pw-pull: learn to do pulls from cover letter

Looks like patchwork is pretty bad at detecting pull requests
which don't fully follow the standard format. Learn to get
the info out of the cover letter ourselves.

Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>

2 files changed, 55 insertions, 11 deletions

diff --git a/README b/README
index 38890b3..d56677d 100644
--- a/README
+++ b/README
@@ -225,11 +225,16 @@ particular PR "patch", PRs don't have series IDs:
   $ pw-pull -p 1393469
 
 When author posted the patches alongside the PR (and patchwork grouped them
-correctly which doesn't always happen, sigh) the script can apply
-those patches to a 'tmp' branch and check the contents match:
+correctly which doesn't always happen) the script can apply those patches
+to a 'tmp' branch and check the contents of series and PR match:
 
   $ pw-pull -p 1393469 -s 12345
 
+pw-pull can also pull from PRs which patchwork does not recognize as PRs
+at all, using -c (AKA --pull-from-cover):
+
+  $ pw-pull -s 12345 -c
+
 I don't mark as applied from the command line (yet) because I double
 check the tests passed in the UI, anyway.
 
diff --git a/pw-pull b/pw-pull
index c25640a..e0bd966 100755
--- a/pw-pull
+++ b/pw-pull
@@ -14,12 +14,13 @@ source $(dirname $0)/lib.sh
 usage()
 {
   cat <<-EOF
-  usage: pw-pull [-p PULL] [-s SERIES] [-h]
+  usage: pw-pull [-p PULL] [-s SERIES] [-c] [-h]
 EOF
   exit
 }
 
 pull=""
+cover=""
 series=""
 series_branch="tmp"
 head_old=$(git rev-parse --verify HEAD)
@@ -27,23 +28,60 @@ while true; do
   case "$1" in
     -p | --pull ) pull="$2"; shift 2 ;;
     -s | --series ) series="$2"; shift 2 ;;
+    -c | --pull-from-cover ) cover="y"; shift ;;
     -h | --help ) usage; break ;;
     -- ) shift; break ;;
     * )  break ;;
   esac
 done
-[ -z "$pull" ]     && usage
+[ -z "$pull" ] && [ -z "$cover" ]     && usage
+[ ! -z "$pull" ] && [ ! -z "$cover" ] && usage
 
 srv=$(git config --get pw.server)
+srv=${srv%/} # strip trailing slash
 
-pull_json=$(curl -s $srv/patches/$pull/)
+if [ ! -z "$pull" ]; then
+  pull_json=$(curl -s $srv/patches/$pull/)
 
-pull_url=$(echo "$pull_json" | jq -r '.pull_url')
-pull_author=$(echo "$pull_json" | jq -r '.submitter.name')
-pull_subject=$(echo "$pull_json" |
+  pull_url=$(echo "$pull_json" | jq -r '.pull_url')
+  pull_author=$(echo "$pull_json" | jq -r '.submitter.name')
+  pull_subject=$(echo "$pull_json" |
 		   jq -r '.name' |
 		   sed -e 's/\[.*\] *//')
-pull_msgid=$(echo "$pull_json" | jq -r '.msgid' | tr -d '<>')
+  pull_msgid=$(echo "$pull_json" | jq -r '.msgid' | tr -d '<>')
+  json=$pull_json
+elif [ ! -z "$cover" ]; then
+  series_json=$(curl -s $srv/series/$series/)
+
+  pull_author=$(echo "$series_json" | jq -r '.submitter.name')
+  pull_subject=$(echo "$series_json" |
+		   jq -r '.name' |
+		   sed -e 's/\[.*\] *//')
+  pull_msgid=$(echo "$series_json" | jq -r '.cover_letter.msgid' | tr -d '<>')
+
+  cover_id=$(echo "$series_json" | jq -r '.cover_letter.id')
+  cover_json=$(curl -s $srv/covers/$cover_id/)
+
+  pull_msg=$(echo "$pull" |
+		   jq -r '.content' |
+		   awk 'BEGIN {e=0}
+                        /^---/ {e=1}
+                        /^Please consider/ {e=1}
+                        {if(!e) print}')
+
+  pull_url=$(echo "$cover_json" |
+	       jq -r '.content' |
+	       awk '/^.+$/                      { if (n) {n=0; print} }
+	            /and are available in the git repository at:/ {n=1}')
+  pull_url=$(echo $pull_url)
+
+  if [ -z "$pull_url" ]; then
+    echo "Unable to parse out pull URL"
+    exit 1
+  fi
+
+  json=$cover_json
+fi
 
 # Download the series from ML and make a local branch
 if [ ! -z "$series" ]; then
@@ -61,16 +99,17 @@ merge_body=$(git show --format="%b" --no-patch)
 # count lines until "* tag", ignore empty and gpg output
 significant_lines=$(echo -e "$merge_body" |
 			awk 'BEGIN {c=0; e=0}
-			     /^\* (tag|git)/ {e=1}
+			     /^\* (tag|git|'"'"')/ {e=1}
 			     /^[^#]/ { if(!e) c++}
 			     END {print c}')
 
 if [ "$significant_lines" -le 1 ]; then
     # Construct the message from email body if tag is empty
-    pull_msg=$(echo "$pull_json" |
+    pull_msg=$(echo "$json" |
 		   jq -r '.content' |
 		   awk 'BEGIN {e=0}
                         /^---/ {e=1}
+                        /^The following changes since commit/ {e=1}
                         /^Please consider/ {e=1}
                         {if(!e) print}')