Is there any way to display a QImage (in memory, not on filesystem) in a QWebFrame without writing the image out to a temporary file?
One option might be using the data URI scheme. You basically base64 encode your picture and write the complete data into the URL.
For example:
data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAABAAAAAQAQMAAAAlPW0iAAAABlBMVEUAAAD///+l2Z/dAAAAM0lEQVR4nGP4/5/h/1+G/58ZDrAz3D/McH8yw83NDDeNGe4Ug9C9zwz3gVLMDA/A6P9/AFGGFyjOXZtQAAAAAElFTkSuQmCC
WebKit should support these URIs either directly, or as part of a HTML source code, like
<img src="data:image/png;base64,blaBLABLA" />
Be aware: if your image is big, you might be running into some constraints. Usually this used more for small stuff like icons.
Damn. Steffen beat me to it! :)
To add, you could convert to base 64 by saving the image to a QBuffer(&byteArray) and then saying byteArray.toBase64()
Related
I need to use some custom font for numbers in my Qt app, but company ask me to use image as font, font file is not allowed(For example: ttf).
the font image like this:
So i wonder is possible to converting the font image to QFont?
like a Qt font generator.
Or any other solution?
There is nothing out of the box that supports what you are looking for. You would need to get around this company rule by building your own TrueTypeFont data programmatically from your image, keeping it in memory only and load that into your application with addApplicationFontFromData(const QByteArray & fontData).
Unfortunately I cannot find a single library that does this, and the TTF format appears dauntingly complex. I would suggest looking into this open source project for font creation. Perhaps you can use this as a starting point? http://fontforge.github.io/en-US/
You may want to find out the reason your company is telling you that you cannot use a custom font. (I just remembered reading that QT has difficulties storing a font as a "resource". If it's not due to copyright issues, then you could suggest not using an image, but storing the entire file in the source code in a text format like base-64. Then convert it back into binary at startup, then load it up as a new font from a QByteArray.
I guess it's about copyright issues ? There are lots of ttf fonts available for completely free commercial use. E.G. have a look at Droid Sans Mono : https://www.google.com/fonts/specimen/Droid+Sans+Mono
Otherwise it would be good to know the reason why ttf is not allowed.
It is possible to create a font from a vector graphic.
If you have resonable high resolution images of every letter (if not already vector graphics) you can try to convert them to vector graphics (SVG).
Afterwards you can use tools to convert SVG to a font. This would be most likely a ttf font which can be used with QFont.
I would like to use SVGs in menus and buttons around a Qt-4.7 application. The images are rendered correctly across linux and windows platforms, however an obnoxious message reading...
couldn't create image from ""
...is printed to the console seemingly as soon as one of these images is loaded or changes state (like highlighting or disabling its container widget). Over the course of the run of the application, many of these lines are printed, leaving a lot of senseless output crufted around reasonable application output.
Poking around the Qt code a bit, this appears to be coming from svg/qsvghandler.cpp:2680 where the line contains the following.
qDebug()<<"couldn't create image from "<<filename;
From the documentation for qDebug, you would think that I could block this by defining QT_NO_DEBUG_OUTPUT at compiletime, but this will only block the application's compiled debug calls, not the one in Qt's SVG library.
So I guess my question is actually two-fold:
As in the title, why is Qt printing this even when rendering SVGs correctly?
Without recompiling Qt or its SVG library, how do I prevent Qt from printing this and crufting the application's output?
I've also posted this question on the QtCentre forums.
To #2: You can prevent qDebug output by using qIntallMsgHandler and writing a handler that just discard the message. Also see this question: How to redirect qDebug, qWarning, qCritical etc output?
Re #1: I'm no expert on SVG but from the Qt source code where that line appears, it looks like Qt is trying to load an image file referenced in an xlink:href attribute. The output you see would seem to indicate that either the value for that attribute is "" or the value is a bunch of white space characters enclosed in "" which is removed by a call to QString.trimmed() which is then used as the name of an image file to be loaded into a QImage. I can't quite discern if the filename is ending up as an empty string or two quotes :)
I would have to guess that either the SVG files you have are filled with empty string xlink:href attributes (probably links or filenames for texture images) or there is something about the SVG file causing Qt to incorrectly parse it. I would lean towards the former.
What is your criteria for concluding that they are being rendered correctly? Perhaps you could open the SVG files in a text editor and look at the xlink:href occurances in it?
I've had a similar issue, I used Adobe Illustrator to create images and then save as SVG. The problem is that hidden layers in Illustrator still gets written to the SVG file, but its display property is set to "none". In my case a hidden layer of a file I "placed" - to get an idea if the background I'm creating would look OK.
This causes your image to render correctly as you say, but Qt complains about the missing invisible layer in the SVG file. If you open the SVG file in a text viewer you might find the hidden layer looks something like this:
<g id="bglogo_x5F_splash" style="display:none;">
<image style="display:inline;overflow:visible;" width="1740" height="438" xlink:href="../logo_preview.png" transform="matrix(1 0 0 1 154 409)">
</image>
</g>
Qt will complain about the missing "../logo_preview.png" file.
Hope this helps
I am looking to write a stream filer in a http module that anyalyze the output html of a page, search through the pages html for any images within the page, check if the source of the image has a full path or relative path. If its a relative path, alter it to have a full path. Anyone ever write anything anything like this? Looking for some examples of such?
I suggest to direct use the WebControlAdapter for the img tag and not go by analyse the full page for many reasons.
Here is an example how you can you this adapter, you can simple change it to just fix the full path:
http://www.codeproject.com/KB/web-image/IePngControlAdapter.aspx
Here is one more example more close to what you ask
Task is simple. Have a tool which will parse CSS on network website, read the links to images from there and download all of them to structured folders to local disc.
The important is that tool should ready to use, and designer should be able to use. Answer "you can write the tool itself is OK", but not acceptible for ready tool .
You can get urls from css manually with regular expressions.
To accompany Roman's answer:
/\burl\(['"]?([^'"\)]+)['"]?\)\b/g
will match as $1 almost all images (almost all meaning, that the central part ['"\)] is a bit too strict and will not match URLs with either of these characters in it).
Note that you might get external stylesheets ('#import'), too, with this regexp.
I have a legacy application that I needed to implement a configuration page for to change text colors, fonts, etc.
This applications output is also replicated with a PHP web application, where the fonts, colors, etc. are configured in a style sheet.
I've not worked with CSS previously.
Is there a programatic way to modify the CSS and save it without resorting to string parsing or regex?
The application is VB6, but I could write a .net tool that would do the css manipulation if that was the only way.
You don't need to edit the existing one. You could have a new one that overrides the other -- you include this one after the other in your HTML. That's what the "Cascading" means.
It looks like someone's already done a VB.NET CSS parser which is F/OSS, so you could probably adapt it to your needs if you're comfortable with the license.
http://vbcssparser.sourceforge.net/
One hack is to create a PHP script that all output is passed through, which then replaces certain parts of CSS with configurable alternatives. If you use .htaccess you can make all output go through the script.
the best way i can think of solving this problem is creating an application that will get some values ( through the URL query ) and generate the appropriate css output based on a css templates
Check this out, it uses ASP.NET and C#.
In my work with the IE control (shadocvw.dll), it has an interesting ability to let you easily manage the CSS of a page and show the effects of modified CSS on a page in realtime. I've never dealt with the details of such implementations myself, but I recommend that as a possible solution worth looking at. Seeing as pretty much everyone is on IE 6 or later nowadays, you can skip the explanations about handling those who only have IE 5,4,3 or 2 installed.
Maybe the problem's solution, which is most simple for the programmer and a user is to edit css via html form, maybe. I suppose, to create css-file, which would be "default" or "standart" for this application, and just to read it, for example, by perl script, edit in html and to write it down. Here is just the simple example.
In css-file we have string like:
border-color: #008a77;
we have to to read this string, split it up, and send to a file, which will write it down. Get something like this in Perl:
tr/ / /s;
($vari, $value) = split(/:/, _$);
# # While you read file, you can just at the time to put this into html form
echo($vari.":<input type = text name = ".$vari." value = ".$value.">");
And here it is, you've got just simple html-form-data, you just shoul overwrite your css-file with new data like this:
...
print $vari[i].": ".$value.";\n";
...
and voila - you've got programmatical way of changing css. Ofcourse, you have to make it more universal, and more close to your particular problem.
Depending on how technically oriented your CSS editors are going to be, you could do it very simply by loading the whole thing up into a TextEdit field to let them edit it - then write it back to the file.
Parsing and creating an interface for all the possibilities of CSS would be an astronomical pain. :-)