ProxyHTMLURLMap not modifying JavaScript file

Apache2mod-proxyreverse-proxy

I'm using Apache 2.4.27 as a Reverse Proxy and I am having trouble with some of my ProxyHTMLURLMap directives.

I have a proxy-server and a backend server, called server1.

I have a JavaScript fragment appear in an HTML as well as a separate JS file. The ProxyHTMLURLMap incorrectly modifies the HTML file, and does not modify the JS file.

How do I define this URLMap to affect both places?

Here's my config fragment:

LogLevel warn proxy_html:trace3

<Location "/server1/">
    ProxyPass https://server1:5443/ ping=2
    ProxyPassReverse https://server1/

    ProxyHTMLEnable On
    ProxyHTMLExtended On

    SetOutputFilter INFLATE;proxy-html;DEFLATE

    ProxyHTMLURLMap /rpc/ /server1/rpc/
</Location>

The JavaScript fragment that appears in both HTML and JS files is:

abc.get({
    url: "/rpc/getdefaultcfg.asp",
    onrcv: function(arg) {
        if (arg.STATUS == 0) {
            default_Lang = WEBVAR_JSONVAR_GETDEFAULTLANG.WEBVAR_STRUCTNAME_GETDEFAULTLANG;
        }
        onload = loadInit();
    }
});

The HTML file is modified by the ProxyServer twice by updating the matching line to:

url: "/server1/server1/rpc/getdefaultcfg.asp",

Notice the duplicate "server1". Why is it being duplicated?

This HTML file includes the JS file as follows:

<script language="Javascript" src="index.js"></script>

However, the index.js does not get modified by the proxy server. It remains untouched as:

url: "/rpc/getdefaultcfg.asp",

Why is the JS file untouched by the URLMap directive? Any ideas much appreciated!

FYI, the httpd_access_log file shows the 404 errors on line 2 and 5 below:

1. 192.1.0.76 - "GET /server1/index.html HTTP/1.1" 200 946
2. 192.1.0.76 - "GET /rpc/getdefaultcfg.asp HTTP/1.1" 404 223
3. 192.1.0.76 - "GET /server1/page/disable_javascript.html HTTP/1.1" 200 212
4. 192.1.0.76 - "GET /server1/page/blank.html HTTP/1.1" 200 -
5. 192.1.0.76 - "GET /server1/server1/rpc/getdefaultcfg.asp HTTP/1.1" 404 126

Finally, the output from proxy_html tracing:

[proxy_html:trace3] mod_proxy_html.c(265): [client 192.1.0.76:51880] C: matched /rpc/, substituting /server1/rpc/
[proxy_html:trace3] mod_proxy_html.c(265): [client 192.1.0.76:51880] C: matched /rpc/, substituting /server1/rpc/
[proxy_html:trace1] mod_proxy_html.c(827): [client 192.1.0.76:51880] Non-HTML content; not inserting proxy-html filter, referer: https://proxy-server/server1/page/header.html
[proxy_html:trace1] mod_proxy_html.c(827): [client 192.1.0.76:51880] Non-HTML content; not inserting proxy-html filter, referer: https://proxy-server/server1/page/header.html
[proxy_html:trace1] mod_proxy_html.c(827): [client 192.1.0.76:51892] Non-HTML content; not inserting proxy-html filter, referer: https://proxy-server/server1/page/login.html
[proxy_html:trace1] mod_proxy_html.c(827): [client 192.1.0.76:51892] Non-HTML content; not inserting proxy-html filter, referer: https://proxy-server/server1/page/login.html
[proxy_html:trace1] mod_proxy_html.c(827): [client 192.1.0.76:51880] Non-HTML content; not inserting proxy-html filter, referer: https://proxy-server/server1/page/header.html
[proxy_html:trace1] mod_proxy_html.c(827): [client 192.1.0.76:51880] Non-HTML content; not inserting proxy-html filter, referer: https://proxy-server/server1/page/header.html

Matt.

Best Answer

I found the ProxyHTMLURLMap directive lacked the flexibility I needed with the substitutions. Instead I used the Substitute directive.

To ensure the substitutions occurred in each of the file types I cared about, I first used the following directive:

     AddOutputFilterByType SUBSTITUTE text/javascript text/html text/css

Then I used many Substitute directives. An example of one of those is:

    # Fixup min.js urlPath() references
    Substitute "s#(urlPath\(\))#$1 + \"$npar/\" #q"

One thing to note is that the whole file is scanned for each Substitute directive, so try to cram as many substitutions as you can into one. For example, I used regex to cover several cases:

    Substitute "s#(\"|\')/(api|app|images|source|style)#$1/$npar/$2#q"

If you have a big file, and lots of individual Substitute directives, then the performance penalty is significant!

Matt.

Related Topic