From 208bf7b187bb61987a128c3a91b1324dc99f2cdb Mon Sep 17 00:00:00 2001
From: bill-auger <mr.j.spam.me@gmail.com>
Date: Wed, 21 Sep 2022 05:39:48 -0400
Subject: [maintenance-tools/parabola-dependents]: account for duplicate
 packages with different deps

---
 src/maintenance-tools/parabola-dependents | 63 +++++++++++++++++++++++++------
 1 file changed, 52 insertions(+), 11 deletions(-)

diff --git a/src/maintenance-tools/parabola-dependents b/src/maintenance-tools/parabola-dependents
index 4e6b7dc..0125f86 100755
--- a/src/maintenance-tools/parabola-dependents
+++ b/src/maintenance-tools/parabola-dependents
@@ -8,6 +8,9 @@ readonly DEBUG=0
 readonly BE_VERBOSE=$( [[ "${1}" == '-v' ]] && echo 1 || echo 0 ) ; (( $BE_VERBOSE )) && shift ;
 readonly PKGBUILD_SED_RX='s|/PKGBUILD$|| ; s|([^/]*/)?([^/]+)/([^/]+)$|\2 \3|'
 readonly DEP=$( sed -E "${PKGBUILD_SED_RX}" <<<${1} | cut -d ' ' -f 2 )
+readonly TRANSITIVE_DEP_RX='( <- [^ ]*)'
+readonly DEP_RX='([a-z]*)/([^ ]*)'
+readonly DEP_CHAIN_RX="\[${DEP}\]${TRANSITIVE_DEP_RX}?${TRANSITIVE_DEP_RX}* <- ${DEP_RX}$"
 
 # readonly REPOS=( nonprism nonsystemd{-testing,} libre{-testing,}
 readonly REPOS=( nonprism{-testing,} nonsystemd{-testing,} libre{-testing,}
@@ -233,24 +236,62 @@ export PATH="/code/pacman-contrib/src:${PATH}"
 
 CollectResults()
 {
-  local dep_chains dep_chain dep_pkg via_pkg repos repo
+  local dep_chains dep_chain
+  local via_pkg transitive_pkg found_repo dep_pkg
+  local repos_deps=() repo_deps dep_repo dep_deps
 
   # query database for dependents
   Log "querying database ...."
-  mapfile -t dep_chains < <(pactree ${PACTREE_OPTS} ${DEP} | sort)
+  mapfile -t dep_chains < <(pactree ${PACTREE_OPTS} ${DEP})
 
   # compile results
   Log "compiling results ...."
   for dep_chain in "${dep_chains[@]}"
-  do  # example dep_chains:
-      #   first-order dep: '[$DEP] <- $dep_pkg'
-      #   high-order dep:  '[$DEP] <- $via_pkg <- transitive_dep <- $dep_pkg'
-      dep_pkg=$(sed 's|.*<- ||'                 <<<${dep_chain})
-      via_pkg=$(sed 's|.*\] <- \([^ ]*\).*|\1|' <<<${dep_chain})
-      repos="$(pacman ${PACMAN_OPTS} -Si ${dep_pkg} | grep Repository | \
-               cut -d ':' -f 2 | tr -d ' '                              )"
-      for repo in ${repos}
-      do  if   IsArchRepo ${repo}
+  do  # `pactree` will only report one dependent package,
+      # even if packages with duplicate names exist in different repos.
+      # Presumably, it selects the first repo in pacman.conf order, as does `pacman`.
+      # This is problematic for the purpose of this script;
+      # because those duplicates do not necessarily have the same dependencies.
+      # In fact, that is a primary motivation for having duplicate packages.
+      # So, in order to ignore packages with duplicate names,
+      # which are not actually in the dependency chain of the input package,
+      # and to report packages with duplicate names
+      # which are actually in the dependency chain of the input package,
+      # we need to capture the first-order dependent ($via_pkg),
+      # the highest-order dependent ($dep_pkg),
+      # and the highest-order transitive dependent ($transitive_pkg).
+      # $via_pkg is the chain root node, and may be empty,
+      # if $dep_pkg is a direct dependent of the input package (example #1).
+      # $transitive_pkg is the direct dependency of the leaf node, and may be empty,
+      # if $dep_pkg is a direct dependent of the input package (example #1),
+      # or if $via_pkg is a direct dependent of $dep_pkg (example #2).
+      # $dep_repo is the repo containing the leaf node package under present scrutiny.
+      # $dep_pkg is the leaf node package under present scrutiny.
+      #
+      # Example dep_chains:
+      #   1. first-order dep:  '[$DEP] <- $found_repo/$dep_pkg'
+      #   2. second-order dep: '[$DEP] <- $via_pkg <- $found_repo/$dep_pkg'
+      #   3. high-order dep:   '[$DEP] <- $via_pkg <- $transitive_pkg <- $found_repo/$dep_pkg'
+      [[ "${dep_chain}" =~ ${DEP_CHAIN_RX} ]] || Exit "script regex error"
+
+      via_pkg=${BASH_REMATCH[1]/ <- /}
+      transitive_pkg=${BASH_REMATCH[2]/ <- /}
+      found_repo=${BASH_REMATCH[3]}
+      dep_pkg=${BASH_REMATCH[4]}
+      mapfile -t repos_deps < <( ( expac ${PACMAN_OPTS} -Ss '%r/%n %D' -l ' ' ${dep_pkg} | \
+                                   grep ^[^/]*/${dep_pkg}\ [^/]*$ || :                   ) | \
+                                   sed "s|/${dep_pkg} | |"                                   )
+
+      for repo_deps in "${repos_deps[@]}"
+      do  dep_repo=$( cut -d ' ' -f 1  <<<${repo_deps})
+          dep_deps="$(cut -d ' ' -f 2- <<<${repo_deps})"
+
+          if   [[ " ${dep_deps} " =~ ' '(${DEP}|${via_pkg}|${transitive_pkg})' ' ]]
+          then dep_chain="$(sed "s| ${found_repo}/${dep_pkg}$| ${dep_repo}/${dep_pkg}|" <<<${dep_chain})"
+          else continue
+          fi
+
+          if   IsArchRepo ${dep_repo}
           then ArchDependents+=(     "${dep_chain}" )
           else ParabolaDependents+=( "${dep_chain}" )
           fi
-- 
cgit v1.2.2