christian/hateimparlament
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

rename some columns to english

Browse Source
This commit is contained in:
christian
2021-08-09 15:31:20 +02:00
parent b02ab91c31
commit 4649658fa7
4 changed files with 30 additions and 32 deletions
Download Patch File Download Diff File Expand all files Collapse all files
R/parse.R
+19 -19
View File
@@ -39,7 +39,7 @@ read_all <- function(path="inst/records/") {
select(-type) ->
comments
filter(commentsandapplause, type == "applause") %>%
select(-type, -kommentator, -content) %>%
select(-type, -commenter, -content) %>%
mutate("CDU_CSU" = str_detect(fraction, "CDU/CSU"),
"SPD" = str_detect(fraction, "SPD"),
"FDP" = str_detect(fraction, "FDP"),
@@ -86,17 +86,17 @@ xml_get <- function(node, name) {
parse_speaker <- function(speaker_xml) {
speaker_id <- xml_attr(speaker_xml, "id")
nm <- xml_child(speaker_xml)
vorname <- xml_get(nm, "vorname")
nachname <- xml_get(nm, "nachname")
prename <- xml_get(nm, "vorname")
lastname <- xml_get(nm, "nachname")
fraction <- xml_get(nm, "fraktion")
titel <- xml_get(nm, "titel")
rolle <- xml_find_all(nm, "rolle")
if (length(rolle) > 0) {
rolle_lang <- xml_get(rolle, "rolle_lang")
rolle_kurz <- xml_get(rolle, "rolle_kurz")
} else rolle_kurz <- rolle_lang <- NA_character_
c(id = speaker_id, vorname = vorname, nachname = nachname, fraction = fraction, titel = titel,
rolle_kurz = rolle_kurz, rolle_lang = rolle_lang)
title <- xml_get(nm, "titel")
role <- xml_find_all(nm, "rolle")
if (length(role) > 0) {
role_long <- xml_get(role, "rolle_lang")
role_short <- xml_get(role, "rolle_kurz")
} else role_short <- role_long <- NA_character_
c(id = speaker_id, prename = prename, lastname = lastname, fraction = fraction, title = title,
role_short = role_short, role_long = role_long)
}
# parse one speech
@@ -165,10 +165,10 @@ parse_comment <- function(comment, speech_id, on_speaker) {
sapply(partial(flip(head), 1) %.% agrep, x=fractionnames, max=0.2, value=T) %>%
str_c(collapse=",") ->
by
c(base, type = "applause", fraction = by, kommentator = NA_character_, content = comment)
c(base, type = "applause", fraction = by, commenter = NA_character_, content = comment)
} else {
ps <- str_match(comment, "(.*) \\[(.*?)\\]: (.*)")[1,]
c(base, type = "comment", fraction = ps[3], kommentator = ps[2], content = ps[4])
c(base, type = "comment", fraction = ps[3], commenter = ps[2], content = ps[4])
}
}
@@ -187,7 +187,7 @@ parse_speechlist <- function(speechlist_xml, date) {
on_speaker = comments["on_speaker",],
type = comments["type",],
fraction = comments["fraction",],
kommentator = comments["kommentator",],
commenter = comments["commenter",],
content = comments["content", ]))
}
@@ -195,12 +195,12 @@ parse_speechlist <- function(speechlist_xml, date) {
parse_speakerlist <- function(speakerliste_xml) {
d <- sapply(speakerliste_xml, parse_speaker)
tibble(id = d["id",],
vorname = d["vorname",],
nachname = d["nachname",],
prename = d["prename",],
lastname = d["lastname",],
fraction = d["fraction",],
titel = d["titel",],
rolle_kurz = d["rolle_kurz",],
rolle_lang = d["rolle_lang",])
title = d["title",],
role_short = d["role_short",],
role_long = d["role_long",])
}
#' Write the parsed and repaired results into separate csv files