Parsing ⅛ and ⅓ Characters from actxserver Outlook Mail object Body and Converting to Floats

Hi all

I am parsing Outlook mails in Matlab by actxserver and regexp.

Some mails contain fraction characters as below

The ½,¼,¾ characters are read ok, but the eighths (⅛,⅜,⅝,⅞) and thirds (⅓,⅔) are present in the body property of the mail object as "?" [char(63)] as per below screenshot from the command-line print of the mail body.

Matlab recognises only ¼ ½ ¾ [char(188:190)] so I guess I need to access non ASCII chars. Its not clear whether the issue is Matlab's 16bit unicode or the actxserver object. The characters are available on Windows Vista Arial font as U+215C,E etc

You can verify this for yourself by emailing yourself a mail with the subjectline

⅛¼⅓⅜½⅝⅔¾⅞

and then running the code below in matlab to regexp this subjectline of the mail in your inbox. Put a breakpoint at the regexp line to inspect what the subject variable looks like, should see "?" in there.

Two questions here:

1. How could I extend Matlab's ASCII set to read these characters

2. Is there a neat way to convert them into equivalent floats (3¼ -- > 3.25) within regexp ?

Grateful for any suggestions here

Mark

% Below function will need to be adapted depending on how your outlook folders are set up:

function myfrac = TestReadFractions

outlook = actxserver('Outlook.Application');
mapi = outlook.GetNamespace('mapi');
folder1 = mapi.Folders(1);
myaccount = folder1.Item(2); 
inboxmails = myaccount.Folders.Item(2).Folders.Item(9).Items;
count = inboxmails.Count;
myfrac = {};
for i = count:-1:count-10
    if strcmp(inboxmails.Item(i).SenderEmailAddress,'[e-mail]')
        subject = inboxmails.Item(i).Subject; % Mail Subject-Line
        myfrac = regexp(subject,'\x215c','match');
    end
end

Answer 1

[Walter Roberson]

regexprep('ABC','B','\x215c')

[Mark Whirdy]

Hi Walter

There are no \x215c characters present in the mail body/subject to replace, only "?" chars.

I think that the solution must involve changing some property of the Mail Item Object or maybe the Outlook object itself.

I've been trying to reset the InternetCodePage property of the mail (a stab in the dark to be honest), but I don't seem to be able to change it from the commandline (its stuck on 65001). I'm not sure this is really even relevant tho.

inboxmails.Item(i).InternetCodepage = 1252;
inboxmails.Item(i).InternetCodepage
>> ans = 
65001

http://msdn.microsoft.com/en-us/library/office/ff860730.aspx

[Walter Roberson]

Ah if it is code page 65001 then that is UTF-8 . You might have to take "subject" and pass it through native2unicode().

For changing locale see http://www.mathworks.com/matlabcentral/answers/96166

Could you show me the result of

subject + 0

? I do not have MATLAB installed on any MS Windows systems to test with.

[Mark Whirdy]

Hi Walter,

emailing myself "⅛¼⅓⅜½⅝⅔¾⅞" and reading as per code above gives

K>> subject+0

ans =

      63   188    63    63   189    63    63   190    63

Since all the question-marks have same 63 integer, I think that passing through nativetounicode will not work.

I can't change the pc locale as this will effect all other applications I think.

Do you know why changing the InternetCodepage property of the outlook mail, doesn't work? (i.e. as above if I set to anything other than 65001, it is still 65001 when I check in then). I guess the property is immutable, perhaps there is a way of setting it in the actxserver constructor? But even if I can do this, I don't know which InternetCodepage value would fix it

http://msdn.microsoft.com/en-us/library/office/aa171439(v=office.11).aspx

Mark

[Walter Roberson]

Sorry, I am not familiar with how InternetCodePage properties work.