shawn

I had made a template for CNET to address Connectify (nifty tool if you're looking for this kind of thing - http://www.connectify.me/), but over the last 48 hours it stopped working. REALLY hoping that your new base template addresses whatever is going on.

Wagner, that was AWESOME. Thank you! It fixed all 30 of the broken SF apps I have. CTS, my guess is that some of your apps are setup without any form of grace for the new "?nnn" format in use.

That works for me, and I'm happy to give you constructive feedback.

Thanks for the tip, but I don't understand. Is it even possible to set a dynamic referer in Ketarin? I've tried using variables in the spoof referer box without luck (for another site, but with similar requirements).

I've found an example situation where I need to specify a user-agent for specific apps. SourceForge has now implemented some form of connection and user-agent monitoring, I suspect it's along the same lines as the CSS trick from the "Cookies for Comments" WP plugin (http://ocaoimh.ie/cookies-for-comments/). Anyway, back to the request. Since the SF site is now tracking user-agents and sessions (you can tell based on the specific content returned in the styles), I went looking for alternatives. I know that most linux admins (including myself) use wget to download various installation packages for our servers. So, I tried that. wget is still working fine, and it uses a user-agent of something like "wget/1.2.3" - this suggests that SF would allow Ketarin to operate fine if it used a wget user-agent. However, I know of some servers (including some I manage) that explicitly forbid certain user-agents, including wget, so it would be best if it were something I could assign dynamically for each app. While not directly related (I don't think), it would also be cool if we could re-order the variables within the variable list and assign the user-agent, referer or other advanced headers directly within the variable editor. Maybe treat variables named certain things as specific values for operations - that is, a variable named "user-agent" would be used as the user-agent for all HTTP requests. A variable named "website" and "notes" could take the place of the "website" and "notes" fields that appear on the "information" tab. Now that I'm using the variables functionality far more, I've really taken a liking to it's capabilities. But I find myself getting stuck when I want to make it do more without hacking the code. The user-agent thing is mentioned briefly here: http://ketarin.canneverbe.com/forum/viewtopic.php?id=266

You can do this with double-click + ALT. http://ketarin.canneverbe.com/forum/viewtopic.php?id=266

Sigh. SourceForge changed their link structure, breaking over 30 of my apps. This is infuriating.

I managed to develop what I believe is a consistent successful download path for SF apps. It requires parsing the target app based on whatever you want to consume. This works fine for most apps with a parse like this (saved to "dl"): URL: http://sourceforge.net/projects/tuxtype/files/ Pattern: [t'"]/projects/([^"']+)/download['"] The captured portion then gets passed to the download as: http://downloads.sourceforge.net{dl:replace:/files/:/}?download The captured portion in bold up there can be modified to help limit the version/type/style of the download (for example, x64, msi, 7z and so on).

Well crap. Thanks for the help and advice in the time I've been here.

I don't think it's your escaping that is the problem, but the rest of the regex pattern. Try this: {file:regex:[^\\]+$}

Here's an example: http://www.bullzip.com/products/pdf/info.php#slice_versionhistory It might not be the best example, but it demonstrates how slices work pretty well. Ideally, this would function by pre-parsing the resulting HTML from the page for the "slice", and excluding all external content.

I've found that with FileHippo you really need to enable the advanced option to "not download the application simultaneously with other applications".

I know the feeling. I put in over 500 hours in May and am on track to do close to the same this month. It's painful. I've been using Ketarin to try to minimize the amount of manual work I have to do - but some sites (like CNET) require extra coding to keep things clean or change things often, and that's a PITA. Contact me by email if you're looking for work.

To my knowledge, regexreplace provides several search strings, but only one *replacement* string. Am I misunderstanding it's operation? I don't see how it could be used to replace the month *name* with it's numeric equivalent in a single operation.

{startuppath} isn't being replaced properly in global commands, so the following: echo {category} /// {appname} {version}>> "{startuppath}\Updates.txt" Actually runs as: echo Utilities /// MyApp 1.0>> "{startuppath}\Updates.txt"

I was using a very similar method to the "url: 'http://downloads.sourceforge.net/....exe'" method detailed above, but the problem is that it only publishes this information for the most recent download, which might be a beta. If you don't want the betas, this is bad.

Per a discussion from here: http://ketarin.canneverbe.com/forum/viewtopic.php?id=150 I'd really find it useful to be able to perform mass replacements on a variable by passing in an array of "find" and an array of "replace" values. This would work something like this: {myvar:arrayreplace:,:Jan,Feb,Mar,Apr,May,Jun,Jul,Aug,Sep,Oct,Nov,Dec:01,02,03,04,05,06,07,08,09,10,11,12} Where "arrayreplace" is the function name, "," is the delimiter, "Jan,Feb,Mar,Apr,May,Jun,Jul,Aug,Sep,Oct,Nov,Dec" is the find array, and "01,02,03,04,05,06,07,08,09,10,11,12" is the replacement array. {myvar:arrayreplace:delimiter:findarray:replacearray} Pushing that against the following values would yield these respective results: 2010-May-08 = 2010-05-08 Dec 11, 2004 = 12 11, 2004 I hate June bugs = I hate 06e bugs Here's another place it would come in handy: {myvar:arrayreplace:|:, Build| Build | part | beta | b:.|.|.|.|-} This would convert: Picasa 3.6, Build 3622 = Picasa 3.6.3622 Audacity 1.3 beta 22 = Audacity 1.3.22 MyApp 2.9 b4 = MyApp 2.9-4

For those with a version number, this is a better regex (I hope this comes thru): h1>[^<>]+\sv([\d\.]+)< And for those without, I've taken to parsing the updated date: Updated\:\s*([^<>]+)\s*< Then I play with the order a bit to try to organize it a little better. "2010May08" (mine) vs "May 8, 2010" (original). Hey, any chance of getting the ability to perform mutliple textual replacements on a variable? I'd like to be able to replace "Jan,Feb,Mar,Apr,May,Jun,Jul,Aug,Sep,Oct,Nov,Dec" with "01,02,03,04,05,06,07,08,09,10,11,12" - in some form of array replacement operation so the above version number could be transformed from "2010May08" to "20100508". I could see this being used on more than just dates, too.

It seems to me that there are several common sites that we end up coding templates for, then plugging in various values to feed them. FileHippo handling is great, but the selection is limited and they only handle the 32bit versions of most apps, and that's not working for me. Since we end up creating a different template for each job even tho we are using the same sources. That means we have to update ALL the templates to fix each job that uses a different source when they change something (like the search form or directory structure). I think it would be better to have the ability to create prototemplates and then select them as the basis for our template, without having to actually rebuild the template from scratch each time. This would minimize most of the functional variables and scraping that had to be done on the backend for each of these templates, and really reduce maintenance time. Instead of a toggle between URL and FileHippo, make it a toggle between URL and a drop-down list to select the prototemplate base. FileHippo, MajorGeeks, CNET Downloads...or any other base XML file within a given folder or imported into the database.

I'm interested.

I've been thru the source now, and it looks like the only way to do what I'm after is to set variables for each of the values (which are enumerated thru in "ReplaceAllInString" on line 908 in Updater.cs), and I don't see a way to make them static. That is, it looks like all variables are assigned from URL content. I'm now able to do this using the "textual content" option from the "variables" page. Hopefully others that need to be able to do this will find this method.

Shouldn't you have a backslash at the end of the destination folder? "C:\Documents and Settings\User\Documents\installers\installable\" vs "C:\Documents and Settings\User\Documents\installers\installable"

I had assumed this was a feature, actually. I have noticed the same behavior and assume that the idea is to ensure that the user can observe changes within groups without having it re-sort during processing. A simple refresh or click of the header re-sorts the list, so it makes sense that the presentation of the formerly "failed" items would let you see the current status after syncing again without losing your place or having to hunt them down.

I'm trying to automate the post process for syncing with Ketarin. I need to be able to access all of the information from the application profile directly within the "default command to execute after updating" box. I've tried "usernotes", "notes", "website", "url", "websiteurl", "link" and others but can't find a way to access the data I stored in the "information" tab for an application profile. Can anyone offer me any tips? Thanks!

Use regex, as so: {yourvariablehere:regex:(...)} Not tested, but should work fine.