[yelp-tools] yelp-check: Add hrefs command to test external links

[Shaun McCance <shaunm src gnome org>]

commit 3182d5625cb0e4f84c9140cc11ad731110012e7b
Author: Shaun McCance <shaunm gnome org>
Date:   Wed Oct 19 20:24:49 2011 -0400

    yelp-check: Add hrefs command to test external links

 tools/yelp-check.in |   59 +++++++++++++++++++++++++++++++++++++++++++++++++++
 1 files changed, 59 insertions(+), 0 deletions(-)
---
diff --git a/tools/yelp-check.in b/tools/yelp-check.in
index a30cf02..b6906df 100755
--- a/tools/yelp-check.in
+++ b/tools/yelp-check.in
@@ -53,12 +53,20 @@ yelp_usage () {
         echo ""
         echo "Commands:"
         echo "  comments      Print the editorial comments in a document"
+        echo "  hrefs         Find broken href links in a Mallard document"
         echo "  links         Find broken xref links in a Mallard document"
         echo "  orphans       Find orphaned pages in a Mallard document"
         echo "  status        Report the status of Mallard pages"
         echo "  validate      Validate files against a DTD or RNG"
     ) 1>&2
 }
+yelp_usage_hrefs () {
+    (
+        echo "Usage: yelp-check hrefs <FILES>"
+        echo ""
+        echo "  Find broken href links in FILES in a Mallard document."
+    ) 1>&2
+}
 yelp_usage_links () {
     (
         echo "Usage: yelp-check links <FILES>"
@@ -123,6 +131,54 @@ if [ $# = 0 ]; then
     exit 1
 fi
 
+yelp_hrefs_page () {
+    page=`(
+        echo '<xsl:stylesheet'
+        echo ' xmlns:xsl="http://www.w3.org/1999/XSL/Transform";'
+        echo ' xmlns:mal="http://projectmallard.org/1.0/";'
+        echo ' version="1.0">'
+        echo '<xsl:output method="text"/>'
+        echo '<xsl:template match="/mal:page">'
+        echo '<xsl:value-of select="@id"/>'
+        echo '</xsl:template>'
+        echo '</xsl:stylesheet>'
+    ) | xsltproc --xinclude - "$1"`
+    (
+        echo '<xsl:stylesheet'
+        echo ' xmlns:xsl="http://www.w3.org/1999/XSL/Transform";'
+        echo ' xmlns:mal="http://projectmallard.org/1.0/";'
+        echo ' version="1.0">'
+        echo '<xsl:output method="text"/>'
+        echo '<xsl:template match="/mal:page">'
+        echo '<xsl:for-each select="//*[ href]">'
+        echo '<xsl:if test="not(starts-with(@href, '\''mailto:'\''))">'
+        echo '<xsl:value-of select="@href"/>'
+        echo '<xsl:text>&#x000A;</xsl:text>'
+        echo '</xsl:if>'
+        echo '</xsl:for-each>'
+        echo '</xsl:template>'
+        echo '</xsl:stylesheet>'
+    ) | xsltproc --xinclude - "$1" | sort | uniq | while read url; do
+        (curl -s -I -L "$url" | grep '^HTTP/' | tail -n 1 | head -n 1 | grep -q 'HTTP/.\.. 200 .*') || echo "$page: $url"
+    done
+}
+
+yelp_hrefs () {
+    if [ "$#" = "0" -o "x$1" = "x--help" ]; then
+        yelp_usage_hrefs
+        exit 1
+    fi
+    for xml in "$@"; do
+        if [ -d "$xml" ]; then
+            for page in "$xml"/*.page; do
+                yelp_hrefs_page "$page"
+            done
+        else
+            yelp_hrefs_page "$xml"
+        fi
+    done
+}
+
 yelp_links_page () {
     (
         echo '<xsl:stylesheet'
@@ -444,6 +500,9 @@ case "x$cmd" in
     "xcomments")
         yelp_comments "$@"
         ;;
+    "xhrefs")
+        yelp_hrefs "$@"
+        ;;
     "xlinks")
         yelp_links "$@"
         ;;