diff options
| author | Sadeep Madurange <sadeep@asciimx.com> | 2026-01-03 12:58:01 +0800 |
|---|---|---|
| committer | Sadeep Madurange <sadeep@asciimx.com> | 2026-01-03 12:58:01 +0800 |
| commit | 8a4da6809cf9368cd6a5dd7351181ea4256453f9 (patch) | |
| tree | 77b2e109ba979332d81799a957bbfa86d010b81b /_site/cgi-bin/seed.sh | |
| download | site-search-bm-8a4da6809cf9368cd6a5dd7351181ea4256453f9.tar.gz | |
Diffstat (limited to '_site/cgi-bin/seed.sh')
| -rwxr-xr-x | _site/cgi-bin/seed.sh | 28 |
1 files changed, 28 insertions, 0 deletions
diff --git a/_site/cgi-bin/seed.sh b/_site/cgi-bin/seed.sh new file mode 100755 index 0000000..9d85108 --- /dev/null +++ b/_site/cgi-bin/seed.sh @@ -0,0 +1,28 @@ +#!/bin/ksh + +# Set the number of files/dirs +TOTAL=10000 +# Approximate size in blocks +BLOCK_SIZE=16 +COUNT=1 + +for i in $(seq 1 $TOTAL); do + # Create a unique directory name + DIR="site_$i" + mkdir -p "$DIR" + + # 1. Generate random valid ASCII (valid UTF-8) text + # We read more from urandom than needed because tr will filter some out + dd if=/dev/urandom bs=1024 count=$BLOCK_SIZE 2>/dev/null | tr -dc 'a-zA-Z0-9 \n' > "$DIR/index.html" + + # 2. Append the necessary HTML structure so your Perl regexes work + # This adds the <title> and <main> tags your script looks for + echo "<html><head><title>Site $i</title></head><body><main><p>Searchable content here for keyword_$i. Lorem ipsum text follows.</p></main></body></html>" >> "$DIR/index.html" + + # Optional: print progress every 100 files + if [ $((i % 100)) -eq 0 ]; then + echo "Created $i files..." + fi +done + +echo "Done! 10000 directories created with valid text." |