From 430d6fa0fe83eae44e9860d065b6d03533aa7ac4 Mon Sep 17 00:00:00 2001
From: Ron Stone <ronald.stone@windriver.com>
Date: Thu, 30 Nov 2023 14:03:40 +0000
Subject: [PATCH] Add duplicate abbr check (r8, r7, r6, r5)

Add a check for duplicate :abbr: anchor strings to prebuild phase
and error on duplicates.

Change-Id: I034c321a3c83aa4c0cbcf551fd90161e9e221c18
Signed-off-by: Ron Stone <ronald.stone@windriver.com>
---
 _utils.sh         | 63 +++++++++++++++++++++++++++++++++++++++++++++--
 dup-abbr-check.sh | 24 ++++++++++++++++++
 tox.ini           |  1 +
 3 files changed, 86 insertions(+), 2 deletions(-)
 create mode 100755 dup-abbr-check.sh

diff --git a/_utils.sh b/_utils.sh
index 5ad1d3bde..3b90c6cce 100644
--- a/_utils.sh
+++ b/_utils.sh
@@ -1,13 +1,14 @@
 
 declare RED='\033[0;31m'
+declare OG="\033[93m"
 declare GR='\033[0;32m'
 declare NC='\033[0m'
 
 # Output functions. Colorize various types of messages. 
 message () { echo -e "$@" 1>&2; }
 confirmation () { message $GR$@$NC; }
-warn () { message $RED$@$NC; }
-error () { message $RED$@$NC; exit 1; }
+warn () { message $OG$@$NC; }
+error () { message $RED$0:$?: $@$NC; exit 1; }
 
 # Check for and exit if file dependancies are not met. Takes a list of full or
 # relative paths.
@@ -52,4 +53,62 @@ trimspaces () {
   echo $_s
 }
 
+# Sets a global hash of acronyms and definitions from rst :abbr: defs. Also
+# sets an array of hash keys to facilitate sorting.
+#
+# Takes path to the file to parse. Optional "1" flag as second option 
+# suppresses plural forms such as "PVCs". 
+get_abbrs () {
+
+  local ABBREVS
+  declare -a -g acro_keys
+  declare -A -g acro_keyvals
+  local regex=":abbr:\`([A-Za-z]+)\s+\((.*)\)\`"
+
+  [[ ! -z ${1+x} ]] && [[ -e $1 ]] && ABBREVS="$1" \
+     || error "Can't find abbrevs file $1"
+
+  [[ ! -z $2{+x} ]] && [[ ${2} == "1" ]] \
+     && local strip_plurals=$2
+
+  while IFS= read -r line
+  do
+    if [[ $line =~ $regex ]]; then 
+  
+      if [[ ${strip_plurals} -eq 1 ]] && [[ ${BASH_REMATCH[1]:0-1} == "s" ]]; then
+        message " Skipping pluralization \"${BASH_REMATCH[1]}\""
+        continue
+      fi
+      acro_keys+=("${BASH_REMATCH[1]}")
+      acro_keyvals["${BASH_REMATCH[1]}"]="${BASH_REMATCH[2]}"
+    fi
+  done < "$ABBREVS" || error "Cannot read $ABBREVS"
+
+}
+
+
+# Report duplicate :abbr: anchor strings. (Duplicate placeholders cause 
+# Sphinx warnings.)
+#
+# Takes an array of anchor strings. Echos duplicates and returns a duplicate
+# count
+check_abbr_dups () {
+
+   local -a _anchors=("$@")
+   declare -a dups; declare -i _dup_count=0
+   IFS=$'\n'; dups=($(sort -f <<<"${_anchors[*]}")); unset IFS
+   
+   message "... Checking for duplicate anchor strings"
+   
+   for ((i=0; i < ${#dups[@]}; i++)); do
+     if [[ ${dups[$i]} == ${dups[$i-1]} ]]; then 
+       warn " Duplicate anchor string \"${dups[$i]}\" found"
+       ((_dup_count=$_dup_count+1))
+     fi
+   done
+
+   echo $_dup_count
+
+}
+
 declare utils_loaded=1
\ No newline at end of file
diff --git a/dup-abbr-check.sh b/dup-abbr-check.sh
new file mode 100755
index 000000000..7c7aa0fb6
--- /dev/null
+++ b/dup-abbr-check.sh
@@ -0,0 +1,24 @@
+#!/usr/bin/env bash
+
+# Read a file (typlically source/shared/abbrevs.txt) and report on duplicate
+# entries
+
+. $(pwd)/_utils.sh
+if [[ -z ${utils_loaded+x} ]]; then echo "Could not load utilities"; exit 1; fi
+
+declare ABBREVS="doc/source/shared/abbrevs.txt"
+# declare ABBREVS=".stx-docs/doc/source/shared/abbrevs.txt" # for testing
+
+message "Checking for duplicate acronyms ..."
+
+# acro_keys: array of anchor strings
+# acro_keyvals: hash of anchors/definitions
+get_abbrs "${ABBREVS}"
+
+# Check for duplicate anchors in :abbr: defs. We've had problems with these
+# in the past.
+dup_count=$(check_abbr_dups "${acro_keys[@]}")
+
+[[ $dup_count -gt 0 ]] && error "Duplicate(s) MUST be fixed in $ABBREVS"
+
+confirmation "... Done"
\ No newline at end of file
diff --git a/tox.ini b/tox.ini
index 5d749b42c..a0d412724 100644
--- a/tox.ini
+++ b/tox.ini
@@ -21,6 +21,7 @@ commands =
   python parser.py -l templates/alarms_template.rst -e tmp/events.yaml -s 100,200,300,400,500,700,800,900 -ts = -type Alarm -outputPath doc/source/fault-mgmt/openstack/ -sort Yes -product openstack -replace "|,OR"  
   python parser.py -l templates/logs_template.rst -e tmp/events.yaml -s 100,200,300,400,500,700,800,900 -ts = -type Log -outputPath doc/source/fault-mgmt/openstack/ -sort Yes -product openstack -replace "|,OR"  
   bash ./normalize-includes.sh
+  bash ./dup-abbr-check.sh
 
 [testenv:postbuild-docs]
 commands =