pandas教程：Interacting with Web APIs API和数据库的交互

数据库的交互"/>

pandas教程：Interacting with Web APIs API和数据库的交互

文章目录

• 6.3 Interacting with Web APIs (网络相关的API交互)
• 6.4 Interacting with Databases(与数据库的交互)

6.3 Interacting with Web APIs (网络相关的API交互)

很多网站都有公开的API，通过JSON等格式提供数据流。有很多方法可以访问这些API，这里推荐一个易用的requests包。

找到github里pandas最新的30个issues，制作一个GET HTTP request, 通过使用requests包：

import pandas as pd
import numpy as np

import requests

url = ''

resp = requests.get(url)

resp

<Response [200]>

response的json方法能返回一个dict，包含可以解析为python object的JSON：

data = resp.json()
data[0]['title']

'Optimize data type'

data[0]

{'assignee': None,'assignees': [],'author_association': 'NONE','body': 'Hi guys, i\'m user of mysql\r\nwe have an "function" PROCEDURE ANALYSE\r\n.5/en/procedure-analyse.html\r\n\r\nit get all "dataframe" and show what\'s the best "dtype", could we do something like it in Pandas?\r\n\r\nthanks!','closed_at': None,'comments': 1,'comments_url': '','created_at': '2017-11-13T22:51:32Z','events_url': '','html_url': '','id': 273606786,'labels': [],'labels_url': '{/name}','locked': False,'milestone': None,'number': 18272,'repository_url': '','state': 'open','title': 'Optimize data type','updated_at': '2017-11-13T22:57:27Z','url': '','user': {'avatar_url': '=4','events_url': '{/privacy}','followers_url': '','following_url': '{/other_user}','gists_url': '{/gist_id}','gravatar_id': '','html_url': '','id': 2468782,'login': 'rspadim','organizations_url': '','received_events_url': '','repos_url': '','site_admin': False,'starred_url': '{/owner}{/repo}','subscriptions_url': '','type': 'User','url': ''}}

data中的每一个元素都是一个dict，这个dict就是在github上找到的issue页面上的信息。我们可以把data传给DataFrame并提取感兴趣的部分：

issues = pd.DataFrame(data, columns=['number', 'title', 'labels', 'state'])
issues

	number	title	labels	state
0	18272	Optimize data type	[]	open
1	18271	BUG: Series.rank(pct=True).max() != 1 for a la...	[]	open
2	18270	(Series|DataFrame) datetimelike ops	[]	open
3	18268	DOC: update Seriesbine/DataFramebine d...	[]	open
4	18266	DOC: updated bine_first doc strings	[{'url': '...	open
5	18265	Calling DataFrame.stack on an out-of-order col...	[]	open
6	18264	cleaned up imports	[{'url': '...	open
7	18263	Tslibs offsets paramd	[]	open
8	18262	DEPR: let's deprecate	[{'url': '...	open
9	18258	DEPR: deprecate (Sparse)Series.from_array	[{'url': '...	open
10	18255	ENH/PERF: Add cache='infer' to to_datetime	[{'url': '...	open
11	18250	Categorical.replace() unexpectedly returns non...	[{'url': '...	open
12	18246	pandas.MultiIndex.reorder_levels has no inplac...	[]	open
13	18245	TST: test tz-aware DatetimeIndex as separate m...	[{'url': '...	open
14	18244	RLS 0.21.1	[{'url': '...	open
15	18243	DEPR: deprecate .ftypes, get_ftype_counts	[{'url': '...	open
16	18242	CLN: Remove days, seconds and microseconds pro...	[{'url': '...	open
17	18241	DEPS: drop 2.7 support	[{'url': '...	open
18	18238	BUG: Fix filter method so that accepts byte an...	[{'url': '...	open
19	18237	Deprecate Series.asobject, Index.asobject, ren...	[{'url': '...	open
20	18236	df.plot() very slow compared to explicit matpl...	[{'url': '...	open
21	18235	Quarter.onOffset looks fishy	[]	open
22	18231	Reduce copying of input data on Series constru...	[{'url': '...	open
23	18226	Patch __init__ to prevent passing invalid kwds	[{'url': '...	open
24	18222	DataFrame.plot() produces incorrect legend lab...	[{'url': '...	open
25	18220	DataFrame.groupy renames columns when given a ...	[]	open
26	18217	Deprecate Index.summary	[{'url': '...	open
27	18216	Pass kwargs from read_parquet() to the underly...	[{'url': '...	open
28	18215	DOC/DEPR: ensure that @deprecated functions ha...	[{'url': '...	open
29	18213	Deprecate Series.from_array ?	[{'url': '...	open

6.4 Interacting with Databases(与数据库的交互)

如果在工作中，大部分数据并不会以text或excel的格式存储。最广泛使用的是SQL-based的关系型数据库（SQL Server，PostgreSQL，MySQL）。选择数据库通常取决于性能，数据整合性，实际应用的可扩展性。

读取SQL到DataFrame非常直观，pandas中有一些函数能简化这个过程。举个例子，这里创建一个SQLite数据库，通过使用python内建的sqlite3 driver：

import sqlite3
import pandas as pd

query = """
CREATE TABLE test
(a VARCHAR(20), b VARCHAR(20),c REAL,        d INTEGER
);"""

con = sqlite3.connect('../examples/mydata.sqlite')

con.execute(query)

<sqlite3.Cursor at 0x1049931f0>

conmit()

然后我们插入几行数据：

data = [('Atlanta', 'Georgia', 1.25, 6),('Tallahassee', 'Florida', 2.6, 3),('Sacramento', 'California', 1.7, 5)]

stmt = "INSERT INTO test VALUES(?, ?, ?, ?)"

con.executemany(stmt, data)

<sqlite3.Cursor at 0x1049932d0>

conmit()

大部分python的SQL驱动(PyODBC, psycopg2, MySQLdb, pymssql, 等)返回a list of tuple，当从一个表格选择数据的时候：

cursor = con.execute('select * from test')

rows = cursor.fetchall()

rows

[('Atlanta', 'Georgia', 1.25, 6),('Tallahassee', 'Florida', 2.6, 3),('Sacramento', 'California', 1.7, 5)]

我们可以把list of tuples传递给DataFrame，但是我们也需要column names,包含cursor的description属性：

cursor.description

(('a', None, None, None, None, None, None),('b', None, None, None, None, None, None),('c', None, None, None, None, None, None),('d', None, None, None, None, None, None))

pd.DataFrame(rows, columns=[x[0] for x in cursor.description])

	a	b	c	d
0	Atlanta	Georgia	1.25	6
1	Tallahassee	Florida	2.60	3
2	Sacramento	California	1.70	5

我们不希望每次询问数据库的时候都重复以上步骤，这样对计算机很不好(逐步对计算机系统或文件做小改动导致大的损害)。SQLAlchemy计划是一个六星的Python SQL工具箱，它能抽象出不同SQL数据库之间的不同。pandas有一个read_sql函数，能让我们从SQLAlchemy connection从读取数据。这里我们用SQLAlchemy连接到同一个SQLite数据库，并从之前创建的表格读取数据：

import sqlalchemy as sqla

db = sqla.create_engine('sqlite:///../examples/mydata.sqlite')

pd.read_sql('select * from test', db)

	a	b	c	d
0	Atlanta	Georgia	1.25	6
1	Tallahassee	Florida	2.60	3
2	Sacramento	California	1.70	5