Upgrade instructions

From versions 0.6.x to Datconv 0.7.x

  • To support iterators Reader and Writer interfaces were extended. Following changes should be done in Reader classes: * If possible, write Iterate method which will work in similar way than Process but should call yield on object returned by writer’s writeRecord method. Typically Iter method is a clone of Process with writeRecord(...) call replaced with yield writeRecord(...). Following changes should be done in Writer classes: * Remember header and footer passed in funtions: writeHeader / writeFooter, and return them in new functions: getHeader / getFooter. * writeRecord class method should now return an object which it pass to OBJECT type output connector (prevoiusly this method did not returned any value.

From versions 0.6.0 to Datconv 0.6.1

  • Output connectors datconv.outconn.*.ddl were changed in non-compatible way: parameters primary_key and not_null were replaced by column_constraints, common_column_constraints and table_constraints.

From versions 0.5.x to Datconv 0.6.x

  • The type of datconv.writers.dcjson parameter convert_values was changed from boolean to int. If you had false change to 0, true - change to 1 or 2. Also default value changed to make conversion by default.

  • Name of of datconv.writers.dcjson parameter ign_rectyp was changed to ignore_rectyp to unify names with xpath writer.

  • Old fashion parameter Reader:PArg:outpath is still supported as fall-back, however it is recommended to explicitly specify output connectior for better clarity. I.e. replace:

     Reader:
     Module: datconv.readers.dcxml
     # ...
     PArg:
         outpath: "out/AcctAgentAdjustment_c5019_s38.xml"

    with:

    OutConnector:
    Module: datconv.outconn.dcfile
    CArg:
        path: "out/AcctAgentAdjustment_c5019_s38.xml"

  • Default debugging level is now INFO (was WARNING) so some information messages are printer to console when program is run without any special logger configuration. To restore previous (silent) behaviour add DefLogLevel: WARNING root key in your configuration file.

From earlier versions to Datconv 0.5.1

New Filter optional functions setHeader() and setFooter():

  • If you managed to write custom Reader, it must be updated in following way: call filter optional functions before calling Writer writeHeader() and writeFooter() functions as shown below:

    if self._flt is not None:
    if hasattr(self._flt, 'setHeader'):
        self._flt.setHeader(self._header)
self._wri.writeHeader(self._header)

    and in similar way for Footer.

  • If you have a filter that does some work at the end of converion process (like writing some statistics) and you implemented it in __del__() fumction, this finalization code can be now moved to setFooter().

From Datconv version 0.2.x to Datconv 0.3.x

Changes in parameters default values

Changed default value for parameter encoding of datconv.writers.dcxml. It was ascii and is now unicode. Note that this parameter is ignored while run against Python3.

From Pandata version 0.1 to Datconv 0.2.x

Changes in files’ layout

Pandata version 0.1 did not had its own installer. Everything was placed in one folder. Now Datconv (renamed from Pandata) modules included in package are being installed in special Python subfolder intended for 3-rd party modules. So all old Pandata core modules may be deleted from current localisation. Recomended Datconv layout is now to keep user own modules in subfolders of one folder, and run datconv script from that folder or add user’s “projects’ folder” to PYTHONPATH environment variable. Datconv core modules should be kept in Python sub-folders where installer placed them.

Modules and classes’ name changes

Names of standard Reader and Writer modules has changed: pdxml -> dcxml, pdcsv -> dccsv, pdxpaths -> dcxpaths. Appropriate changes should be done in user’s YAML configuration files.

Fixed names of Reader, Filter and Writer classes has changed: PDReader -> DCReader, PDFilter -> DCFilter, PDWriter -> DCWriter. Appropriate changes should be done in user’s filters’ definition files.

All version 0.1 modules were moved to common datconv meta package as subpackages. So in user’s YAML configuration files and in import statements in filters datconv. specification should be added.

Example 1 (YAML file):
Module: writers.pdcsv
should be changed to:
Module: datconv.writers.dccsv
Example 2 (filter definition):
from filters import WRITE
should be changed to:
from datconv.filters import WRITE

Change in main program invocation

Datconv main script is now included into system PATH. On Linux .py file extension was removed, so it can be called like (without ./ prefix): datconv [options]. On Windows script was renamed, so it can be called like (without python explicit invocation): datconv-run.py [options].

Note

Windows Python installer should create file-type entries that allow user to directly call python script (without python explicit invocation).

To check that, run following connads from command box:
C:\>assoc .py
should give:
.py=Python.File

and:
C:\>ftype Python.File
should give:
Python.File="c:\python27\python.exe" "%1" %*

Important is %\* at end — what allows to pass additional arguments to program.