Stata 15 help for filefilter

[D] filefilter -- Convert ASCII or binary patterns in a file


filefilter oldfile newfile , { from(oldpattern) to(newpattern) | ascii2ebcdic | ebcdic2ascii } [options]

where oldpattern and newpattern for ASCII characters are

"string" or string

string := [char[char[char[...]]]] char := regchar | code regchar := ASCII 32-91, 93-127, or extended ASCII 128, 161-255; excludes '\' code := \BS backslash \r carriage return \n newline \t tab \M Classic Mac EOL, or \r \W Windows EOL, or \r\n \U Unix or Mac EOL, or \n \LQ left single quote, ` \RQ right single quote, ' \Q double quote, " \$ dollar sign, $ \###d 3-digit [0-9] decimal ASCII \##h 2-digit [0-9,A-F] hexadecimal ASCII

options Description ------------------------------------------------------------------------- * from(oldpattern) find oldpattern to be replaced * to(newpattern) use newpattern to replace occurrences of from() * ascii2ebcdic convert file from ASCII to EBCDIC * ebcdic2ascii convert file from EBCDIC to ASCII replace replace newfile if it already exists ------------------------------------------------------------------------- * Both from(oldpattern) and to(newpattern) are required, or ascii2ebcdic or ebcdic2ascii is required.


filefilter reads an input file, searching for oldpattern. Whenever a matching pattern is found, it is replaced with newpattern. All resulting data, whether matching or nonmatching, are then written to the new file.

Because of the buffering design of filefilter, arbitrarily large files can be converted quickly. filefilter is also useful when traditional editors cannot edit a file, such as when unprintable ASCII characters are involved. In fact, converting end-of-line characters between Macintosh, Windows, and Unix is convenient with the EOL codes.

Unicode is not directly supported, but UTF-8 encoded files can be operated on by using byte-sequence methods in some cases.

Although it is not mandatory, you may want to use quotes to delimit a pattern, protecting the pattern from Stata's parsing routines. A pattern that contains blanks must be in quotes.


from(oldpattern) specifies the pattern to be found and replaced. It is required unless ascii2ebcdic or ebcdic2ascii is specified.

to(newpattern) specifies the pattern used to replace occurrences of from(). It is required unless ascii2ebcdic or ebcdic2ascii is specified.

ascii2ebcdic specifies that characters in the file be converted from ASCII coding to EBCDIC coding. from(), to(), and ebcdic2ascii are not allowed with ascii2ebcdic.

ebcdic2ascii specifies that characters in the file be converted from EBCDIC coding to ASCII coding. from(), to(), and ascii2ebcdic are not allowed with ebcdic2ascii.

replace specifies that newfile be replaced if it already exists.

Technical note

Unicode is not directly supported, but you can try to operate on a UTF-8 encoded Unicode file by working on the byte sequence representation of the UTF-8 encoded Unicode character. For example, the Unicode character é, the Latin small letter "e" with an acute accent (Unicode code point \u00e9), has the byte sequence representation (195,169). You can obtain the byte sequence by using tobytes("é"). Although you may use 195 and 169 in regchar and code, they will be treated as two separate bytes instead of one character é (195 followed by 169). In short, this goes beyond the original design of the command and is technically unsupported. If you try to use filefilter in this way, you might encounter problems.


Convert Classic Mac-style EOL characters to Windows-style . filefilter macfile.txt winfile.txt, from(\M) to(\W) replace

Convert left quote (`) characters to the string "left quote" . filefilter auto1.csv auto2.csv, from(\LQ) to("left quote")

Convert the character with hexidecimal code 60 to the string "left quote" . filefilter auto1.csv auto2.csv, from(\60h) to("left quote")

Convert the character with decimal code 96 to the string "left quote" . filefilter auto1.csv auto2.csv, from(\096d) to("left quote")

Convert strings beginning with hexidecimal code 6B followed by "Text" followed by decimal character 100 followed by "Text" to an empty string (remove them from the file) . filefilter file1.txt file2.txt, from("\6BhText\100dText") to("")

Convert file from EBCDIC to ASCII encoding . filefilter ebcdicfile.txt asciifile.txt, ebcdic2ascii

Stored results

filefilter stores the following in r():

Scalars r(occurrences) number of oldpattern found r(bytes_from) # of bytes represented by oldpattern r(bytes_to) # of bytes represented by newpattern

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index