From 9bce34f7c6debb03e9419e0145f114eb4817a493 Mon Sep 17 00:00:00 2001
From: Eric van der Vlist <vdv@dyomedea.com>
Date: Fri, 27 Apr 2012 18:29:15 +0200
Subject: [PATCH] Rewriting links in HTML and CSS resources within WARC
 archives

---
 .../pipelines/actions/mediatypes/warc-css.xpl | 68 ++++++++++++++++
 .../actions/mediatypes/warc-html.xpl          | 80 +++++++++++++++++++
 .../actions/package-heritrix-warc.xpl         | 57 ++++++-------
 .../pipelines/actions/resource-index.xslt     | 10 +++
 4 files changed, 187 insertions(+), 28 deletions(-)
 create mode 100644 archiver/pipelines/actions/mediatypes/warc-css.xpl
 create mode 100644 archiver/pipelines/actions/mediatypes/warc-html.xpl

diff --git a/archiver/pipelines/actions/mediatypes/warc-css.xpl b/archiver/pipelines/actions/mediatypes/warc-css.xpl
new file mode 100644
index 0000000..dde3f50
--- /dev/null
+++ b/archiver/pipelines/actions/mediatypes/warc-css.xpl
@@ -0,0 +1,68 @@
+<p:config xmlns:p="http://www.orbeon.com/oxf/pipeline" xmlns:oxf="http://www.orbeon.com/oxf/processors" xmlns:xsl="http://www.w3.org/1999/XSL/Transform" xmlns:saxon="http://saxon.sf.net/">
+
+  <p:param name="record" type="input"/>
+  <p:param name="index-entry" type="input"/>
+  <p:param name="index" type="input"/>
+  <p:param name="rewritten" type="output"/>
+
+
+  <!-- Store the document -->
+  <p:processor name="oxf:file-serializer">
+    <p:input name="config">
+      <config>
+        <scope>session</scope>
+      </config>
+    </p:input>
+    <p:input name="data" href="#record#xpointer(/record/content/document)"/>
+    <p:output name="data" id="url-written"/>
+  </p:processor>
+
+  <!-- And read it as CSS -->
+  <p:processor name="oxf:url-generator">
+    <p:input name="config" transform="oxf:xslt" href="#url-written">
+      <config xsl:version="2.0">
+        <url>
+          <xsl:value-of select="/*"/>
+        </url>
+        <content-type>text/css</content-type>
+        <mode>text</mode>
+      </config>
+    </p:input>
+    <p:output name="data" id="css" debug="css"/>
+  </p:processor>
+
+
+  <!-- Update the links -->
+  <p:processor name="oxf:unsafe-xslt">
+    <p:input name="data" href="#css"/>
+    <p:input name="index-entry" href="#index-entry"/>
+    <p:input name="index" href="#index"/>
+    <p:input name="config">
+      <xsl:stylesheet version="2.0">
+        <xsl:variable name="index" select="doc('input:index')/*"/>
+        <xsl:variable name="resource" select="doc('input:index-entry')/resource"/>
+        <xsl:variable name="base" select="$resource/uri"/>
+        <xsl:template match="/document">
+          <xsl:copy>
+            <xsl:copy-of select="@*"/>
+            <xsl:analyze-string select="." regex="url\([&quot;']?([^)'&quot;]+)[&quot;']?\)" flags="">
+              <xsl:matching-substring>
+                <xsl:text>url(</xsl:text>
+                <xsl:variable name="abs" select="substring-before(concat(resolve-uri(regex-group(1), $base), '#'), '#')"/>
+                <xsl:value-of select="$index/resource[(uri, same-as) = $abs]/local-name"/>
+                <xsl:text>)</xsl:text>
+              </xsl:matching-substring>
+              <xsl:non-matching-substring>
+                <xsl:copy-of select="."/>
+              </xsl:non-matching-substring>
+            </xsl:analyze-string>
+          </xsl:copy>
+        </xsl:template>
+      </xsl:stylesheet>
+    </p:input>
+    <p:output name="data" ref="rewritten" debug="rewritten"/>
+  </p:processor>
+
+
+
+</p:config>
diff --git a/archiver/pipelines/actions/mediatypes/warc-html.xpl b/archiver/pipelines/actions/mediatypes/warc-html.xpl
new file mode 100644
index 0000000..7d30e57
--- /dev/null
+++ b/archiver/pipelines/actions/mediatypes/warc-html.xpl
@@ -0,0 +1,80 @@
+<p:config xmlns:p="http://www.orbeon.com/oxf/pipeline" xmlns:oxf="http://www.orbeon.com/oxf/processors" xmlns:xsl="http://www.w3.org/1999/XSL/Transform" xmlns:saxon="http://saxon.sf.net/">
+
+  <p:param name="record" type="input"/>
+  <p:param name="index-entry" type="input"/>
+  <p:param name="index" type="input"/>
+  <p:param name="rewritten" type="output"/>
+
+
+  <!-- Store the document -->
+  <p:processor name="oxf:file-serializer">
+    <p:input name="config">
+      <config>
+        <scope>session</scope>
+      </config>
+    </p:input>
+    <p:input name="data" href="#record#xpointer(/record/content/document)"/>
+    <p:output name="data" id="url-written"/>
+  </p:processor>
+
+  <!-- And read it as HTML -->
+  <p:processor name="oxf:url-generator">
+    <p:input name="config" transform="oxf:xslt" href="#url-written">
+      <config xsl:version="2.0">
+        <url>
+          <xsl:value-of select="/*"/>
+        </url>
+        <mode>html</mode>
+      </config>
+    </p:input>
+    <p:output name="data" id="html" debug="html"/>
+  </p:processor>
+
+  <!-- Update the links -->
+  <!-- TODO: support links in inline CSS -->
+  <!-- TODO: support iframes and objects -->
+
+
+  <p:processor name="oxf:unsafe-xslt">
+    <p:input name="data" href="#html"/>
+    <p:input name="index-entry" href="#index-entry"/>
+    <p:input name="index" href="#index"/>
+    <p:input name="config">
+      <xsl:stylesheet version="2.0">
+        <xsl:variable name="index" select="doc('input:index')/*"/>
+        <xsl:variable name="resource" select="doc('input:index-entry')/resource"/>
+        <xsl:variable name="base" select="$resource/uri"/>
+        <xsl:template match="@*|node()">
+          <xsl:copy>
+            <xsl:apply-templates select="@*|node()"/>
+          </xsl:copy>
+        </xsl:template>
+        <xsl:template match="link[@rel='stylesheet']/@href|img/@src|script/@src">
+          <xsl:attribute name="{name(.)}">
+            <xsl:variable name="abs" select="substring-before(concat(resolve-uri(., $base), '#'), '#')"/>
+            <xsl:value-of select="$index/resource[(uri, same-as) = $abs]/local-name"/>
+          </xsl:attribute>
+        </xsl:template>
+        <xsl:template match="link[@rel!='stylesheet']/@href|a/@href">
+          <xsl:attribute name="{name(.)}">
+            <xsl:value-of select="resolve-uri(., $base)"/>
+          </xsl:attribute>
+        </xsl:template>
+      </xsl:stylesheet>
+    </p:input>
+    <p:output name="data" id="html-rewritten" debug="rewritten"/>
+  </p:processor>
+
+  <p:processor name="oxf:xml-converter">
+    <p:input name="config">
+      <config>
+        <content-type>application/xml</content-type>
+        <encoding>utf-8</encoding>
+        <version>1.0</version>
+      </config>
+    </p:input>
+    <p:input name="data" href="#html-rewritten"/>
+    <p:output name="data" ref="rewritten"/>
+  </p:processor>
+
+</p:config>
diff --git a/archiver/pipelines/actions/package-heritrix-warc.xpl b/archiver/pipelines/actions/package-heritrix-warc.xpl
index e999dad..194cd3d 100644
--- a/archiver/pipelines/actions/package-heritrix-warc.xpl
+++ b/archiver/pipelines/actions/package-heritrix-warc.xpl
@@ -93,12 +93,31 @@
             <p:output name="data" id="index-entry" debug="index-entry"/>
         </p:processor>
         <p:choose href="#index-entry">
-            <p:when test="/entry/embeds">
+            <p:when test="/resource/embeds">
                 <!-- The resource has embedded content and must be rewritten -->
-                <p:processor name="oxf:identity">
-                    <p:input name="data" href="current()#xpointer(/record/content/document)"/>
-                    <p:output name="data" id="document"/>
+
+                <!-- Call the corresponding pipeline -->
+                <p:processor name="oxf:url-generator">
+                    <p:input name="config" transform="oxf:xslt" href="#index-entry">
+                        <config xsl:version="2.0">
+                            <url>
+                                <xsl:text>oxf:/actions/mediatypes/warc-</xsl:text>
+                                <xsl:value-of select="/resource/type"/>
+                                <xsl:text>.xpl</xsl:text>
+                            </url>
+                        </config>
+                    </p:input>
+                    <p:output name="data" id="pipeline"/>
                 </p:processor>
+
+                <p:processor name="oxf:pipeline">
+                    <p:input name="config" href="#pipeline"/>
+                    <p:input name="record" href="current()"/>
+                    <p:input name="index" href="#index"/>
+                    <p:input name="index-entry" href="#index-entry"/>
+                    <p:output name="rewritten" id="document" debug="rewritten"/>
+                </p:processor>
+
             </p:when>
             <p:otherwise>
                 <!-- The resource can be stored  -->
@@ -123,10 +142,6 @@
         </p:processor>
     </p:for-each>
 
-    <p:processor name="oxf:null-serializer">
-        <p:input name="data" href="#loop" debug="loop"/>
-    </p:processor>
-
 
 
     <!-- Store the WARC in a temp file -->
@@ -141,30 +156,16 @@
     </p:processor>
 
     <p:processor name="oxf:zip">
-        <p:input name="data" transform="oxf:unsafe-xslt" href="aggregate('root', #warc-location)">
+        <p:input name="data" transform="oxf:unsafe-xslt" href="aggregate('root', #warc-location, #loop)">
             <files xsl:version="2.0" file-name="archive.zip">
                 <file name="archive.warc">
                     <xsl:value-of select="/root/url"/>
                 </file>
-                <!--<xsl:for-each select="/root/files/file[url]">
-                    <xsl:choose>
-                        <xsl:when test="position()=1">
-                            <!-\- TODO: support non HTML documents... -\->
-                            <file name="rewritten/index.html">
-                                <xsl:value-of select="url"/>
-                            </file>
-                        </xsl:when>
-                        <xsl:otherwise>
-                            <xsl:variable name="tokens" select="tokenize(archive/@url, '/')"/>
-                            <xsl:variable name="last-token" select="$tokens[last()]"/>
-                            <xsl:variable name="tokens2" select="tokenize($last-token, '\.')"/>
-                            <xsl:variable name="extension" select="$tokens2[last()]"/>
-                            <file name="rewritten/{saxon:string-to-hexBinary(substring(archive/@url, 1, string-length(archive/@url) - string-length($extension) - 1), 'utf-8')}.{$extension}">
-                                <xsl:value-of select="url"/>
-                            </file>
-                        </xsl:otherwise>
-                    </xsl:choose>
-                </xsl:for-each>-->
+                <xsl:for-each select="/root/root/doc">
+                    <file name="rewritten/{resource/local-name}">
+                        <xsl:value-of select="url"/>
+                    </file>
+                </xsl:for-each>
             </files>
         </p:input>
         <p:output name="data" id="zip"/>
diff --git a/archiver/pipelines/actions/resource-index.xslt b/archiver/pipelines/actions/resource-index.xslt
index a6dadc9..82d7036 100644
--- a/archiver/pipelines/actions/resource-index.xslt
+++ b/archiver/pipelines/actions/resource-index.xslt
@@ -58,6 +58,16 @@
             <local-name>
                 <xsl:value-of select="owk:unique-local-name(.)"/>
             </local-name>
+            <type>
+                <xsl:choose>
+                    <xsl:when test="content-type = 'text/html'">html</xsl:when>
+                    <xsl:when test="content-type = 'application/xhtml+xml'">html</xsl:when>
+                    <xsl:when test="content-type = 'text/plain'">text</xsl:when>
+                    <xsl:otherwise>
+                        <xsl:value-of select="substring-after(content-type, '/')"/>
+                    </xsl:otherwise>
+                </xsl:choose>
+            </type>
             <xsl:apply-templates select="." mode="redirect"/>
             <xsl:apply-templates select="/log/entry[referer = current()/uri and ends-with(discovery-path, 'E')]" mode="embedding"/>
         </resource>