From 5730c668eadf10abf15e3cfb802c16e22a127336 Mon Sep 17 00:00:00 2001
From: adomani <adomani@gmail.com>
Date: Sat, 18 May 2024 20:23:38 +0200
Subject: [PATCH] remove percentage, small cleanup

---
 scripts/mathlib_stats.sh | 35 +++++++++++------------------------
 1 file changed, 11 insertions(+), 24 deletions(-)

diff --git a/scripts/mathlib_stats.sh b/scripts/mathlib_stats.sh
index ef5343782e77b..0858eef412700 100755
--- a/scripts/mathlib_stats.sh
+++ b/scripts/mathlib_stats.sh
@@ -21,12 +21,6 @@ date=${hashAndDate/#* /}
 ## 'x files changed, y insertions(+), z deletions(-)'
 gdiff="$(git diff --shortstat "${oldCommit}"...${currentCommit})"
 
-## percentage breakdown of changes
-percent="$(printf '|%%|Folder|\n|-:|:-|\n'; git diff --dirstat "${oldCommit}"...HEAD |
-  sed 's=^ *=|=; s=  *=|`=g; s=$=`|='
-#awk 'BEGIN{FS=" "; OFS="|"} {for(i=1; i<=NF; i++){printf $i}}'
-)"
-
 printf -v today '%(%Y-%m-%d)T\n' -1
 
 ## insertions-deletions
@@ -42,8 +36,12 @@ net=$(awk -v gd="${gdiff}" 'BEGIN{
 # Lean-based reports #
 ######################
 
-newDeclsTots="$(sed 's=^--\(count_decls\)=\1=' scripts/count_decls.lean | lake env lean --stdin |
-  sed -z 's=, *=,\n=g; s=[ [#]==g; s=]=,=g; s=\n\n*=\n=g')"
+getCountDecls () {
+  sed 's=^--\(count_decls\)=\1=' scripts/count_decls.lean | lake env lean --stdin |
+    sed -z 's=, *=,\n=g; s=[ [#]==g; s=]=,=g; s=\n\n*=\n=g'
+}
+
+newDeclsTots="$(getCountDecls)"
 newDecls="$(echo "${newDeclsTots}" | awk 'BEGIN{ count=0 }
   /[^,]$/ { count++; type[count]=$0; acc[count]=0; }
   /,$/ { acc[count]++ } END{
@@ -62,8 +60,10 @@ lake build --quiet
 # update the `count_decls` and `mathlib_stats` scripts to the latest version
 git checkout -q origin/adomani/periodic_reports_dev_custom_action scripts/count_decls.lean scripts/mathlib_stats.sh
 
-oldDeclsTots="$(sed 's=^--\(count_decls\)=\1=' scripts/count_decls.lean | lake env lean --stdin |
-  sed -z 's=, *=,\n=g; s=[ [#]==g; s=]=,=g; s=\n\n*=\n=g')"
+# the output of `count_decls`
+oldDeclsTots="$(getCountDecls)"
+
+# the tally of the output of `count_decls`
 oldDecls="$(echo "${oldDeclsTots}" | awk 'BEGIN{ count=0 }
   /[^,]$/ { count++; type[count]=$0; acc[count]=0; }
   /,$/ { acc[count]++ } END{
@@ -71,6 +71,7 @@ oldDecls="$(echo "${oldDeclsTots}" | awk 'BEGIN{ count=0 }
 }')"
 # Definitions 73152...
 
+# produce the `+X -Y` report for the declarations in each category
 plusMinus="$(for typ in $(echo "$newDeclsTots" | grep "[^,]$" | tr '\n' ' ');
 do
   comm -123 --total <(echo "${newDeclsTots}" |
@@ -84,18 +85,6 @@ do
       ($0 ~ typ) { found=1 }' | sort) | awk '{ printf("%s %s\n", $1, $2)}'
 done)"
 
-#awk 'BEGIN{ count=0 }
-#  ((NFR == NR) && (/[^,]$/))  { count++; newType[count]=$0 }
-#  ((NFR == NR) && (!/[^,]$/)) { newAcc[$0]=count }
-##
-#  ((NFR != NR) && (/[^,]$/))  { for(t in type) { if(type[t]==$0) { count=t } }; oldType[count]=$0 }
-#  ((NFR != NR) && (!/[^,]$/)) { oldAcc[$0]=count }
-##
-#  END{
-#
-#  }
-#  ' <(echo "${newDeclsTots}") <(echo "${oldDeclsTots}")
-
 declSummary="$(paste -d' ' <(echo "${newDecls}") <(echo "${oldDecls}") <(echo "${plusMinus}") |
   LC_ALL=en_US.UTF-8 awk 'BEGIN{ print "|Type|New|+-|%|\n|:-:|:-:|:-:|:-:|" }{
     printf("| %s | %'"'"'d | +%'"'"'d -%'"'"'d | %4.2f%% |\n", $1, $2, $5, $6, ($2-$4)*100/$2)
@@ -109,6 +98,4 @@ printf -- ' Reference commits: old %s, new %s.\n\n' "${oldCommitURL}" "${current
 
 printf -- '%s, %s total(insertions-deletions)\n\n' "${gdiff}" "${net}"
 
-#printf -- '---\n\n%s\n\n' "${percent}"
-
 printf -- 'Declarations:\n%s\n\nTake also a look at the [`Mathlib` stats page](%s).\n' "${declSummary}" "${statsURL}"