def parse_data(src):
# HTML解析
et = html.fromstring(src)
# 整理数据
product_items_list = et.xpath(
"//div[@class='list-product']//div[@class='plp-slide']")
final_list = []
for i in product_items_list:
data = {}
data["img_box_src"] = i.xpath(".//div[@class='img-box']//img/@lazysrc")
data["img_box_src"] = data["img_box_src"][0] if data[
"img_box_src"] else ""
data["goods_tit"] = i.xpath(".//p[@class='goods-tit']/a/text()")
data["goods_tit"] = data["goods_tit"][0] if data["goods_tit"] else ""
data["goods_introudce"] = i.xpath(
".//p[@class='goods-introudce']/a/text()")
data["goods_introudce"] = data["goods_introudce"][0] if data[
"goods_introudce"] else ""
goods_classify = i.xpath(".//div[@class='goods-classify']//span")
gc_list = data["goods_classify"] = []
for gc in goods_classify:
dgc = {}
dgc["title"] = gc.xpath("./img/@title")
dgc["title"] = dgc["title"][0] if dgc["title"] else ""
dgc["title"] = dgc["title"].replace('\xa0', ' ')
dgc["code"] = gc.xpath("./@data-code")
dgc["code"] = dgc["code"][0] if dgc["code"] else ""
dgc["saleprice"] = gc.xpath("./@data-saleprice")
dgc["saleprice"] = dgc["saleprice"][0] if dgc["saleprice"] else ""
dgc["img_src"] = gc.xpath("./img/@src")
dgc["img_src"] = dgc["img_src"][0] if dgc["img_src"] else ""
# 解析SKU颜色值
if dgc["img_src"]:
req_img = requests.get(dgc["img_src"], verify=False)
img_data = req_img.content
bio = BytesIO()
bio.write(img_data)
bio.seek(0)
pimg = Image.open(bio) # 读入PIL图像
pimg.thumbnail((1, 1)) # 转换为1x1像素的图片
r, g, b = pimg.getcolors(
pimg.size[0] *
pimg.size[1])[0][1] # 形式：[(1, (223, 218, 212))]
dgc["img_color"] = '#%02x%02x%02x' % (r, g, b)
pimg.close()
bio.close()
else:
dgc["img_color"] = ""
gc_list.append(dgc)
final_list.append(data)
return final_list
class BaseBitcoinClient(object):
def __init__(self, socket):
self.socket = socket
self.buffer = BytesIO()
self.stop_client = False
def close_stream(self):
self.socket.close()
def send_message(self, message):
self.socket.sendall(message.to_bytes())
def handshake(self):
# Send a "version" message to start the handshake
msg = msg_version()
# See BIP 111 (https://github.com/bitcoin/bips/blob/master/bip-0111.mediawiki)
msg.nVersion = 70011
msg.fRelay = False # If false then broadcast transactions will not be announced until a filter{load,add,clear} command is received
self.send_message(msg)
def handle_version(self, _):
# Respond with a "verack" message to a "version" message
msg = msg_verack()
self.send_message(msg)
def handle_ping(self, ping_message):
# Respond with a pong message to a ping message
msg = msg_pong()
msg.nonce = ping_message.nonce
self.send_message(msg)
def run(self):
while self.stop_client != True:
# Read and store the data from the socket
data = self.socket.recv(64)
self.buffer.write(data)
try:
# Go at the beginning of the buffer
self.buffer.seek(0)
# Deserialize the message
message = MsgSerializable().stream_deserialize(self.buffer)
# Reset the buffer
remaining = self.buffer.read()
self.buffer = BytesIO()
self.buffer.write(remaining)
# Call handle function
if message is not None:
handle_func_name = "handle_" + message.command.decode(
"utf-8")
handle_func = getattr(self, handle_func_name, None)
if handle_func:
handle_func(message)
except SerializationTruncationError:
# Read more data from the socket
pass
def history_tick_content(contract, date):
response = requests.get(hist_tick_url(contract, date), stream=True)
if response.status_code == 200:
disposition = response.headers['Content-Disposition']
bio = BytesIO(b"")
chunk_size = 2**16
with click.progressbar(response.iter_content(chunk_size),
label=disposition) as bar:
for content in bar:
bio.write(content)
bio.seek(0)
return bio.read()
else:
raise IOError(response.status_code)
def func4():
"""
StringIO顾名思义就是在内存中读写str。
"""
f = StringIO("可以这样初始化#\t#\t")
# f = StringIO()
f.write("HelloWorld!") # 后面写入会覆盖初始化
print(f.getvalue()) # getvalue()方法用于获得写入后的str。
"""
StringIO操作的只能是str，如果要操作二进制数据，就需要使用BytesIO
"""
fb = BytesIO()
# f = BytesIO(b'\xe4\xb8\xad\xe6\x96\x87')#也可以这样初始化
fb.write("测试中文".encode(encoding='utf_8'))
print(fb.getvalue())
pass
def decompress(self, input):
header = lz4Header().read(input)
table = lz4Table().read(input, header.chunkCount)
input.seek(header.headerSize, 0)
data = bytes(header.headerSize + sum(chunk.decompressedChunkSize
for chunk in table))
memoryStream = BytesIO(data)
memoryStream.seek(header.headerSize, 0)
for chunk in table:
saveChunk = chunk.read(input)
#print(saveChunk)
memoryStream.write(saveChunk)
memoryStream.seek(header.headerSize, 0)
return memoryStream
async def run_wsgi_feeder(self, scope: Scope, receive: Receive,
fut: Awaitable[None], input_io: BytesIO):
while not fut.done():
event = await receive()
if event["type"] == "http.disconnect":
input_io.close()
break
# Use exponential back-off until each attempt is 30 seconds apart.
b = 0
while len(input_io.getbuffer()) > 1024 * 1024:
await asyncio.sleep(0.1 * (1.1**b))
if b < 60:
b += 1
input_io.write(event.get("body", b""))
if not event.get("more_body", False):
input_io.close()
break
def useByteIO():
b = BytesIO()
b.write('中文'.encode(encoding='utf_8'))
printStrIO(b)
'''
Created on 2017年6月21日
@author: admin
'''
from io import StringIO #StringIO模块的作用是在内存中读写
from _io import BytesIO
f = StringIO()
s = f.write('wskd fkdsa ')
print(f.getvalue()) #获取写入的值
print(s) #获取长度
n = BytesIO()
n.write('中文'.encode('utf_8')) #写入的不是str类型，而是经过utf-8编码的。
#n.write('中文'.encode(encoding='utf_8', errors='strict'))
print(n)
print(n.getvalue())
data = '人闲桂花落，夜静春山空。月出惊山鸟，时鸣春涧中。'.encode('utf-8')
n = BytesIO(data)
print(n)
print(n.read())
def create_instream(self, bdata):
buff = BytesIO()
buff.write(bdata)
buff.seek(0)
return buff
from _io import StringIO, BytesIO
f = open('E:/a.txt', 'r')
for line in f.readlines():
print(line.strip())
f.close()
#readline 和 readlines
#像open()函数返回的这种有个read()方法的对象，
#在Python中统称为file-like Object
with open('E:/a.txt', 'w') as f:
f.write("en")
f = StringIO()
f.write("hello")
f.write(" ")
f.write("world")
print(f.getvalue())
b = BytesIO()
b.write("中文".encode(encoding='utf_8', errors='strict'))
str = b.getvalue()
print(str)
def useByteIO():
b = BytesIO()
b.write('中文'.encode(encoding='utf_8'))
printStrIO(b)
r.setex("MessageIndex" + localuu + "Timestamp",
datey.isoformat() + "T" + timestamp, 90)
CurResult = r.get("MessageIndex" + localuu + "ResultDataPayload")
timestamp = time.strftime(" %H:%M:%S")
timefinished = time.time()
timediff = timefinished - timestart
r.setex("MessageIndex" + localuu + "ResultDataPayload",
"CONTENT REMOVED: " + "duration ; {%f}" % timediff, 480)
r.setex("MessageIndex" + localuu + "Status",
"FINISHED_CONTENT_EXTRACTION", 90)
r.setex("MessageIndex" + localuu + "Timestamp",
datey.isoformat() + "T" + timestamp, 90)
#finaldata = base64.b64decode(CurResult)
bin_fromstr = binascii.a2b_base64(CurResult)
uiee = BytesIO()
uiee.write(bin_fromstr)
new_img = Image.open(uiee)
new_img.save("Blu_QR_RESULT" + localuu[1:6] + ".png", format='PNG')
#new_qr = qrcode.QRCode(
#version=1,
#error_correction=qrcode.constants.ERROR_CORRECT_L,
#box_size=24,
#border=4,
#)
#newqr_i = new_qr.make()
#finaldata = binascii.a2b_base64(CurResult)
#fs = io.open('cn.txt','w')
#print (finaldata)
#fs.flush
#fs.close
#qr = qrcode.QRCode(
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
'''
Created on 2018年11月13日
@author: zzk
'''
# StringIO和BytesIO是在内存中操作str和bytes的方法，使得和读写文件具有一致的接口。
# StringIO顾名思义就是在内存中读写str。
from io import StringIO
from _io import BytesIO
f=StringIO('hello\nworld\n zzk')
f2=StringIO()
f2.write('my\nname\nis\nzzk')
print(f2.getvalue())
print('====================')
while True:
s=f.readline()
if s=='':
break
print(s.strip())
print('====================')
# BytesIO
fb = BytesIO()
fb.write('哈哈哈'.encode('utf-8'))
print(fb.getvalue())
fb2 = BytesIO(b'\xe5\x93\x88\xe5\x93\x88\xe5\x93\x88')
print(fb2.read().decode('utf-8'))
'''
@author: xilh
@since: 20200128
'''
from _io import BytesIO
from demo.tools.tool import pline
# 创建对象
f = BytesIO()
f.write('hello'.encode(encoding='utf_8'))
f.write(' '.encode(encoding='utf_8'))
f.write('world'.encode(encoding='utf_8'))
f.write(' '.encode(encoding='utf_8'))
f.write('张三'.encode(encoding='utf_8'))
# 获取值
ret = f.getvalue()
print(ret)
pline()
f = BytesIO('张三是程序员'.encode('utf-8'))
print(f.readline())
# 关闭
f.close()
