Posted by Jerry Stuckle on 09/01/07 01:54

The Natural Philosopher wrote:
> Jerry Stuckle wrote:
>> The Natural Philosopher wrote:
>>> ELINTPimp wrote:
>>>> On Aug 31, 2:18 pm, giloosh <giloos...@gmail.com> wrote:
>>>>> On Aug 30, 9:11 am, Jan Thomä <k...@insomnia-hq.de> wrote:
>>>>>
>>>>>
>>>>>
>>>>>> Jerry Stuckle wrote:
>>>>>>> Jan Thomä wrote:
>>>>>>>> I always use the placeholder notation for doing SQL. Concatening
>>>>>>>> SQL
>>>>>>>> strings from input values is almost certainly a safe path to SQL
>>>>>>>> injection.
>>>>>>> Not if you properly cleanse your input. Ensure numeric values are
>>>>>>> really numeric, and string values are processed through
>>>>>>> mysql_real_escape_string(), for instance.
>>>>>> Thanks for the input, and yes I agree, you should definitely
>>>>>> cleanse your
>>>>>> input before feeding it to the database. My point was simply, that
>>>>>> when you
>>>>>> give this kind of work to a framework and always use the ?
>>>>>> notation, you
>>>>>> are safe from injection, even if you forget to check a single input
>>>>>> variable (which surely happens from time to time). Also you don't
>>>>>> have to
>>>>>> do the conversions to different formats manually, so you save a
>>>>>> bit of time
>>>>>> and effort.
>>>>>> Best regards,
>>>>>> Jan
>>>>>> --
>>>>>> _________________________________________________________________________
>>>>>>
>>>>>> insOMnia - We never sleep...http://www.insOMnia-hq.de
>>>>> thanks for the feedback...
>>>>> isn't is safe to just put quotes around the variable
>>>>> $id = $_POST['id']
>>>>>
>>>>> $q = "select * from table1 where id = '$id'";
>>>>>
>>>>> even if id holds none numeric characters, it's still safe... no?
>>>>
>>>> No. That introduces sql injection and cross site scripting
>>>> vulnerabilities. mysql_real_escape_strings, as previously suggested,
>>>> will help prevent against this.
>>>>
>>>
>>> Dunno what that means, but since Ive been working all day on this
>>> sort of stuff..
>>>
>>> 1/. $q = "select * from table1 where id = '".$id."'";
>>>
>>> is what you want. Note the order of quotes and dots.
>>>
>>> 2/. On my server - virgin php5 apache2 setup, all POST data gets
>>> escaped with backslashes into a format that is completely digestible
>>> by MySql. So no special processing is needed before inserting or
>>> updating data from $_POST[] derived variables.
>>>
>>
>> mysql_real_escape_string is much safer - it takes into consideration
>> the charset currently in use, for one thing. Also, being a MySQL
>> function, it knows what MySQL needs or uses.
>>
> Sager than what? Nothing? All characters that are entered in the fields
> make their way into the database unaltered. That's how I like it.
>

Safer than using addslashes().

> Ter is no nede to change anything in THAT area.
>

No, not when you're ignorant about the realities, there isn't.

>
>>> 3/. One coding bug showed up interestingly when I had done
>>>
>>> $new_id=mysql_insert_id;
>>>
>>> instead of
>>>
>>> $new_id=mysql_insert_id();
>>>
>>> The insert of what surprisinlgly was NOT a syntax error, but a string
>>> called "mysql_insert_id" into an integer field resulted in the value
>>> zero being put in. (Javascript gives you a NaN response when you try
>>> and do maths on a string like (1,000 * 1,000) = Not a Number ;-)).
>>>
>>
>> You would have gotten an E_NOTICE if you had them turned on (which you
>> should have in development).
>
> You should not teach your grandmother to suck eggs.
>

You shouldn't complain about not getting messages when you have those
very messages disabled.

> I had my OWN debugging on. I looked at the echoed query strings and
> realised that what was going in was weird and didn't match what was in
> the database subseqeuntly, Simple innit?
>

Yes, and if you would have had the E_NOTICE turned on, you would have
gotten the message you bitched about not getting. You have no one to
blame but yourself.

> Now thing about all the bugs that E_NOTICE would NOT catch, like having
> the data fields transposed so that the guys name goes in his street
> address and so on.
>

That's not a bug. That's a programmer deficiency.

> Yu would have looked an seen no bugs, I would have looked three imes and
> double checked it and fund it. Because I wasn't relying on anything more
> than hard concentration and nitpicking over the details.
>

And if you would have had E_NOTICE enabled, you would have known the
exact line number of the error the first time you executed the code.

> Why do I get the feeling that you are someone who enjoys displaying a
> superfcial knowledge about programming, but doesn't actually do very
> much OF it in real world situations?
>
>

Been programming for 40 years, asshole. Almost 20 different languages
over that time. And been a successful consultant for the last 17 years.

I probably write more code in a week than you do in a year. At least
*good code*.

Learn from your elders, asshole.

>>
>>> 4/. I did have an issue with redisplaying data that had come from a
>>> POST derived form..now normally I update the database, then read the
>>> data from the database back into the form: In this case I was testing
>>> 'failed to update, re-enter some data' and the backslashed stuff gave
>>> me issues with quotes and backslashes.
>>>
>>> Basically a variable that might be "Fred Bloggs" became \"Fred
>>> Bloggs\" after being POSTED. This went into mysql fine using the sort
>>> of query I outlined above. And checking using direct database access
>>> showed the data in *there* was in fact "Fred Bloggs"
>>>
>>> However if I wanted to reinsert that into e.g. text box as i.e.
>>> value="\"Fred Bloggs\"" it didn't work too well at all. Repeated
>>> updates left me with truncated data and/or multiple backlashes
>>>
>>
>> mysql_real_escape_string() solves that problem, also.
>>
>>> The key is as follows - in my particular apache php configs anyway -
>>> to do the following.
>>>
>>> Any POST data that needs to be inserted into input fields and the like -
>>> goes through this:-
>>>
>>> function sanitise($string)
>>> {
>>> $string=stripslashes($string); // remove any backslashes
>>> $string=htmlspecialchars($string); // turn oddities that HTML barfs
>>> // on into ampersand stuff
>>> return $string;
>>> }
>>
>> You shouldn't be doing this on data going into the database.
>
> If you read what I *said*, I specifically said that I *wasn't*. I was
> doing it on data coming from POSTed variables before applying to the
> BROWSER.
>

No, you weren't at all clear about that.

> What part of
>
> "all POST data gets
> escaped with backslashes into a format that is completely digestible
> by MySql. So no special processing is needed before inserting or
> updating data from $_POST[] derived variables."
>
> Do you NOT understand?
>

I understand when someone is clear with their words.

> You are heading for a plonk.

Thank you. From assholes like you, it's an honor.

>
>> htmlspecialchars() is a display function. Use it after reading out of
>> the database.
>>
>
> That is precisely what I said, dickhead.
>

Then be clear about what you write, asshole.

>> What happens, if for instance, you want to create a report to be
>> printed, say using PHP's CLI? Or even MySQL's CLI.
>>
>
> I'll write one. However, in this instance I don't.
>

But the key is you don't know if you'll need one tomorrow, or six months
from now, or never. You've obviously never been on a project which
lasted more then four hours or you'd know that.

> God, a whole post that merely confirms you didn't even read mine before
> jumping in to show off how fucking smart you are (not!).
>

Bullshit. It just goes to show how unclear you were with your posting,
asshole.

>
>>> If it's data derived from MySQL it comes out 'as written' with
>>> everything correct BUT that can confuse the browser..broswers don't
>>> like <> chars or '", but just running it through htmlspecialchars()
>>> is enough to sort THAT all out. Inverted commas and single quotes get
>>> the &blah; type treatment, and they display fine, and gets the
>>> correct stuff into the POST data when submitting the form - I know
>>> cos I sent three hours testing it.
>>>
>>> i.e. when I did a view source my input text box had
>>>
>>> value="&wahtever;Fred Bloggs&whatever;"
>>>
>>> inside it, but that showed up as "Fred Bloggs" IN the box, and POSTed
>>> up correctly as \"Fred Bloggs\" and went into SQL as "Fred Bloggs"
>>>
>>> So the correct php to generate THAT fragment would be
>>>
>>> echo ("value=\"".htmlspecialcars($my_value)."\"");
>>>
>>> And bugger., Ive just noted that the hardlinked copies of my library
>>> - one and the same file - have been busted into two different ones by
>>> my editor. It must be one of those that does a 'create tmp file,
>>> write it, unlink original, rename temp to original' that I so much
>>> hate... The quickest way to break SUNOS4 linked config files was to
>>> use EMACS to edit them..
>>>
>>> man diff..;-)
>>>
>>>
>>>
>>>
>>> }
>>>
>>>
>>
>>

This is really funny. You're totally ignorant, but trying to give
advice. I'd recommend you stick your head back up your ass where it
normally is, and learn to program before giving advice.

Because your advice is shit.

--
==================
Remove the "x" from my email address
Jerry Stuckle
JDS Computer Training Corp.
jstucklex@attglobal.net
==================

[Back to original message]