README.Rmd

---
output: github_document
---

<!-- README.md is generated from README.Rmd. Please edit that file -->

```{r, include = FALSE}
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%",
  message = FALSE,
  warning = FALSE
)
library(functiondepends)
```

# functiondepends <img src="man/figures/logo.png" align="right" width="120" />

<!-- badges: start -->
[![R build status](https://github.com/jakubsob/functiondepends/workflows/R-CMD-check/badge.svg)](https://github.com/jakubsob/functiondepends/actions)
[![license](https://img.shields.io/badge/license-mit-lightgrey.svg)](https://choosealicense.com/)
[![CRAN_Status_Badge](https://www.r-pkg.org/badges/version/functiondepends)](https://cran.r-project.org/package=functiondepends)
[![CRAN_latest_release_date](https://www.r-pkg.org/badges/last-release/functiondepends)](https://cran.r-project.org/package=functiondepends)
[![CRAN status](https://cranlogs.r-pkg.org/badges/grand-total/functiondepends)](https://CRAN.R-project.org/package=functiondepends)
<!-- badges: end -->

The goal of functiondepends is to allow for tidy exploration of unstructured codebase without evaluation of code. 

## Installation

One can install `functiondepends` from CRAN:

```{r, eval = FALSE}
install.packages("functiondepends")
```

or development version from GitHub:

```{r, eval = FALSE}
# install.packages("devtools")
devtools::install_github("jakubsob/functiondepends")
```

## Examples

```{r, include = FALSE}
envir <- functiondepends:::envir
functions <- functiondepends:::functions
```

```{r, eval = FALSE}
library(functiondepends)
# Create environment for loaded functions 
envir <- new.env()
# Search recursively current directory
functions <- find_functions(".", envir = envir, recursive = TRUE)
```

```{r}
functions
```

Search for dependencies of function `find_functions` within parsed functions:

```{r}
dependency <- find_dependencies("find_functions", envir = envir, in_envir = TRUE)
dependency
```

Note that `SourceNamespace` column has value `user-defined` as the functions are searched within source of the package.

Search for all dependencies of `find_functions` function:

```{r functions_in_path, fig.height=4}
library(ggplot2)
library(dplyr)

dependency <- find_dependencies("find_functions", envir = envir, in_envir = FALSE)
dependency %>% 
  slice_max(SourceRep, n = 10) %>% 
  mutate(Source = reorder(Source, SourceRep)) %>% 
  ggplot(aes(x = Source, y = SourceRep, fill = SourceNamespace)) +
  geom_col() +
  coord_flip() +
  labs(caption = "Top 10 most repeated calls in 'find_functions'.")
```

Note that name `df` is often used to store object of type `data.frame`. `df` is also a name of F distribution density function from `stats` package. If you suspect that given function ought not to use a specific package, see the source code of function to check the context. To do so, one can execute `find_dependencies` function with `add_info` argument set to `TRUE`.

```{r}
library(tidyr)

dependency <- find_dependencies("find_functions", envir = envir, in_envir = FALSE, add_info = TRUE)
dependency %>% 
  filter(SourceNamespace == "stats") %>% 
  select(Source, SourcePosition, SourceContext) %>% 
  unnest(c(SourcePosition, SourceContext)) 
```

One can see that indeed `df` is not a call to function `stats::df`.

```{r target_degree, fig.height=4}
dependency <- find_dependencies(unique(functions$Function), envir = envir, in_envir = FALSE)
dependency %>% 
  distinct(Target, TargetInDegree) %>%
  mutate(Target = reorder(Target, TargetInDegree)) %>%
  ggplot(aes(x = Target, y = TargetInDegree)) +
  geom_col() +
  coord_flip() + 
  labs(caption = "Functions with most function calls.")
```

```{r namespace_count, fig.height=4}
dependency <- find_dependencies(unique(functions$Function), envir = envir, in_envir = FALSE)
dependency %>% 
  group_by(SourceNamespace) %>% 
  tally(name = "Count") %>% 
  slice_max(Count, n = 10) %>% 
  mutate(SourceNamespace = reorder(SourceNamespace, Count)) %>% 
  ggplot(aes(x = SourceNamespace, y = Count)) +
  geom_col() +
  coord_flip() +
  labs(caption = "Top 10 used namespaces.")
```

See which user-defined functions depend most on other user-defined functions within searched codebase.

```{r}
dependency <- find_dependencies(unique(functions$Function), envir = envir, in_envir = TRUE)
dependency %>% 
  distinct(Target, TargetInDegree) %>% 
  arrange(-TargetInDegree)
```

```{r network_env, fig.height=4}
library(igraph)

edges <- dependency %>% 
  select(Source, Target) %>% 
  na.omit()

vertices <- unique(c(dependency$Source, dependency$Target))
vertices <- vertices[!is.na(vertices)]

g <- graph_from_data_frame(d = edges, vertices = vertices)
deg <- degree(g, mode = "in")
V(g)$size <- deg * 10 + 5
V(g)$label.cex <- (degree(g, mode = "in", normalized = TRUE) + 1)

plot(
  g,
  vertex.color = "grey",
  edge.color = "grey",
  edge.arrow.size = .4,
  main = "Functions dependency graph"
)
```


```{r network, fig.height=4}
dependency <- find_dependencies(unique(functions$Function), envir = envir, in_envir = FALSE)
edges <- dependency %>% 
  select(Source, Target) %>% 
  na.omit()
vertices <- unique(c(edges$Source, edges$Target))

g <- graph_from_data_frame(edges)
deg <- degree(g, mode = "in")
V(g)$size <- deg
V(g)$label.cex <- (degree(g, mode = "in", normalized = TRUE) + 1) / 1.8

plot(
  g,
  vertex.color = "grey",
  edge.color = "grey",
  edge.arrow.size = .4,
  main = "Full functions dependency graph"
)
```