无法导入系统或 io

Question

我正在尝试导入sys和io:

import sys
import io

但我得到以下错误：

Error while compiling (offending file last):
File '/Users/username/worktest/index.py', line 11, at import of:
File '/Users/username/worktest/env/lib/python3.7/site-packages/numpy/__init__.py', line 293, at import of:
File '/Users/username/opt/anaconda3/lib/python3.7/os.py', line 1026, at import of:
File 'io', line 95, namely:

Import error, can't find any of:
/Users/username/worktest/_io/_WindowsConsoleIO.py
/Users/username/worktest/_io/_WindowsConsoleIO.js
/Users/username/worktest/env/lib/python3.7/site-packages/transcrypt/modules/_io/_WindowsConsoleIO.py
/Users/username/worktest/env/lib/python3.7/site-packages/transcrypt/modules/_io/_WindowsConsoleIO.js
/Users/username/worktest/_io/_WindowsConsoleIO.py
/Users/username/worktest/_io/_WindowsConsoleIO.js
/Users/username/opt/anaconda3/lib/python3.7/_io/_WindowsConsoleIO.py
/Users/username/opt/anaconda3/lib/python3.7/_io/_WindowsConsoleIO.js
/Users/username/opt/anaconda3/lib/python3.7/lib-dynload/_io/_WindowsConsoleIO.py
/Users/username/opt/anaconda3/lib/python3.7/lib-dynload/_io/_WindowsConsoleIO.js
/Users/username/worktest/env/lib/python3.7/site-packages/_io/_WindowsConsoleIO.py
/Users/username/worktest/env/lib/python3.7/site-packages/_io/_WindowsConsoleIO.js
/Users/username/worktest/_io.py
/Users/username/worktest/_io.js
/Users/username/worktest/env/lib/python3.7/site-packages/transcrypt/modules/_io.py
/Users/username/worktest/env/lib/python3.7/site-packages/transcrypt/modules/_io.js
/Users/username/worktest/_io.py
/Users/username/worktest/_io.js
/Users/username/opt/anaconda3/lib/python3.7/_io.py
/Users/username/opt/anaconda3/lib/python3.7/_io.js
/Users/username/opt/anaconda3/lib/python3.7/lib-dynload/_io.py
/Users/username/opt/anaconda3/lib/python3.7/lib-dynload/_io.js
/Users/username/worktest/env/lib/python3.7/site-packages/_io.py
/Users/username/worktest/env/lib/python3.7/site-packages/_io.js

Aborted

👀  ./index.py
Module build failed (from ./node_modules/transcrypt-loader/__target_es5__/index.js):
Error: Command failed: python3 -m transcrypt --nomin --map --verbose "index"
at checkExecSyncError (child_process.js:630:11)
at Object.execSync (child_process.js:666:15)
at Object.main (/Users/username/worktest/node_modules/transcrypt-loader/__target_es5__/index.js:56:67)
Error: webpack returned an error. Try configuring `entry` in your webpack config relative to the current working directory, or setting `context = __dirname` in your webpack config.

我正试图使用Transcrypt和Wrangler让我的Python脚本在Cloudflare的Workers上运行，但无法导入这些模块(pypi上不存在(的问题阻止了我这么做。

在pypi中，有没有任何替代方案可以让我使用？如果没有，如何在我的路径中本地安装它们。

它们的实际用途如下：

old_stdout = sys.stdout # Memorize the default stdout stream
sys.stdout = buffer = io.StringIO()
for row in query_job:
# Row values can be accessed by field name or index.
print("col1={}, col2={}, col3={}".format(row[0], row[1], row[2]))
sys.stdout = old_stdout # Put the old stream back in place
out1 = buffer.getvalue() # Return a str containing the entire contents of the buffer.

其被执行以收集来自Google BigQuery查询的输出。这是我能找到的最简单的方法来实际检索输出，但我也愿意修改它。

Answer 1

您使用非常非常奇怪的方法来创建字符串。

您不需要使用print()将其作为字符串保存在变量中，因此您不需要io.String()

---

更流行的是将所有字符串保留在列表中，然后使用"n"加入它们

all_rows = []
for row in query_job:
all_rows.append( "col1={}, col2={}, col3={}".format(row[0], row[1], row[2]) )
out1 = 'n'.join(all_rows)

哪个可以写为列表理解

all_rows = [ "col1={}, col2={}, col3={}".format(row[0], row[1], row[2]) for row in query_job ]
out1 = 'n'.join(all_rows)

---

不太流行的是直接将字符串与+=和n一起使用

out1 = ""
for row in query_job:
out1 += "col1={}, col2={}, col3={}".format(row[0], row[1], row[2])
out1 += "n"

您甚至可以将"n"直接放在"col1={}, col2={}, col3={}n"中

out1 = ""
for row in query_job:
out1 += "col1={}, col2={}, col3={}n".format(row[0], row[1], row[2])

---

BTW:如果row中只有3个元素，则可以在.format(*row)中使用*

out1 = ""
for row in query_job:
out1 += "col1={}, col2={}, col3={}n".format(*row)

或具有列表理解

all_rows = [ "col1={}, col2={}, col3={}".format(*row) for row in query_job ]
out1 = 'n'.join(all_rows)

甚至在一行中

out1 = 'n'.join([ "col1={}, col2={}, col3={}".format(*row) for row in query_job ])

---

BTW:

如果必须使用带缓冲区(或带文件处理程序(的print()，则不必替换sys.stdout，但可以使用print(..., file=buffer)

buffer = io.StringIO()
for row in query_job:
print("col1={}, col2={}, col3={}".format(row[0], row[1], row[2]), file=buffer)
out1 = buffer.getvalue() # 

但我不知道它是否适用于Transcrypt(和Wrangler(，因为Transcrypt可能只有标准模块的一部分——只有可以转换为JavaScript的模块。有些函数可能无法转换，因为JavaScript无法访问您的磁盘(出于安全原因(，也无法使用您的文件。

Answer 2

能够重新使用pandas收集输出的以前(效率很低(的方法：

df = pd.read_gbq(sql, project_id=project_id)

更快、更简单的解决方案，这是我以前不知道的。目前仍然与Transcrypt不兼容，但至少解决了这一部分。