`liam erd build` fails with “syntax error at end of input” when parsing large `structure.sql` files

[Said-MZ]

Self Checks

• [x] This is only for bug report, if you would like to ask a question, please head to Discussions.
• [x] I have searched for existing issues search for existing issues, including closed ones.

Version Type

CLI Version (npm package)

Version (only for Self Hosted)

0.7.22

Steps to reproduce

• Run liam erd build --input db/structure.sql --format postgres --output-dir erd
• Ensure the SQL file is large (tens of thousands of lines) with a long INSERT INTO schema_migrations section.
• Observe the “syntax error at end of input” warning.

Expected Behavior

The parser should handle large files

Actual Behavior

When generating an ERD from a large Rails structure.sql, the CLI throws: WARN: Error during parsing schema file: syntax error at end of input

After testing, it’s not an actual SQL syntax issue. The error happens because the structure.sql file (specifically the INSERT INTO schema_migrations VALUES (...) block at the end) is extremely long. If I manually delete from any part ~700 lines from that block, the build completes successfully.

Additional Context

Environment: OS: macOS (can add yours) DB: PostgreSQL liam version: 0.7.22 Node.js version: v22.19.0

Temporary Workaround: Manually remove INSERT INTO schema_migrations

[MH4GF]

@Said-MZ Thanks for creating issue!

Would you be able to provide a reproducible schema that exhibits this behavior? I'm imagining a schema with a very long INSERT INTO schema_migrations block.

That would be super helpful for debugging.

[Said-MZ]

Hey @MH4GF, thanks for the quick response! Unfortunately, I can’t share the actual structure.sql since it’s from a private company project and contains sensitive data.

From my testing, the issue seems related to input size, specifically by the very long INSERT INTO schema_migrations VALUES (...) statement at the end.

When I remove a few hundred lines from the file, the build succeeds, so it might be hitting a parser or buffer limit rather than an SQL syntax issue.

but i got it reproduced with some dummy data by generating about 3k INSERT INTO schema_migrations entries, and it triggers the same warning:

WARN: Error during parsing schema file: syntax error at end of input
For more information, see https://liambx.com/docs/parser/troubleshooting

You can reproduce it easily with this script:

#!/usr/bin/env bash
set -e

OUTPUT="structure.sql"
LINES=3000

echo "Creating dummy structure.sql with $LINES migration entries..."

cat > "$OUTPUT" <<'SQL'
-- Dummy PostgreSQL structure.sql for reproduction
CREATE EXTENSION IF NOT EXISTS "plpgsql" WITH SCHEMA pg_catalog;

CREATE TABLE public.users (
    id SERIAL PRIMARY KEY,
    name VARCHAR(255),
    email VARCHAR(255)
);

CREATE TABLE public.posts (
    id SERIAL PRIMARY KEY,
    user_id INTEGER REFERENCES public.users(id),
    title VARCHAR(255),
    body TEXT
);

-- Insert migrations
INSERT INTO schema_migrations VALUES
SQL

for i in $(seq 1 $LINES); do
  if [ "$i" -eq "$LINES" ]; then
    echo "('migration_$i');" >> "$OUTPUT"
  else
    echo "('migration_$i')," >> "$OUTPUT"
  fi
done

echo "✅ structure.sql generated successfully!"
echo "File: $OUTPUT (contains $LINES migration entries)"

Then run:

liam erd build --input structure.sql --format postgres --output-dir erd

It should reproduce the exact same parsing error.

if it didn't trigger the error just increase the lines to something like 5000-10000

[Said-MZ]

I tested further and confirmed that the issue isn’t caused by file length alone.

I generated another large dummy structure.sql with 40k+ lines containing a mix of statements — CREATE TABLE, ALTER TABLE, COMMENT, and DO blocks — but no large single-line INSERT. That file built perfectly fine using liam erd build.

This strongly suggests the problem is specific to very long single-line INSERT statements, not overall file size.

---

to try the script for the working 40k+ line version:

#!/usr/bin/env bash
set -e

OUTPUT="structure.sql"
LINES=40000

echo "Generating large dummy structure.sql with $LINES lines..."

cat > "$OUTPUT" <<'SQL'
-- Large dummy PostgreSQL structure.sql for testing parser limits
CREATE EXTENSION IF NOT EXISTS "plpgsql" WITH SCHEMA pg_catalog;

CREATE TABLE users (
  id SERIAL PRIMARY KEY,
  username VARCHAR(100),
  email VARCHAR(255),
  created_at TIMESTAMP DEFAULT NOW()
);

CREATE TABLE orders (
  id SERIAL PRIMARY KEY,
  user_id INTEGER REFERENCES users(id),
  total NUMERIC(10,2),
  status VARCHAR(50),
  created_at TIMESTAMP DEFAULT NOW()
);

CREATE INDEX idx_orders_user_id ON orders(user_id);
SQL

for i in $(seq 1 $LINES); do
  case $((i % 5)) in
    0) echo "CREATE TABLE t_${i} (id SERIAL PRIMARY KEY, value TEXT);" >> "$OUTPUT" ;;
    1) echo "ALTER TABLE orders ADD COLUMN extra_field_${i} TEXT;" >> "$OUTPUT" ;;
    2) echo "COMMENT ON TABLE users IS 'User table #${i}';" >> "$OUTPUT" ;;
    3) echo "CREATE INDEX idx_${i}_orders_status ON orders(status);" >> "$OUTPUT" ;;
    4) echo "DO \$\$ BEGIN PERFORM pg_sleep(0); END \$\$;" >> "$OUTPUT" ;;
  esac
done

echo "✅ Done! Created $OUTPUT (~$LINES lines)."

[MH4GF]

Hi @Said-MZ Thank you for the very important information! The PostgreSQL parser may have an issue there because it parses after chunking due to a problem with the WASM being used. I'll look into this matter (Devin will), please wait a moment.

[Said-MZ]

@MH4GF nice work, appreciate how fast you tackled this.