lib/sh/util.sh: $sudo_cmd_wrapper: use esc() so this does not need to be embedded with [@] array notation

bugfix: lib/runscripts/util.run: $wrap_fn: this always needs to contain `is_runscript=1` because $is_runscript is not (and cannot be) exported

bugfix: lib/sh/util.sh: sudo(): when don't need sudo: do need `command` to have effects of invoking external command

fix: lib/sh/db.sh: pg_snapshot(): first need to run pg_stop_backup() in case a backup was not stopped cleanly

fix: lib/sh/db.sh: pg_stop_backup(): ignore benign "a backup is not in progress" error

lib/sh/util.sh: sudo(): only run sudo if needed

lib/sh/util.sh: sudo(): factored PATH-preserving code out into separate wrapper var that can also be used separately, eg. with `sudo -u`

lib/sh/util.sh: sudo(): moved addition of -E (which should always happen) to separate, overridden sudo() function

lib/sh/util.sh: moved -E to is_callable section because it just applies to that

lib/sh/util.sh: added become_sudo alias

lib/sh/util.sh: added $is_root

lib/sql_io.py: cleanup_table(): trim(): documented that this also converts character varying fields to text

bugfix: lib/sql_io.py: null_strs_str_default: removed "NA" because this is the abbr for a Spanish province (Navarra). this fixes the 2nd bug of #955, geovalidation duplicated rows.

lib/sql_io.py: cleanup_table(): debug-print null_strs

lib/sql_io.py: null_strs: made it customizable from an env var, since the same list of null_strs doesn't work for all datasources (see #957)

fix: *Makefile: changed line endings to \n so that `patch` can work with pasted input. use `svn di --extensions --ignore-eol-style` to verify no diff.

lib/sql_io.py: null_strs: made it customizable from an env var, since the same list of null_strs doesn't work for all datasources (see #957)

bugfix: lib/tnrs.py: encode_map: also need to encode + because TNRS removes it from the morphospecies (vegpath.org/wiki/CVS_validation#Bobs-revised-document > issue #4)

lib/sql_io.py: null_strs: added N/A and NA (this will remove a common abbr for North America, but we don't use the continent, so this is OK)

lib/runscripts/table.run: added check_headers()

bugfix: lib/runscripts/table.run: srcs: missing []

lib/runscripts/table.run: added header() and use it in header.txt()

lib/runscripts/table.run: 1st_src(): use a variable for this instead, to avoid needing to run this function each time it's used, and to make @srcs available

lib/sh/util.sh: wildcard.(): clarified that it only removes . .. when at then beginning of the list

lib/runscripts/table.run: added header.txt()

lib/runscripts/table.run: added 1st_src()

lib/runscripts/subdir.run: subdir_make(): use new $datasrc_dir

lib/runscripts/subdir.run: added $datasrc_dir

lib/sh/make.sh: make(): added support for $output_data mode which uses --silent

fix: lib/sh/util.sh: already_exists_msg(): changed calling convention to avoid it seeming like `return 0` is run if already_exists_msg() throws an error, when in fact already_exists_msg() is just a command that should be run before returning/errexiting

bugfix: lib/common.Makefile: $(wildcard/): need final pass with $(wildcard) to support inputs without wildcard chars

bugfix: lib/sh/local.sh: $sync_remote_url: jupiter user is always aaronmk, not the local user

*{.sh,run}: use standard WARNING syntax for warning labels

*{.sh,run}: use standard WARNING syntax for warning labels

fix: lib/sh/util.sh: verbosity_compat(): documented that this should not be run until right before executing an external command, so that it doesn't mess up the logging mechanism

bugfix: lib/sh/util.sh: functions called by pst(): commented out/removed logging calls that would have caused infinite recursion when pst()'ing a logging function

lib/sh/util.sh: echo_func(): get call context before wrapper(s), which is more useful for debugging. this uses skip_stack_frames()'s lookahead=1 mode.

fix: lib/sh/util.sh: skip_stack_frames() callers: updated for new skip_stack_frames rather than get_stack_frame behavior

lib/sh/util.sh: skip_stack_frames(): added lookahead support, which looks at entry after current to deterine whether to skip current. this is useful for skipping wrappers, by looking at the calling function's name.

fix: lib/sh/util.sh: skip_stack_frames(): for new skip_stack_frames rather than get_stack_frame behavior, stack frames must be skipped in the caller to preserve the stack frame pointer

lib/sh/util.sh: get_stack_frame_after(): renamed to skip_stack_frames() for clarity

lib/sh/util.sh: added skip_stack_frame_in_caller, unskip_stack_frame_in_caller

lib/sh/util.sh: added prev_stack_frame

lib/sh/util.sh: echo_func(): use new format_stack_frame, which adds call context information to what was provided by func_loc

fix: lib/sh/util.sh: format_stack_frame(): need to hide canon_rel_path() info using log+

lib/sh/util.sh: added get_stack_frame_after()

lib/sh/util.sh: added matches()

lib/sh/util.sh: next_stack_frame: documented usage

fix: lib/runscripts/util.run: runscript template: all(): moved example commands to target(), where they would more likely be located

lib/sh/util.sh: format_stack_frame(): support including args

lib/sh/util.sh: debugging: added pst() (print_stack_trace)

lib/sh/util.sh: added stack_trace(), print_stack_trace()

lib/sh/util.sh: added format_stack_frame()

lib/sh/util.sh: added get_stack_frame() and helpers

lib/sh/util.sh: terminal: moved before errors so it can be used by it

lib/sh/util.sh: errors, debugging: moved after datatype sections so their functions can use these

bugfix: lib/sh/util.sh: canon_rel_path() stub: proper no-op requires passing through original path

lib/sh/util.sh: canon_rel_path(): fall back to original path if can't resolve, instead of errexiting

bugfix: lib/sh/util.sh: canon_rel_path(): don't re-localize $path because this clears it

lib/sh/util.sh: canon_rel_path(): import $1 to $path before function body, so that the function body can be moved to a nested function

lib/sh/util.sh: added canon_rel_path() stub for use by debugging functions

lib/sh/util.sh: moved func_loc() to before debugging section so it can be used by debugging functions

bugfix: lib/sh/util.sh: command__exec(): need to restore $verbosity before calling die_e

fix: lib/sh/local.sh: $sync_remote_url: need $USER so user can be overridden when running as root

lib/Firefox_bookmarks.reformat.csv: label page's self-description as such: also support quotations enclosed in '

lib/sh/util.sh: echo_vars(): merge repeated flags so there aren't flags in between the vars (which is also not valid declare syntax)

lib/sh/db.sh: pg_cmd(): log vars on same line to avoid clutter

lib/sh/util.sh: echo_vars(): put all the vars on the same line so they don't clutter up the call graph generated at the default verbosity

lib/tnrs.py single_tnrs_request(), bin/tnrs_client: use_tnrs_export: default to False because this mode uses incorrect selected matches (vegpath.org/issues/943), and the JSON mode that fixes this is now available

bugfix: lib/csvs.py: JsonReader: need to pass col_order to row_dict_to_list_reader

bugfix: lib/tnrs.py: JSON output: need to stringify arrays so they match what is output in TSV-export mode

lib/csvs.py: JsonReader: added support for values that are arrays

lib/csvs.py: MultiFilter: inherit from WrapReader instead of Filter to avoid needing to define a no-op filter_() function

bugfix: lib/csvs.py: row_dict_to_list_reader: need to override next() directly instead of just using Filter, because Filter doesn't support returning multiple rows for one input row (in this case, prepending a header row). this caused the 1st data row to be missing.

lib/csvs.py: Filter: inherit from WrapReader, which separates out the CSV-reader API code

lib/csvs.py: added WrapReader

lib/csvs.py: added Reader

lib/csvs.py: JsonReader: factored out row-dict-to-list into new row_dict_to_list_reader so that JSON-specific preprocessing is kept separate from the row format translation

lib/csvs.py: added MultiFilter, which enables applying multiple filters by nesting

lib/tnrs.py: single_tnrs_request(): JSON mode: implemented output of JSON data

lib/tnrs.py: single_tnrs_request(): factored out wrapping in TnrsOutputStream, since this is done for both modes

fix: lib/tnrs.py: JSON mode: TSV export columns: need to translate these to JSON column names before they can be used with the JSON data

lib/csvs.py: added JsonReader, which reads parsed JSON data as row tuples

lib/csvs.py: added row_dict_to_list(), which translates a CSV dict-based row to a list-based one

lib/csvs.py: RowNumFilter: added support for filtering the header row as well

lib/csvs.py: ColInsertFilter: added support for filtering the header row as well

lib/csvs.py: InputRewriter: documented that this is also a stream (in addition to inheriting from StreamFilter)

bugfix: lib/csvs.py: InputRewriter: accept a reader, as would be expected, instead of a custom stream whose lines are tuples

fix: lib/sql_io.py: append_csv(): use new csvs.ProgressInputFilter instead of streams.ProgressInputStream(csvs.StreamFilter(__)), so that the input to csvs.InputRewriter is a reader, not a stream. this avoids the need for csvs.InputRewriter to accept a stream whose lines are tuples, instead of the expected reader.

lib/csvs.py: added ProgressInputFilter, analogous to streams.ProgressInputStream

lib/sql_io.py: added commented-out debug statement used to troubleshoot copy_expert() errors

lib/dicts.py: added pair_keys(), pair_values()

bugfix: lib/streams.py: CaptureStream: end_idx must also be > start_idx

lib/tnrs.py: single_tnrs_request(): use_tnrs_export=False: need to obtain export columns

lib/csvs.py: added header(stream)

fix: lib/tnrs.py: single_tnrs_request(): need to `assert name_ct >= 1`, because with no names, TNRS hangs indefinitely

bugfix: lib/sh/archives.sh: compress(): don't include dir prefix in zip archive

lib/sh/util.sh: cd(): use echo_run instead of a manual echo_cmd call

fix: lib/sh/util.sh: cd(): indent after running cd rather than before

lib/sh/util.sh: cd(): support rebasing path vars for the new dir