inputs/VegBank/: Added vegbank.~.utils.sql (which runs after vegbank.sql), for use by tables' create.sql scripts

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

inputs/VegBank/: Added observation_/

inputs/VegBank/: Added vegbank.~.clean_up.sql (which runs after vegbank.sql), to prevent "cannot alter type of a column used by a view or rule" errors

inputs/VegBank/: Added plot_/

inputs/VegBank/: Added plot_/

inputs/VegBank/: Added logs

input.Makefile: Staging tables installation: `%/install: %/create.sql`: Log the output to the install log, just like for other %/install targets

vegbien_dest: schemas: Added public explicitly, even though it's already in the default search_path, in order to shadow any datasource's tables of the same name as a VegBIEN table (such as in VegBank). (VegBIEN tables are referenced without a schema, while datasource tables are referenced with a schema, so collisions are not a problem after this fix.)

input.Makefile: Staging tables installation: sql/install: Fixed bug where needed space before \ at end of line, because one is not automatically added in a recipe command (although it's added elsewhere)

sql.py: run_query(): DuplicateException: Also match "of relation" part of error message, so that parsed column name does not contain "of relation"

subtract: Made it case- and punctuation-insensitive

mappings/: Removed no longer needed Veg+.cs-VegBIEN.csv, which is now the same as Veg+-VegBIEN.csv which was derived from it

join: Documented that it's case- and punctuation-insensitive.

bin/map: map_table(): Refactored to map simplified to original column names first and then determine column index for each original name, in order to avoid trying to recover the original name from a simplified name where multiple original names might collide onto the same simplified name. Documented that it's case- and punctuation-insensitive.

intersect, union: Made case- and punctuation-insensitive. mappings/Veg+-VegBIEN.csv: Removed no longer needed duplicate entries for each first letter case, which must now be removed for case- and punctuation-insensitive intersect/union to work. Note that the SpeciesLink `svn diff` hides _alt entry 0, which contains one of the removed duplicate columns that appears in the diff.

bin/map: map_table(): Resolve all mappings and prefixes after applying maps.simplify()

inputs/SpeciesLink/Specimen/map.csv: _alt all scientificNameAuthorship synonyms together in one _alt

schemas/functions.sql: _alt(): Added extra numbered parameters. Eventually these will need to be converted to variadic args, but this will require special support from column-based import.

join: Use new maps.simplify()

maps.py: Added simplify()

join: Match terms with non-alphanumeric chars removed

join: Match terms case-insensitively

Added inputs/TEAM/

sql_io.py: put_table(): Creating the into table: into_out_pkey: If is_function, just use "result" as the output column name, without prefixing the function name. This shortens the table names of function calls on function calls, which need a fixed column name to detect which columns are function results and use just the table names for those columns.

input.Makefile: Documentation: $(steps): Fixed bug where import make target needed to be changed to new single-table import target

schemas/vegbien.sql: analytical_db_view: Changed LEFT JOINs to JOINs where tables contain information that's required for the analytical DB. This should also enable the PostgreSQL query planner to make additional join optimizations, in the hopes of avoiding disk-space-intensive hash joins.

Replaced repr() with strings.urepr() (or equivalent) everywhere needed, to avoid future UnicodeEncodeErrors

Replaced str() with strings.ustr() (or equivalent) everywhere needed, to avoid future UnicodeEncodeErrors

sql.py: map_expr(): Replacing without quotes: Don't match unquoted name where it's preceded or followed by '.', because this could be a '.' embedded in a punctuation-containing column name, such as those frequently used by column-based import. Note that because database-internal names currently do not contain punctuation, this situation only occurs when a database-internal expression (such as a check constraint condition) is replaced in two steps, and the first step introduces punctuation-containing column names into the expression.

schemas/vegbien.sql: project: Don't require projectname to be specified when sourceaccessioncode is provided

sql_gen.py: ensure_not_null(): If type_ is set, cast the column to it if needed

README.TXT: Data import: Added testing steps to perform on local machine before running the import

README.TXT: Documentation: Redmine-formatted list of steps for column-based import: Updated make command for new table subdir name

sql.py: run_query(): Parse "types cannot be matched" error as MissingCastException to type text

sql_io.py: put_table(): Creating the into table: Fixed bug where in_pkey and out_pkey names would collide if the output and input pkeys have the same name (as is the case for SALVIAS.projects). This entails changing out_pkey to new into_out_pkey wherever the into table's out_pkey is created or referenced.

sql_io.py: put_table(): Combining output and input pkeys in inserted order: Changed sql_gen.Table to sql_gen.Col when creating the column references (they have a similar effect, so using the wrong type did not cause any tests to fail)

README.TXT: Added steps before the import to `svn up` and update the schemas

README.TXT: Merged Backups > After a new import and Data import sections into one Data import section that contains the steps to perform and back up an import. Note that many `svn diff` lines result from a change in indentation.

sql_io.py: put_table(): Combining output and input pkeys in inserted order: Fixed bug where column references would be ambiguous if the output and input pkeys have the same name (as is the case for SALVIAS.projects)

schemas/functions.sql: Added _nullIf() overload where the type param has type text, to handle cases where row-based import auto-casts all args to text in response to a 'could not determine polymorphic type because input has type "unknown"' error

schemas/vegbien.sql: party: Removed party_datasource unique index because it was causing problems with column-based import (due to multiple unique indexes covering the same columns in different ways), and because it prevented creation of more than one party per organization

xml_func.py: _if(): Documented that it must be run to remove conditions that functions._if() can't handle

README.TXT: Datasource setup: Testing: Added step to test column-based import (by_col=1), because it is stricter about types than row-based import and sometimes fails when row-based import succeeds

schemas/functions.sql: _nullIf(): Polymorphically support other datatypes besides text

bin/map: Clearing errors table: Fixed bug where needed to check if sql_io.errors_table() returned None (indicating that the errors table didn't exist) before calling sql.drop_table()

bin/map: Clearing errors table: Fixed bug where needed to use sql.drop_table() instead of sql.truncate() now that errors tables are not created until column-based import runs

input.Makefile: Maps validation: $(missingMappingsCmd): Fixed bug where need to use system's sort, not bin/sort, now that bin/ is added to the PATH by this makefile

inputs/SALVIAS/verify/plots.ref: Regenerated on PostgreSQL staging tables. The orders have changed slightly because this is derived from a PostgreSQL translation of the queries, with corresponding changes in collations and NULL sort orders. The counts have also changed slightly, possibly due to the changes Brad made to the salvias_plots database on nimoy after the initial version was downloaded. (The current counts are correct according to the current salvias_plots database.)

inputs/SALVIAS/verify/plots.ref.sql: # locations: Fixed bug where a NULL value in LatDec or LongDec would propagate to the concatenated value, reducing its uniqueness

inputs/SALVIAS/verify/plots.ref.sql: Retrofitted to work with PostgreSQL staging tables

schemas/vegbien.sql: project: Added project_unique_name_date unique index for projects that don't have a sourceaccessioncode

inputs/SALVIAS/plotMetadata/map.csv: Remapped project_id to project.sourceaccessioncode

inputs/SALVIAS/: Added projects/

input.Makefile: Sources: $(catSrcs): Fixed bug where needed to use cat_csv even if subdir was not actually a CSV table, because this also cats the header.csv file created for a subdir that references an already-installed staging table

input.Makefile: Existing maps discovery: Fixed bug where top-level logs dir needed to be excluded from list of subdirs that are treated as tables

my2pg: Prepend 'SET standard_conforming_strings = off;' because this defaults to on starting with PostgreSQL 9.1

schemas/vegbien.sql: locationevent: Made location_id optional when sourceaccessioncode is provided, since a sourceaccessioncode is globally unique and does not require a location to scope it

input.Makefile: Staging tables installation: Store install logs for full-DB exports in new logs subdir of main dir. This also fixes a bug where the install log itself was considered a DB export, because its extension was .log.sql.

Added inputs/SALVIAS/logs/

input.Makefile: SVN: add: Also add logs subdir of main dir, to store install logs for full-DB exports

mappings/VegCore-VegBIEN.csv: if subplot: Also forward locationID and plotName to the location of the parent locationevent (in addition to the parent location of the location), in order to "complete the diamond" connecting subplot locationevent -> (parent plot locationevent, subplot location) -> parent plot location

sql_io.py: cleanup_table(): NullValueException: Log the caught exception so it's clear that the update is being retried

input.Makefile: Staging tables installation: %/install: Fixed bug where $(if $(isRef)) needed to be checked before $(if $(nonXml)) because a subdir referencing an already-installed staging table must be treated specially by ignoring its autogenerated header.csv file, and not trying to install that file as if it were itself CSV data

my2pg, my2pg.data: Fixed bug where replacement for '0000-00-00' date needed to be wrapped in single quotes

input.Makefile: sql/install: Log the installation of a full-DB export to a log file in the main dir

input.Makefile: Staging tables installation: %/install: Factored out stderr logging into $(logInstall)

input.Makefile: Support empty subdirs referencing an already-installed staging table everywhere, by replacing $(isCsv) with new $(nonXml) where needed

inputs/SALVIAS/: Switched to using the DB export's staging tables instead of the exported CSVs

input.Makefile: Staging tables installation: Treat empty subdirs as referencing an already-installed staging table, and run cleanup and header export operations on them

input.Makefile: Staging tables installation: `%/install: %/create.sql`: Factored out cleanup and header export operations for reuse in other types of table subdirs

input.Makefile: Staging tables installation: `%/install: %/create.sql`: Removed deprecated (but benign) errors_table_only option to csv2db. Run csv2db without a command in order to clean up the created staging table.

sql_io.py: cleanup_table(): Removed no longer used cols param

csv2db: When no command is specified, just clean up the specified table

sql_io.py: cleanup_table(): Always clean up all columns in the table

sql_io.py: cleanup_table(): Handle NullValueExceptions (due to setting values to NULL in a NOT NULL column) by dropping the NOT NULL constraint

sql.py: Added drop_not_null()

sql_gen.py: is_text_col(): Also consider character varying to be a text type

csv2db: Removed no longer used errors_table_only option

README.TXT: Schema changes: Removed step to reinstall errors tables, because they are now created automatically by column-based import

csv2db: Removed no longer needed creation of errors table, because it is now created automatically by column-based import

input.Makefile: Staging tables installation: $(dbExports): Fixed bug where it would be non-empty even when the input contains no DB exports, because += adds extra whitespace. This caused sql/install to be incorrectly included as part of $(allInstalls).

db_xml.py: put_table(): Create errors table if it doesn't exist

sql_io.py: Added mk_errors_table()

inputs/Makefile: Input data: $(rsyncSrcs): Also exclude logs subdirs located at more than one level below the root, which occurs for example when a table subdir is moved into _archive/

input.Makefile: Staging tables installation: sql/install: Fixed bug where _always was part of $+, causing cat to try to cat this nonexistent file

Added inputs/SALVIAS/salvias_plots.schema.sql

Added inputs/SALVIAS/_MySQL/

input.Makefile: Staging tables installation: MySQL exports: Run all non-data-only exports through my2pg, not just schema-only exports. This supports transforming a combined schema+data export.

my2pg: Also perform data-only replacements, since default values can contain data-specific replacements. This also allows my2pg to transform a combined schema+data export.

input.Makefile: Staging tables installation: Also translate MySQL data to PostgreSQL

Added my2pg.data

input.Makefile: Staging tables installation: Place MySQL exports in separate _MySQL/ subdir so they don't clutter up the main dir, which will contain PostgreSQL translations

Added my2pg

input.Makefile: Staging tables installation: DB exports: Concatenate all exports together, with schemas first, so that any config options which were applied only in the schema export will remain active when the data is imported. Changed `%.pg.sql: .my.sql` to `.schema.sql: %.schema.my.sql` so there doesn't need to be a .pg suffix for PostgreSQL schemas and only the schema gets translated.

input.Makefile: Staging tables installation: $(dbExports): Don't consider MySQL DB exports as part of the DB exports that get installed, because they are not directly installable

input.Makefile: Staging tables installation: Added `%.pg.sql: %.my.sql` to translate MySQL DB schemas to PostgreSQL

inputs/SALVIAS/_src/: Added salvias_plots.sql.url to provide a link to where salvias_plots.sql was exported from (it was not a raw file given to us by the data provider)

Added cc_tty

inputs/input.Makefile: `%: %.make`: Don't automatically redirect stderr to a log file, because some .make scripts need to display password prompts, etc. on the TTY and output them to stderr instead of /dev/tty