TEXTSCAN is my personal nemesis: Quotes strings with commas

조회 수: 11(최근 30일)
Dear all,
I thought I acquired some Matlab knowledge over the years. But textscan really makes me feel dumb. I can not figure out the syntax to read a quoted text file.
Let's say I have a text file with the following format:
"L1: quoted text, with commas","111.111",123\n
"L2: quoted text, with commas","222.222",234\n
(the field width are the same just for the example; consider them variable in the real world).
Now my understanding of textscan's formatSpec would be:
format = '%q,"%f",%f';
delim = '\n';
res = textscan(fileId,format,'Delimiter',delim,'ReturnOnError',false);
This is because I am assuming that %q considers everything between two quotes as part of the same string, as the help states
"If the string begins with a double quotation mark ("), omit the leading quotation mark
and its accompanying closing mark, which is the second instance of a lone double
quotation mark"
However this fails with the ominous error message:
Error using textscan
Mismatch between file and format string.
Trouble reading 'Literal' field from file (row number 1, field number 2) ==>
"L2: quoted text, with commas","222.222",234\n
So the commas in the quoted string somehow screw-up the parsing, since using the regexp-like
format = '"%[^"]","%f",%f';
instead of %q works perfectly fine.
So my questions are:
  1. What am I doing wrong?
  2. Why is there the %q option if it does not treat everything in between as string (as this would than simply be a shorthand for "%s")?
  3. Am I using the delimiter option correctly?
Any help appreciated!
Greetings, David
  댓글 수: 2
David J. Mack
David J. Mack 2017년 2월 13일
You mean every superhero needs a supervillain?

댓글을 달려면 로그인하십시오.

채택된 답변

Stephen23 2017년 2월 13일
편집: Stephen23 2017년 2월 13일
fmt = '%q"%f"%f';
fid = fopen('temp5.txt','rt');
C = textscan(fid,fmt,'Delimiter',',');
Gives this:
>> C{:}
ans =
'L1: quoted text, with commas'
'L2: quoted text, with commas'
ans =
ans =
And your questions:
  1. you are over-specifying the delimiter, and also confusing the delimiter with the end of line character |
  댓글 수: 6
Stephen23 2017년 2월 13일
편집: Stephen23 2017년 2월 13일
@David J. Mack: the task is complicated by having numeric fields surrounded by double quotes, which is rather erroneous because double quotes always indicate strings. Two solution for this are:
  1. Change the file writing so that it correctly does not put double quotes around numeric values.
  2. Import the numeric values are strings using %q, and quickly convert them using str2double once inside MATLAB.
In both of these cases you could go back to defining the delimiter as just the comma, which would avoid collapsing fields together.

댓글을 달려면 로그인하십시오.

추가 답변(2개)

Jeremy Hughes
Jeremy Hughes 2017년 2월 16일
If you have access to R2016b, try using detectImportOptions with readtable.
opts = detectImportOptions(yourfile);
T = readtable(yourfile,opts)
I think this will do what you want without a lot of fuss.
  댓글 수: 1
David J. Mack
David J. Mack 2017년 2월 17일
Dear Jeremy,
just installed R2016b and will have a look. Judging by the name of the function, I implemented that already (as readtable uses textscan if I remeber it correctly). But its good to know!

댓글을 달려면 로그인하십시오.

David J. Mack
David J. Mack 2017년 2월 13일
편집: David J. Mack 2017년 2월 13일
Thanks Stephen for staying with me (you where right in the first place; reading the doc carefully helps)!
I finally found my kryptonite! Don't think REGEX when using textscan!
The solution is not using quotes directly in the format, as this is not supported by textscan. Instead create a matching group for each quote. For the example above, the correct format is (using %q instead of the datetime expression for simplicity).
fmt = '%*["]%f%*["]%*["]%f%*["]%q';
instead of
fmt = '"%f""%f"%q';
So just in case, somebody encounters the same problem.
  댓글 수: 1
Serge 2018년 5월 27일
Any suggestions for this question:

댓글을 달려면 로그인하십시오.

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by