Telegram.R

## https://github.com/lbraglia/telegram
# install.packages('telegram')


# Load Libraries ----------------------------------------------------------
library(telegram)
library(httr)
library(dplyr)
library(tidyr)
library(tibble)
library(glue)



# Tokens ------------------------------------------------------------------

Notion_secret <- Sys.getenv("NOTION_SECRET")
Notion_database <- Sys.getenv("NOTION_DATABASE")
Telegram_token <- Sys.getenv("R_TELEGRAM_BOT_RBot")
Telegram_chatID <- Sys.getenv("R_TELEGRAM_USER_me")

# Pull 'Quotes' DB from Notion --------------------------------------------

getNotionDatabase <- function(secret, database, filters = NULL, show_progress = FALSE, all_pages = TRUE, cover_icon = FALSE){
  options(dplyr.summarise.inform = FALSE) # to supress all the grouping warnings!
  
  # +++++++++ construct headers
  headers = c(`Authorization` = secret, `Notion-Version` = '2022-02-22', `Content-Type` = 'application/json' )
  
  # +++++++++ api call -------------------------------------------------------------
  callAPI <- function(database, headers, filters, cursor){
    res <- httr::POST(url = paste0('https://api.notion.com/v1/databases/', database, '/query'),
                      httr::add_headers(.headers = headers),
                      body = list("filters" = filters,
                                  "start_cursor" = cursor),
                      # start_cursor = cursor,
                      encode = "json")
    if(show_progress){ print(paste0("!! API Call: https://api.notion.com/v1/databases/", database, "/query")) }
    
    return( httr::content(res) )
  }
  
  # +++++++++ this function "flattens" the results into a usable data.frame with 1 row per page (like the real database)
  getItemsAndFlattenIntoDataFrame <- function(results, cover_and_icon = cover_icon){
    if(show_progress){ print(paste0("- flattening into data.frame")) }
    
    # the results (i.e., rows) are extracted into a simple data.frame with value being a list of each item's properties and id's
    items <- tibble::enframe(results)
    
    # now, for each item, we will extract a tidy data.frame where we have all of the columns
    dd <- NULL
    for(i in 1:nrow(items)){
      ## before we tidy up,
      ## add NA's if there is no cover or icon AND we want to based on the option in the parameters of the function
      
      if(cover_and_icon){
        if(is.null(  items[[2]][[i]][["cover"]] )){
          items[[2]][[i]][["cover"]] <- as.logical("FALSE")
        }
        if(is.null( items[[2]][[i]][["icon"]] )){
          items[[2]][[i]][["icon"]] <- as.logical("FALSE")
        }
      }
      
      # this is a tidy dataset with column 1 = name (i.e., value.object.type, etc) and col2 = value (i.e,. d3f0ee76-fc3b-426c-8d23-cff84800b0d6)
      tmp <- tibble::enframe(unlist(items[[i, 2]]))
      
      # to avoid duplicates, (such as two relationships tied to a page) I will condense them into 1 separated by a pipe
      tmp <- tmp %>%
        group_by(name) %>%
        summarise("value" = paste(value, collapse = " | "))
      
      # now, I want to keep this as 1 row in a big data set, so I will pivot_wider
      tmp <- tidyr::pivot_wider(tmp)
      
      # now, I will create one big dataset, I will use dplyr in case columns are not exactly the same, which could be the case if one or various of the properties are missing
      dd <- dplyr::bind_rows(dd, tmp)
    }
    
    return(dd)
  }
  
  if(all_pages){
    if(show_progress){ print(paste0("++++ PAGINATING CALLS: ")) }
    
    # if using all_pages, I will run all of the pagination available.
    new_cursor <- TRUE
    dd <- NULL
    cursor <- NULL
    
    while( new_cursor ){
      if(show_progress){ print(paste0("- cursor: ", cursor, " / new_cursor: ", new_cursor )) }
      
      r <- callAPI(database = database,
                   headers = headers,
                   filters = filters,
                   cursor = cursor)
      
      new_cursor <- r$has_more
      cursor <- r$next_cursor
      
      # stack the data.frames together
      tmp <- getItemsAndFlattenIntoDataFrame( r$results )
      
      dd <- dplyr::bind_rows(dd, tmp)
      
      if(show_progress){ print(paste0("- nrow of downloads: ", nrow(dd) )) }
    }
  }else{
    # no pagination, just the top 100
    if(show_progress){ print(paste0("++++ NO PAGINATION: ")) }
    cursor <- NULL
    r <- callAPI(database = database,
                 headers = headers,
                 filters = filters,
                 cursor = cursor)
    
    dd <- getItemsAndFlattenIntoDataFrame( r$results )
  }
  
  return(dd)
}

df <- getNotionDatabase(secret = Notion_secret, database = Notion_database)



# Sample 1 quotes ---------------------------------------------------------

data_s1 <- df[sample(1:nrow(df), 1), ]  # Sample rows of data with Base R
quotes <- data_s1$properties.Quote.title.plain_text

d <- as.data.frame(quotes)

Quote1 <- d$quotes[1]
# Quote2 <- d$quotes[2]

# Telegram_message <- glue({Quote1},{Quote2}, .sep = ". \n \n")

Telegram_message <- Quote1

# Telegram Message --------------------------------------------------------

## Create the bot object
bot <- TGBot$new(token = Telegram_token)
## Set Chat ID
bot$set_default_chat_id(Telegram_chatID)

#Send Message

bot$sendMessage(Telegram_message)