From 174a7426057dfdebe78d525f59c621d0665e454e Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Fran=C3=A7ois=20Briatte?= <briatte@gmail.com>
Date: Tue, 14 Mar 2017 09:40:30 +0100
Subject: [PATCH] URL checks

---
 check.log | 60 +++++++++++++++++++++++++++----------------------------
 check.r   | 24 ++++++++++------------
 2 files changed, 40 insertions(+), 44 deletions(-)

diff --git a/check.log b/check.log
index 77f74da..dd54c66 100644
--- a/check.log
+++ b/check.log
@@ -1,36 +1,36 @@
-2017-01-02 19:46:05 : checking 11 URLs
+2017-03-14 09:04:07 : checking 11 URLs
 
-http://derekgreene.com/gephitutorial/      
-Status code: 404 
-
-http://nexus.igraph.org/      
-Failed to access
-
-http://www.derekruths.com/      
-Status code: 403 
-
-http://www.nature.com/nature/journal/v445/n7127/full/445489a.html      
-Status code: 401 
-
-http://www.ssc.wisc.edu/~emirbaye/Mustafa_Emirbayer/ARTICLES_files/manifesto%20for%20a%20relational%20sociology.pdf      
-Status code: 404 
-
-http://www.ssc.wisc.edu/~emirbaye/Mustafa_Emirbayer/ARTICLES_files/network%20analysis,%20culture,%20and%20the%20problem%20of%20agency.pdf      
-Status code: 404 
-
-https://410f84824e101297359cc81c78f45c7c079eb26c.googledrive.com/host/0Bz6WHrWac3FrWnA5MjZqb3lWa2c/      
-Status code: 404 
-
-https://dlist.server.uni-frankfurt.de/mailman/listinfo/sna-de      
+http://asonam.cpsc.ucalgary.ca/    
 Status code: 400 
 
-https://lra.le.ac.uk/bitstream/2381/36068/2/Draft%20BJM%20Revised%20(3rd%20iteration)%20Manuscript.pdf      
-Status code: 501 
+http://comnet.oxfordjournals.org/    
+Failed to access
 
-https://www.linkedin.com/in/mcculloh      
-Status code: 999 
+http://comnet.oxfordjournals.org/content/4/3/457.short    
+Failed to access
 
-https://www.puf.com/content/R%C3%A9seaux_sociaux_et_structures_relationnelles      
-Status code: 501 
+http://netzwerkerei.org/    
+Status code: 403 
 
-2017-01-02 19:46:09 : done.
+http://nexus.igraph.org/    
+Failed to access
+
+http://www.clementlevallois.net/gephi.html    
+Status code: 404 
+
+http://www.cmh.pro.ens.fr/reseaux-sociaux/    
+Failed to access
+
+http://www.derekruths.com/    
+Status code: 403 
+
+http://www.esri.com/software/arcgis/extensions/networkanalyst    
+Status code: 403 
+
+http://www.melissaclarkson.com/resources/R_guides/    
+Status code: 500 
+
+http://www.nature.com/nature/journal/v445/n7127/full/445489a.html    
+Status code: 401 
+
+2017-03-14 09:08:15 : done.
diff --git a/check.r b/check.r
index 836d81c..bd4356a 100644
--- a/check.r
+++ b/check.r
@@ -28,8 +28,7 @@ if (!file.exists(f)) {
     unlist
   
   # total number of links
-  t <- str_count(u, "http") %>%
-    sum
+  t <- sum(str_count(u, "http"))
   
   cat(t, "URLs, ")
   
@@ -50,15 +49,13 @@ if (!file.exists(f)) {
   
   cat("Source:", f, "\n")
   
-  l <- readLines(f) %>%
-    str_subset("^http")
+  l <- str_subset(readLines(f), "^http")
   
   cat(length(l), "URLs, ")
   
 }
 
-l <- unique(l) %>%
-  sort
+l <- sort(unique(l))
 
 cat(length(l), "unique\n")
 
@@ -68,9 +65,7 @@ sink()
 
 for (i in l) {
   
-  x <- try(GET(i) %>%
-             status_code,
-           silent = TRUE)
+  x <- try(status_code(GET(i)), silent = TRUE)
   
   if (!"try-error" %in% class(x) && x != 200) {
     
@@ -80,7 +75,7 @@ for (i in l) {
     cat(i, "\nStatus code:", x, "\n\n")
     sink()
     
-  } else if("try-error" %in% class(x)) {
+  } else if ("try-error" %in% class(x)) {
     
     cat("?")
     
@@ -94,8 +89,11 @@ for (i in l) {
     
   }
   
-  if (!which(l == i) %% 50)
+  if (!which(l == i) %% 50) {
+    
     cat("", length(l) - which(l == i), "left\n")
+    
+  }
   
 }
 
@@ -103,6 +101,4 @@ sink(f, append = TRUE)
 cat(as.character(Sys.time()), ": done.\n")
 sink()
 
-cat("\nFound", str_count(readLines(f), "^http") %>%
-      sum,
-    "problems.\n")
+cat("\nFound", sum(str_count(readLines(f), "^http")), "problems.\n")