Delete script.R
Browse files
script.R
DELETED
|
@@ -1,196 +0,0 @@
|
|
| 1 |
-
library(tidyverse)
|
| 2 |
-
|
| 3 |
-
# ----- #
|
| 4 |
-
read_captions_from_directory <- function(directory_path) {
|
| 5 |
-
# ディレクトリ内の.txtファイルのリストを取得
|
| 6 |
-
txt_files <- list.files(directory_path, pattern = "\\.txt$", full.names = TRUE)
|
| 7 |
-
|
| 8 |
-
# 各.txtファイルからキャプションを読み込み
|
| 9 |
-
data <- lapply(txt_files, function(file) {
|
| 10 |
-
captions <- readLines(file, warn = FALSE)
|
| 11 |
-
captions_list <- strsplit(captions, ",")[[1]]
|
| 12 |
-
captions_list <- trimws(captions_list) # 余分な空白を取り除く
|
| 13 |
-
tibble(
|
| 14 |
-
image_path = gsub(".txt$", ".png", file),
|
| 15 |
-
caption_order = 1:length(captions_list),
|
| 16 |
-
caption = captions_list
|
| 17 |
-
)
|
| 18 |
-
})
|
| 19 |
-
|
| 20 |
-
# データフレームに変換
|
| 21 |
-
bind_rows(data)
|
| 22 |
-
}
|
| 23 |
-
|
| 24 |
-
get_caption_frequency <- function(data) {
|
| 25 |
-
data %>%
|
| 26 |
-
group_by(caption) %>%
|
| 27 |
-
summarise(frequency = n()) %>%
|
| 28 |
-
arrange(-frequency)
|
| 29 |
-
}
|
| 30 |
-
|
| 31 |
-
search_by_caption <- function(data, target_caption) {
|
| 32 |
-
data %>%
|
| 33 |
-
filter(caption == target_caption) %>%
|
| 34 |
-
group_by(image_path) %>%
|
| 35 |
-
distinct()
|
| 36 |
-
}
|
| 37 |
-
|
| 38 |
-
remove_caption_and_adjust_order <- function(data, target_image_path, target_caption) {
|
| 39 |
-
# キャプションが存在するか確認
|
| 40 |
-
if (!any(data$image_path == target_image_path & data$caption == target_caption)) {
|
| 41 |
-
cat(sprintf("The caption '%s' does not exist for image '%s'.\n", target_caption, target_image_path))
|
| 42 |
-
return(data)
|
| 43 |
-
}
|
| 44 |
-
|
| 45 |
-
# 削除するキャプションのcaption_orderを取得
|
| 46 |
-
removed_order <- data$caption_order[data$image_path == target_image_path & data$caption == target_caption]
|
| 47 |
-
# キャプションを削除
|
| 48 |
-
data <- data %>% filter(!(image_path == target_image_path & caption == target_caption))
|
| 49 |
-
# caption_orderを調整
|
| 50 |
-
data$caption_order[data$image_path == target_image_path & data$caption_order > removed_order] <- data$caption_order[data$image_path == target_image_path & data$caption_order > removed_order] - 1
|
| 51 |
-
|
| 52 |
-
return(data)
|
| 53 |
-
}
|
| 54 |
-
|
| 55 |
-
remove_low_frequency_captions <- function(data, threshold) {
|
| 56 |
-
# キャプションの頻度を取得
|
| 57 |
-
caption_freq <- get_caption_frequency(data)
|
| 58 |
-
|
| 59 |
-
# 指定された頻度以下のキャプションのリストを作成
|
| 60 |
-
low_freq_captions <- caption_freq %>%
|
| 61 |
-
filter(frequency <= threshold) %>%
|
| 62 |
-
pull(caption)
|
| 63 |
-
|
| 64 |
-
# 低頻度のキャプションを削除し、caption_orderを調整
|
| 65 |
-
for (caption in low_freq_captions) {
|
| 66 |
-
unique_images <- unique(data$image_path[data$caption == caption])
|
| 67 |
-
for (image in unique_images) {
|
| 68 |
-
data <- remove_caption_and_adjust_order(data, image, caption)
|
| 69 |
-
}
|
| 70 |
-
}
|
| 71 |
-
|
| 72 |
-
return(data)
|
| 73 |
-
}
|
| 74 |
-
|
| 75 |
-
edit_captions_interactively <- function(data, target_caption) {
|
| 76 |
-
# キャプションで画像を検索
|
| 77 |
-
image_paths <- search_by_caption(data, target_caption)$image_path
|
| 78 |
-
|
| 79 |
-
for (path in image_paths) {
|
| 80 |
-
# OSに応じて画像を開く
|
| 81 |
-
if (Sys.info()["sysname"] == "Windows") {
|
| 82 |
-
cmd <- sprintf('start "" "%s"', path)
|
| 83 |
-
shell(cmd, intern = TRUE)
|
| 84 |
-
} else if (Sys.info()["sysname"] == "Darwin") { # macOS
|
| 85 |
-
cmd <- sprintf('open "%s"', path)
|
| 86 |
-
system(cmd)
|
| 87 |
-
} else { # Linux
|
| 88 |
-
cmd <- sprintf('xdg-open "%s"', path)
|
| 89 |
-
system(cmd)
|
| 90 |
-
}
|
| 91 |
-
|
| 92 |
-
# ユーザーにキャプションの削除を選択させる
|
| 93 |
-
cat(sprintf("Do you want to remove the caption '%s' from image '%s'? (yes/no/end): ", target_caption, path))
|
| 94 |
-
response <- readline()
|
| 95 |
-
|
| 96 |
-
if (tolower(response) == "end") {
|
| 97 |
-
break
|
| 98 |
-
} else if (tolower(response) == "yes") {
|
| 99 |
-
data <- remove_caption_and_adjust_order(data, path, target_caption)
|
| 100 |
-
}
|
| 101 |
-
}
|
| 102 |
-
|
| 103 |
-
return(data)
|
| 104 |
-
}
|
| 105 |
-
|
| 106 |
-
add_caption_at_order <- function(data, target_image_path, target_caption, target_order = NULL) {
|
| 107 |
-
# 指定された画像の最大のcaption_orderを取得
|
| 108 |
-
max_order <- max(data$caption_order[data$image_path == target_image_path], na.rm = TRUE)
|
| 109 |
-
|
| 110 |
-
# キャプションの重複チェック
|
| 111 |
-
if (target_caption %in% data$caption[data$image_path == target_image_path]) {
|
| 112 |
-
return(data) # 重複がある場合、データをそのまま返す
|
| 113 |
-
}
|
| 114 |
-
|
| 115 |
-
# target_orderが指定されていない場合、キャプションを表示してユーザーに選ばせる
|
| 116 |
-
if (is.null(target_order)) {
|
| 117 |
-
print_image_captions_as_csv(data, target_image_path)
|
| 118 |
-
cat("Enter the position (order) to insert the new caption (1 to", max_order + 1, "): ")
|
| 119 |
-
target_order <- as.numeric(readline())
|
| 120 |
-
|
| 121 |
-
# 不適切な値が入力された場合、最大のorder + 1で追加
|
| 122 |
-
if (target_order <= 0 || target_order > max_order + 1) {
|
| 123 |
-
target_order <- max_order + 1
|
| 124 |
-
}
|
| 125 |
-
}
|
| 126 |
-
|
| 127 |
-
# 指定されたorder以降のcaption_orderを増加
|
| 128 |
-
data <- data %>%
|
| 129 |
-
mutate(caption_order = ifelse(image_path == target_image_path & caption_order >= target_order, caption_order + 1, caption_order))
|
| 130 |
-
|
| 131 |
-
# 新しいキャプションを追加
|
| 132 |
-
new_caption <- tibble(
|
| 133 |
-
image_path = target_image_path,
|
| 134 |
-
caption_order = target_order,
|
| 135 |
-
caption = target_caption
|
| 136 |
-
)
|
| 137 |
-
data <- bind_rows(data, new_caption)
|
| 138 |
-
|
| 139 |
-
return(data)
|
| 140 |
-
}
|
| 141 |
-
|
| 142 |
-
move_caption_order <- function(data, target_image_path, target_caption, new_order) {
|
| 143 |
-
|
| 144 |
-
# キャプションが存在するか確認
|
| 145 |
-
if (!any(data$image_path == target_image_path & data$caption == target_caption)) {
|
| 146 |
-
cat(sprintf("The caption '%s' does not exist for image '%s'.\n", target_caption, target_image_path))
|
| 147 |
-
return(data)
|
| 148 |
-
}
|
| 149 |
-
|
| 150 |
-
# キャプションを削除
|
| 151 |
-
data_after_removal <- remove_caption_and_adjust_order(data, target_image_path, target_caption)
|
| 152 |
-
|
| 153 |
-
# 新しい位置にキャプションを追加
|
| 154 |
-
data_after_addition <- add_caption_at_order(data_after_removal, target_image_path, target_caption, new_order)
|
| 155 |
-
return(data_after_addition)
|
| 156 |
-
}
|
| 157 |
-
|
| 158 |
-
# キャプションが存在するか確認 (使わないかも)
|
| 159 |
-
is_caption_present <- function(data, target_image_path, target_caption) {
|
| 160 |
-
return(any(data$image_path == target_image_path & data$caption == target_caption))
|
| 161 |
-
}
|
| 162 |
-
|
| 163 |
-
# すべてのキャプションを表示
|
| 164 |
-
print_all_unique_captions_as_csv <- function(data) {
|
| 165 |
-
# 重複なく全てのキャプションを取得
|
| 166 |
-
unique_captions <- unique(data$caption)
|
| 167 |
-
# CSV形式で表示
|
| 168 |
-
cat(paste(unique_captions, collapse = ", "), "\n")
|
| 169 |
-
}
|
| 170 |
-
|
| 171 |
-
print_image_captions_as_csv <- function(data, target_image_path) {
|
| 172 |
-
captions <- filter(data, image_path == target_image_path) %>%
|
| 173 |
-
arrange(caption_order) %>%
|
| 174 |
-
pull(caption)
|
| 175 |
-
|
| 176 |
-
cat(paste(captions, collapse = ", "), "\n")
|
| 177 |
-
}
|
| 178 |
-
|
| 179 |
-
# 代表するキャプションに集約
|
| 180 |
-
remove_related_captions_except_representative <- function(data, related_captions, representative_caption, target_image_path) {
|
| 181 |
-
|
| 182 |
-
# representative_captionがtarget_image_pathに紐づいているか確認
|
| 183 |
-
if (!any(data$image_path == target_image_path & data$caption == representative_caption)) {
|
| 184 |
-
cat(sprintf("The representative caption '%s' is not associated with image '%s'.\n", representative_caption, target_image_path))
|
| 185 |
-
return(data)
|
| 186 |
-
}
|
| 187 |
-
|
| 188 |
-
# target_image_pathに関連するキャプションを削除
|
| 189 |
-
for (caption in related_captions) {
|
| 190 |
-
if (caption != representative_caption) {
|
| 191 |
-
data <- remove_caption_and_adjust_order(data, target_image_path, caption)
|
| 192 |
-
}
|
| 193 |
-
}
|
| 194 |
-
|
| 195 |
-
return(data)
|
| 196 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|