ASDM backend for VIPER Framework

pyasdm is still under active development

• API
  • pyasdm
  • BDF
  • enumerations
  • exceptions
  • types
  • utils
• Examples
  • Example 1
  • Example 2

Overview

pyasdm provides a pure-python interface to the ASDM content (tables and rows) and the binary data (BDF) that is the part of that ASDM.

It follows the class design of the Java and c++ ASDM interface that is part of ALMA software. The same c++ classes are also incorported into CASA.

Most of the pyasdm code is generated from the ASDM UML by the ALMA software build process in the same way that the Java and c++ code is generated. The code follows the Java implementation most closely although it does borrow from the c++ implementation in several places.

The BDF classes in pyasdm are hand written following some of the design of the c++ code that reads and parses BDF files. There is no Java code in ALMA that reads BDF files.

The ASDM code can write out an ASDM except for any BDF data. BDF write functionality is planned but has not yet been started.

Documents

• SDM Tables Short Description.

• Science Data Model Binary Data Format.

Todo

DataViews

The c++ BDF code includes classes and structures that are useful when filling a MeasurementSet v2. Those date data views are unlikely to be useful for the users of pyasdm. Useful views need to be defined an implemented. This layer also includes useful methods to link back from a BDF to the relevant meta data in the ASDM. This layer may also include some data selection methods and criteria.

Binary Tables

Tables can be stored in binary form. These are read by the current classes but it follows the Java implementation which reads of the data into memory at once. For some tables (e.g. Pointing) that can use quite a lot of memory and can take a long time to read. See also the next item. The design of the ASDM allows for the size of a table stored in the ASDM.xml file to be different from what is actually found in that table. Since optional fields in a row of a table can come and go from row to row it’s not possible to skip to a random row in a table without reading all of the previous rows (although it’s not necessary to store them, see the next item) and it’s not possible to be sure how many rows are in a table until all rows are read. The current behavior is unacceptable and so a way will be found to provide access to chunks of rows at a time without having all rows in memory. Stepping through a table in order will be efficient. Jumping to a random row will not be.

Row uniqueness checks

The Java classes always check on row uniqueness for the required key fields when a table is ready. The c++ classes can optionally skip that step. Skipping that step can save considerable time when reading large binary tables (see the previous item). This needs to be implemented.

Read early SDM versions

The current code only reads versions (schemaVersion in ASDM.xml) >= 3. Earlier versions are handled in the c++ code by using the libxslt library and two xsl (stylesheet) files packaged with casarundata that describe the transformations from version 2 to version 3 for both ALMA and EVLA data (the transformations are different). That transformation is not yet available in pyasdm. Attempting to read an ASDM without a schemaVersion or with a schemaVersion value < 3 will raise a ConversionException. That transformation is also not available in the Java code (so pyasdm currently behaves as the Java code does).

BDF write.

This will be added.

Test code

This will include equivalent test code for the c++ ALMA classes used by CASA and ALMA to ensure that all of the classes agree. There is currently no test code for the c++ (or Java) ASDM classes.

Documentation

This obviously needs much work.