Return indexes when .schema is run

[kracekumar]

Description

1. As per .schema: line breaks and missing index information #149. the .schema command doesnot return index associated with the table.
2. The sqlite special command, .schemacommand returns so.
3. This PR calls the special command list_indexes to adhere to the similar behaviour
4. Breaking Change: Before when .index command is run only name was returned, now both name and SQL is returned that is similar to sqlite

Examples

1. Sqlite

sqlite> .schema
CREATE TABLE u (name TEXT) STRICT;
CREATE TABLE u1 (name TEXT) STRICT;
CREATE TABLE u2 (name TEXT) STRICT;
CREATE TABLE u3 (name TEXT) STRICT;
CREATE TABLE u4 (name TEXT) STRICT;
CREATE TABLE people (
  person_id VARCHAR PRIMARY KEY,
  name VARCHAR,
  born INTEGER,
  died INTEGER
);
CREATE INDEX ix_people_name ON people (name);
CREATE TABLE people1(
  person_id VARCHAR PRIMARY KEY,
  name VARCHAR,
  born INTEGER,
  died INTEGER
);
CREATE INDEX ix_people_name_1 ON people (name);
CREATE INDEX ix_people_name_2 ON people1 (name);

sqlite> .schema people
CREATE TABLE people (
  person_id VARCHAR PRIMARY KEY,
  name VARCHAR,
  born INTEGER,
  died INTEGER
);
CREATE INDEX ix_people_name ON people (name);
CREATE INDEX ix_people_name_1 ON people (name);

2. LiteCLI

./test.db> .schema
+-------------------------------------------------+
| sql                                             |
+-------------------------------------------------+
| CREATE TABLE people (                           |
|   person_id VARCHAR PRIMARY KEY,                |
|   name VARCHAR,                                 |
|   born INTEGER,                                 |
|   died INTEGER                                  |
| )                                               |
| CREATE INDEX ix_people_name ON people (name)    |
| CREATE INDEX ix_people_name_1 ON people (name)  |
| CREATE TABLE people1(                           |
|   person_id VARCHAR PRIMARY KEY,                |
|   name VARCHAR,                                 |
|   born INTEGER,                                 |
|   died INTEGER                                  |
| )                                               |
| CREATE INDEX ix_people_name_2 ON people1 (name) |
| CREATE TABLE u (name TEXT) STRICT               |
| CREATE TABLE u1 (name TEXT) STRICT              |
| CREATE TABLE u2 (name TEXT) STRICT              |
| CREATE TABLE u3 (name TEXT) STRICT              |
| CREATE TABLE u4 (name TEXT) STRICT              |
+-------------------------------------------------+
Time: 0.004s

+------------------+-------------------------------------------------+
| name             | sql                                             |
+------------------+-------------------------------------------------+
| ix_people_name   | CREATE INDEX ix_people_name ON people (name)    |
| ix_people_name_1 | CREATE INDEX ix_people_name_1 ON people (name)  |
| ix_people_name_2 | CREATE INDEX ix_people_name_2 ON people1 (name) |
+------------------+-------------------------------------------------+

../test.db> .schema people;
+----------------------------------+
| sql                              |
+----------------------------------+
| CREATE TABLE people (            |
|   person_id VARCHAR PRIMARY KEY, |
|   name VARCHAR,                  |
|   born INTEGER,                  |
|   died INTEGER                   |
| )                                |
+----------------------------------+
Time: 0.004s

+------------------+-------------------------------------------------+
| name             | sql                                             |
+------------------+-------------------------------------------------+
| ix_people_name   | CREATE INDEX ix_people_name ON people (name)    |
| ix_people_name_1 | CREATE INDEX ix_people_name_1 ON people (name)  |
| ix_people_name_2 | CREATE INDEX ix_people_name_2 ON people1 (name) |
+------------------+-------------------------------------------------+
Time: 0.003s
../test.db>

Checklist

• I've added this contribution to the CHANGELOG.md file.

[kracekumar]

Without these my tests were breaking.

[kracekumar]

Added an option to return SQL of the index. This is breaking change.

[kracekumar]

As per my understanding, l1 + l2or l1.extend(l2)or [*l1, l2]` have similar performance hence used the most intutive one.

[amjith]

@kracekumar I see that .schema without a table name returns the table schema and the indexes. The indexes are missing only when we call .schema table_name.

The implementation in this PR shows the indices twice when we run .schema. That's a bit confusing. I'd prefer to match what sqlite3 is doing and show the CREATE TABLE... and CREATE INDEX ... statements in the same output instead of the separate tables as your output shows.

I believe what we need is to change these lines to this query:

            SELECT sql FROM sqlite_master
            WHERE tbl_name=? AND sql IS NOT NULL
            ORDER BY tbl_name, type DESC, name

That should do what we want. Thoughts?

[kracekumar]

That's a good call and consistent. I have updated the code to use tbl_nameover name. Can you take a look? Thanks!

[amjith]

Very nice! Thank you for the quick turnaround. 🥂