[yelp-tools] yelp-check: Support Mallard Sites for hrefs

[Shaun McCance <shaunm src gnome org>]

commit 1ad0853ee662ae67d52107949f2cebb7f13ec772
Author: Shaun McCance <shaunm gnome org>
Date:   Thu Jun 18 20:31:02 2015 -0400

    yelp-check: Support Mallard Sites for hrefs

 tools/yelp-check.in |  118 ++++++++++++++++++++++++++------------------------
 1 files changed, 61 insertions(+), 57 deletions(-)
---
diff --git a/tools/yelp-check.in b/tools/yelp-check.in
index 91b7289..05de8fc 100755
--- a/tools/yelp-check.in
+++ b/tools/yelp-check.in
@@ -128,6 +128,9 @@ yelp_usage_hrefs () {
         echo ""
         echo "  Find broken href links in FILES in a Mallard document, or"
         echo "  broken ulink or XLink links in FILES in a DocBook document."
+        echo ""
+        echo "Options:"
+        echo "  -s            Treat pages as belonging to a Mallard site"
     ) 1>&2
 }
 yelp_usage_ids () {
@@ -274,54 +277,60 @@ yelp_check_iter_args () {
        elif [ "x$check_db" != "x" -a \( "x$ext" = "xdocbook" -o "x$ext" = "xxml" \) ]; then
            $check_db "$arg" || yelp_check_retval="$?"
         else
-            echo "Unrecognized page $arg" 2>&1
+            echo "Unrecognized page $arg" 1>&2
             exit 1
         fi
     done
     return $yelp_check_retval
 }
 
-yelp_hrefs_db () {
-    (
-        echo '<xsl:stylesheet'
-        echo ' xmlns:xsl="http://www.w3.org/1999/XSL/Transform";'
-        echo ' xmlns:db="http://docbook.org/ns/docbook";'
-        echo ' xmlns:xlink="www.w3.org/1999/xlink"'
-        echo ' version="1.0">'
-        echo '<xsl:output method="text"/>'
-        echo '<xsl:template match="/">'
-        echo '<xsl:for-each select="//ulink/@url | //*/xlink:href">'
-        echo '<xsl:if test="not(starts-with(string(.), '\''mailto:'\''))">'
-        echo '<xsl:value-of select="(ancestor-or-self::*/@id | ancestor-or-self::*/@xml:id)[last()]"/>'
-        echo '<xsl:text> </xsl:text>'
-        echo '<xsl:value-of select="string(.)"/>'
-        echo '<xsl:text>&#x000A;</xsl:text>'
-        echo '</xsl:if>'
-        echo '</xsl:for-each>'
-        echo '</xsl:template>'
-        echo '</xsl:stylesheet>'
-    ) | xsltproc --xinclude - "$1"
-}
-
 yelp_hrefs_page () {
+    base=$(dirname "$1")
+    if [ "x$check_site" = "x1" ]; then
+        sdir=$(cd $(dirname "$1") && pwd)
+        sdir=${sdir##${check_site_root}}/
+    fi
     (
         echo '<xsl:stylesheet'
         echo ' xmlns:xsl="http://www.w3.org/1999/XSL/Transform";'
         echo ' xmlns:mal="http://projectmallard.org/1.0/";'
+        echo ' xmlns:db="http://docbook.org/ns/docbook";'
+        echo ' xmlns:xlink="www.w3.org/1999/xlink"'
         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="/mal:page/@id"/>'
-        echo '<xsl:text> </xsl:text>'
-        echo '<xsl:value-of select="@href"/>'
-        echo '<xsl:text>&#x000A;</xsl:text>'
-        echo '</xsl:if>'
-        echo '</xsl:for-each>'
+        echo ' <xsl:for-each select="//*[ href]">'
+        echo '  <xsl:if test="not(starts-with(@href, '\''mailto:'\''))">'
+        echo '   <xsl:value-of select="/mal:page/@id"/>'
+        echo '   <xsl:text> </xsl:text>'
+        echo '   <xsl:value-of select="@href"/>'
+        echo '   <xsl:text>&#x000A;</xsl:text>'
+        echo '  </xsl:if>'
+        echo ' </xsl:for-each>'
+        echo '</xsl:template>'
+        echo '<xsl:template match="/*[namespace-uri(.) = '\'\''] | /db:*">'
+        echo ' <xsl:for-each select="//ulink/@url | //*/xlink:href">'
+        echo '  <xsl:if test="not(starts-with(string(.), '\''mailto:'\''))">'
+        echo '   <xsl:value-of select="(ancestor-or-self::*/@id | ancestor-or-self::*/@xml:id)[last()]"/>'
+        echo '   <xsl:text> </xsl:text>'
+        echo '   <xsl:value-of select="string(.)"/>'
+        echo '   <xsl:text>&#x000A;</xsl:text>'
+        echo '  </xsl:if>'
+        echo ' </xsl:for-each>'
         echo '</xsl:template>'
         echo '</xsl:stylesheet>'
-    ) | xsltproc --xinclude - "$1"
+    ) | xsltproc --xinclude - "$1" | sort | uniq | \
+        while read id url; do
+            colon=`echo "$url" | cut -d: -f1`
+            if [ "x$colon" = "x$url" ]; then
+                test -f "$base/"$(urldecode "$url") || echo "$sdir$id: $url"
+            else
+                (curl -s -I -L "$url" | \
+                        grep '^HTTP/' | tail -n 1 | head -n 1 | \
+                        grep -q 'HTTP/.\.. 200 .*') || \
+                    echo "$sdir$id: $url"
+            fi
+        done
 }
 
 yelp_hrefs () {
@@ -329,32 +338,27 @@ yelp_hrefs () {
         yelp_usage_hrefs
         exit 1
     fi
+    while [ "$#" != "0" ]; do
+        case "$1" in
+            "-s")
+                check_site="1"
+                check_site_root=$(pwd)
+                shift
+                ;;
+            *)
+                break
+                ;;
+        esac
+    done
+    if [ "$#" = "0" -o "x$1" = "x--help" -o "x$1" = "x-h" ]; then
+        yelp_usage_hrefs
+        exit 1
+    fi
     check_out_file=`mktemp "${TMPDIR:-/tmp}"/yelp-XXXXXXXX`
-    for xml in "$@"; do
-        if [ -d "$xml" ]; then
-            for page in "$xml"/*.page; do
-                yelp_hrefs_page "$page"
-            done
-        else
-            ext=`echo "$1" | sed -e 's/.*\.//'`
-            if [ "x$ext" = "xxml" -o "x$ext" = "xdocbook" ]; then
-                yelp_hrefs_db "$xml"
-            else
-                yelp_hrefs_page "$xml"
-            fi
-        fi | while read id url; do
-            echo "`dirname $xml`" "$id" "$url"
-        done
-    done | sort | uniq | while read base id url; do
-        colon=`echo "$url" | cut -d: -f1`
-        if [ "x$colon" = "x$url" ]; then
-            test -f "$base/`urldecode $url`" || echo "$id: $url"
-        else
-            (curl -s -I -L "$url" | grep '^HTTP/' | tail -n 1 | head -n 1 | grep -q 'HTTP/.\.. 200 .*') ||
-            echo "$id: $url"
-        fi
-    done > "$check_out_file"
-    ret=`cat "$check_out_file" | wc -l`
+    check_db=yelp_hrefs_page
+    check_page=yelp_hrefs_page
+    yelp_check_iter_args $@ > "$check_out_file"
+    ret=$(cat "$check_out_file" | wc -l)
     if test "x$ret" != "x0"; then ret="1"; fi
     cat "$check_out_file"
     rm "$check_out_file"