bash - Extract image URI from markdown files using sed/grep containing duplicates in a single line

Question

I have some markdown files to process which contain links to images that I wish to download. e.g. a markdown file:

[![](https://imgs.xkcd.com/comics/git.png)](https://imgs.xkcd.com/comics/git.png)

a lot of text 
some more text...

[![](https://1.bp.blogspot.com/-Ze2SiBflkZ4/XbtF1TjELcI/AAAAAAAALL4/IDC6W-b5moU0eGu2eN60aZ4pxfXW1ybmQCLcBGAsYHQ/s320/take_a_break_git.gif)](https://1.bp.blogspot.com/-Ze2SiBflkZ4/XbtF1TjELcI/AAAAAAAALL4/IDC6W-b5moU0eGu2eN60aZ4pxfXW1ybmQCLcBGAsYHQ/s1600/take_a_break_git.gif)


some more text

another URL but not image
[https://github.com]

so on

I am trying to parse through this file and extract the list of image URLs, which I can later pass on wget command to download.

So far I have used grep and sed and have got results:

$ sed -nE "/https?://[^ ]+.(jpg|png|gif)/p" $path
[![](https://imgs.xkcd.com/comics/git.png)](https://imgs.xkcd.com/comics/git.png)
[![](https://1.bp.blogspot.com/-Ze2SiBflkZ4/XbtF1TjELcI/AAAAAAAALL4/IDC6W-b5moU0eGu2eN60aZ4pxfXW1ybmQCLcBGAsYHQ/s320/take_a_break_git.gif)](https://1.bp.blogspot.com/-Ze2SiBflkZ4/XbtF1TjELcI/AAAAAAAALL4/IDC6W-b5moU0eGu2eN60aZ4pxfXW1ybmQCLcBGAsYHQ/s1600/take_a_break_git.gif)

$ grep -Eo "https?://[^ ]+.(jpg|png|gif)" $path
https://imgs.xkcd.com/comics/git.png)](https://imgs.xkcd.com/comics/git.png
https://1.bp.blogspot.com/-Ze2SiBflkZ4/XbtF1TjELcI/AAAAAAAALL4/IDC6W-b5moU0eGu2eN60aZ4pxfXW1ybmQCLcBGAsYHQ/s320/take_a_break_git.gif)](https://1.bp.blogspot.com/-Ze2SiBflkZ4/XbtF1TjELcI/AAAAAAAALL4/IDC6W-b5moU0eGu2eN60aZ4pxfXW1ybmQCLcBGAsYHQ/s1600/take_a_break_git.gif

The regex is essentially working fine, but the issue is that as the same URL is present twice in the same line, the text selected is the first occurrence of https and last occurrence of jpg|png|gif. But I want the first occurrence of https and first occurrence of jpg|png|gif

How can fix this?

P.S. I have also tried lynx -dump -image_links -listonly $path but this prints the entire file.

I am also open to other options that solve the purpose, and as long as I can hook the code up in my current shell script.

Answer 1

You may add square brackets into the negated bracket expression:

grep -Eo "https?://[^][ ]+.(jpg|png|gif)"

See the online demo. Details:

• https?:// - http:// or https://
• [^][ ]+ - one or more chars other than ], [ and space
• . - a dot
• (jpg|png|gif) - either of the three alternative substrings.