Update check.r

This commit is contained in:
François Briatte 2023-04-27 00:04:41 +02:00
parent dc5d1af045
commit cfd8deddda

60
check.r
View File

@ -13,46 +13,46 @@ library(stringr)
f <- "check.log"
if (!file.exists(f)) {
u <- str_c(
"https://raw.githubusercontent.com/",
"briatte/awesome-network-analysis/",
"master/README.md"
)
cat("Source:", u, "\n")
u <- GET(u) %>%
content("text") %>%
str_split("\\n") %>% # so as to find [foo]: bar links
unlist
unlist()
# total number of links (made to match web.archive.org links only once)
t <- sum(str_count(u, "(?<!/)http"))
cat(t, "URLs, ")
l <- c(
# [foo](bar)
str_extract_all(u, "\\(http(.*?)\\)") %>%
lapply(str_replace_all, "^\\(|\\)$", "") %>%
unlist,
unlist(),
# [foo]: bar
str_extract_all(u, "^\\[(.*)\\]: (.*)") %>%
unlist %>%
str_extract_all(u, "^\\[(.*)\\]: (.*)") %>%
unlist() %>%
str_replace("^\\[(.*)\\]: (.*)", "\\2")
)
stopifnot(length(l) == t)
} else {
cat("Source:", f, "\n")
l <- str_subset(stringi::stri_read_lines(f), "^http")
cat(length(l), "URLs, ")
}
l <- str_squish(sort(unique(l)))
@ -67,37 +67,37 @@ cat(as.character(Sys.time()), ": checking", length(l), "URLs\n\n")
sink()
for (i in l) {
x <- try(status_code(GET(i)), silent = TRUE)
if (!"try-error" %in% class(x) && x != 200) {
cat("X")
sink(f, append = TRUE)
cat(i, "\nStatus code:", x, "\n\n")
sink()
} else if ("try-error" %in% class(x)) {
cat("?")
sink(f, append = TRUE)
cat(i, "\nFailed to access\n\n")
sink()
} else {
cat(".")
}
if (!which(l == i) %% 50) {
cat("", length(l) - which(l == i), "left\n")
}
}
sink(f, append = TRUE)