Re: Recursively scraping web pages for embedded links and files

Tim- thank you again for your response. I may send some specific questions
via email, but right now my questions are general enough that responses in
the public record may help others.

I've managed to get the HTML of a target page (the top page of interest)
using the code at the bottom of this post. I have two questions;

(1) The returned object (SourceHTMLText)- what exactly is it? I tried
setting Sheet1.range("A1").text equal to it and got an error; either it
isn't text, or maybe there is some limit where I can't assign a string over
a certain size to a cell? I pushed it to a msgbox, and it did show on the
screen (although truncated, because the HTML is long for a messagebox).

(2) What is the best way to parse the content for each instance that starts
with "<A HREF= " and ends with the next ">"?
In the past, I've read flat files but I could read a line at a time and look
for what I wanted. Now I have the whole text at once, so I have to handle it
all together. Also, the text may span more than one line, which wouldn't
work with the way I used to look for short strings anyway. Should I be using
Regex? I saw some "simple" tutorials that tried to cover every aspect of
regex, but since they cover a lot more than I need to know, I found it
difficult to extract how to look for strings that have a specific start
sequence ("<A HREF="), and always terminate with the very next instance of
another (">"). Pointers to any tutorials that are relatively easy to
understand would be very helpful.

(2b) In case any referred tutorials don't include the info- if I do use
Regex, and I do expect multiple matches within the document, what is best
practice for storing those matches and using them- does Regex automatically
build an array of matches?

Many thanks,
Keith
XL2003


Sub GrabPage()

Set objHTTP = CreateObject("MSXML2.ServerXMLHTTP")
URL = "http://ourdomain/targetpage";
objHTTP.Open "GET", URL, False
objHTTP.setRequestHeader "User-Agent", "Mozilla/4.0 (compatible; MSIE
6.0; Windows NT 5.0)"
objHTTP.send ("")

SourceHTMLText = objHTTP.responseText

End Sub


"Tim Williams" <timjwilliams at gmail dot com> wrote in message
news:u55Lug4bIHA.4684@xxxxxxxxxxxxxxxxxxxxxxx
Starting from the main page you could identify all of the "folder" links
by looking at the URL: each could be clicked in order to drill down into
subfolders, and each of these listed etc etc.

Grabbing URL's to the files will be more difficult: you'll have to
deconstruct the "openDocument()" javascript code to see how it determines
what URL to open. You can't use the javascript href directly in Excel: it
depends on having the js function available.

If you're new to working with HTML docs from Excel then it may be a long
haul. I can help you with specific points but can't provide a solution.
If you prefer you can follow up via email (tim j williams at gmail dot
com: no spaces, etc.).

Tim


"Ker_01" <ker_01@xxxxxxxxxxxxxxxx> wrote in message
news:OnBOfG1bIHA.5164@xxxxxxxxxxxxxxxxxxxxxxx
This is a followup to a post from yesterday (Thanks to Tim Williams for
responding). I have more information now, and felt it warranted a second
try to see if there is way to do this now that we've gotten the documents
exposed via the web interface. Using XL2003 on WinXP.

We have a corporate web application that exposes various documents in
multiple levels of subdirectories. My belief is that these are stored in
a database, but now they are directly accessible via web links through
this web application, so where they come from hopefully doesn't affect
what I am trying to accomplish.

Starting from the main page of the web application, I need to scrape the
entire directory tree and capture some of the details (javascript links
to .doc and .pdf files that can be opened through IE6 via 'dedicated'
URLs for each document). I'm sure I'll have more questions once I start
dissecting the HTML, but for starters I need to understand how to even
scrape multiple levels within the directory tree of a website. I've
copied in some of the URLS (changed slightly for corporate security) to
give a sense of what I'm working with.

Top of tree:
http://ourserver.com/rtsa-bin/PermaSite.dll/aaumain.htm?site=omatcone&pagetitle=M%20S%20-%20L

I can click a link to go to the next level of subfolder:
http://ourserver.com/rtsa-bin/PermaSite.dll/aaudisplayfolder.htm?folderpathID=0b00043d8000cdb6&foldername=M+R&se=omatcone&pagetitle=M

Third level of folder:
http://ourserver.com/rtsa-bin/PermaSite.dll/aaudisplayfolder.htm?folderpathID=0b00043d8000d337&foldername=EPD+100000-109999&site=omatconec&pagetitle=M

and so on.

A sample link for a single document within one of the pages in the web
tree/directory is:
javascript:openDocument('0900043d802b3528');

where clicking that link ultimately opens:
http://ourserver.com/Documentation/03451TRs142.pdf

Ultimately I need to recreate all the links in an Excel workbook so users
can click on a hyperlink and access the relevant document. An Excel
hyperlink that uses the javascript:opendocument command is totally fine
with me, but first I need to collect them all. Alternatively I'll have to
figure out how to cycle through each javascript command anyway, then
identify the URL it opened (which sounds harder).

Any advice or code snippets greatly appreciated- I haven't done anything
with HTML at all.

Thanks,
Keith





.