Jane Doe", "amount": 120, "id": "rec5"},
    {"payee_name": "Jon Smtih", "amount": 75, "id": "rec6"}, # 拼写错误示例
    {"payee_name": "J Smith", "amount": 80, "id": "rec7"},
    {"payee_name": "Johnny Smith", "amount": 90, "id": "rec8"},
]

matched_contributions = []

print(f"开始匹配 '{target_candidate_name}' 的捐款记录 (相似度阈值 >= {similarity_threshold})...")

for record in api_data_results:
    payee_name_from_api = record.get("payee_name")
    if payee_name_from_api:
        # 转换为小写进行不区分大小写匹配，提高匹配鲁棒性
        score = fuzz.ratio(target_candidate_name.lower(), payee_name_from_api.lower())

        if score >= similarity_threshold:
            matched_contributions.append({
                "original_payee_name": payee_name_from_api,
                "amount": record.get("amount"),
                "similarity_score": score,
                "record_id": record.get("id")
            })

# 打印匹配结果
if matched_contributions:
    print("\n匹配到的捐款记录:")
    for contribution in matched_contributions:
        print(f"- 姓名: {contribution['original_payee_name']}, 金额: {contribution['amount']}, 相似度: {contribution['similarity_score']}")
else:
    print("未找到符合条件的匹配记录。")

# 另一个高级用法：使用 process.extract 从一个列表中查找最佳匹配
# 假设我们想从所有已知的候选人姓名中找出与“John Smith”最接近的几个
all_payee_names_from_api = [record.get("payee_name") for record in api_data_results if record.get("payee_name")]

if all_payee_names_from_api:
    print(f"\n使用 process.extract 查找与 '{target_candidate_name}' 最接近的几个姓名:")
    # limit=3 表示返回前3个最相似的匹配
    top_matches = process.extract(target_candidate_name, all_payee_names_from_api, limit=3, scorer=fuzz.ratio)
    for match in top_matches:
        # match 是一个元组 (匹配到的字符串, 相似度分数)
        print(f"- {match[0]} (相似度: {match[1]})")
else:
    print("\nAPI数据中没有可供匹配的姓名列表。")