I have a set of lines say "She is a little girl". I want to highlight "She" when voice "she" reads, then highlight "is" when voice "is" reads and so on...How to do this.
This requires two things:
You're going to have to remember the location of each word as you put it on the screen.
You will have to measure the timing of each word in your sound file so the highlights are in sync. There's probably a fancy way to do that by detecting the sound level while it's playing, but if I were doing it, I would simply create a list of timing values and emit the text at those times. I'd test it over and over again until I got it right.
Related
I'm using a CRichEditCtrl to edit a computer language, and on every change to it I'm calling SetSelectionCharFormat on the current line of text (as reported by LineFromChar(-1)) to highlight the syntax. (EG: comments in green, section headings in a bigger font, compilation errors in red, etc.) Note this language doesn't have multi-line features such as a C comment where typing /* on one line makes following lines part of a comment too; for any given character change I only need to change the color of the current line.
It all looks like its working fine.
However there are some weird issues. One is, when multiple lines of text is selected from somewhere else, and pasted. My OnUpdate() is called but is naively assuming that the only line that potentially needs re-formatting is the one returned by LineFromChar(). That suffices when the user is typing character by character, but it means that after receiving a multi-line paste, the program only reformats the last line of the pasted text. How can it know where the start of the insert was?
OnUpdate is called inside the Paste operation.
It should be possible to subclass the RTF control and to intercept the WM_PASTE message. If WM_PASTE is not used internally it might be possible to use EM_PASTESPECIAL. If even tis message isn't sent, you have to interecept the Ctrl+V that causes the paste Operation.
Than you can determine the starting position of the paste operation.
Spy++ might be helpful to determine the message flow in the RTF control.
I posted a question similar to this earlier, however, after thinking about it and testing the answers, I believe I misinterpreted the answers and the answerer(s) misinterpreted me. The original question is here. I think people believed that I just wanted to highlight strings, I didn't state my exact purpose. So, I will now:
What I've been trying to do lately is create a 100% from scratch text box in C++ CLR using GDI+. I've gotten to the challange of placing the caret when the user clicks in the textbox. Doing simple math (Where they clicked divided by line width) I can figure out which line they clicked. But in order to get the character clicked, I need (unless there are better ways) to compare the bounding rectangles of all the characters in the line and place the caret before the one the mouse fits into. In order to do this, I need to get the exact bounds of each individual character, not an entire string.
I've already tried a few things, none of which seemed to work:
Graphics::MeasureString is not recommended by anyone, nor does
it give what I want
TextRenderer::MeasureText is more accurate, but for this not accurate enough
Graphics::MeasureCharacterRanges has a 32 character
cap, and I'm expecting lines to be over 32 characters long in some
cases
I believe I can't use these methods, unless there are ways around their limitations. I hope I made my problem and expected solution a lot more clear than I previously did.
Because of the way text is kerned and anti-aliased, the boundary of a character depends on all of the characters to the left of it. However you don't need to know every character boundary, only the ones on either side of your click point. You can find those with a binary search - split your string in half, measure that (using TextRenderer::MeasureText), and determine if it's to the left or right of your click point. Keep narrowing down the size of the string until there's only one possibility remaining.
There's a completion type that isn't listed in the vim help files (notably: insert.txt), but which I instinctively feel the need for rather often. Let's say I have the words "Awesome" and "SuperCrazyAwesome" in my file. I find an instance of Awesome that should really be SuperCrazyAwesome, so I hop to the beginning of the word, enter insert mode, and then must type "SuperCrazy".
I feel I should be able to type "S", creating "SCrazy", and then simply hit a completion hotkey or two to have it find what's to the left of the cursor ("S"), what's to the right ("Crazy"), regex this against all words in the file ("/S\w*Crazy/"), and provide me with a completion popup menu of choices, or just do the replace if there's only one match.
I'd like to use the actual completion system for this. There exists a "user defined" completion which uses a function, and has a good example in the helps for replacing from a given list. However, I can't seem to track down many particulars that I'd need to make this happen, including:
How do I get a list of all words in the file from a vim function?
Can I list words from all buffers (with filenames), as vim's complete does?
How do I, in insert mode, get the text in the word before/after the cursor?
Can completion replace the entire word, and not just up to the cursor?
I've been at this for a couple of hours now. I keep hitting dead ends, like this one, which introduced me to \%# for matching with the cursor position, which doesn't seem to work for me. For instance, a search for \w*\%# returns only the first character of the word I'm on, regardless of where I'm in it. The \%# doesn't seem to anchor.
Although its not exactly following your desired method in the past I've written https://github.com/mjbrownie/swapit which might perform your task if you are looking for related keywords. It would fall down in this scenario if you have hundreds of matches.
It's mainly useful for 2-10 possible sequenced matches.
You would define a list
:SwapList awesomes Awesome MoreAwesome SuperCrazyAwesome FullyCompletelyAwesome UnbelievablyAwesome
and move through the matches with the incrementor decrementor keys (c+a) (c+x)
There are also a few other cycling type plugins like swap words that I know of on vim.org and github.
The advantage here is you don't have to group words together with regex.
I wrote something like that years ago when working with 3rd party libraries with rather long CamelCasePrefixes in every function different for each component. But it was in Before Git Hub era and I considered it a lost jewel, but search engine says I am not a complete ass and posted it to Vim wiki.
Here it is: http://vim.wikia.com/wiki/Custom_keyword_completion
Just do not ask me what 'MKw' means. No idea.
This will need some adaptation to your needs, as it is looking up only the word up to the cursor, but the idea is there. It works for current buffer only. Iterating through all buffers would be sluggish as it is not creating any index. For those purposes I would go with external grep.
I'm making a game gui api and I'm wondering how to implement tabs. I'm using freetype for text. When I try to render '\t' It looks like a square. I'm wondering how tabs are implemented because they are not a fixed width.
Thanks
For a fixed-width font you could compute how many spaces to the next tab stop, but the general solution is to stop rendering when you hit a tab, move to the next tabstop, and then render the text that comes after the tab character starting from there. Where the tabstops are is up to you, but a good default is probably something like every 8 ems.
The simplest strategy is to swap a tab out for a some number of spaces. I.e. when the user pushes tab, pretend they pushed tab four times instead. (Or just print out four spaces, whatever works for you)
I am working with iCal entries:
BEGIN:VEVENT
UID:944f660b-01f8-4e09-95a9-f04a352537d2
ORGANIZER;CN=******
DTSTART;TZID="America/Chicago":20100802T080000
DTEND;TZID="America/Chicago":20100822T170000
STATUS:CONFIRMED
CLASS:PRIVATE
X-MICROSOFT-CDO-INTENDEDSTATUS:BUSY
TRANSP:OPAQUE
X-MICROSOFT-DISALLOW-COUNTER:TRUE
DTSTAMP:20100802T212130Z
SEQUENCE:0
END:VEVENT
BEGIN:VEVENT
UID:aa132e2b-8a8d-4ffc-9e54-b75249e78c72
RRULE:FREQ=DAILY;COUNT=12;INTERVAL=1
SUMMARY:***********
X-ALT-DESC;FMTTYPE=text/html:<html><body><div style='font-family:Times New R
oman\; font-size: 12pt\; color: #000000\;'></div></body></html>
LOCATION:Map Room
ORGANIZER;CN=*********
DTSTART;TZID="America/Chicago":20100730T080000
DTEND;TZID="America/Chicago":20100730T170000
STATUS:CONFIRMED
CLASS:PUBLIC
X-MICROSOFT-CDO-INTENDEDSTATUS:BUSY
TRANSP:OPAQUE
X-MICROSOFT-DISALLOW-COUNTER:TRUE
DTSTAMP:20100727T025231Z
SEQUENCE:0
EXDATE;TZID="America/Chicago":20100810T080000
EXDATE;TZID="America/Chicago":20100807T080000
BEGIN:VALARM
ACTION:DISPLAY
TRIGGER;RELATED=START:-PT5M
DESCRIPTION:*********
END:VALARM
END:VEVENT
I need to parse out starting and ending times. I have a comparison function that determines if the passed in event is between the two times. Due to the increased complexity in calculating the times I plan on not supporting the recurrance series. I would like to play the safe side and make sure my code only reads the first event as a match and not the second. So I have the following RegEx with the single-line option:
BEGIN:VEVENT.+?
DTSTART;.+?([0-9]{8})T([0-9]{6})
DTEND;.+?([0-9]{8})T([0-9]{6}).+?
END:VEVENT
This gets me the start and end times of both entries. My thought was to only match ones that don't have FREQ= between the BEGIN:VEVENT and DTSTART. I don't understand how to do this, however. I was wondering if someone could help me out here?
I realize at a certain point a full blown parser is a better option, but I am unskilled with parsers and I am under a slight time constraint. I have tried using the !? operator without success.
It's harder to write a regex to match for things you don't want then to match the things you do want. Usually when I run into this situation, I find it easier and faster to do things in two steps. In this case, I'd probably find all events that do contains FREQ=, remove those events, then continue matching on the result for the start and end times I want. Could you post the regex you tried with !?, because maybe it's easy to fix... Also, I assume this is in Objective-C, and I'm guessing the environment you're using does support !? (but not all of them do)...
UPDATE
Ok, try this one:
BEGIN:VEVENT.+?
(?<!FREQ=.+)DTSTART;.+?([0-9]{8})T([0-9]{6})
DTEND;.+?([0-9]{8})T([0-9]{6}).+?
END:VEVENT
Why not use a PHP iCalendar parser?
http://www.phpclasses.org/browse/file/16660.html