xml_func.py: Removed no longer needed local XML functions that have been translated to SQL functions

input.Makefile: Testing: Removed VegBIEN.%.xml test because the import.%.xml test output includes the template tree that it's inserting, so there is no need to generate the XML tree in a separate test. This will also remove the need to maintain local XML functions that have already been translated to DB functions for the sole purpose of this automated test.

schemas/vegbien.sql: Made datasource_id required on every table that has it, to trigger the automatic population of it by sql_io.put_table()'s col_defaults

Moved importing of col_defaults from db_xml.put_table() to bin/map, so that it also happens in row-based mode. Note that this causes a DB entry for the datasource to always be created, even if the datasource has no mappings or no rows.

Use new exc.reraise() where exc.raise_() was used, so that the stack trace is preserved when the exception is rethrown

exc.py: reraise(): Take optional exception argument so it can be invoked in the same way as raise_(). Interestingly, this missing parameter does not produce the usual "...() takes no arguments (1 given)" error when the function is called inside an except block.

exc.py: Added reraise()

db_xml.py: put(): Inserting node: Wrap sql_io.put_table() call in catch-all exception handler that calls on_error_() (wrapper for error handler provided by caller) and returns None. This both adds additional debugging info to the exception (in on_error_()) and allows recovery from arbitrary exceptions that happen in sql_io.put_table(), so that an exception does not abort the import.

exc.py: get_e_tracebacks_str(): Use the current system traceback if the exception doesn't contain its own traceback(s)

schemas/vegbien.sql: specimenreplicate: Added locationevent fkey, since fkeys are not inherited from parent tables

schemas/vegbien.sql: Added datasource_id fkey constraints to all tables that needed it

bin/map: out_is_db: Use col_defaults in row-based mode as well

db_xml.py: Renamed put_table_special_funcs to put_special_funcs because it is now used by put() as well

db_xml.py: Moved put() before the functions that use it

db_xml.py: Renamed _put_table_part() to put(), replacing the existing put() whose functionality it now performs

db_xml.py: _put_table_part(): Reordered params to match put(), so that it can eventually be substituted for it

db_xml.py: _put_table_part(): Allow being invoked directly by adding defaults for parameters

db_xml.py: put(): Use _put_table_part(). This will ensure that all the put-related functionality is in one place, rather than duplicated.

db_xml.py: _put_table_part(): Append the node to errors handled with on_error()

sql_io.py: Added own SyntaxError class to replace built-in SyntaxError because it stringifies to only the first line

input.Makefile: Testing: Removed $(via).%.xml tests because they require the via format (DwC/VegX) to be XML, but we want to flatten VegX into a DwC-like set of CSV column names

Removed inputs/NY/test/VegX.specimens.xml.ref because NY is not mapped via VegX

input.Makefile: Testing: Renamed import.*.out tests to end in .xml because they now contain XML import trees for validation, and this extension turns on XML syntax highlighting in a text editor

bin/map: out_is_db: Output the put template to stdout so it will be validated in the automated testing

xml_func.py: process(): If local XML function can't be found, just replace with last param instead of returning an error. This allows DB-only functions to be ignored in XML output mode.

sql_gen.py: ColDict.__setitem__(): Fixed bug where None value should not be replaced with column default value if column has no underlying table

sql.py: DbConn.col_info(): If column does not exist, raise sql_gen.NoUnderlyingTableException

sql_io.py: put_table(): In log messages, use `.to_str(db)` instead of repr() where possible to use the SQL syntax of the DB driver

sql_io.py: put_table(): ignore(): Replacing invalid value with NULL in nullable column: Corrected log message to "Replacing invalid value ... with NULL in column ..." because the rows with that value are not ignored in that case

sql.py: run_query(): InvalidValueException: Parse any exception ending in "out of range", not just "field value out of range", in order to support errors that the timezone is out of range

schemas/py_functions.sql: _dateRange*(): Made functions STRICT because they return NULL on NULL input

sql_io.py: put(): Use a simple case of put_table(), which now supports everything put() needs. This will enable all row-based and column-based processing to be maintained in the same function, put_table(), and avoids the need to reimplement any column-based functionality (like SQL functions) in put().

xml_dom.py: NodeTextEntryIter: Allow empty values through as None, and instead filter them out in TextEntryOnlyIter using new helper function non_empty(). This allows XML functions to decide for themselves whether empty values should be filtered out, because process() will now no longer automatically remove them. This will enable process() to work with SQL functions, which must not have empty values filtered out because this will remove required, but nullable, arguments.

xml_func.py: Use conv_items() in every XML function that needs empty (NULL) entries removed, so that they are not dependent on what process() does to the items

sql_io.py: put_table(): ignore(): Support invalid literals in addition to invalid column values. This also allows put_table() to fully support being called by put().

xml_func.py: process(): In row-based mode, if function is not explicitly a relational function but does not exist as a local XML function, treat it as a relational function. This will help in merging sql_io.put() and put_table(), since put() did not support SQL functions but put_table() does, and this ensures that a SQL function is always used if the local XML function has been removed in favor of it.

sql_io.py: put_table(): Removed into param to set a custom into table name because put_table() now has all the info it needs to generate this name automatically, and callers are no longer providing it

bin/map: by_col: db_xml.put_table() call: Use new col_defaults param to automatically set datasource_id to the in_label (datasource name)

xpath.py: path2xml(): Skip to tree created inside root, since that is how callers want to use the returned node

db_xml.py: put_table(): Import col_defaults to translate nodes to pkeys

db_xml.py: _put_table_part(): Support no in_table, for iterations with only literal values

sql_io.py: put_table(): is_literals: When ignoring all rows, return default value instead of always None

db_xml.py: put_table(): Removed parent_ids_loc and next params since these are only used in the recursion

db_xml.py: put_table(): Split into an outer function that sets up the database environment and subsets in_table, and a (recursive) inner function that imports the data

db_xml.py: put_table(): Subsetting and partitioning in_table: Documented that it's OK to do this even if table already the right size because it takes <1 sec

sql_io.py: put_table(): Use is_function where caller-provided is_func was used, since is_function determines whether something is a function based on whether it actually exists as a SQL function instead of just whether its name starts with "_". Removed now-unneeded is_func param.

sql_io.py: put_table(): Added col_defaults param and use it if there's a missing mapping for a NOT NULL column. This requires callers passing arguments by position to add an empty value for this parameter.

bin/map: by_col: Only clear errors table if doing full re-import starting from row 0, not if restarting import at a later row

input.Makefile: Import to VegBIEN: Fixed bug where `&>>` was used to append stdout and stderr to the log file, but is not supported on Mac OS X. Replaced with `&>` (overwrite instead of append) because log file is unique by date/time the import runs, so there won't be an existing log file that would be overwritten.

schemas/vegbien.sql: Added datasource_id to all tables with a sourceaccessioncode (and corresponding *_unique_datasource constraint on these columns) so they can be directly looked up using just the input table's own fkey to parent. This will enable loading hierarchical (plots) data without "breadcrumbs", a huge benefit! Also added sourceaccessioncode wherever there was a datasource_id, to standardize on these names as being the columns that link directly to the input table rows.

README.TXT: Datasource setup: Installing the staging tables: View the logs: Fixed bug in tail syntax to also work on Linux

Added inputs/Madidi/ with empty mappings

README.TXT: Datasource setup: Populating the src/ subdir with input data: Added step to make sure each header in multiple part files for a table is EXACTLY the same

README.TXT: Datasource setup: Installing the staging tables: Added steps to deal with colliding column names in the flat file headers. Added command to view the logs.

csv2db: log(): sys.stderr.write(): Run strings.to_raw_str() on message to handle Unicode chars

csv2db: Run strings.to_unicode() on column names to handle Unicode chars

csv2db: esc_name(): Use db.esc_name()

Added inputs/BIEN2.datasources.xlsx (formerly bien_data_sources.xlsx in nimoy:/home/bien/raw_data/)

exc.py: e_msg(): Added assertions to check that e.args is compatible with this function

exc.py: Use new e_str() where its definition was used

exc.py: Use new Unicode-safe e_msg() instead of strings.ustr() on exceptions

exc.py: e_msg(): Run strings.ustr() on the returned string so it will be appendable to other Unicode strings

exc.py: Added e_msg(), e_str() (from SQL py_functions._date())

db_xml.py: put_table(): Adding fkey to parent: Fixed bug where should only add parent_ids_loc table to list of tables not to truncate if it's a column, because it is sometimes just a pkey value when that iteration contained only literals

inputs/import.stats.xls: Updated with stats from latest import

inputs/import.stats.xls: Corrected date of last import

sql_gen.py: plpythonu_error_handler: Fixed bug where PL/Python exceptions could not be filtered by strings after the first line, because only the "message" portion of the exception is available in SQLERRM

schemas/py_functions.sql: _date(): YMD parsing: Fixed bug where exception for ValueError needed to be stored in local var so its message could be parsed

sql_gen.py: plpythonu_error_handler: Always raise PL/Python exceptions as data_exception so they go in the errors table, instead of aborting the iteration

sql_gen.py: plpythonu_error_handler: Fixed bug where not all PL/Python exceptions start with "PL/Python: " (e.g. on PostgreSQL 9.1 on vegbiendev), so the PL/Python prefix must be optional. Refactored to put IF clause for non-PL/Python exception at end for a more logical ordering of the conditions.

Added inputs/CVS/

README.TXT: Datasource setup: Added steps to place the relevant files under version control

README.TXT: Datasource setup: Accepting the test cases: Don't auto-accept the initial tests because there could be bugs in the initial mappings that would be revealed upon inspecting the test output

sql_gen.py: plpythonu_error_handler: Added section comment before handler block, so that it's clear in the (very long) wrapper function definition what the block is doing

input.Makefile: Documentation: import/steps.by_col.sql: Added -s to make to avoid echoing make commands to the log file

README.TXT: Moved Reinstall all datasources at once to Schema changes and renamed it to Reinstall staging tables to reflect that it is only necessary when the staging table format is changed

README.TXT: Datasource setup: Updating vegbiendev: Added step to also install the staging tables on vegbiendev

README.TXT: Datasource setup: Moved Install the staging tables before Map each table's columns because the install can run in the background while you're mapping. It must, however, come after Auto-create the map spreadsheets because it uses the filenames of the created maps to determine which staging tables to create.

README.TXT: Datasource setup: Adding a new datasource: Changed <short_name> to <name> to match usage elsewhere. Documented that it may not contain spaces, and should be abbreviated.

README.TXT: Datasource setup: Added steps to update vegbiendev

inputs/Makefile: Input data: Added upload target

README.TXT: Datasource setup: Added steps to accept the test cases and commit

README.TXT: Datasource setup: Added step to install the staging tables

bin/map: in_is_xml: doc2rows(): "Root not found in input" warning: Changed "error" to "warning" to match the type of error condition signaled

bin/map: map_rows(): out_is_db: Changed `id_node != None` assertion to a warning because this is a normal circumstance in the base case where there are no mappings

input.Makefile: Testing: Added test/accept-all

csv2db: COPY FROM: Fixed %-injection bug where column names' %s were not escaped prior to cursor.mogrify(), by changing the code to use inline db.esc_value() instead

bin/map: in_is_xml: doc2rows(): "Root not found in input" error: Changed SystemExit to a warning because this is a normal circumstance in the base case where the input XML file contains no rows

README.TXT: Datasource setup: Documented how to map each table's columns

README.TXT: Datasource setup: Changed "Auto-create the src column spreadsheets" to "Auto-create map spreadsheets" and updated command to bootstrap all maps, including newly-autogeneratable via maps

input.Makefile: Maps building: maps/$(via).%.csv: Auto-create by copying the src map if doesn't exist. Existing maps discovery: Look up via format in src maps' roots if no via map already exists.

src_map: Fixed bug where non-header rows needed to be materialized with empty fields for each column in the header

input.Makefile: Maps building: Via maps cleanup: Match maps/$(via).%.csv with pattern instead of $(viaMaps) var so that a non-existing via map will have the recipe run, too. When auto-creating via maps is later added, this will be required.

inputs/*/maps/src.*.csv: Regenerated using new src_map output format

parallelproc.py: MultiProducerPool: Removed warning if not using parallel processing because this also gets generated when it's explicitly turned off, which is currently the case and clutters up stderr when testing

src_map: Also add columns for the output mappings and comments, so that the src map can be directly copied for use as the via map (DwC.specimens.csv, etc.). The output mapping column name must be provided by the caller, which input.Makefile maps/src.%.csv provides using the new mappings roots.

Added mappings/roots for use in creating src maps

input.Makefile: Maps building: maps/src.%.csv: Clean up by passing through `$(bin)/cols '*'` whenever it's changed. This ensures that the CSV dialect is always consistently Python's Excel dialect. (Note that this dialect actually uses \r\n as the line ending. The \n line endings were from src maps generated by a previous version of bin/src_map.)

input.Makefile: Maps building: maps/$(via).%.full.csv: Removed alternate rule when $(srcMap) doesn't exist, because this effect is actually achieved by the no-prereqs rule for maps/src.%.csv, which causes make to think it exists when matching pattern rules even if its recipe doesn't actually create it

input.Makefile: Maps building: maps/$(via).%.full.csv: Added alternate rule when $(srcMap) doesn't exist