Merge pull request #257 from chrisr-diffblue/apex_naming_conventions

Add partial support for Rational APEX file naming conventions to list_unsupported.sh

Author: martin-cs

experiments/list_unsupported.sh

@@ -3,14 +3,23 @@
 # Usage info
 usage()
 {
-   echo "Usage:\n\nlist_unsupported.sh path_to_ada_source_folder\n"
-   echo "Run GNAT2Goto on an Ada repository.\n"
+   echo "Usage:list_unsupported.sh [--help] [--apex] [--adc pragmas.adc] path_to_ada_source_folder"
+   echo
+   echo "Run GNAT2Goto on an Ada repository."
+   echo
    echo "The output is an ordered list of currently unsupported features"
-   echo "with the number of times they occur in the input repository.\n"
-   echo "The script builds a parsing program using collect_unsupported.cpp and expects"
-   echo "this file to be in the same folder.\n"
+   echo "with the number of times they occur in the input repository."
+   echo
+   echo "Options:"
+   echo "  --help             Display this usage information"
+   echo "  --apex             Use Rational APEX style naming convention .1.ada and .2.ada"
+   echo "  --adc pragmas.adc  Use the specified 'pragmas.adc' file during compilation"
 }
 
+# File extensions to expect for Specification files and Body files
+spec_ext="${SPEC_EXT:-ads}"
+body_ext="${BODY_EXT:-adb}"
+
 # First check some environment prerequisites
 echo >&2 "Checking environment..."
 
@@ -93,46 +102,82 @@ echo >&2 "...environment is OK."
 
 # Command line processing....
 
-if [ "$1" = '--help' ]; then
+if [ "$#" -eq 0 ]; then
    usage
    exit
 fi
 
-if [ "$#" -ne 1 ]; then
+while [ -n "$1" ] ; do
+   case "$1" in
+      --apex)
+         spec_ext="1.ada"
+         body_ext="2.ada"
+         pragma_file="$(dirname ${0})/rational-apex.adc"
+         ;;
+      --adc)
+         shift
+         if [ -r "${1}" ] ; then
+            pragma_file="${1}"
+         else
+            usage >&2
+            echo >&2 "--adc option must specify a configuration pragma file"
+            exit 2
+         fi
+         ;;
+      --help)
+         usage
+         exit
+         ;;
+      *)
+         if [ -n "$project_dir" ]; then
+            usage >&2
+            echo >&2 "Only a single folder name may be specified"
+            exit 2
+         fi
+         project_dir="$1"
+   esac
+   shift
+done
+
+if [ -z "$project_dir" ]; then
    usage >&2
-   echo >&2 "Only a single folder name may be specified"
+   echo >&2 "A project folder name must be specified"
    exit 2
 fi
 
 # Finally start work
 
-echo >&2 "Project to build: $1"
-path="$(cd "$(dirname "$1")"; pwd)/$(basename "$1")"
+echo >&2 "Project to build: ${project_dir}"
+file_name=$(basename "${project_dir}")
+path="$(cd "$(dirname "${project_dir}")"; pwd)/${file_name}"
 include_path=""
-file_name=$(basename "$1")
 
 for foldername in $(find ${path} -type d -name "*" | LC_ALL=posix sort ); do
-   count=`ls -1 ${foldername}/*.ads 2>/dev/null | wc -l`
+   count=`ls -1 ${foldername}/*.${spec_ext} 2>/dev/null | wc -l`
    if [ $count != 0 ]
    then
       include_path="${include_path} -I ${foldername}"
    fi
 done
 
-echo "$1: Unsupported features\n" > "$file_name".txt
+echo "${project_dir}: Unsupported features\n" > "$file_name".txt
 
 # Enumerate all the sub directories of ADA_INCLUDE_PATH
 for include_folder in `echo "$ADA_INCLUDE_PATH" | tr ':' ' '` ; do
    echo "Expanding $include_folder..."
    for foldername in $(find ${include_folder} -type d -name "*" | LC_ALL=posix sort); do
-      count=`ls -1 ${foldername}/*.ads 2>/dev/null | wc -l`
+      count=`ls -1 ${foldername}/*.${spec_ext} 2>/dev/null | wc -l`
       if [ $count != 0 ]
       then
          ADA_INCLUDE_PATH="${ADA_INCLUDE_PATH}:${foldername}"
       fi
    done
 done
 
+if [ -n "${pragma_file}" ]; then
+   pragma_file_opt="-gnatec=${pragma_file}"
+fi
+
 # Before attempting to start compiling, make a clear log of the environment
 # we will be using:
 echo >&2 "-------------------------------------------------------"
@@ -147,10 +192,10 @@ echo >&2 "-------------------------------------------------------"
 # some other error that caused the compiler to exit with non-zero
 # exit code
 compile_error_occured=0
-for filename in $(find ${path} -name '*.adb' | LC_ALL=posix sort); do
+for filename in $(find ${path} -name "*.${body_ext}" | LC_ALL=posix sort); do
    printf "Compiling %s..." "${filename}" >&2
    echo "---------- COMPILING: $filename" >>"$file_name".txt
-   "${GNAT2GOTO}" -gnatU ${include_path} "${filename}" > "$file_name".txt.compiling 2>&1
+   "${GNAT2GOTO}" ${pragma_file_opt} -gnatU ${include_path} "${filename}" > "$file_name".txt.compiling 2>&1
    result=$?
    cat "$file_name".txt.compiling >> "$file_name".txt
 
@@ -176,11 +221,15 @@ for filename in $(find ${path} -name '*.adb' | LC_ALL=posix sort); do
    fi
 done
 
+# Need to use ${spec_ext} and ${body_ext} inside some regex's here,
+# so add any quoting necessary
+quoted_spec_ext=$(printf "%s" "${spec_ext}" | sed 's/\./\\./g')
+quoted_body_ext=$(printf "%s" "${body_ext}" | sed 's/\./\\./g')
 # This redacting system is really pretty crude...
 sed '/^\[/ d' < "$file_name".txt | \
    sed 's/"[^"][^"]*"/"REDACTED"/g' | \
-      sed 's/[^ ][^ ]*\.adb/REDACTED.adb/g' | \
-         sed 's/[^ ][^ ]*\.ads/REDACTED.ads/g' \
+      sed "s/[^ ][^ ]*\.${quoted_body_ext}/REDACTED.${body_ext}/g" | \
+         sed "s/[^ ][^ ]*\.${quoted_spec_ext}/REDACTED.${spec_ext}/g" \
    > "$file_name"_redacted.txt
 
 # Collate and summarise unsupported features

experiments/rational-apex.adc

@@ -0,0 +1,4 @@
+pragma Source_File_Name
+  (Spec_File_Name => "*.1.ada", Dot_Replacement => "-");
+pragma Source_File_Name
+  (Body_File_Name => "*.2.ada", Dot_Replacement => "-");