In urswilke/pyramidi: Generate and Manipulate Midi Data in R Data Frames

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

pyramidi::install_miditapyr(envname = "r-reticulate")

The following diagram depicts the structure of the pyramidi package and its python counterpart miditapyr.

Hover over nodes to see additional information. Many of the nodes are also clickable for further information.

library(pyramidi)
library(stringr)
library(purrr)
library(DiagrammeR)

mt_funs <- c("frame_midi", "unnest_midi", "nest_midi", "write_midi")

mt_fun_descrs <- map_chr(mt_funs, ~miditapyr[[.x]]$`__doc__`) %>% 
  str_split("\\n\\n") %>% 
  map_chr(1) %>% 
  str_replace_all("\\n", " ") %>% 
  str_squish() %>% 
  str_remove_all(":.*?:") %>% 
  str_remove_all("[`~]") 


pm_funs <- c(
  "pivot_long_notes",
  "pivot_wide_notes",
  "tab_measures",
  "merge_midi_frames",
  "split_midi_frame"
)
# inspired by:
# https://stackoverflow.com/a/8928894 &
# https://stackoverflow.com/a/23897916
get_help <- function(...){
    thefile <- help(...)
    capture.output(
        tools:::Rd2txt(utils:::.getHelpFile(thefile))
    )
}


pm_fun_descrs_raw <- map(pm_funs, get_help) %>% 
  map(~.x[-c(1:4)])


title_until_line_ints <- pm_fun_descrs_raw %>% 
  map_dbl(~str_which(.x, "^$") %>% min() %>% {. - 1})

pm_fun_descrs <- map2_chr(
  pm_fun_descrs_raw,
  title_until_line_ints,
  ~ .x[1:.y] %>% paste(collapse = " ") %>% str_squish() %>% str_remove_all("[`'\"’‘]")
)
gv <- grViz("package_workflow.gv")

# chunk options from here: https://stackoverflow.com/a/43213737
gv

miditapyr can be used to load the midi file midi_in into a mido.MidiFile. The chain down to midi_out shows, how miditapyr's functions can be used to transform the data into different formats and finally save it to a new midi file. Have a look here to see in detail how this can be done.
mf shows the structure of a MidiFrames object and how its fields are related to the low-level functions. Have a look here to see how MidiFrames can be used in python.
The pyramidi subgraph illustrates how the functions in this package work. Starting from the dataframe midi_frame_unnested, the data is transformed to different formats, finally resulting in the same format.
The dashed borders of the nodes in the graph depict the fields that are in a MidiFramer object (for clarity not included in the graph).
The MidiFramer method update_nodes_wide() can be used to modify the field df_notes_wide. The red arrows and borders show the other fields of the MidiFramer object that are automatically updated when using this method.