On testing the relationships for LiteratureSource and CreationInfo, it
was found that they were not one to one, although they were one to one
on the schema level. An extra parameter was added to the relationships
on the parent side, backref=().

The unit tests have now been fixed to run without error and some
information has been added to the readme file, but this has gaps. The
app file has had an extra warning added correctly informing the user
that a config file does not exist rather than the config section not
existing.

Exploring the usage of the WS with Postgresql, it was found that the
column names for flattened tables LiteratureSource etc. were too long
for the character limit, and these were truncated so that the db could
not be built.
The >63 char limit meant that the only sensible solution was to
reimplement the table structure as being nested tables.

Test files have also been moved to appropriately named folders.

Also some cleaning up of moved files was done.

The datetimes output from the schema will always be utc, with
a timezone of 0, so the timestamp has been removed.

There is a specific case where if hydraulic query parameters are
used and these filter out all the hydraulic children, then nothing
was returned. We thought the expection should be that if there
are no matching children, the parents should still be returned if
they are valid.

Validation on query parameters has also been updated in accordance
with the Ramsis core code so that these match.

When datetime query parameters were used to filter hydraulics,
The sections of the borehole would all be returned regardless
of the section epoch time. Now logic has been added to also
filter sections on epoch time if this does not overlap with
the hydraulic search criteria.

A test case was added to the populate_db_cases.py to test.

Previously this route would always return a list, although more
than one JSON boject would never be returned.

Flask migrate functionality was added as this might be required in
the future of using postresql. Some print statements were removed
with this commit.

The SQLite db and the code that generates the content for this
populate_db_test_cases.py has been added.

Some small changes made to comments and docstrings in files.

Addition of manage.py to manage migration within the postgresql db.
This is not finished at present.

Some refactoring of code was required to make the code cleaner.

Added more validation to the schema based on what input data is allowed.

Centralised the column prefix m_

Added more docstrings.

Cross-code change to the way the schemas have been implemented and where
they are used. Each fields that will be output from a schema is uniquely
referenced so that the same schema can be used for validataion on
deserialization and also for producing json output.

Added measured_depth to the borehole level.

Using a custom parser with strict validation on the input query
parameters, as when using webargs, inputing a parameter that is not
included in the schema would not return an error. I have added the tests
for the strict.py parser file but these do not pass yet.

It was decided that it would be sensible if the column key
closely matched to the attribute name. Previously this
was deeply nested and made very long and confusing attribute
names. Have also added the ability to give a different column
name to the attribute name.

As it didn't belong in an existing file, this has been
moved. The tests and the code are currently unfinished,
further testing required.

It was decided that it was a cleaner implementation to name the
ORM columns and ORM attributes differently.

This is cleaner without having to reference the m_ prefix all the
time, and the user should not have to know about this.

To avoid having m_ prefix on the attributes but still maintaining
prefix on column keys required an extra arg to be added to most
functions. Also fixed the bug where if an attribute was
previously called:
m_literaturesource_m_literaturesource_creator_m_\
literaturesource_creator_person_givenname

it would now be called:
m_literaturesource_literaturesource_creator_\
literaturesource_creator_person_givenname

(m_'s removed from within the name)

Question raised about whether we want the attribute to be called:
m_literaturesource_creator_person_givenname
but on attempting to modify this, an mapping error is called.

Implemented cleaner way of doing column comparison to input query
parameters. No functionality currently to do specialised filters which
use joins, but this use will be added if and when required.

Unit tests writen for new class.

sarsonl authored 5 years ago and sarsonl committed 5 years ago

Refactoring code to try and minimise use of same code multiple times. Have to discuss the use of None filtering in query as I am not sure about the purpose.

When evaluating time constraint parameters validate the
HydraulicSample's datetime value, as well. As a consequence, the
response returns only data, when there is at least a single valid sample
available.

Note, when creating a QuakeML quantity field the type of the '*_value'
field was not set properly.

Start implementation of BoreholeHydraulicDataListResource. Currently,
the only supported output format is JSON.

GeneralSchema hosts the parser's HYDWS common query parameters.

Make it executable as hydws-db-init.