Using the YAML output fails for non ascii characters.

dbexport user@livecopy/database/competition?link_name=Swisscom_Snow_Cup -i competition_definition,competition_definition.competition_instance.safari_route_id.user_2_route -Y

Actual

dbexport: 'ascii' codec can't encode characters in position 47-48: ordinal not in range(128)

Expected

<YAML>

Even in the the simplest case, which means that I'm exporting only one table without references, the Exporter does not include a restriction in the update statement.

E.g. the given statement is:

update section set s_id = 3824, id = 131055131155, duration = 200, length = 600 [...] where ;

instead of

update section set id = 131055131155, duration = 200, length = 600 [...] where s_id=3824;

The number of columns is to damn high!

When updating SQLAlchemy to a a version >= 1.0.8, dbexporter has NULL values only, in its generated statements.
The error can be mitigated by downgrading to an older version of SQLAlchemy.
The changelog (http://docs.sqlalchemy.org/en/latest/changelog/changelog_10.html) states that the query result class uses slots instead of dict now.
There are various locations throughout the code, where either dict is used, or at least these objects are treated as dictionaries.

Add a --daemon option that works similar to the Gradle --daemon option. The daemon should stay connected to the database for a few seconds or minutes, allowing to reuse the connection and gathered information from reflecting the database.

It could work that way:

When the option is set, use a separate DaemonDatabaseConnection (DDC) instead of the specific database connection. The DDC should then start the daemon (if not already running), connect to it, and use that daemon to execute the statements.

The daemon needs a separate writer that writes to a socket instead of stdout. It should also behave as a web server, the transport layer should be JSON (easy to debug with Postman, SoapUI, ...).

i would like to use placeholders in the -s option to reference the original value.

| id | name | value |
| 123456 | Berti Bauer | 3 |

-s value="myFunction('{value}')"

or even other column values:

-s value="myFunction('{id}')"

Use an in-memory database (like H2) for tests. As a requirement, add H2 to the list of available databases.

Calling the following leads to an error:

dbnav skiline@localhost/skiline/spot?name~%nigs%

Error:

dbnav: 'ascii' codec can't encode character u'\xf6' in position 1: ordinal not in range(128)

Title says it all.

Add an option to allow a dry-run.

On Ubuntu running in daemon mode does not have default value for hostname

Command:

echo "update chip_number_to_external_id_mapping set valid_from = ski_date, valid_until = ski_date where valid_from is null and valid_until is null and ski_date is not null and ski_date < '2015-09-02';" | dbexec skiline@localhost/skiline_test

Result:

dbexec: 'str' object has no attribute 'close'

Expected:

Changed rows: 217326

Use tabulate to display results, which creates a much cleaner table.

in.yaml:

includes:
    name="'Sölden 2015/16'"

dbargs in.yaml

Error:

dbargs: 'ascii' codec can't encode character u'\xf6' in position 17: ordinal not in range(128)

When exporting a table string values with a % sign get escaped and end up as %%. This leads to i.e. corrupt string formats (%%d:%%02d:%%02dh).

The exporter should reverse the order of the delete statements, i.e. delete article before deleting user.

dbexport -D dbnav.sqlite/article?id=1 -i user_id

Actual

delete from "user" where id = 558;
delete from article where id = 1;

Expected

delete from article where id = 1;
delete from "user" where id = 558;

Steps for reproduction:

• Install on a system without python database connector(s) but existing postgres connection configuration (.pgpass)
• execute dbexport user@connection/database/table
• Error message is "Specify the complete URI to a table"

Error message should be: "It was not possible to connect to the database. If the connection is not among the following connections, please install connector (see [location of description]):
[list of connection]

Add a pattern to filter the resulting query list. Only queries that match the given pattern will be displayed.

By adding a separate table (_comment) comments for all tables may be stored for databases that don't support table comments (i.e. SQLite).

Add support for python 3.

Add support for MySQL.

ijson egg file was missing
workaround sudo easy_install ijson,

The part that takes longest in executing is retrieving metadata from the database. The result should be cached, and the cache should not expire.

Additionally, add a CLI parameter to refresh the cache (--refresh-cache, invalidates cache and forces reload of metadata), and another to disable caching at all (--disable-cache, does not invalidate cache).