Quick script to generate Open Know-How manifests out of a Google Form

manifest-maker.R

require(googledrive)

url <- ""


# Download form responses to a .csv file
filename <- "okh_data.csv"
drive_download(file = as_id(url),
               path = filename,
               overwrite = TRUE)

# Read file and rename columns 
# to something more readable
okh_row_names <- c(
  "timestamp",
  "doc_url",
  "project_url",
  "title",
  "description",
  "date_published",
  "doc_language",
  "username",
  "doc_contains",
  "built",
  "sdg",
  "image_filename",
  "intended_use",
  "designer_name",
  "affiliation",
  "country",
  "version",
  "built_independently",
  "derivative_of",
  "variant_of",
  "schematics_url",
  "bom_url",
  "manufacturing_files_url",
  "tool_list_url",
  "main_materials",
  "estimated_cost",
  "estimated_cost_currency"
)
okh_data <- read.csv(filename,
                     header=T,
                     fill=T,
                     stringsAsFactors=F
)
colnames(okh_data) <- okh_row_names

# Transforming timestamp into a workable date
okh_data$date <- as.Date(strptime(okh_data$timestamp, "", tz="America/El_Salvador"))
okh_data$date_published_2 <- as.Date(strptime(okh_data$date_published, "%m/%d/%Y", tz="America/El_Salvador"))
okh_data$date_published_3 <- format(okh_data$date_published_2, "%Y")


# Images
image_links <- paste0("https://appropedia.org/File:", okh_data$image_filename)

# Web crawler to retrieve an image url
# http://mazamascience.com/WorkingWithData/?p=912
image_url <- vector()
for(i in 1:length(image_links)){
  image_url[i] <- tryCatch({
    session <- html_session(image_links[i])
    session %>%
      html_nodes(".fullImageLink a") %>% 
      html_attr("href") %>% 
      head(1)
  }, warning = function(w){
    NA
  }, error = function(x){
    NA
  })
}

okh_data$image_filename_full <- ifelse(!is.na(image_url), paste0("https://appropedia.org", image_url), NA)

# This function concatenates content into 
# a manifest as a YAML file.
apply(okh_data, 1, function(test) cat("# Open know-how manifest v0.1
---
", 
  "date-created: ", ifelse(is.na(test["date"]), "", as.character(test["date"])), "\n",
  "date-updated: ", as.character(Sys.Date()), "\n",
  "manifest-author:\n  name: Kathy Nativi\n  affiliation: Appropedia\n  email: info@appropedia.org\nmanifest-language: eng-us\n",
  "documentation-language: ", test["doc_language"], "\n",
  "\n# Properties\n",
  "title: ", gsub('"', '', test["title"]), "\n",
  if(nchar(test["description"]) > 0 && !is.na(test["description"])){paste0("description: |\n  ", test["description"], "\n")},
  if(nchar(test["intended_use"]) > 0 && !is.na(test["intended_use"])){paste0("intended-use: |\n  ", test["intended_use"], "\n")},
  if(nchar(test["variant_of"]) > 0 && !is.na(test["variant_of"])){paste0("keywords: \n  - ", gsub(", ", "\n  - ", tolower(test["variant_of"])), "\n")},
  if(nchar(test["project_url"]) > 0 && !is.na(test["project_url"])){paste0("project-link: ", test["project_url"], "\n")},
  if(nchar(test["username"]) > 0 && !is.na(test["username"])){paste0("contact:\n  name: ", test["username"], "\n")},
  # if(nchar(test["username"]) > 0 && !is.na(test["username"])){paste0("contact:\n  name: ", gsub(" ", "_", test["username"]), "\n")},
  if(nchar(test["affiliation"]) > 0 && !is.na(test["affiliation"])){paste0("  affiliation: ", test["affiliation"], "\n")},
  if(!is.na(test["image_filename_full"])){paste0("image: ", test["image_filename_full"], "\n")},
  ifelse(test["built"]=="Yes", "made: FALSE\n", "made: TRUE\n"),
  ifelse(test["built_independently"] == "TRUE", "", paste0("made-independently: TRUE\n")),
  if(nchar(test["derivative_of"]) > 0 && !is.na(test["derivative_of"])){paste0("derivative-of: \n  ", test["derivative_of"], "\n")},
  "license:\n  documentation: CC-BY-SA 3.0\n\n",
  "# Documentation\n",
  if(nchar(test["doc_url"]) > 0 && !is.na(test["doc_url"])){paste0("documentation-home: ", test["doc_url"], "\n\n")},
  "# Appropedia-specific fields\n",
  if(nchar(test["date_published"]) > 0 && !is.na(test["date_published"])){paste0("date-published: ", test["date_published_2"], "\n")},
  if(nchar(test["sdg"]) > 0 && !is.na(test["sdg"])){paste0("sustainable-development-goals: \n  - ", gsub(", SDG", "\n  - SDG", test["sdg"]), "\n")},
sep = "", file = paste0("R/manifests/okh-", gsub('"|/', '', test["title"]), ".yml")
))