[whimsy] branch member_data updated: Sync with master

[sebb]

This is an automated email from the ASF dual-hosted git repository.

sebb pushed a commit to branch member_data
in repository https://gitbox.apache.org/repos/asf/whimsy.git


The following commit(s) were added to refs/heads/member_data by this push:
     new 2835afc  Sync with master
2835afc is described below

commit 2835afc878350e9f7b3d476a1dca501e141bcee4
Author: Sebb <s...@apache.org>
AuthorDate: Tue Mar 22 22:51:43 2022 +0000

    Sync with master
---
 tools/download_check.rb | 15 ++++++++++-----
 1 file changed, 10 insertions(+), 5 deletions(-)

diff --git a/tools/download_check.rb b/tools/download_check.rb
index 80eea42..17d9a81 100755
--- a/tools/download_check.rb
+++ b/tools/download_check.rb
@@ -56,9 +56,7 @@ $versions = Hash.new {|h1, k1| h1[k1] = Hash.new {|h2, k2| 
h2[k2] = Array.new} }
 # match an artifact
 # TODO detect artifacts by URL as well if possible
 # $1 = base, $2 = extension
-# Skywalking uses .crate
-# TODO: better method for detecting artifacts that does not rely on 
enumerating types
-ARTIFACT_RE = 
%r{/([^/]+\.(pom|tar|tar\.xz|tar\.gz|deb|nbm|dmg|sh|zip|tgz|tar\.bz2|jar|whl|war|msi|exe|rar|rpm|nar|xml|crate))([&?]action=download)?$}
+ARTIFACT_RE = 
%r{/([^/]+\.(pom|tar|tar\.xz|tar\.gz|deb|nbm|dmg|sh|zip|tgz|far|tar\.bz2|jar|whl|war|msi|exe|rar|rpm|nar|xml))([&?]action=download)?$}
 
 def init
   # build a list of validation errors
@@ -342,7 +340,10 @@ end
 def text2ext(txt)
   # need to strip twice to handle ' [ asc ] '
   # TODO: perhaps just remove all white-space?
-  tmp = txt.downcase.strip.sub(%r{^\.}, '').sub(%r{^\[(.+)\]$}, '\1').sub('-', 
'').sub(/ ?(digest|checksum)/, '').sub(/ \(tar\.gz\)| \(zip\)| /, '').strip
+  tmp = txt.downcase.strip.sub(%r{^\.}, '').sub(%r{^\[(.+)\]$}, '\1').sub('-', 
'').
+        sub(/ ?(digest|checksum)/, '').sub(/ \(tar\.gz\)| \(zip\)| /, '').
+        sub('(opens new window)', ''). # doris
+        strip
   return 'sha256' if tmp =~ %r{\A[A-Fa-f0-9]{64}\z}
   return 'sha512' if tmp =~ %r{\A[A-Fa-f0-9]{128}\z}
   ALIASES[tmp] || tmp
@@ -692,6 +693,10 @@ def _checkDownloadPage(path, tlp, version)
                 break
               end
             end
+            if bdy.include? 'The object is in our archive'
+                W "File is archived: '#{name}' in page: '#{h}'"
+                next
+            end
           end
         end
         if path
@@ -707,7 +712,7 @@ def _checkDownloadPage(path, tlp, version)
             E "NAK: ct='#{ct}' cl='#{cl}' #{path}"
           end
         else
-          E "Could not find link for '#{name}' in page: '#{h}' (missing or 
archived)"
+          E "Could not find link for '#{name}' in page: '#{h}' (missing)"
         end
       end
     elsif h =~ %r{\.(md5|sha\d*)$}