build_cannonical (3)

#!/bin/ash


[ -z "$alias_file" ] && alias_file="$(realpath ~/cat_test_alias.in)"

PKG_FIX_CATEGORIES_in_REPO_DB=true
PKG_use_ALIASES_in_DB_CAT_Lookup=true

CATEGORIES_INIT_BLOCK='FILENAME==CATFILE {parse_cats($0)
                    next} #https://www3.physnet.uni-hamburg.de/physnet/Tru64-Unix/HTML/APS32DTE/WKXXXXXX.HTM'
[ -z "$PKG_FIX_ALIASES" ] && PKG_FIX_ALIASES=true

function cat_macro_fn(){
    if [ "$PKG_FIX_ALIASES" = true ]; then
      echo "$1"
    else
      echo ""
    fi
}
function alias_macro_fn(){
    if [ "$PKG_use_ALIASES_in_DB_CAT_Lookup" = true ]; then
      echo "$1"
    else
      echo ""
    fi
}
function comment_macro(){
      echo ""
}

ALIAS_FNS="$(awk -v  ALIASFILE="$alias_file" '
function add_get_canonical_hdr(){
  print "function get_canonical_name(s,v){                  "
  print "  if (s in CANONICAL_ARY){                       "
  print "   return CANONICAL_ARY[s]                       "
  print "  } else {                                       "
  print "    switch(s \"-\" v){                                   "
}
function add_get_canonical_footer(){
  print "    }                                            "
  print "  }                                              "
  print "  return s                                       "
  print "}                                                "
}
function add_canonical_case(s,name){

  #TODO: Do this untill there is no astrixs left to replace.
  s=gensub(/^(.*[^.*])(\*)(.*)$/,"\\1.*\\3","G",s) # replace "*" with ".* "
  s=gensub(/^(.*[^\\+])(\+)(.*)$/,"\\1\\\\+\\3","G",s) # replace "+" with "\+"
  print "      case /" s "/:                              "
  print "        return \"" name "\"                      "
}
#function write_CANONICAL_ARY_init_hdr(){
#  print "function init_CANONICAL_ARY(){                            "
#}
function write__init_CANONICAL_ARY(){
  print "function init_CANONICAL_ARY(){                            "
  for (p in CANONICAL_ARY){
    print "  CANONICAL_ARY[\"" p "\"]=\"" CANONICAL_ARY[p] "\""
  }
  print "}"
}


#This writes the start of the function get_canonical_name()
#this function will return a single name for all equivalent alias values.
BEGIN { FS = ","; add_get_canonical_hdr()}

#If wild cards are used in an alias get_canonical_name() uses regX case statments to do the match.
#Otherwise get_cannonical_name() looks up the value in an associative array (AKA dictonary or hashmap)
{
  canonical_NAME=$1
  sub(/\*/,"",canonical_NAME) #This isn'"'"'t necissary but makes the key slightly nicer looking
  for(i=NF;i>=1;i--){
    if(index($i,"*")>0){
      add_canonical_case($i,canonical_NAME) #alias has a wild card so wirite case statement to match alias
    }
    else{
      CANONICAL_ARY[$i]=canonical_NAME #alias has no wild card so add to lookup array)
    }
  }
}
END {add_get_canonical_footer()
     write__init_CANONICAL_ARY()
    } ' "$alias_file")"

#echo "$ALIAS_FNS"


CATEGORY_FNS=\
'function cat_fallback(s){
  if (length(s)>0){
    switch(s){
        case /vcs.*/:
          return "Utility;development"
          break
        case /doc.*/:
          return "Help"
          break
        case /admin.*/:
          return "Setup"
          break
        case /games.*/:
          return "Fun"
          break
        case /.{7,}/:
          return "science"
          break
       default:
          return "BuildingBlock"
          break

    }
  }
  else
  {
      return "BuildingBlock"
  }
}
function get_category(s1,v1,c1,     out,s2,sc1,sc2){
  if ( s1 in pkg_categories){
    return pkg_categories[s1]
  } '"
$(alias_macro_fn \
'  sc1=get_canonical_name(s1,v1)
  if (sc1 in pkg_categories_canonical){
      return pkg_categories_canonical[sc1]
  }
  s2=gensub(/^(.+[^0-9])([0-9]*)$/,"\\1","g",s1)
  if ( s2 in pkg_categories){
    return pkg_categories[s2]
  }
  sc2=get_canonical_name(s2,v1)
  if (sc2 in pkg_categories_canonical){
      return pkg_categories_canonica[sc2]
  }')""$(comment_macro "****End: alias_macro_fn()"****)"'
  #TODO: consider also trying to look up category based on the full name.
  return cat_fallback(c1)
}
function parse_cats(s){
  i=index(s,"=")
  if (i>0){
    #catagory=( package1 package2 ... packageN)
    category=substr(s,0,i-1)
    sub(/PKGCAT_/,"",category) #e.g. category=PKGCAT_Utility => category=Utility
    sub(/_/,";",category) #e.g. category=Filesystem_filemanager => category=Filesystem;filemanager

    #e.g catagory=( package1 package2 ... packageN) => package1 package2 ... packageN
    s=gensub(/^([^=]+="\s*)([^\s].*[^\s])(\s*"\s*)$/,"\\2","g",s)

    split(s,pkg_names," ")
    for (pkg_i in pkg_names){
      pkg=pkg_names[pkg_i]
      pkg_categories[pkg]=category
      pkg_categories_canonical[pkg]=get_canonical_name(category)
    }
  }
}'