Using http remote MD5 hash

Question

I would like to use the md5sums function of pkgbuild without keeping track of new hashsum at each update, and just change the pkgver variable to build a new version of my package.

md5 are available on the official website of the project i'm interested in near the source archive download in a span of class value of md5sum

As I know the hashsum will always be in a span of class md5sum right after the span of class releasedate , itself behind two a href having tar.xz in them, how can I extract the hashsum from the website using a shell script that may call tools usually preinstalled on an archlinux installation into a variable that will strictly contain only the hashsum ?

Answer 1

If you are intent on scraping the hash off the page, then something like this will work. It's the sort of one-liner I write for one-off use, because it's likely to break as soon as the layout changes:

hashsum=$(curl http://php.net/downloads.php |
 grep -A1 -F ">php-${pkgver}.tar.xz<" |
 sed 1d |
 tr '>' '<' |
 cut -d '<' -f3 )

---

To understand why this is an incredibly inflexible approach, it may help to look at the output of each stage.

• curl fetches the page and sends it to the standard output. (If you don't have curl available, wget -O - will also work.)

  The output includes this:

  ...
  <ul>
            <li>
        <a href="/get/php-7.0.7.tar.bz2/from/a/mirror">php-7.0.7.tar.bz2</a> <a href="/get/php-7.0.7.tar.bz2.asc/from/a/mirror">(sig)</a> [13,776Kb]          <span class="releasedate">26 May 2016</span>
        <span class="md5sum">cc231de15146ca14a69610f695995ab8</span>
        <span class="sha256">474f2925c4782b94016e3afbb17b14ff9cc6f4fdb6f6e231b36a378bb18a3d1a</span>
                </li>
            <li>
        <a href="/get/php-7.0.7.tar.gz/from/a/mirror">php-7.0.7.tar.gz</a> <a href="/get/php-7.0.7.tar.gz.asc/from/a/mirror">(sig)</a> [17,792Kb]          <span class="releasedate">26 May 2016</span>
        <span class="md5sum">197ec66d134da8968ddd0b89f1cfac2a</span>
        <span class="sha256">66282ff4a9f88fe9607d9574e15bf335885b964245591a1740adb3f79c514a67</span>
                </li>
            <li>
        <a href="/get/php-7.0.7.tar.xz/from/a/mirror">php-7.0.7.tar.xz</a> <a href="/get/php-7.0.7.tar.xz.asc/from/a/mirror">(sig)</a> [11,219Kb]          <span class="releasedate">26 May 2016</span>
        <span class="md5sum">75f8d1693a470cefe2a50abd283eb291</span>
        <span class="sha256">9cc64a7459242c79c10e79d74feaf5bae3541f604966ceb600c3d2e8f5fe4794</span>
                </li>
          <li>
      <a href="http://windows.php.net/download#php-7.0">
        Windows downloads
      </a>
    </li>
  </ul>
  ...
  

• grep -A1 returns the line containing that pattern (the name of the source file you need), plus 1 line following it:

    <a href="/get/php-7.0.7.tar.xz/from/a/mirror">php-7.0.7.tar.xz</a> <a href="/get/php-7.0.7.tar.xz.asc/from/a/mirror">(sig)</a> [11,219Kb]          <span class="releasedate">26 May 2016</span>
    <span class="md5sum">75f8d1693a470cefe2a50abd283eb291</span>
  

• We only care about the second line, so sed 1d deletes the first. (If you want to be sure it's an md5sum, use grep 'class="md5sum"' instead.)

    <span class="md5sum">75f8d1693a470cefe2a50abd283eb291</span>
  

• The tr command turns all > to <. This is not useful on its own, but turns the line into something that cut can work with:

    <span class="md5sum"<75f8d1693a470cefe2a50abd283eb291</span<
  

• Finally, cut treats those <s (including the ones that were previously >s) as delimiters, and extracts the 3rd item, ie. the hash:

  75f8d1693a470cefe2a50abd283eb291