Files
webdoc-analysis/run.sh
2026-02-04 17:58:51 +09:00

106 lines
2.8 KiB
Bash
Executable File

#!/usr/bin/env bash
OUTPUT_DIR="out"
mkdir -p "$OUTPUT_DIR"
if [[ ! -f "$OUTPUT_DIR/cgi-paths.txt" ]]; then
find \
/home/pvv/?/*/web-docs \
-type f \
\( \
-name '*.cgi' \
-o -name '*.php' \
-o -name '*.php3' \
-o -name '*.pm' \
-o -name '*.pl' \
-o -name '*.sh' \
-o -name '*.bash' \
-o -name '*.phtml' \
-o -name '*.shtml' \
-o -name '*.lisp' \
-o -name '*.cl' \
\) \
2>/dev/null \
| tee "$OUTPUT_DIR/cgi-paths.txt"
fi
if [[ ! -f out/perl-libs.txt ]]; then
find \
/home/pvv/?/*/web-docs \
-type f \
\( \
-name '*.pm' \
-o -name '*.pl' \
\) \
-exec rg '^use ([^;]+);' {} -N -o -r '$1' \; \
2>/dev/null \
| tee "$OUTPUT_DIR/perl-libs.txt"
fi
cat "$OUTPUT_DIR/perl-libs.txt" | sort > "$OUTPUT_DIR/perl-libs-tmp1.txt"
# Remove import specifiers
sed -E \
-e 's|\s*(qw)?\s*\(.*\)||g' \
-e 's|\s*(qw)?\s*\/.*\/||g' \
"$OUTPUT_DIR/perl-libs-tmp1.txt" | sort > "$OUTPUT_DIR/perl-libs-tmp2.txt"
# Remove standard Perl modules
comm -2 -3 "$OUTPUT_DIR/perl-libs-tmp2.txt" <(cat PERL_STANDARD_MODULES.txt | sort) > "$OUTPUT_DIR/perl-libs-tmp3.txt"
# Remove pragmas
readarray -t PERL_PRAGMAS < PERL_PRAGMAS.txt
remove_pragmas_regex=$(printf '|^%s' "${PERL_PRAGMAS[@]}")
remove_pragmas_regex="${remove_pragmas_regex:1}" # remove leading '|
sed -E "/${remove_pragmas_regex}/d" "$OUTPUT_DIR/perl-libs-tmp3.txt" > "$OUTPUT_DIR/perl-libs-filtered.txt"
cat "$OUTPUT_DIR/perl-libs-filtered.txt" | uniq -c | sort -gr > "$OUTPUT_DIR/perl-libs-overview.txt"
if [[ ! -f "$OUTPUT_DIR/php-libs.txt" ]]; then
find \
/home/pvv/?/*/web-docs \
-type f \
\( \
-name '*.php' \
-o -name '*.php3' \
-o -name '*.phtml' \
\) \
-exec rg '^use ([^;]+);' {} -N -o -r '$1' \; \
2>/dev/null \
| tee "$OUTPUT_DIR/php-libs.txt"
fi
cat "$OUTPUT_DIR/php-libs.txt" | sort | uniq -c | sort -gr > "$OUTPUT_DIR/php-libs-overview.txt"
if [[ ! -f "$OUTPUT_DIR/cgi-progs.txt" ]]; then
readarray -t BIN_LOCATIONS < BIN_LOCATIONS.txt
bin_locations_regex=$(printf '|%s' "${BIN_LOCATIONS[@]}")
bin_locations_regex="${bin_locations_regex:1}" # remove leading '|'
bin_locations_regex="(?:${bin_locations_regex})"
find \
/home/pvv/?/*/web-docs \
-type f \
\( \
-name '*.cgi' \
-o -name '*.php' \
-o -name '*.php3' \
-o -name '*.pm' \
-o -name '*.pl' \
-o -name '*.sh' \
-o -name '*.bash' \
-o -name '*.phtml' \
-o -name '*.shtml' \
-o -name '*.lisp' \
-o -name '*.cl' \
\) \
-exec rg "$bin_locations_regex/(?:env\s*)?(\w+(?:/\w+)*)" {} -N -o -r '$1' \; \
2>/dev/null \
| tee "$OUTPUT_DIR/cgi-progs.txt"
fi
cat "$OUTPUT_DIR/cgi-progs.txt" | sort | uniq -c | sort -gr > "$OUTPUT_DIR/cgi-progs-overview.txt"
# TODO find lisp libraries