问题
What would be a proper way to extract function calls and their occuring line in one or more R scripts? Is there a parsing base function or package that allows me to do this or should I build a solution with regular expressions?
For example:
function_calls("project1/exploratory_analysis.R")
should output a dataframe like:
## function line filename
## 1 tapply 35 exploratory_analysis.R
## 2 qplot 80 exploratory_analysis.R
What I want to achieve finally is to build a reverse index of function calls and loaded packages as used in one or more R scripts, for educational and reference purposes. (e.g. used as a repository with examples of usage). For example:
--------------------------------------------------------
| function | source_file | line | package |
|:--------:|:----------------------:|:-----:|:--------:|
| tapply | exploratory_analysis.R | 35 | base |
| qplot | exploratory_analysis.R | 80 | ggplot2 |
| cor | regression.R | 15 | stats |
| cor | regression.R | 27 | stats |
| tapply | regression.R | 12 | base |
| fromJSON | load_dataset.R | 5 | jsonlite |
| %>% | transformation.R | 10 | magrittr |
--------------------------------------------------------
One could use some regular expressions to extract function calls but I was wondering whether there exists a parsing, static code analysis or reflection facility supporting this task.
I suppose combining parse(), substitute(), getParseData(), deParse(), srcfile() (or maybe functions in the packages mvbutils, codetools) would make it possible, but I'm not familiar enough with their usage to figure it out on my own.
来源：https://stackoverflow.com/questions/33064376/extract-information-on-usage-of-functions-and-packages-in-r-scripts