Program djvuserve is a CGI program that can be executed by a HTTP server for serving DjVu documents. This program is able to convert a bundled multi-page document into an indirect document on the fly.
Program djvuserve must first be installed as a CGI program for your web server. There are several ways to achieve this. The Apache web server, for instance, often defines a specific directory for CGI programs using the ScriptAlias directive. Assume that the file httpd.conf contains the following line:
ScriptAlias /cgi–bin/ "/var/www/cgi–bin"
It is then sufficient to create a small executable shell script /var/www/cgi–bin/djvuserve containing the following lines:
Suppose that a large bundled multi-page DjVu document is available at the following URL.
The CGI program djvuserve lets you access this same document as an indirect multi-page DjVu document using the following URL.
Serving indirect multi-page DjVu documents provides for efficiently browsing large document without transferring unnecessary pages over the network. See djvu(1) for more information.
Furthermore djvuserve searches certain keywords among the CGI arguments of the URL. The keyword bundled forces serving a bundled document using
The keyword download inserts a content disposition HTTP header that suggests to display a save dialog instead of displaying the document.
Using Djvuserve As a Handler
The Apache web server provides a way to automatically execute djvuserve for all DjVu documents. This can be achieved using the following directives in either the Apache configuration file or the .htaccess files.
Action djvu-server /cgi–bin/djvuserve/
AddHandler djvu-server .djvu
Apache then executes program djvuserve for serving all DjVu files. Providing the URL of DjVu file serves this DjVu file as usual, except that bundled multipage documents are converted to indirect documents on the fly. This convenience comes at the expense of the computational cost of executing djvuserve whenever a DjVu file is requested.
Program djvuserve provides a mean to directly access any component of a bundled multi-page DjVu document can be accessed using an extended URL. Suppose that the component file representing page 1 is named p0001.djvu. The following URL provides a direct access to this page:
It is preferred however to access individual pages using the CGI style arguments described in nsdejavu(1), as in the following URL.
The special component file name index.djvu is recognized as a request for the index of the corresponding indirect multi-page document. In fact, when you access a bundled document using djvuserve, the browser gets redirected to the following URL:
and then behaves as if the bundled file was a directory containing the various component files of an equivalent indirect document.
Program djvuserve, like many CGI programs, bypasses a number of access protections established in a web server. Assume for instance that your web site contains DjVu files protected by a password. Program djvuserve knows nothing about this protection and will happily serve any DjVu file associated with a valid URL.
Access control with djvuserve can be implemented by first remembering that the web server always executes program djvuserve via shell script /var/www/cgi–bin/djvuserve.
This script can decide to execute the real program djvuserve on the basis of the target filename available in the environment variable PATH_TRANSLATED.
There can be several such scripts providing access to various collections of DjVu files. Each of these scripts can be password protected using the usual methods supported by your web server.
Hyperlinks specified using a relative URL may not work with djvuserve. These URLs are relative to the URL of the DjVu document. Yet djvuserve changes the apparent document URL http://server/dir/doc.djvu into the more complicated URL http://server/cgi–bin/djvuserve/dir/doc.djvu/index.djvu. The extra components change the interpretation of relative URLs.
This program was written by Leon Bottou <email@example.com>.
djvu(1), djvmcvt(1), nsdejavu(1)