ref: bcc3f2acb5d11bc917c96519c8a810fc06058a24 pg2sqlite/README.md -rw-r--r-- 4.9 KiB
bcc3f2ac — Louis Brauer Update README 3 months ago


Go Report Card License Downloads

pg2sqlite can migrate tables from PostgreSQL to SQLite3.


#From source

$ go get -u github.com/louis77/pg2sqlite

#Download pre-built binaries

Platform Intel/AMD ARM
MacOS Intel M1
Linux amd64 arm
FreeBSD amd64 arm64
Windows amd64 arm


$ pg2sqlite -h


  -h, --help
      display help information
      *Postgres connection string (i.e. postgres://localhost:5432/mydb)
      *Path to SQLite database file (i.e. mydatabase.db)
  -t, --table
      *Name of table to export
      comma-separated list of columns to ignore
      DANGER: Drop target table if it already exists
      Confirm prompts with Y, useful if used in script
      Verify that the number of rows inserted into SQLite equals the number of rows loaded from Postgres. In case of failure, exits with status code 2


$ pg2sqlite --pg-url postgres://localhost:5432/defaultdb \
            --sqlite-file mysqlite.db \
            --table daily_sales \
            --ignore-columns raw_hash

Schema of table "daily_sales"
Column                     | Type                     | Ignore
-------------------------- | ------------------------ | ------
reference_id               | integer                  | No
checkin                    | date                     | No
checkout                   | date                     | No
price                      | numeric                  | No
raw_hash                   | character                | Yes
currency                   | character                | No
ts                         | timestamp with time zone | No
Creating Table statement:
CREATE TABLE daily_sales (         
        reference_id INTEGER, 
        checkin TEXT, 
        checkout TEXT, 
        price REAL, 
        currency TEXT,  
        ts TEXT )
Does this look ok? (Y/N) y

Estimated row count: 50042260
Loading data with this statement:
SELECT "reference_id", "checkin", "checkout", "price", "currency", "external_id", "ts" FROM results_y2021m02 T

  24s [==>-----------------------------------------------------------------]   3%



#Release history

  • 1.0.0
    • Initial version


pg2sqlite works with a single connection to PostgreSQL and SQLite. To keep memory consumption low, rows are transferred without buffering.


pg2sqlite will try to complete these steps in the following order:

  1. Validate the connection to Postgres
  2. Validate the existence of the specified SQLite file
  3. Fetch the table schema from Postgres
  4. Display a CREATE TABLE statement for the SQLite table
  5. Ask for your confirmation (can be suppressed with --confirm)
  6. Drop target table if it already exists (if --drop-table-if-exists is set)
  7. Estimate number of rows in the Postgres table for progress display
  8. Load source table rows and inserts them while they come in
  9. Verify that the number of inserted rows equals the number if loaded rows (if --verify is set)
#pg2sqlite doesn't do:

pg2sqlite creates the bare table with its columns. No primary keys, foreign keys, constraints or indexes are created in the SQLite table.


  • [ ] Create SQLite file if it doesn't exist


Be careful with the --drop-table-if-exists option. It will drop your SQLite-table without confirmation. This is useful if you use pg2sqlite in a scripted context. By default, if the table already exists in SQLite, pg2sqlite will terminte the process with an error message to avoid data loss.

#Type mapping

SQLite supports only a limited number of types aka. storage classes. Postgres' complex types like JSON, ARRAY etc. are not supported. These fields will be converted to TEXT fields by using Postgres' ROW() constructor.

Here is a table of explicit mappings:

PG Type SQLite Type
integer, smallint INTEGER
numeric REAL
All other types TEXT


Copyright © 2021 by Louis Brauer.

Software released under GNU GPLv3 license. See LICENSE.