added parameters, computation output, and output length to stats

athalhammer · Nov 22, 2023 · 104f1bb · 104f1bb
1 parent 3961313
commit 104f1bb
Show file tree

Hide file tree

Showing 2 changed files with 15 additions and 8 deletions.
diff --git a/danker/danker.py b/danker/danker.py
@@ -317,6 +317,7 @@ def _main():
         "-i", "--int_only", action="store_true", help="All nodes are integers (flag)"
     )
     args = parser.parse_args()
+    param_out = f"[iterations ({args.iterations}), damping ({args.damping}), start value({args.start_value})]"
     if (
         args.iterations < 0
         or args.damping > 1
@@ -325,12 +326,16 @@ def _main():
     ):
         print(
             "ERROR: Provided PageRank parameters\n\t"
-            f"[iterations ({args.iterations}), damping ({args.damping}), "
-            f"start value({args.start_value})]\n\tout of allowed range.\n\n",
+            f"{param_out}\n\tout of allowed range.\n\n",
             file=sys.stderr,
         )
         parser.print_help(sys.stderr)
         sys.exit(1)
+    print(
+        f"danker: starting computation of PageRank on '{args.left_sorted}' with parameters\n\t"
+        f"{param_out}",
+        file=sys.stderr,
+    )
     start = time.time()
     dictionary = init(
         args.left_sorted, args.start_value, args.right_sorted, args.int_only
@@ -351,8 +356,7 @@ def _main():
         danker_bigmem(dictionary, args.iterations, args.damping)
 
     print(
-        f"Computation of PageRank on '{args.left_sorted}' with danker took "
-        f"{time.time() - start:.2f} seconds.",
+        "danker: PageRank computation took " f"{time.time() - start:.2f} seconds.",
         file=sys.stderr,
     )
 

diff --git a/script/dank.sh b/script/dank.sh
@@ -67,6 +67,8 @@ fi
 
 if [ "$1" == "ALL" ]; then
     filename=$(date +"%Y-%m-%d").all"$project".links
+    > "$filename.files.txt"
+    > "$filename.stats.txt"
     if languages=$(./script/get_languages.sh "$project"); then
 
 	# collect
@@ -79,7 +81,6 @@ if [ "$1" == "ALL" ]; then
 
         # collect stats
 	xargs wc -l < "$filename.files.txt" | grep -v "total" | sed "s/^[[:space:]]\+//" > "$filename.stats.txt"
-        wc -l "$filename" >> "$filename.stats.txt"
 
         # clean up
 	xargs rm < "$filename.files.txt"
@@ -90,25 +91,27 @@ if [ "$1" == "ALL" ]; then
     fi
 else
     filename=$(./script/create_links.sh "$1" "$project" "$dump_time" "$folder")
+    > "$filename.stats.txt"
 fi
 
 # "extract links only" option
 if [ "$links" ]; then
     echo "$filename"
     exit 0
 fi
+wc -l "$filename" >> "$filename.stats.txt"
 
 if [ "$bigmem" ]; then
     python3 -m danker  "$filename" "$damping" "$iterations" "$start_value" -i \
-        | sed "s/\(.*\)/Q\1/" \
+       2>> "$filename.stats.txt" | sed "s/\(.*\)/Q\1/" \
     > "$filename".rank
 else
     sort -k 2,2n -T . -S "$MEM_PERC" -o "$filename"".right" "$filename"
     python3 -m danker  "$filename" -r "$filename"".right" "$damping" "$iterations" "$start_value" -i \
-        | sed "s/\(.*\)/Q\1/" \
+        2>> "$filename.stats.txt" | sed "s/\(.*\)/Q\1/" \
     > "$filename".rank
     rm "$filename"".right"
 fi
 sort -k 2,2nr -T . -S "$MEM_PERC" -o "$filename"".rank" "$filename"".rank"
+wc -l "$filename"".rank" >> "$filename.stats.txt"
 bzip2 "$filename"
-wc -l "$filename"".rank"