add basic analysing utils

NAMESPACE

@@ -1,6 +1,8 @@
 # Generated by roxygen2: do not edit by hand
 
 export(fetch_all)
+export(find_word)
+export(join_redner)
 export(read_all)
 export(read_from_csv)
 export(repair)

R/analyze.R

@@ -0,0 +1,13 @@
+#' @export
+find_word <- function(res, word) {
+    talks <- res$talks
+    mutate(talks, occurences = sapply(str_match_all(talks$content, regex(word, ignore_case = TRUE)),
+           nrow))
+}
+
+#' @export
+join_redner <- function(tb, res, fraktion_only = F) {
+    joined <- left_join(tb, res$redner, by=c("redner" = "id"))
+    if (fraktion_only) select(joined, "fraktion")
+    else joined
+}

vignettes/funwithdata.Rmd

@@ -57,3 +57,15 @@ left_join(reden, redner, by=c("redner" = "id")) %>%
     ggplot(aes(x = fraktion, y = n)) +
     geom_bar(stat = "identity")
 ```
+
+### Count a word occurence
+
+```{r}
+find_word(res, "hitler") %>%
+    filter(occurences > 0) %>%
+    join_redner(res) %>%
+    select(content, fraktion) %>%
+    group_by(fraktion) %>%
+    summarize(n = n()) %>%
+    arrange(desc(n))
+```