Pass up .limit in all functions

Author: bschilder

NAMESPACE

@@ -3,6 +3,7 @@
 export(add_owner_repo)
 export(description_extract)
 export(github_branches)
+export(github_code)
 export(github_commits)
 export(github_dependencies)
 export(github_dependents)
@@ -13,12 +14,15 @@ export(github_pages)
 export(github_pages_files)
 export(github_pages_vignettes)
 export(github_permissions)
+export(github_repositories)
 export(github_traffic)
+export(github_user_events)
 export(github_workflows)
 export(r_repos)
 export(r_repos_data)
 export(r_repos_downloads)
 export(r_repos_opts)
+export(r_repos_upset_queries)
 importFrom(RCurl,url.exists)
 importFrom(data.table,":=")
 importFrom(data.table,.SD)
@@ -31,6 +35,7 @@ importFrom(data.table,merge.data.table)
 importFrom(data.table,rbindlist)
 importFrom(data.table,setnafill)
 importFrom(data.table,setnames)
+importFrom(data.table,setorderv)
 importFrom(gh,gh)
 importFrom(gh,gh_token)
 importFrom(methods,is)

NEWS.md

@@ -8,6 +8,11 @@
     - `echogithub::readme_header` --> `rworkflows::use_badges`
 * `r_repos_data`
     - Add new arg `add_hex` to find hex URL with `rworkflows::get_hex`.
+* `r_repos`
+    - New arg `queries`
+* New functions:
+    - `github_user_events`
+* Expose `.limit` arg wherever relevant.
 
 ## Bug fixes
 

R/github_code.R

@@ -0,0 +1,40 @@
+#' GitHub code
+#' 
+#' Search for GitHub code using specific queries.
+#' @source 
+#' \href{https://docs.github.com/en/search-github/searching-on-github/searching-code}{
+#' GitHub Docs: Searching code}
+#' @inheritParams github_files
+#' @inheritParams gh::gh
+#' 
+#' @export
+#' @importFrom gh gh gh_token
+#' @importFrom data.table :=
+#' @importFrom methods is
+#' @examples
+#' \dontrun{
+#' ## easily exceeds API limit
+#' repos <- github_code(query="Package path:DESCRIPTION", .limit=5)
+#' }
+github_code <- function(query,
+                        token = gh::gh_token(),
+                        .limit = Inf,
+                        verbose = TRUE){
+    owner_repo <- repo <- NULL;
+    
+    endpoint <- "https://api.github.com/search/code"
+    res <-  gh::gh(endpoint,
+                   .token = token,
+                   .limit = .limit,
+                   q = query,
+                   #### only beta version supports full-on regex ####
+                   # q = "/(?-i)Package/ path:/(?-i)^DESCRIPTION$/", 
+                   per_page = 100)  
+    dt <- gh_to_dt(gh_response = res$items,
+                   verbose = verbose)
+    dt[,owner_repo:=mapply(repo, FUN=function(x){x$name})] 
+    messager("Returning",formatC(nrow(dt),big.mark = ","),
+             "GitHub code files.",v=verbose)
+    return(dt)
+}
+ 

R/github_commits.R

@@ -2,16 +2,19 @@
 #'
 #' Get metadata for all commits to a given GitHub repo.
 #' @inheritParams github_files 
+#' @inheritParams gh::gh
 #' @return A nested list of metadata
 #'
 #' @export 
 #' @importFrom gh gh gh_token
 #' @examples
 #' commits <- github_commits(owner="RajLabMSSM", 
-#'                           repo="echolocatoR")
+#'                           repo="echolocatoR",
+#'                           limit=100)
 github_commits <- function(owner,
                            repo, 
                            token = gh::gh_token(),
+                           .limit = Inf,
                            verbose = TRUE) {
 
     #devoptera::args2vars(github_commits)
@@ -20,7 +23,7 @@ github_commits <- function(owner,
                       owner,repo,"commits",sep="/")
     gh_response <- gh::gh(endpoint = endpoint,
                           .token = token,
-                          .limit = Inf,
+                          .limit = .limit,
                           per_page = 100) 
     dt <- gh_to_dt(gh_response)
     return(dt)

R/github_files.R

@@ -31,6 +31,7 @@
 #' @param verbose Print messages.
 #' @inheritParams github_files_download
 #' @inheritParams base::grepl
+#' @inheritParams gh::gh
 #'
 #' @return A list of paths.
 #'
@@ -47,6 +48,7 @@ github_files <- function(owner,
                          ignore.case = FALSE,
                          method = "gh", 
                          token = gh::gh_token(),
+                         .limit = Inf,
                          download = FALSE,
                          download_dir = tempdir(),
                          overwrite = FALSE,
@@ -77,6 +79,7 @@ github_files <- function(owner,
                               repo = repo,
                               branch = branch,
                               token = token,
+                              .limit = .limit,
                               verbose = verbose)
     } 
     #### Return NULL early ####

R/github_files_gh.R

@@ -1,11 +1,12 @@
 #' List files in GitHub repo: via \pkg{gh}
 #'
 #' Search for files within a public GitHub repository and return their paths.
-#' @inheritParams github_files
 #' @source \href{https://docs.github.com/en/rest/overview/endpoints-available-for-github-apps}{
 #' GitHub endpoints}
 #' @source \href{https://docs.github.com/en/rest/git/trees#get-a-tree-recursively}{
 #' GitHub trees API}
+#' @inheritParams github_files
+#' @inheritParams gh::gh
 #' @returns A \link[data.table]{data.table} fo file paths. 
 #'
 #' @keywords internal
@@ -15,6 +16,7 @@ github_files_gh <- function(owner,
                             repo,
                             branch = c("master","main"),
                             token = gh::gh_token(), 
+                            .limit = Inf,
                             verbose = TRUE) {
     path <- link <- NULL;
 
@@ -29,6 +31,7 @@ github_files_gh <- function(owner,
     )
     gh_response <- gh::gh(endpoint = endpoint,
                           .token = token,
+                          .limit = .limit,
                           per_page = 100) 
     dt <- gh_to_dt(gh_response$tree)
     dt[,link:=paste("https://github.com",owner,repo,

R/github_metadata.R

@@ -4,6 +4,7 @@
 #' @param add_traffic Add traffic metadata 
 #' with \link[echogithub]{github_traffic}.
 #' @inheritParams github_files 
+#' @inheritParams gh::gh
 #' @return A \link[data.table]{data.table}.
 #'
 #' @export
@@ -16,6 +17,7 @@ github_metadata <- function(owner,
                             repo, 
                             add_traffic = FALSE, 
                             token = gh::gh_token(),
+                            .limit = Inf,
                             verbose = TRUE) { 
 
     #devoptera::args2vars(github_metadata)
@@ -37,6 +39,7 @@ github_metadata <- function(owner,
                       owner,repo,sep="/")
     gh_response <- gh::gh(endpoint = endpoint,
                           .token = token, 
+                          .limit = .limit,
                           per_page = 100)
     dt <- cbind(
         owner=owner,

R/github_pages.R

@@ -5,6 +5,7 @@
 #' @param error Throw an error when the GitHub repository does not exist 
 #' (default: \code{TRUE}).
 #' @inheritParams github_files
+#' @inheritParams gh::gh
 #' 
 #' @export
 #' @importFrom gh gh gh_token
@@ -15,13 +16,15 @@ github_pages <- function(owner,
                          repo,
                          error = TRUE,
                          token = gh::gh_token(),
+                         .limit = Inf,
                          verbose = TRUE){
 
     endpoint <- paste("https://api.github.com","repos",
                       owner,repo,"pages",sep="/")
     res <- tryCatch({
         gh::gh(endpoint = endpoint,
                .token = token,
+               .limit = .limit,
                per_page = 100)
     }, error = function(e){return(e)})
     #### Stop if the repo doesn't exist #####

R/github_pages_files.R

@@ -7,11 +7,14 @@
 #' @param local_repo Search a cloned local folder of a git repo instead of 
 #' the remote GitHub repo. If not \code{NULL}, expects a path to the local repo.
 #' @inheritParams github_files
+#' @inheritParams gh::gh
 #' 
 #' @export
 #' @importFrom data.table := fwrite data.table
-#' @examples 
-#' dt <- github_pages_files(owner="RajLabMSSM", repo="echolocatoR")
+#' @examples
+#' dt <- github_pages_files(owner="RajLabMSSM",
+#'                          repo="echolocatoR", 
+#'                          .limit=5)
 github_pages_files <- function(owner,
                                repo,
                                branch = "gh-pages",
@@ -20,6 +23,7 @@ github_pages_files <- function(owner,
                                local_repo = NULL,
                                save_path = NULL,
                                token = gh::gh_token(),
+                               .limit = Inf,
                                verbose = TRUE) { 
 
     link_ghpages_index <- link_ghpages <- link_html <- path <- NULL;
@@ -35,6 +39,7 @@ github_pages_files <- function(owner,
                            query = query,
                            branch = branch, 
                            token = token,
+                           .limit = .limit,
                            verbose = verbose)
     } else {
         dt <- data.table::data.table(

R/github_repositories.R

@@ -0,0 +1,65 @@
+#' GitHub repositories
+#' 
+#' Search for GitHub repositories using specific queries.
+#' @source 
+#' \href{https://docs.github.com/en/rest/search#search-repositories}{
+#' GitHub Docs: Search repositories}
+#' @source
+#' \href{https://docs.github.com/en/rest/search#constructing-a-search-query}{
+#' GitHub Docs: Constructing a search query}
+#' @source 
+#' \href{https://docs.github.com/en/search-github/searching-on-github/searching-for-repositories}{
+#' Searching GitHub repos}
+#' @source 
+#' \href{https://github.com/orgs/community/discussions/9759}{
+#' Case-sensitive GitHub searches}
+#' @source \href{https://github.com/r-lib/gh/pull/136}{
+#' Examples of using gh to search repositories}
+#' @inheritParams github_files
+#' @inheritParams gh::gh
+#' 
+#' @export
+#' @importFrom gh gh gh_token
+#' @importFrom data.table :=
+#' @importFrom methods is
+#' @examples
+#' repos <- github_repositories(query="language:r", .limit=100)
+github_repositories <- function(query,
+                                token = gh::gh_token(),
+                                .limit = Inf,
+                                verbose = TRUE){
+    owner_repo <- repo <- NULL;
+    
+    endpoint <- "https://api.github.com/search/repositories"
+    res <-  gh::gh(endpoint,
+                   .token = token,
+                   .limit = .limit,
+                   per_page = 100,
+                   q = paste(query, collapse = " ")
+                   )  
+    dt <- gh_to_dt(gh_response = res$items,
+                   verbose = verbose)
+    dt[,owner_repo:=mapply(repo, FUN=function(x){x$name})] 
+    messager("Returning",formatC(nrow(dt),big.mark = ","),
+             "GitHub repositories.",v=verbose)
+    return(dt)
+}
+
+# make_query <- function(query, 
+#                        sep = ":", 
+#                        collapse = " ",
+#                        encode=FALSE) {
+#     txt <- paste(names(query),
+#           query,
+#           sep = sep, collapse = collapse)
+#     if(encode) utils::URLencode(txt, reserved=TRUE) else {txt}
+# }
+# endpoint <- paste0("https://api.github.com/",
+#                   "search?q=", make_query(query, encode = TRUE))
+# url <- httr::modify_url("https://api.github.com",
+#                         path = "search/",
+#                         query = list(q = make_query(query)))
+# req <- httr::GET(url = url)
+# cont <- httr::content(req)
+# dt <- gh_to_dt(gh_response = cont$items,
+#                verbose = verbose)

R/github_traffic.R

@@ -3,6 +3,7 @@
 #' Get traffic info on a target GitHub repository.
 #' @param na_fill Value to fill NAs with.
 #' @inheritParams github_files 
+#' @inheritParams gh::gh
 #' @return A \link[data.table]{data.table}.
 #'
 #' @export
@@ -14,6 +15,7 @@
 github_traffic <- function(owner,
                            repo, 
                            token = gh::gh_token(),
+                           .limit = Inf,
                            na_fill = NULL,
                            verbose = TRUE) {
 
@@ -38,7 +40,7 @@ github_traffic <- function(owner,
                       owner,repo,"traffic","clones",sep="/")
     ghr_clones <- gh::gh(endpoint = endpoint,
                           .token = token,
-                          .limit = Inf,
+                          .limit = .limit,
                           per_page = 100)
     clones <- data.table::data.table(
         clones_count=ghr_clones$count,