I have a lot of Jupyter notebooks that contain the following pattern

try:
    import tensorflow as tf
except:
    %pip install tensorflow
    import tensorflow as tf

According to pylint, I should provide a more specific exception object, i.e.

try:
    import tensorflow as tf
except ModuleNotFoundError:
    %pip install tensorflow
    import tensorflow as tf

That's the basic idea, but to be more precise, the notebooks are essentially JSON files, and they actually contain something like

"try:\n",
"    import tensorflow\n",
"except:\n",
"    %pip install tensorflow\n",
"    import tensorflow\n",

Since there are hundreds of notebooks, I cannot go over them manually, so I planned to do a find-and-replace with ag and sed, e.g.

ag -l '^"except:\\\\n",$' | xargs sed -i '' 's/except:/except ModuleNotFoundError:/g'

However, not all except: blocks contain %pip install statements. How can I replace all except: with except ModuleNotFoundError: only if it's followed by a %pip install?

CodePudding user response：

Another option using sed might be:

sed '/except:/{N;/:\n[[:space:]]*%pip install/{s// ModuleNotFoundError&0/}}' file

• When matching except:
• Read (append) the next line to the pattern space using N
• If the current pattern space matches :\n[[:space:]]*%pip install
• Use the last matched pattern // and replace with ModuleNotFoundError followed by the full match &0

This pattern :\n[[:space:]]*%pip install matches the : a newline and optional spaces and then %pip install

Output

try:
    import tensorflow as tf
except ModuleNotFoundError:
    %pip install0 tensorflow
    import tensorflow as tf

CodePudding user response：

Using sed

$ sed '/except:/{N;/%pip install/{s/except:/except ModuleNotFoundError:/}}' input_file
try:
    import tensorflow as tf
except ModuleNotFoundError:
    %pip install tensorflow
    import tensorflow as tf

• Match lines with except:
• If the N; next line contains the string %pip install then
• Match only the except: with %pip install as the next line and change it to except ModuleNotFoundError: