Added 'noindex' meta tag and cleaning up the rss code

patperu · Feb 25, 2016 · 012d447 · 012d447
1 parent cfd0397
commit 012d447
Show file tree

Hide file tree

Showing 6 changed files with 18 additions and 13 deletions.
diff --git a/DESCRIPTION b/DESCRIPTION
@@ -2,7 +2,7 @@ Package: albopop
 Title: Read some 'Albo Pretorio' pages and build a RSS feeds
 Description: Provides functions to read some 'Albo Pretorio' pages and export a
     RSS feed
-Version: 0.1.0
+Version: 0.1.1
 Authors@R: c(
     person("Patrick", "Hausmann", role = c("aut", "cre"), email = "patrick.hausmann@covimo.de")
     )

diff --git a/NAMESPACE b/NAMESPACE
@@ -7,6 +7,7 @@ export(ap_parse_page)
 export(ap_send_telegram)
 export(disk)
 importFrom(XML,addNode)
+importFrom(XML,newXMLNode)
 importFrom(XML,saveXML)
 importFrom(XML,xmlParse)
 importFrom(XML,xmlRoot)

diff --git a/R/albopop-package.R b/R/albopop-package.R
@@ -12,7 +12,7 @@
 #' @importFrom purrr flatten_chr map_chr
 #' @importFrom stringr str_replace
 #' @importFrom XML xmlTree addNode saveXML xpathSApply
-#'                 xmlRoot xmlParse xmlValue xmlTreeParse
+#'                 xmlRoot xmlParse xmlValue xmlTreeParse newXMLNode
 #' @importFrom xml2 read_html
 NULL
 

diff --git a/R/ap_build_rss.R b/R/ap_build_rss.R
@@ -22,11 +22,14 @@ ap_build_rss <- function(x, rss_title, rss_link, rss_file, store = disk()) {
   x$Fino      <- Set_Date(x$Fino)
   x$timestamp <- x$Inizio
 
-  feed <- xmlTree("rss")
+  feed <- xmlTree("rss", attrs=list(version = "2.0"))
   feed$addNode("channel", close=FALSE)
   feed$addNode("title", rss_title)
   feed$addNode("description", paste0(rss_title, " Albo POP RSS"))
   feed$addNode("link", rss_link)
+  feed$addNode(newXMLNode("xhtml:meta",
+                          namespaceDefinitions = list(xhtml= "http://www.w3.org/1999/xhtml"),
+                          attrs = list(name = "robots", content = "noindex")))
   feed$addNode("language", "it")
 
   for (i in 1:nrow(x)) {
@@ -42,8 +45,7 @@ ap_build_rss <- function(x, rss_title, rss_link, rss_file, store = disk()) {
 
   feed$closeTag()
 
-  rss <- str_replace(saveXML(feed, prefix = '<?xml version="1.0" encoding="UTF-8"?>\n',
-                             encoding = "UTF-8"), "<rss>", '<rss version="2.0">')
+  rss <- saveXML(feed, prefix = '<?xml version="1.0" encoding="UTF-8"?>\n')
 
   #writeLines(rss, con = file.path("feed", rss_file))
 

diff --git a/README.Rmd b/README.Rmd
@@ -26,6 +26,7 @@ The following functions are implemented:
 
 ### News
 
+- Version 0.1.1 : added `noindex` meta tag ([#90 in aborruso/albo-pop ](https://github.com/aborruso/albo-pop/issues/90)) 
 - Version 0.1.0 released
 
 ### Installation

diff --git a/README.md b/README.md
@@ -11,6 +11,7 @@ The following functions are implemented:
 
 ### News
 
+-   Version 0.1.1 : added `noindex` meta tag ([\#90 in aborruso/albo-pop](https://github.com/aborruso/albo-pop/issues/90))
 -   Version 0.1.0 released
 
 ### Installation
@@ -28,7 +29,7 @@ library("rvest")
 
 # current version
 packageVersion("albopop")
-## [1] '0.1.0'
+## [1] '0.1.1'
 ```
 
 The function `ap_parse_page` parse the content for different sites. The parameter `site` specify the type of input. Currently four providers are supported:
@@ -43,14 +44,14 @@ To parse the announcements for the [Comune di Lu](http://albo.studiok.it/lu/albo
 ``` r
 res <- ap_parse_page(url = "http://albo.studiok.it/lu/albo", site = "studiok")
 str(res)
-## Classes 'albopop_srs' and 'data.frame':  39 obs. of  7 variables:
-##  $ APNumero   : chr  "60 / 2016" "59 / 2016" "58 / 2016" "57 / 2016" ...
+## Classes 'albopop_srs' and 'data.frame':  44 obs. of  7 variables:
+##  $ APNumero   : chr  "67 / 2016" "66 / 2016" "65 / 2016" "64 / 2016" ...
 ##  $ Descrizione: chr  "" "" "" "" ...
-##  $ Tipo       : chr  "DELIBERA DI GIUNTA" "VARIE COMUNE DI LU" "DELIBERA DI GIUNTA" "DELIBERA DI GIUNTA" ...
-##  $ Oggetto    : chr  "[60 / 2016] DELIBERA GIUNTA COMUNALE N. 10 DEL 26/01/2016" "[59 / 2016] CONTRIBUTI REGIONALI PER RACCOLTA, TRASPORTO E SMALTIMENTO DI PICCOLI QUANTITATIVI DI AMIANTO (ETERNIT)" "[58 / 2016] DELIBERA GIUNTA COMUNALE N. 109 DEL 12/11/2015" "[57 / 2016] DELIBERA GIUNTA COMUNALE N. 56 DEL 07/07/2015" ...
-##  $ Inizio     : chr  "19/02/2016" "18/02/2016" "18/02/2016" "18/02/2016" ...
-##  $ Fino       : chr  "05/03/2016" "24/02/2016" "04/03/2016" "04/03/2016" ...
-##  $ link       : chr  "http://albo.studiok.it/lu/albo/dettaglio.php?id=MES00000000602016" "http://albo.studiok.it/lu/albo/dettaglio.php?id=MES00000000592016" "http://albo.studiok.it/lu/albo/dettaglio.php?id=MES00000000582016" "http://albo.studiok.it/lu/albo/dettaglio.php?id=MES00000000572016" ...
+##  $ Tipo       : chr  "DETERMINA" "DETERMINA" "DELIBERA DI CONSIGLIO" "DELIBERA DI CONSIGLIO" ...
+##  $ Oggetto    : chr  "[67 / 2016] DETERMINA RESPONSABILE SERVIZIO FINANZIARIO N. 3 DEL 17/02/2016" "[66 / 2016] DETERMINA RESPONSABILE SERVIZIO FINANZIARIO N. 2 DEL 15/02/2016" "[65 / 2016] DELIBERA CONSIGLIO COMUNALE N. 08 DEL 10/02/2016" "[64 / 2016] DELIBERA CONSIGLIO COMUNALE N. 07 DEL 10/02/2016" ...
+##  $ Inizio     : chr  "24/02/2016" "24/02/2016" "24/02/2016" "24/02/2016" ...
+##  $ Fino       : chr  "10/03/2016" "10/03/2016" "10/03/2016" "10/03/2016" ...
+##  $ link       : chr  "http://albo.studiok.it/lu/albo/dettaglio.php?id=MES00000000672016" "http://albo.studiok.it/lu/albo/dettaglio.php?id=MES00000000662016" "http://albo.studiok.it/lu/albo/dettaglio.php?id=MES00000000652016" "http://albo.studiok.it/lu/albo/dettaglio.php?id=MES00000000642016" ...
 ```
 
 The function `ap_build_rss` generates an RSS feed and writes a simple XML file in the local default directory `feed`. The input must contain at least three columns named `Oggetto`, `Inizio` and `link`. To write the file "Lu.xml":