| 1 |
#!/bin/sh |
| 2 |
# Build a libFuzzer seed corpus from the existing tests/ data and |
| 3 |
# the bundled hyph_en_US.dic. Each seed file has the format: |
| 4 |
# |
| 5 |
# [word bytes][NUL][dictionary bytes] |
| 6 |
# |
| 7 |
# matching what fuzz_hyphen.c expects. |
| 8 |
|
| 9 |
set -eu |
| 10 |
|
| 11 |
cd "$(dirname "$0")/.." |
| 12 |
|
| 13 |
CORPUS_DIR=${1:-fuzz/corpus_seed} |
| 14 |
mkdir -p "$CORPUS_DIR" |
| 15 |
|
| 16 |
count=0 |
| 17 |
|
| 18 |
# tests/<name>.pat + tests/<name>.word -> one seed per word line |
| 19 |
for pat in tests/*.pat; do |
| 20 |
name=$(basename "$pat" .pat) |
| 21 |
wordfile="tests/${name}.word" |
| 22 |
[ -f "$wordfile" ] || continue |
| 23 |
|
| 24 |
line_no=0 |
| 25 |
while IFS= read -r word || [ -n "$word" ]; do |
| 26 |
line_no=$((line_no + 1)) |
| 27 |
# strip CR if present |
| 28 |
word=$(printf '%s' "$word" | tr -d '\r') |
| 29 |
out="$CORPUS_DIR/${name}_${line_no}" |
| 30 |
# concat: word, NUL, dictionary |
| 31 |
{ printf '%s' "$word"; printf '\0'; cat "$pat"; } > "$out" |
| 32 |
count=$((count + 1)) |
| 33 |
done < "$wordfile" |
| 34 |
done |
| 35 |
|
| 36 |
# A few generic seeds against hyph_en_US.dic (much larger dictionary, |
| 37 |
# helps the FSM walk lots of states). |
| 38 |
if [ -f hyph_en_US.dic ]; then |
| 39 |
for w in hyphenation example creating absolutely beautifully \ |
| 40 |
antidisestablishmentarianism photograph supercalifragilistic \ |
| 41 |
"" "a" "ab" "abc" ".a" "a." ".." "a-b" "12345"; do |
| 42 |
out="$CORPUS_DIR/en_$(printf '%s' "$w" | tr -c 'a-zA-Z0-9' '_')" |
| 43 |
{ printf '%s' "$w"; printf '\0'; cat hyph_en_US.dic; } > "$out" |
| 44 |
count=$((count + 1)) |
| 45 |
done |
| 46 |
fi |
| 47 |
|
| 48 |
# A pure-parser seed: empty word, full hyph_en_US dictionary. |
| 49 |
if [ -f hyph_en_US.dic ]; then |
| 50 |
cat hyph_en_US.dic > "$CORPUS_DIR/parser_only_en" |
| 51 |
count=$((count + 1)) |
| 52 |
fi |
| 53 |
|
| 54 |
# Tiny edge cases for the parser path |
| 55 |
{ printf '%s' "x"; printf '\0'; printf 'UTF-8\n1.a\n'; } > "$CORPUS_DIR/tiny_boundary" |
| 56 |
{ printf '%s' "x"; printf '\0'; printf 'UTF-8\n'; } > "$CORPUS_DIR/tiny_empty_dict" |
| 57 |
{ printf '%s' "x"; printf '\0'; printf 'UTF-8\nNOHYPHEN \n1a1\n'; } > "$CORPUS_DIR/tiny_nohyphen_empty" |
| 58 |
{ printf '%s' "x"; printf '\0'; printf 'UTF-8\nNEXTLEVEL\nUTF-8\n1a1\n'; } > "$CORPUS_DIR/tiny_nextlevel" |
| 59 |
count=$((count + 4)) |
| 60 |
|
| 61 |
echo "[corpus] wrote $count seed files to $CORPUS_DIR" |
| 62 |
du -sh "$CORPUS_DIR" |