From 1463f6092e131bc7c29d3113d9819a7cac24271c Mon Sep 17 00:00:00 2001
From: flavis <christian@flavigny.de>
Date: Wed, 28 Jul 2021 21:23:06 +0200
Subject: [PATCH] add basic analysing utils

---
 NAMESPACE                 |  2 ++
 R/analyze.R               | 13 +++++++++++++
 vignettes/funwithdata.Rmd | 12 ++++++++++++
 3 files changed, 27 insertions(+)
 create mode 100644 R/analyze.R

diff --git a/NAMESPACE b/NAMESPACE
index b52ac75..9b21836 100644
--- a/NAMESPACE
+++ b/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)
diff --git a/R/analyze.R b/R/analyze.R
new file mode 100644
index 0000000..b818005
--- /dev/null
+++ b/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
+}
diff --git a/vignettes/funwithdata.Rmd b/vignettes/funwithdata.Rmd
index 098ce4c..17c6ff4 100644
--- a/vignettes/funwithdata.Rmd
+++ b/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))
+```