libhxl-python v4.22

• consolidate recent bugfix releases
• use same input code for XLS (old Excel) and XLSX (new Excel) via xlrd3 library
• more-informative exception for an out-of-range Excel sheet index

libhxl-python v4.21.3

• Bug-fix release: switch to xlrd3 library, since xlrd has dropped support for XLSX files
• be more careful about detecting non-XLSX zip files
• be more efficient with memory usage for Excel files
• for web-security reasons, unless allow_local is True, block fetching datasets from localhost, *.localdomain, or any dotted quad

libhxl-python v4.21.2

Bugfix release: avoid rare Unicode error and end of lookahead buffer for CSV delimiter detection.

libhxl-python v4.21.1

Bugfix release: correct issue handling HXL Proxy URLs as a data source, when a saved recipe has a URL override.

libhxl-python v4.21

• BACKWARDS-INCOMPATIBLE: rename "whitelist" to "includes" in
  hxl.model.Dataset.with_columns() and JSON format for loading a
  ColumnFilter
• BACKWARDS-INCOMPATIBLE: rename "blacklist" to "excludes" in
  hxl.model.Dataset.without_columns() and JSON format for loading a
  ColumnFilter
• added a new ExpandListsFilter and hxlexpand command-line script
• added --skip-untagged option to hxlcut script
• improved docstrings for hxl.io, hxl.datatypes, hxl.converters, and hxl.geo modules
• added hxlspec command-line script to process a HXL JSON spec
• added URL munging for HXL Proxy links (non-CSV/JSON)
• added URL munging for Kobo survey links (requires an Authorization: header)
• if there's an error parsing a formula, return the value "** ERROR **" and log the error
• added requests_cache to requirements (to disable caching for API calls)

libhxl-python v4.20

• switch to jsonpath-ng library to support more JSONPath features
• add support for row formulas as right side of row queries (making it possible to compare one column to another)
• add today() function for row formulas (today's UTC date in ISO YYYY-mm-dd format)
• add support for parsing Unix-style epoch timestamps (nanoseconds, milliseconds, or seconds)
• add long description in setup.py for PyPi (from README.md)
• work around xlrd date-handling bug (for malformed Excel dates)
• more cleanup on docstrings

libhxl-python v4.19

• allow selector for JSON data to be a JSON path as well as just a simple token
• add --selector option to all command-line scripts
• added toupper() and tolower() row functions
• fixed bugs with date heuristics and with JSONpath parsing of non-strings
• various code cleanup
• add a generic Makefile to simplify common testing and git tasks

libhxl-python v4.18

• added optional encoding parameter to hxl.io.data() etc. to force a character encoding
• fix bug opening an Excel file when the server sends a
• content-type: zip MIME type
• fix bug peeking at UTF-8 files with many extended characters
• fix minor stability issues (e.g. unclosed file resources)
• fix CSV delimiter detection to ensure "," is always tried first

libhxl-python v4.17

• added option to flatten multiple JSONPath results, separated by " | "

libhxl-python v4.16

• significant improvements to Excel-file handling (even when the server sends a zipfile MIME type)
• fix bug peeking at UTF-8 files with extended characters
• fix CSV delimiter detection to ensure "," is always tried first
• allow forcing a character encoding in hxl.io.data()
• fix minor stability issues