Of course, as the post states, I was originally looking for a way to make a graph showing the relationship among PHP files in regard to include statements.

Well, I'm home sick and after a few hours of trying to find an easy, out-of-box solution I gave up and rolled my own Python script to make me a DOT graph file.

I didn't have anything better to do.

The results are pretty simplistic, but I'm happy enough with it for now.

The Python script takes three arguments: the directory in which the PHP files exist, whether to search recursively or not (0=no, 1=yes), and the name of the output file as such:

$ python makeDOT.py blog/wordpress 1 wordpressIncludes.dot


#####
#importing modules
import glob, re, sys, os, fnmatch
br = "\n"
tab = "\t"

#####
#exiting if all 3 arguments are not passed via command line
def fail():
    print ("ERROR: " + str(len(sys.argv)-1) + " of 3 required arguments provided.")
    sys.exit()

#####
#getting arguments passed via command line

#testing for root DIRECTORY string
try: myDir = sys.argv[1]
except: fail()

#testing for RECURSION boolean
try: myRec = sys.argv[2]
except: fail()

#testing for OUTPUT filename string
try: myFile = sys.argv[3]
except: fail()

#####
#making list of PHP files within DIRECTORY
if myRec == "0": #without recursion
    myDir2 = myDir + "/*.php"
    PHP_list = glob.glob(myDir2)
elif myRec == "1": #with recursion
    PHP_list = []
    for dirname, dirnames, filenames in os.walk(myDir):
        for filename in filenames:
            if fnmatch.fnmatch (filename,("*.php")):
                match = os.path.join(dirname,filename)
                PHP_list.append(match)

#make an empty list;
#tuples will go in the list;
#each tuple will contain a PHP filename and a PHP filename it includes
includeList = []

#iterate through each PHP file and place tuples in the list
for phpFile in PHP_list:
    fileOpen = open(phpFile, "r")
    #for each line in a PHP file
    for line in fileOpen:
            m = re.match(r"(.*)include(.*\()(.*)\)", line) #for include(),include_once()
            if m:
                matchFile = m.group(3)[1:-1]
                if matchFile[-4::] == ".php": #only PHP files
                    phpFile = phpFile.replace("\\","/")
                    matchFile = matchFile.replace("\\","/")
                    matchFile = matchFile.replace("\"","")
                    matchFile = matchFile.replace('\'',"")
                    includeList.append([phpFile[len(myDir)+1:], matchFile])
            else: pass

            m = re.match(r'(.*)require(.*\()(.*)\)', line) #for require(), require_once()
            if m:
                matchFile = m.group(3)[1:-1]
                if matchFile[-4::] == '.php': #only PHP files
                    phpFile = phpFile.replace("\\","/")
                    matchFile = matchFile.replace("\\","/")
                    matchFile = matchFile.replace("\"","")
                    matchFile = matchFile.replace('\'',"")
                    includeList.append([phpFile[len(myDir)+1:], matchFile])
            else: pass

#####
#creating DOT file
dot = open(myFile, "w")

#writing to DOT file
dot.write("digraph {" + br)
for a,b in includeList:
    dot.write(tab)
    dot.write("\"")
    dot.write(a)
    dot.write("\"")
    dot.write(" -> ")
    dot.write("\"")
    dot.write(b)
    dot.write("\"")
    dot.write(";")
    dot.write(br)
dot.write("}")
dot.close()

#####
#exiting
sys.exit()

I ran the Python script on the PHP scripts for MXMLiszt.

Then I used the "circo" layout engine in Graphviz – specifically the Gvedit.exe application – on this resultant DOT file.

Here's the result:

Obviously, I want to extend that experience to running XQueries online. So far, I know this can easily be done with a native XML database server like eXist. That is to say, when I installed eXist on my local machine and made it go live: bam! – instant XML server with built-in XQuery functionality, accessible from anywhere.

Another way is to use one of the more popular web-scripting languages to execute XQuery syntax. I nearly killed myself this weekend trying to install the Zorba PHP binding for XQuery (i.e. run XQuery natively from within a PHP script). I just couldn't get all the dependencies successfully installed on my virtual install of Ubuntu Netbook Remix (BTW: I use Sun's VirtualBox for virtualization). Perhaps I'll be able to make it work another time.

Now, even though it makes all the sense in the world to stick with a native XML server like eXist if I want to make a large collection of searchable XML documents online (and I do), I'm feeling non-sensical. What I decided to try was to run my computer as a more traditional server using the X-Apache-MySql-PHP model, specifically WampServer.

From there, I placed my sample document, "books.xml" and my "test.xquery" file from last time:

<ul>
{
for $x in doc("books.xml")/bookstore/book/title
order by $x
return <li>{data($x)}</li>
}
</ul>

in WAMP's "www" directory – i.e. the directory which is accessible from the browser, the place where one would put all their HTML files, etc. for the world to see.

Of course, I was still missing the actual XQuery processor at this point. What I tried is to put the relevant executables for Saxon in the "www" directory as well.

… Now, I'm sure this is totally unsafe or something, but I was just testing and I only make my computer "go live" as a server for short periods of time.

Anyway, from there I used PHP to call the Saxon processor and to display the results in the browser. It actually worked!

Here's the code:

<?php
echo "Hi. I'm going to use XQuery to list the books alphabetically.";
exec("query.exe test.xquery !indent=yes", $results);
foreach ($results as $value)
    {
       echo "$value";
    }
?> 

You can see that the PHP "exec" command called the Saxon executable named "query.exe" and executes the "test.xquery" file.

It saves the results in a variable called "results" and then prints each value of the results in the browser.

For now, I'm OK with this, but I need to eventually do the same thing with the Java version of Saxon, I suppose, if I'm ever going to run this on a Linux server. It shouldn't present any new hurdles, but I need to try it to make sure, of course.

If anyone out there has any thoughts or recommendations on a more elegant method to achieve these results – and what the security risks of the approach I've outlined presents (since I didn't use a CGI-bin), please speak up. I'm all ears.

…

Update, November, 2009: No problems with using the Java version of Saxon. I did place it in a "bin" directory on my local server so that the Java file wouldn't reside in the directory that users have direct access to.