Getting values from a Huge String format= ("value : number")

[kamhagh]

I'm not sure how i should do this, i'm really bad at strings
but after i execute a linux shell command using QProcess it gives me this(with diffrent values):

Default Adapter - AMD Radeon HD 7700 Series
Core (MHz) Memory (MHz)
Current Clocks : 1020 1125
Current Peak : 1020 1125
Configurable Peak Range : [300-1300] [150-1300]
GPU load : 23%

and i want to make the gpu load appear on a widget and do the same for current clocks, I have already made the widget that displays percent but the problem is i'm not sure how to get the GPU load value so i can give it to my widget

also the percent might change to 100% so its not always 2 characters, the same is for current clocks otherwise i would use indexOf and use mid(indexOf,2);
also i have to make it an int, which i guess toInt does the job

can anyone help me with this?

[p3c0]

@kamhagh If the number of lines are fixed then you can get the exact line i.e here 6th and then split it using section with : as seperator.

[kamhagh]

@p3c0 Thanks for your answer
The problem is sometimes the GPU name might be Long or the user might have More than 1 graphics card. I don't think its a reliable solution!

is there anyway i can take out the line as a separate String if i know the first word index? indexOf can find the "GPU Load :"

[p3c0]

@kamhagh Well then find the line which will have the string "GPU Load" using contains.

[kamhagh]

@p3c0 thanks!
but how can i take out that line using indexOf?(contains only returns true or false)

[p3c0]

@kamhagh If that string is new line separated ("\n"), as it seems, split it into different strings, iterate and then check in it using contains

QStringList list = outstr.split("\n");
foreach(QString str, list) {
    if(str.contains("GPU load"))
        qDebug() << str.section(":",-1);
}

[Chris Kawa]

Split and a loop can be a slow operation if the searched string is long. It also creates a lot of copies.
On a large inputs a faster, more memory friendly and a little more permissive of the format solution would be to use a regexp, e.g.:

auto options = QRegularExpression::CaseInsensitiveOption;
auto m = QRegularExpression(".*GPU load[\\s:]+([\\d.]+)%", options).match(text);
if(m.hasMatch()) {
    float value = m.capturedRef(m.lastCapturedIndex()).toFloat();
    // do something with value
}

This allows varying amount of whitespace between the "GPU load" and the actual number and will tolerate a number in any "C" locale form e.g. 7%, 42%, 100%, 4.54534345% etc.
Note that QStringref::toFloat will work only for "C" locale i.e XXX.XXX% format.
If you expect anything else, e.g. comma instead of dot or thousand separators you should use QLocale::toFloat instead and modify the regexp accordingly.

[kamhagh]

@Chris-Kawa thanks! that looks wierd, and like alien language to me but i'm googling about regexp! and than i guess i'll figure it out :D

[Chris Kawa]

@kamhagh said:

thanks! that looks wierd, and like alien language to me

haha. To you and me both ;) Regexps should not be the first thing to go to in most cases but they are powerful and should land in every programmer's toolbox at some point. They are great at parsing blocks of text and flexibly extracting pieces of it.

Basically in this case the interesting part of the regexp is the ([\\d.]+) part. The () is a capture group. Anything matched in that will be placed in an array of results accessed by capturedRef(). The []+ part says "anything inside the brackets at least once". \\d means "any digit" and . is, well, a dot character.
Other stuff: .* means "any character 0 or more times", [\\s:]+ means "at least one whitespace or :".
The \\ are unfortunate result of c++ standard. The actual characters needed by regexp are just \, but since it's in a c++ string literal we need to escape it and thus it becomes \\.

[kamhagh]

@Chris-Kawa
thanks! now I understand!

But I get this while trying to compile:

**C:\Qt\Qt5.3.0\Tools\QtCreator\bin\regexTemp\main.cpp:14: error: 'options' does not name a type
auto options = QRegularExpression::CaseInsensitiveOption;
^

C:\Qt\Qt5.3.0\Tools\QtCreator\bin\regexTemp\main.cpp:15: error: 'm' does not name a type
auto m = QRegularExpression(".*GPU load[\s:]+([\d.]+)%", options).match(outstr);
^

C:\Qt\Qt5.3.0\Tools\QtCreator\bin\regexTemp\main.cpp:16: error: 'm' was not declared in this scope
if(m.hasMatch()) {
^
^**

after i add c++11 option to .pro file i get these

C:\Qt\Qt5.3.0\Tools\QtCreator\bin\regexTemp\main.cpp:14: error: incomplete type 'QRegularExpression' used in nested name specifier
auto options = QRegularExpression::CaseInsensitiveOption;
^

C:\Qt\Qt5.3.0\Tools\QtCreator\bin\regexTemp\main.cpp:15: error: invalid use of incomplete type 'class QRegularExpression'
auto m = QRegularExpression(".*GPU load[\s:]+([\d.]+)%", options).match(outstr);
^
C:\Qt\Qt5.3.0\5.3\mingw482_32\include\QtCore\qstring.h:78: error: forward declaration of 'class QRegularExpression'
class QRegularExpression;
^

[Chris Kawa]

You are missing an include: #include <QRegularExpression>.

[kamhagh]

@Chris-Kawa
oh..... im going to kill myself :| how could I forget about that:D sorry!

thanks a lot!

but one last question, Do stuff like Program save files(quake arena if you've ever played it, or warcraft, basically old games or metro!) or Raspberry pi's setting(these are all i can think of + MSI Afterburner) use regexps to read save files?

[SGaist]

Hi,

If you want to toy around with QRegularExpression, a new version of the regexp tool can be found in examples/widgets/tools/regularexpression. It might be available only in the dev branch currently but you can build it with any Qt version including QRegularExpression.

Hope it helps

[Chris Kawa]

There's no way to say how these programs parse their files without looking at their source but I doubt they use regexps for that.

Regexps are best when the content you want to parse is fuzzy e.g. it can be uppercase, lowercase, spaces or tabs, comma or dot, unknown or partially known data patterns etc.
A user input of some sort or a spill from some diagnostic tool are good examples, where you want to just grab a piece of data you know the format of and don't care about its surrounding.

If you design your own app and thus define the format of your files you know what is where and you can assume stuff to make the parsing faster e.g "ABC tag always comes after XYZ tag so don't bother to check for it before".
In that case an incremental or structured parser can be a better and a lot faster solution. Regexp are flexible but come at a cost. They're sort of a heavy artillery of parsers.
An example of structural parser is QSettings used with QSettings::IniFormat to parse .ini files or a QDomDocument to parse an xml.
An example of incremental parser is a QXmlStreamReader that provides way to read an xml file "a token at a time" and decide what to do with it.

[kamhagh]

@Chris-Kawa
oh thanks, I just thought it's always faster than doing that!

and thanks @SGaist i will take a look at that!