2

I have a huge text file on an Ubuntu system and there are many lines of pattern of "Document" followed by 25 random characters, i.e.

cussion. But we cancelled. That's correct," Fasel said.
The 2021 IIHF Women's World Championships is scheduled for the Russian city of Ufa.
Document TASS0000202asd07eg370012y
Fasel said that the IIHF had cancelled all women's international tournaments this year, including the IIHF Ice Hockey Women's World Championship Division I Group A in Angers, France on April 12-18.
Document TaSS0asfd0200307eg370012y
Nevertheless, the IIHF president pointed out that there was no decision yet about the men's world championships set to open in Switzerland in May.
Document aASS000020200307eg370012y
"We are working normally with the Swiss association and everybody is thinking and hoping that we can organize the world championship in May," Fasel said when asked about new information on that tournament.
Canada reported the first coronavirus case on January 26. Up to now, 54 cases have been confirmed in the country. In late December 2019, a pneumonia outbreak caused by the COVID-19 virus (previously known as 2019-nCoV) was reported in China's city of Wuhan, an economic and industrial megacity with a population of 12 million. The World Health Organization declared the new coronavirus outbreak a public health emergency of international concern, characterizing it as an epidemic with multiple locations. Outside China, the worst affected countries are Iran, Italy and South Korea. Overall, more than 90 other countries, including Russia, have reported confirmed coronavirus cases. WHO says that new coronavirus cases outside China have passed 21,000, and there are over 400 deaths.
Document TASS0fgs20200307eg370012y

I want to find all the matching lines and replace that pattern with a specified string as below:

sed -i 's/^Document\s{1}\w{25}\n$/MYLINEBREAK/' textfile.txt

However, it doesn't work at all.

Jeff Schaller
  • 67,283
  • 35
  • 116
  • 255
fishfree
  • 23

1 Answers1

7

By default, sed uses POSIX basic regular expressions and doesn't understand \s or \w or even {}. It also doesn't know how to match a \n since that is the end of the line. The portable way of doing this would be:

sed 's/^Document [a-zA-Z0-9-]\{25\}$/MYLINEBREAK/' file

Almost as portable is using -E to enable extended regular expressions:

sed -E 's/^Document\s[a-zA-Z0-9-]{25}$/MYLINEBREAK/' file

Which, at least on GNU sed (the one on Linux) lets you further simplify to almost what you had in the beginning:

sed -E 's/^Document\s\w{25}$/MYLINEBREAK/' file

See Why does my regular expression work in X but not in Y? for more details on the different regular expression flavors.

terdon
  • 242,166
  • Since you're discussing portability - the only 2 seds that support -E and -i AFAIK are GNU sed and the BSD one on MacOS. The MacOS one won't work without specifying a backup file name after -i even if it's null, i.e. sed -i'' ..., and I could be wrong but I don't think it'll recognize \s either even with the -E - I'm pretty sure thats a GNU-ism. – Ed Morton May 22 '21 at 21:17