From 880c654c43620eb872fba30cea3f7a303eb4d613 Mon Sep 17 00:00:00 2001
From: mxmehl <mail@mehl.mx>
Date: Fri, 28 Jul 2017 17:12:21 +0200
Subject: [PATCH] add debugging and dynamic date ranges to improve speed of
 finding and tagging mails

---
 astroid/poll.sh             |  2 +-
 astroid/scripts/move-tag.sh | 40 ++++++++++++++++++++++++++++++++++---
 2 files changed, 38 insertions(+), 4 deletions(-)

diff --git a/astroid/poll.sh b/astroid/poll.sh
index 8346773..4e20a14 100755
--- a/astroid/poll.sh
+++ b/astroid/poll.sh
@@ -7,7 +7,7 @@ source "$TMP"
 
 # Move selectedly tagged mails to folders if they are only in INBOX
 echo "[DEBUG] Move tagged mails in INBOX to folders"
-bash $CURDIR/scripts/move-tag.sh
+bash $CURDIR/scripts/move-tag.sh ${var_conn}
 
 # If offline mode, do a quick sync
 if [ "$var_conn" == "off" ]; then
diff --git a/astroid/scripts/move-tag.sh b/astroid/scripts/move-tag.sh
index 518acb6..416bf87 100755
--- a/astroid/scripts/move-tag.sh
+++ b/astroid/scripts/move-tag.sh
@@ -5,6 +5,8 @@
 
 CURDIR=$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )
 
+DEBUG=false    # use 'true' to echo verbose debug messages
+
 CSV="$CURDIR"/tag-list.csv    # Table of tags/folders
 MAILDIR="$HOME/Mails/FSFE"    # Maildir
 TRASHTAG="del"    # Tag of trashed mails
@@ -17,10 +19,13 @@ LOG="$CURDIR"/move-tag.log  # Location of logfile
 USETAGS=".+"  # List of tags which shall be moved
 THREADTAGS=".+"  # List of tags for which single, but connected messages in INBOX shall be tagged
 
-BL_USETAGS=""
-BL_THREADTAGS="$TRASHTAG|$SPAMTAG|$ARCHTAG|draft"
+BL_USETAGS=""   # Tags which are *not* considered to be moved
+BL_THREADTAGS="$TRASHTAG|$SPAMTAG|$ARCHTAG|draft"   # Tags for which their whole thread will not be tagged if they are still in INBOX
+
 touch "$LOG"  # Create logfile or update time stamp
 
+LASTRUN="$CURDIR"/move-tag-lastrun.log  # file where the date of the last successful run of this script is saved
+var_conn=$1   # supposed to be "on" or "off"
 
 function pdate {
   DATE=$(date +%y-%m-%d_%H:%M:%S)
@@ -33,6 +38,21 @@ function logrun {
   eval "$@" 2>> "$LOG"
 }
 
+function debugmsg {
+  if ${DEBUG}; then
+    echo "$(pdate) $@"
+  fi
+}
+
+# Set date until when files are searched for, depending on current connection status
+if [ "${var_conn}" == "off" ]; then
+  DATELIMIT=$(date -d "$(cat ${LASTRUN}) - 1 days" +%Y-%m-%d)
+  logrun echo "Datelimit for search set to $DATELIMIT"
+else
+  DATELIMIT=$(date -d "$(cat ${LASTRUN}) - 1 weeks" +%Y-%m-%d)
+  logrun echo "Datelimit for search set to $DATELIMIT"
+fi
+
 while read line; do
   TAG=$(echo $line | cut -d";" -f1)
   FOL=$(echo $line | cut -d";" -f2)
@@ -42,16 +62,19 @@ while read line; do
 
     # TRASH tag "del"
     if [ "$TAG" = "$TRASHTAG" ]; then
+      debugmsg "search trash tagged mails"
       HITFILES=$(notmuch search --output=files --exclude=false tag:$TRASHTAG and not path:$TRASHFOL/**) # Files which are tagged with "del" but not in Trash bin
       FOL="$TRASHFOL"   # Set destination folder to Trash
 
     # SPAM tag "spam"
     elif [ "$TAG" = "$SPAMTAG" ]; then
+      debugmsg "search spam tagged mails"
       HITFILES=$(notmuch search --output=files --exclude=false tag:$SPAMTAG and not path:$SPAMFOL/**) # Files which are tagged with "spam" but not in Spam folder
       FOL="$SPAMFOL"   # Set destination folder to INBOX.Junk
 
     # ARCHIVE tag "archive"
     elif [ "$TAG" = "$ARCHTAG" ]; then
+      debugmsg "search archive tagged mails"
       HITFILES=$(notmuch search --output=files --exclude=false tag:$ARCHTAG and not path:$ARCHFOL/**) # Files which are tagged with "archive" but not in ARCHIVE folder
       # Folder will be derived from the current position later
 
@@ -62,9 +85,11 @@ while read line; do
         # If one file is tagged (except del, spam, archive), also tag the other mails of its thread respectively, IF they are located in INBOX [TODO: is this the best approach?]. Only then they can be moved by this script
 
         # 1. Get thread ids of all entries with the searched tag
-        THREADS=$(notmuch search --exclude=false tag:$TAG and date:$(date -d "-6 weeks" +%Y-%m-%d)..$(date -d "+1 year" +%Y-%m-%d) | cut -d" " -f1 | cut -d":" -f2)
+        debugmsg "search for all threads tagged with $TAG"
+        THREADS=$(notmuch search --exclude=false tag:$TAG and date:$DATELIMIT..$(date -d "+1 year" +%Y-%m-%d) | cut -d" " -f1 | cut -d":" -f2)
 
         # 2. Get message ids of all messages in these threads
+        debugmsg "search for all IDs of these threads"
         MID=
         for t in $THREADS; do
           MID="$MID "$(notmuch search --output=messages --exclude=false thread:$t)
@@ -72,19 +97,23 @@ while read line; do
         done
 
         # 3. Tag all message ids of messages which are still in INBOX
+        debugmsg "tag all these IDs if they are in INBOX:" $(echo "$MID" | wc -w) "files"
         for m in $MID; do
           if $(notmuch search --exclude=false --output=files "$m" | grep -q "$MAILDIR/INBOX/"); then
+            debugmsg "tag $m with $TAG"
             logrun notmuch tag +"$TAG" -- "$m"
           fi
         done
         
         # Search for these lone messages in INBOX
+        debugmsg "search for these messages which are still in INBOX"
         HITFILES=$(notmuch search --output=files --exclude=false path:INBOX/** and tag:$TAG | grep "$MAILDIR/INBOX/")
       fi
       
     fi
 
     # Move files to their destination folders
+    debugmsg "move the files of tag $TAG"
     for m in $HITFILES; do
 
       # Destination folder is derived from the current location
@@ -99,5 +128,10 @@ while read line; do
       logrun mv "$m" "$DESTFOL"
     done
   fi
+  
+  debugmsg "Tag $TAG finished"
+  debugmsg ""
 
 done < "$CSV"
+
+echo $(date +%Y-%m-%d) > ${LASTRUN}