yt.R (6330B)
1 #!/usr/bin/env Rscript
2 # SPDX-License-Identifier: AGPL-3.0-or-later
3 # SPDX-FileCopyrightText: 2021-2024 JayVii <jayvii[AT]posteo[DOT]de>
4
5 # Stop script after 300 seconds, whether is is done or not
6 setTimeLimit(elapsed = 300)
7
8 # Load Packages ----------------------------------------------------------------
9 if (!require("tidyRSS")) {
10 install.packages("tidyRSS")
11 library("tidyRSS")
12 }
13 if (!require("textutils")) {
14 install.packages("textutils")
15 library("textutils")
16 }
17
18 # Load URLs --------------------------------------------------------------------
19 channels <- as.character(
20 read.csv(file = "./url.csv", header = TRUE, sep = ",")$url
21 )
22 data <- matrix(
23 data = NA,
24 nrow = 1,
25 ncol = 7,
26 dimnames = list(
27 NULL,
28 c("title", "url", "author", "date", "time", "vid", "img")
29 )
30 )
31
32 # functions --------------------------------------------------------------------
33
34 fetch.yt <- function(channel) {
35
36 # fetch RSS feed
37 channel_data <- tidyRSS::tidyfeed(
38 channel,
39 clean_tags = TRUE,
40 list = TRUE,
41 parse_dates = FALSE
42 )
43
44 # strip information that is required for constructing HTML
45 video_dat <- data.frame(
46 cid = gsub(
47 x = channel,
48 pattern = "^.*channel_id=",
49 replacement = ""
50 ),
51 title = channel_data$entries$entry_title,
52 url = channel_data$entries$entry_link,
53 author = channel_data$meta$feed_title,
54 date = gsub(
55 x = channel_data$entries$entry_published,
56 pattern = "T.*$",
57 replacement = ""
58 ),
59 time = gsub(
60 x = channel_data$entries$entry_published,
61 pattern = "^.*T|\\+.*$",
62 replacement = ""
63 ),
64 vid = gsub(
65 x = channel_data$entries$entry_link,
66 pattern = "^.*\\?v=",
67 replacement = ""
68 ),
69 img = paste0(
70 gsub(
71 x = channel_data$entries$entry_link,
72 pattern = "^.*\\?v=",
73 replacement = "https:\\/\\/i4.ytimg.com\\/vi\\/"
74 ),
75 "/hqdefault.jpg"
76 )
77 )
78
79 # return video data
80 return(video_dat)
81 }
82
83 # fetch data -------------------------------------------------------------------
84
85 video_dat <- list()
86 for (i in seq_along(channels)) {
87 cat(paste("Fetching:", as.character(channels[i]), "\n"))
88 video_dat[[i]] <- tryCatch(fetch.yt(channels[i]), error = function(e) NULL)
89 }
90 data <- do.call(rbind, video_dat)
91
92 # encode text for ASCII compatibility
93 for (var in c("author", "title")) {
94 data[, var] <- textutils::HTMLencode(data[, var])
95 }
96
97 # edit data --------------------------------------------------------------------
98
99 # sorting according to date and time
100 dates <- as.numeric(
101 gsub(x = data[, "date"], pattern = "-", replacement = "")
102 )
103 times <- as.numeric(
104 gsub(x = data[, "time"], pattern = ":", replacement = "")
105 )
106 data <- data[rev(order(dates, times, na.last = FALSE)), ]
107
108 # construct per channel HTML ---------------------------------------------------
109
110 # unique, vectors of channels that returned some data
111 channel_id <- unique(data[, "cid"])
112 channel_name <- unique(data[, "author"])
113
114
115 # initilise entry-per-channel object
116 entry_pc <- list()
117
118 # fill in entry-per-channel object
119 for (chan in seq_along(channel_id)) {
120
121 # choose entries for the current channel
122 entries <- which(data[, "cid"] == channel_id[chan])
123
124 # fill entry-per-channel-object with contents from current channel
125 entry_pc[[chan]] <- paste0(
126 "<section id=\"entry_", seq_along(entries), "\">",
127 "<h2>",
128 "<a href=\"", data[entries, "url"], "\">",
129 data[entries, "title"],
130 "</a>",
131 "</h2>",
132 "<div class=\"thumbnails\">",
133 "<img src=\"", data[entries, "img"], "\"",
134 " id=\"thumbnail_", seq_along(entries), "\" loading=\"lazy\"",
135 " onclick=embed_yt(\"",
136 seq_along(entries), "\",\"", data[entries, "vid"],
137 "\")",
138 ">",
139 "</div>",
140 "<p>",
141 "<a class=\"button\" href=\"./", data[entries, "cid"], ".html\">",
142 data[entries, "author"],
143 "</a>",
144 " on ", data[entries, "date"], " ", data[entries, "time"],
145 "</p>",
146 "</section>"
147 )
148
149 }
150
151 names(entry_pc) <- channel_id
152
153 # construct main HTML ---------------------------------------------------------
154
155 # only list first "n" entries
156 n <- seq_len(500)
157
158 entry <- paste0(
159 "<section id=\"entry_", n, "\">",
160 "<h2>",
161 "<a href=\"", data[n, "url"], "\">",
162 data[n, "title"],
163 "</a>",
164 "</h2>",
165 "<div class=\"thumbnails\">",
166 "<img src=\"", data[n, "img"], "\"",
167 " id=\"thumbnail_", n, "\" loading=\"lazy\"",
168 " onclick=embed_yt(\"", n, "\",\"", data[n, "vid"], "\")",
169 ">",
170 "</div>",
171 "<p>",
172 "<a class=\"button\" href=\"./", data[n, "cid"], ".html\">",
173 data[n, "author"],
174 "</a>",
175 " on ", data[n, "date"], " ", data[n, "time"],
176 "</p>",
177 "</section>"
178 )
179
180 # additional HTML -------------------------------------------------------------
181
182 # Load template and fill in page title
183 template <- paste0(readLines("./template.html", encoding = "UTF-8"), "\n")
184 template <- sub(x = template, pattern = "%%TITLE%%", replacement = "Video-Feed")
185
186 # insert content for top of the page
187 top <- paste0("<p>Last Updated: ", Sys.time(), "</p><hr>")
188 top_mainfeed <- paste0(
189 "<details>",
190 "<summary>Channel List</summary>",
191 paste0(
192 "<a href=\"", channel_id, ".html\">",
193 channel_name,
194 " (", sapply(X = entry_pc, FUN = length), ")",
195 "</a>",
196 collapse = "<br>"
197 ),
198 "</details><hr>"
199 )
200 bottom <- paste0("</body></html>")
201
202 # print files -------------------------------------------------------------------
203
204 html_output <- file("index.html", open = "wt", encoding = "UTF-8")
205 sink(html_output)
206 cat(
207 template, "\n",
208 top, "\n",
209 top_mainfeed, "\n",
210 entry, "\n",
211 bottom, "\n"
212 )
213 sink()
214 close(html_output)
215
216 for (cid in channel_id) {
217 html_output <- file(
218 paste0(cid, ".html"), open = "wt", encoding = "UTF-8"
219 )
220 sink(html_output)
221 cat(
222 template, "\n",
223 top, "\n",
224 top_mainfeed, "\n",
225 entry_pc[[cid]], "\n",
226 bottom, "\n"
227 )
228 sink()
229 close(html_output)
230 }
231
232 # EOF yt.R
233
