From e038d86339e35bf5a6b61214c3e6113032209727 Mon Sep 17 00:00:00 2001
From: flavis <christian@flavigny.de>
Date: Tue, 10 Aug 2021 12:42:41 +0200
Subject: [PATCH] add check for valid structure of tables

---
 DESCRIPTION |  1 +
 R/analyze.R |  3 +++
 R/helpers.R | 35 +++++++++++++++++++++++++++++++++++
 3 files changed, 39 insertions(+)

diff --git a/DESCRIPTION b/DESCRIPTION
index a0170ad..3ef2aa8 100644
--- a/DESCRIPTION
+++ b/DESCRIPTION
@@ -16,6 +16,7 @@ Roxygen: list(markdown = TRUE)
 RoxygenNote: 7.1.1
 Imports: 
     dplyr,
+    lubridate,
     pbapply,
     purrr,
     rvest,
diff --git a/R/analyze.R b/R/analyze.R
index 4570a02..984c897 100644
--- a/R/analyze.R
+++ b/R/analyze.R
@@ -7,6 +7,7 @@
 #' 
 #' @export
 find_word <- function(res, word) {
+    is_valid_res(res)
     talks <- res$talks
     mutate(
         talks, 
@@ -28,6 +29,7 @@ find_word <- function(res, word) {
 #'
 #' @export
 join_speaker <- function(tb, res, fraction_only = F) {
+    is_valid_res(res)
     joined <- left_join(tb, res$speaker, by=c("speaker" = "id"))
     if (fraction_only) select(joined, "fraction")
     else joined
@@ -145,6 +147,7 @@ bar_plot_fractions <- function(tb,
 #'
 #' @export
 word_usage_by_date <- function(res, patterns, tidy=F) {
+    is_valid_res(res)
     tb <- res$talks
     nms <- names(patterns)
     for (i in seq_along(patterns)) {
diff --git a/R/helpers.R b/R/helpers.R
index 863852c..db097dd 100644
--- a/R/helpers.R
+++ b/R/helpers.R
@@ -24,3 +24,38 @@ make_directory_path <- function(path) {
     if (!str_ends(path, .Platform$file.sep)) str_c(path, .Platform$file.sep)
     else path
 }
+
+# check if res is of expected format
+is_valid_res <- function(res) {
+    stopifnot("Data is missing relevant tables. Is this a return value of read_all or repair?"
+              = all(c("speaker", "speeches", "talks", "comments", "applause") %in% names(res)))
+    stopifnot("Some entries of res are no tibbles."
+              = all(sapply(res, typeof) == "list" & "tbl" %in% sapply(res, class)))
+    stopifnot("Speaker table is of wrong format."
+              = all(c("id", "prename", "lastname", "fraction", "title", "role_short", "role_long")
+                    %in% names(res$speaker)) &&
+                all(sapply(res$speaker, is.character)))
+    stopifnot("Speeches table is of wrong format."
+              = all(c("id", "speaker", "date") %in% names(res$speeches)) &&
+                is.character(res$speeches$id) &&
+                is.character(res$speeches$speaker) &&
+                lubridate::is.Date(res$speeches$date))
+    stopifnot("Talks table is of wrong format."
+              = all(c("speech_id", "speaker", "content") %in% names(res$talks)) &&
+                all(sapply(res$talks, is.character)))
+    stopifnot("Comments table is of wrong format."
+              = all(c("speech_id", "on_speaker", "fraction", "commenter", "content")
+                    %in% names(res$comments)) &&
+                all(sapply(res$comments, is.character)))
+    stopifnot("Applause table is of wrong format."
+              = all(c("speech_id", "on_speaker", "CDU_CSU", "SPD", "FDP", "DIE_LINKE", "BUENDNIS_90_DIE_GRUENEN", "AfD")
+                    %in% names(res$applause)) &&
+                is.character(res$applause$speech_id) &&
+                is.character(res$applause$on_speaker) &&
+                is.logical(res$applause$`CDU_CSU`) &&
+                is.logical(res$applause$`SPD`) &&
+                is.logical(res$applause$`FDP`) &&
+                is.logical(res$applause$`DIE_LINKE`) &&
+                is.logical(res$applause$`AfD`) &&
+                is.logical(res$applause$`BUENDNIS_90_DIE_GRUENEN`))
+}