From c0a1e2d301801f1016e3329dffe593849071113f Mon Sep 17 00:00:00 2001
From: Dessalines <tyhou13@gmx.com>
Date: Thu, 24 Jan 2019 14:45:18 -0800
Subject: [PATCH] Using new tracker health for torrent scanner. Fixes #34

---
 scripts/prune.sh         |   9 ++++
 scripts/scan_torrents.sh | 105 ++++++++++++---------------------------
 2 files changed, 41 insertions(+), 73 deletions(-)

diff --git a/scripts/prune.sh b/scripts/prune.sh
index 609e44c..64e5c40 100755
--- a/scripts/prune.sh
+++ b/scripts/prune.sh
@@ -2,6 +2,8 @@
 echo "Pruning torrents.csv ..."
 cd ..
 torrents_csv="`pwd`/torrents.csv"
+scanned_out="`pwd`/infohashes_scanned.txt"
+
 
 # Remove lines that don't have exactly 7 ';'
 rg "^([^;]*;){7}[^;]+$" $torrents_csv > tmp_adds
@@ -22,6 +24,13 @@ sed -i '1d' $torrents_csv
 sort -u -t';' -k1,1 -o $torrents_csv $torrents_csv
 sort -u -t';' -k2,2 -k3,3 -o $torrents_csv $torrents_csv
 
+# Same for the infohashes scanned
+sort -u -o $scanned_out $scanned_out
+
+# Remove torrents with zero seeders
+awk -F';' '$5>=1' $torrents_csv> tmp
+mv tmp $torrents_csv
+
 # Sort by infohash asc
 sort --field-separator=';' --key=1 -o $torrents_csv $torrents_csv
 
diff --git a/scripts/scan_torrents.sh b/scripts/scan_torrents.sh
index 97b5b4e..e491a16 100755
--- a/scripts/scan_torrents.sh
+++ b/scripts/scan_torrents.sh
@@ -5,8 +5,12 @@
 cd ..
 torrents_csv="`pwd`/torrents.csv"
 scanned_out="`pwd`/infohashes_scanned.txt"
-
+tmp_torrent_dir="`pwd`/tmp_torrents"
+names_out="`pwd`/names.out"
+health_out="`pwd`/health.out"
 touch $scanned_out
+touch $names_out
+touch $health_out
 
 help="Run ./scan_torrents.sh [TORRENTS_DIR] \nor goto https://gitlab.com/dessalines/torrents.csv for more help"
 if [ "$1" == "-h" ] || [ -z "$1" ]; then
@@ -29,79 +33,34 @@ fi
 # Loop over all torrents
 pushd $torrents_dir
 # for torrent_file in *.torrent; do
+# Copy the unscanned torrent files to a temp dir
+mkdir $tmp_torrent_dir
 find `pwd` -name "*.torrent" | sort -n | grep -vFf $scanned_out | while read torrent_file ; do
-
-  file_infohash=$(basename $torrent_file | cut -d'.' -f 1)
-
-  # Scrape it
-  health_text=$(torrent-tracker-health --torrent $torrent_file --timeout 1000 --addTrackers={udp://tracker.coppersurfer.tk:6969/announce,udp://tracker.internetwarriors.net:1337/announce,udp://tracker.opentrackr.org:1337/announce,udp://exodus.desync.com:6969/announce,udp://explodie.org:6969/announce})
-
-  echo -e "$health_text"
-
-  infohash=$(jq  -r '.hash' <<< $health_text)
-  name=$(jq -r '.name' <<< $health_text)
-  size_bytes=$(jq -r '.length' <<< $health_text)
-  seeders=$(jq -r '.seeds' <<< $health_text)
-  leechers=$(jq -r '.peers' <<< $health_text)
-  completed=$(jq -r '.completed' <<< $health_text)
-  date_string=$(jq -r '.created' <<< $health_text)
-  if [ "$date_string"  == "null" ]; then
-    echo "Date was null, setting to now"
-    created_date=$(date +%s)
-  else
-    created_date=$(date -d "${date_string}" +"%s")
-  fi
-  scraped_date=$(date +%s)
-
-  # Construct add line
-  add_line="$infohash;$name;$size_bytes;$created_date;$seeders;$leechers;$completed;$scraped_date"
-  # echo -e $add_line
-
-  if (( $seeders > 0 )); then
-
-    found_line=$(rg -n $infohash $torrents_csv)
-    found_seeders=$(echo -e $found_line | cut -d';' -f 5)
-
-    # Only re-add if the infohash doesn't exist, or 
-    if [ ! -z "$found_line" ]; then
-
-      # Seeder counts are different
-      if [ "$found_seeders" != "$seeders" ]; then
-
-        # Delete the original infohash line
-        rg -Nv "$infohash" $torrents_csv > torfile.tmp.2; mv torfile.tmp.2 $torrents_csv
-
-        # Append the add lines to the torrents.csv file
-        echo -e "\n$add_line" >> $torrents_csv
-        # truncate -s -1 $torrents_csv # Removing last newline
-        echo -e "Updating Seeders:  $torrent_file | $name | $infohash | $seeders"
-      else
-        echo -e "Not adding $name, had identical seeders"
-      fi
-    else
-      # Append the add lines to the torrents.csv file
-      echo -e "\n$add_line" >> $torrents_csv
-      # truncate -s -1 $torrents_csv # Removing last newline
-      echo -e "New Torrent: $torrent_file | $name | $infohash | $seeders"
-    fi
-
-  else 
-    # Deleting the line if it existed
-    if [ ! -z "$infohash" ]; then
-      # This removes lines that have no seeders
-      echo -e "$name has no seeders"
-      if rg -Nq $infohash $torrents_csv; then
-        echo "Removing $name from $torrents_csv"
-        rg -Nv "$infohash" $torrents_csv > torfile.tmp.2; mv torfile.tmp.2 $torrents_csv
-      fi
-    fi
-  fi
-
-  echo $file_infohash >> $scanned_out
-
+  cp "$torrent_file" "$tmp_torrent_dir"
+  echo $(basename "$torrent_file" .torrent) >> $names_out
 done
 
-popd
-cd scripts
-. prune.sh
+if [ -z "$(ls -A $tmp_torrent_dir)" ]; then
+  echo "No new torrents."
+else
+  # Scrape it
+  torrent-tracker-health --torrent "$tmp_torrent_dir" > $health_out
+
+  echo -e "$health_out"
+
+  results=$(jq -r '.results | map([.hash, .name, .length, (.created | strptime("%Y-%m-%dT%H:%M:%S.000Z") | mktime), .seeders, .leechers, .completed, (now | floor)] | join(";")) | join("\n")' $health_out)
+
+  echo -e "$results" >> $torrents_csv
+  cat "$names_out" >> $scanned_out
+
+  popd
+  cd scripts
+  . prune.sh
+fi
+
+# Remove the temp dir
+rm -rf "$tmp_torrent_dir"
+rm "$names_out"
+rm "$health_out"
+