CHANGELOG.md

A changelog for the lib/hyperglot language database and CLI tool (dd.mm.yyyy)

0.8.1 (11.03.2026):

• DATA: "unknown" allowed for Language speaker counts where cannonically there is no data (note: by contrast, omitting the speaker attribute communicates the data is incomplete in that regard)
• DATA: No commas in language names/preferred_names
• DATA: Nepali and Zaza added as macrolanguages
• DATA: Removed stray Latin chars in Japanese Hiragana orthography
• TWEAK: Upgraded to pyproject.toml config
• TWEAK: Checker._get_checks_for_orthography redefined as class method, to allow for easier inspecting of what checks are opted in an orthgraphy
• TWEAK: Check now have precheck method to initialize and indicate pass early on

0.8.0 (17.02.2026)

• LICENSE: Relicensed under Apache License 2.0
• FIX: Shaper.check_joining refined to not be more lenient and not fail fonts with other than one-to-one positional substitutions or general sequence manipulating ccmp code
• FIX Using a --status that does not include 'living' no correctly omits these languages
• TWEAK: Better output with -v: Consequently log skipped languages and orthographies plus the skipping the reason
• FEATURE: Added -t/--shaping-threshold that allows fine-tuning conjunct check failures by accounting for conjunct frequency
• FEATURE: Added --no-shaping flag to disable shaping checks entirely (on by default)
• DATA: Modified multitudes of design_requirements, dropped design_alternates
• DATA: jpn Latin orthography marked as secondary
• DATA: Introduced combinations orthography attribute
• DATA: hin and mai include syllable combinations with frequencies distributions (1: most common, 0: least common)
• TWEAK: design_requirments can now be either a string, or a dict of note + alternates (detailing which characters are affected)
• TWEAK: Minor tweak to logging in Orthography
• TWEAK: Checker._check_shaping with better pre-check to skip mark attachment checks for glyphs not in the font (obvious)
• TWEAK: Cleaned up multiple CLI options:
  • Added --check option to replace --supportlevel. --check takes any value of base,auxiliary,punctuation,numerals,currency,all or a comma-separated list of those
  • Removed --include-historical and --include-historical in favor or --status which accepts any combination of LanguageStatus or "all", defaults to "living"
  • Removed --include-all-orthographies in favor of --orthography which accepts any combination of OrthographyStatus or "all", defaults to "primary"
• TWEAK: Better logging output and logging strategies in the CLI/modules, -v provides basic language in/out and config logs, -vv gives very detailed support logs
• TWEAK: Removed the deprecated STATUSES, ORTHOGRAPHY_STATUSES and SUPPORTLEVELS from the codebase
• TWEAK: Dropped python 3.8 & 3.9 from supported environments, added 3.13, 3.14

0.7.3 (13.10.2025)

• DATA: Fixed to cbi (thanks @moyogo)
• TWEAK: Fixed hyperglot-export command for dumping expanded database

0.7.2 (15.01.2025)

• DATA: Minor refinements to fin, ces and nav
• DATA: Design requirements updated for bos and srb as well as some Cyrillic breve mentiones
• TWEAK: Improved inherited type where the original value is a yaml list
• TWEAK: Added parameter to instantiate a Language and force a reload of the data / ignoring the .hyperglot-cache

0.7.1 (07.10.2024)

• FIX: Fixed hyperglot-data error

0.7.0 (07.10.2024)

• DATA: All language yaml documents now have their contributors listed, some have reviewers listed
• DATA: *Massive improvement of language sources with proper source citations where possible
• DATA: Added punctuation, numerals and currency attributes to orthographies - checking for these attributes will be added in the next update!
• DATA: Added lib/hyperglot/extra_data/default.yaml to include inheritable defaults per script
• DATA: Refined jpn, ryu and ain Katakana orthographies
• FEATURE: Orthography attributes can inherit from other languages with <iso> syntax, see README
• TWEAK: Improved loading time for repeat access by saving parsed language data cache file
• TWEAK: Orthographies can no longer have an inherit attribute
• TWEAK: Improved loading speed for repeat queries and indivudal language queries
• TWEAK: Refactored Languages, Language and Orthography object instantiation to always return parsed and defaulted nested objects
• TWEAK: Removed the --speakers and --autonym CLI options
• TWEAK: Removed the --comparison CLI option (see examples instead)
• TWEAK: Removed the --languages CLI option, use hyperglot-info LanguageName/ISO instead
• TWEAK: Removed the --strict_iso CLI option; use the python library to access this option, particularly Language.get_name(script, strict_iso=True)

0.6.4 (20.06.2024)

• FIX: Fixed an issue where trying to log missing shaping glyphs would crash in FontChecker
• FIX: Improved mark shaping detection to interpret ccmp substitutions of base + mark as correctly shaping (thanks @arialcrime)
• TWEAK: Cleaned up hyperglot.language.Language class and added attribute properties for dict properties with computed defaults (as opposed to writing defaults for missing attributes) as well as more code annotation
• TWEAK: hyperglot.orthography.Orthography object has script_iso attribute returning the mapped ISO 15924 script tag
• DATA: Added lib/extra_data/script-names.yaml with a list of all current Hyperglot scripts and a mapping to their ISO 15924 code equivalent
• DATA: Added di/tri-graphs to Czech and Hungarian orthographies and fixed their order
• DATA: Added Squamish (squ) (thanks @justinpenner)
• DATA: Unified "Geʽez" script with reversed comma, as opposed to previous mixed use of "Ge'ez/Fidel" and "Ge'ez"
• DATA: Amended spelling "Tai Viet" script in title case to match other script names
• DATA: Corrected spelling of "Bamum" script and language (instead of less used "Bamun" used in Hyperglot)
• DATA: Use "Coptic" instead of "Coptic/Numbian" script name
• DATA: Use "Burmese" script for language "Mon"
• DATA: Use "Baybayin" script name instead of "Tagalog (Baybayin, Alibata)"
• DATA: Fixed Toki Pona (tok) file name
• TWEAK: Make sure Orthography.base_chars and Orthography.aux_chars return no duplicates for decomposed character sequences
• TWEAK: Define Languages, Language and Orthography as module top level exports for easier importing, e.g. now: from hyperglot import Language

0.6.3 (08.04.2024)

• FIX: Set correct default values for Language.status and Orthography.preferred_as_group and provide validation and tests for these.
• TWEAK: Deprecated plain list SUPPORTLEVELS, VALIDITYLEVELS, STATUSES, ORTHOGRAPHY_STATUSES and replaced them with SupportLevel, LanguageValidity, LanguageStatus, OrthographyStatus enums throughout the code base. The deprecated values will be removed in the next minor version.
• TESTS: Added simple tox config for running test on all supported minor python versions

0.6.2 (22.3.2024)

• FIX: Fixed type hinting issue causing failure on python 3.8.x
• DATA: Added Banjar (bjn) (thanks @mahalisyarifuddin)
• DATA: Expanded Xavánte (xav) data (thanks @moyogo)

0.6.1 (30.1.2024)

• DATA: Refined Romanian by adding design_alternates explicitly

0.6.0 (16.01.2024)

• DATA: Refined Klingon (tlh) orthography and added a draft version of Toki Pona (tok)
• FEATURE: Implemented shaping checks for mark positioning when required by unencoded base + mark combinations or --decompose
• FEATURE: Implemented shaping checks for connecting scripts to detect presence of required positional forms
• FEATURE: Implemented hyperglot-report command with same options as hyperglot and additional --report-missing n, --report-marks n and --report-joining n — or --report-all n to toggle all aforementioned — parameters/flags for outputting languages almost supported by the font
• TWEAK: Support checking is now done via hyperglot.checker objects for cleaner separation between language data and checking fonts
• TWEAK: Various python APIs and objects changed and refactored
• TWEAK: Bumped required python version to 3.8.0

0.5.3 (13.12.2023)

• DATA : Added Tlingit tli language data (thanks @jcrippen)
• DATA: Fixed inconsistent note about Ŋ in various languages (thanks @moyogo)
• TWEAK: Improved hyperglot-validate to spot lookalike characters in the wrong script, e.g. a (Latin U+0061) vs а (Cyrillic U+0430)
• TWEAK: Explicitly ignore non-yaml files (e.g. operating system or other) in the data when parsing
• TWEAK: Improved hyperglot-validate command to better catch yaml issues (thanks for reporting @jcrippen)

0.5.2 (23.11.2023)

• DATA: Removed orthography status deprecated and using historical for those instances
• DATA: Added Ethiopic languages awn, byn, gez, har, sgw, tig, xan and updated tir (thanks @dyacob and @NeilSureshPatel)
• DATA: Added Avestan
• DATA: Corrections to jbo (thanks @berrymot)
• DATA: Updated sco primary orthography (thanks @moyogo)
• DATA: Some fixes to kkj orthography (thanks @moyogo)
• DATA: Small note added to Dagbani (thanks @clauseggers and @moyogo)
• DATA: Fix to Shan (shn) containing some stray Latin characters
• FIX: Fix issue with file name conflicts on Windows systems

0.5.1 (21.06.2023)

• FIX: Fix pypi missing data files

0.5.0 (21.06.2023)

• FEATURE: Added -l/--language flag to show supported/not supported glyphs of a font for specific languages
• DATA: Restructured hyperglot.yaml into individual files for each language in hyperglot/data/xxx/xxx.yaml
• DATA: Fix two auxiliary glyphs in Georgian which where swapped uppercase / lowercase by mistake
• DATA: Small charset fixes to Kom bkm and Southern Samo sbd (thanks @moyogo)
• DATA: Small tweak to Afrikaans afr (thanks @iandoug)

0.4.5 (28.3.2023)

• DATA: Added languages and scripts for: Ainu, Akkadian, Ancient Egyptian, Mycenaean Greek, Linear A, Linear B, Minoan, Pontic Greek, Okinawan, Sumerian, Klingon, Minaen, Hadramautic, Qatabanian and Sabaean (big thanks to @gusbemacbe !)
• DATA: Added Kayah autonym
• DATA: Added design requirement note for Ŋ
• DATA: Improved Georgian, added Mtavruli and auxiliary
• DATA: Added historical orthographies for German and English that use ſ

0.4.4 (10.12.2022)

• FIX: Fixed orthography of Thai to not require ◌̍ ◌̎ in base checks

0.4.3 (10.12.2022)

• FIX: Fixed missing script attribute in 'lee' orthography
• FIX: Fixed typo in 'Oriya' script name

0.4.2 (25.11.2022)

• FEATURE: Implemented hyperglot-data CLI command to search and display language information returned by Hyperglot
• FEATURE: Implemented more convenient language access via attributes on hyperglot.languages.Languages, e.g. Languages().eng to access a hyperglot.language.Language object for "eng"
• DATA: Fix in Standard Malay encoding of ' (thanks M. Mahali Syarifuddin and Caleb Maclennan)
• DATA: Added numerous Burkina Faso and other African languages (another huge thanks to @moyogo !)
• DATA: Added Oriya
• DATA: Added Kartvelian languages (kat, sva, xmf, lzz) (thanks Ana)

0.4.1 (18.08.2022)

• DATA: Dozens of African and North-American languages added and refined (thanks @moyogo !)
• DATA: Refined English auxiliary

0.4.0 (29.06.2022)

• DATA: Fix for Pinyin
• CLI: Introduced --sort (alphabetic, default, or speakers) and --sort-dir (asc, default, or desc)

0.3.9 (21.06.2022)

• DATA: Fix for Skolt Sami (soft sign)
• DATA: Fix for Hawaiian (okina)
• DATA: Fix for Thai including several missing marks and letters
• DATA: Fix in Buginese
• DATA: Updates to Indonesian and Standard Malay

0.3.8 (01.03.2022)

• DATA: Fix for Turkish orthography

0.3.7 (04.01.2022)

• DATA: Fix for Afrikaans orthography
• DATA: Corrected ISO code for Gen language

0.3.6 (11.11.2021)

• DATA: Added Benin languages
• DATA: Small fix to Portuguese

0.3.5 (07.09.2021)

• DATA: Revised Tamil orthography
• DATA: Added Apinayé, Karo and Awetí languages

0.3.4 (06.09.2021)

• FIX: Fixed an encoding issue affecting Windows environments
• DATA: Fixed typos in Buginese
• DATA: Reviewed Minangkabau orthography

0.3.3 (23.04.2021)

• DATA: Added Batak languages and refined Balinese
• FIX: Further improvement to detection of orthographies with unencoded base + mark combinations
• TWEAK: Refined the returned properties of hyperglot.language.Orthography to include base and auxiliary lists of encoded characters as well as required marks for
• TOOLS: Added scraper for fetching a mapping of Opentype language systems to ISO codes and saving them in other/languagesystems.yaml

0.3.2 (13.04.2021)

• DATA: Renamed design_note to design_requirements and made its data structure a list
• DATA: Introduced design_alternates - a list of characters which may require special design in a font supporting an orthography
• DATA: Added design_alternates for several Cyrillic and Latin languages

0.3.1 (12.04.2021)

• DATA: Corrected speaker count for Manipuri
• DATA: Updates to Andaandi and Old Nubian
• DATA: Minor formatting and duplicate fixes
• FIX: Fixed parsing issue that led for some languages to require marks in their support as if the --marks flag was used
• TWEAK: hyperglot.language.Language no longer prunes or parses any character lists, but this is instead done on running the support checks by instantiating a Orthography object and using it for checking, leaving the dict representation of the yaml data in the Language untouched
• FEATURE: Introduced hyperglot.language.Orthography abstraction for easier access of parse lists vs yaml raw character strings
• TESTS: More refactored Languages, Language and new Orthography tests

0.3.0 (09.04.2021)

• DATA: Changed the way marks and decomposition are handled in the data entry and saving
• DATA: base and auxiliary may now contain unencoded base + mark character combinations without those getting decomposed on saving
• DATA: Updated approximately 50-100 languages which previously had unencoded base + mark combinations not saved in their character sets, since those were not unicode characters - this update added and retains those unencoded combinations for more comprehensive listing of the orthographies
• DATA: Marks are now always placed on ◌ in the data for easier readability
• CLI: Default checking (without -m) no longer requires implicit combining marks, meaning those which are retrieved from decomposing the characters - the default check will still require those marks, which are explicitly listed in marks and are not the result of decomposing the characters
• CLI: Introduced -m/--marks as a flag to require all marks for a support level check
• CLI: Changed -m/--mode to -c/--comparison
• TWEAK: Removed hyperglot.parse.prune_superflous_marks as no longer needed
• TWEAK: Introduced hyperglot.parse.parse_marks
• TWEAK: Removed prune and pruneRetainDecomposed flags from Languages() and changed default call to Languages() to no longer prune or parse its dict contents
• TWEAK: Only calls to Language() now parse the orthography data (with default True for argument parse)
• TWEAK: Renamed methods hyperglot.languages.get_support_from_chars to supported and hyperglot.languages.has_support to supported
• TWEAK: Added warnings and validation checks for multiple inheritance levels (e.g. A inherits from B inherits from C should instead be A inherits from C)

0.2.12 (06.04.2021)

• Data: Updated Ter Sami orthography as inheriting from Kildin Sami
• Data: Fixes to Kildin Sami
• Data: Some fixes to Marshallese
• Data: Added Ottoman Turkish and a transliteration orthography for it
• Data: Added Hanunoo
• Data: Replaced Single right comma (and other variants) with Modifier letter apostrophe for some Sami languages
• Data:
• FIX: Fixed issue that caused to parse some fonts (#24)
• TWEAK: Allow inheriting an orthography without explicitly having a script present in the orthography, this will inherit the primary script orthography of the parent

0.2.11 (29.03.2021)

• DATA: Updated language data for Nubian languages and Japanese
• DATA: Introduced transliteration orthography status (started in 0.2.10)

0.2.10 (26.03.2021)

• DATA: Updated language data for Minang (xrg), Tamil (tam), Cherokee (chr), Tagalog (tgl), Aja (ajg), Khmer (khm), Madurese (mad), Javanese (jav) and others
• FIX: Reverted hotfix from 0.2.9 and implemented validation to use iso yaml file only for editable package installs and emit warning
• FIX: Refined --decompose and fixed an issue where the decompose option ended up returning more stringent matches than teh default
• FIX: --output output refactored to no longer expect the result to be structured by support levels
• TWEAK: Refactored multiple file input result intersection and union
• TESTS: Better tests relating to deomposed output
• TESTS: Added tests for multiple file input intersection and union results

0.2.9 (24.03.2021)

• HOTFIX: Prevent error message about missing file in CLI use

0.2.8 (24.03.2021)

• FIX: Fixed inheritence when it chains, e.g. Algerian Arabic inheriting from Tunisian Arabic which inherits from Standard Arabic
• FIX: Fixed inheritence missing marks, design_notes and note
• TWEAK: Make sure marks are saved in ordered form, so saving does not arbitrarily alter the order
• TESTS: Added tests for orthography inheritance
• DATA: Constrained speaker counts to integers only

0.2.7 (24.03.2021)

• DATA: Fixed various speaker counts containing malformed data
• DATA: More design notes for Latin-script languages
• DATA: Khmer added as draft, Armenian, Buginese, Georgian, Burmese, Lao and Thai refined
• TWEAK: Implemented validation for speaker count data

0.2.6 (23.03.2021)

• DATA: Various status updates, notes and reviewed orthographies
• DATA: Introduced marks attribute containing all combinging marks needed for an orthography
• FEATURE: Automatically extract and save marks from base data, plus retain any explicitly added marks in the data
• TWEAK: For default hyperglot-save calls automatically run validation to flag any remaining issues
• TWEAK: Flag legacy marks being used in charset data

0.2.5 (23.03.2021)

• DATA: Introduced design_note parameter
• DATA: Various language data updates and smaller fixes
• DATA: Several orthography fixes, thanks Denis Moyogo Jacquerye
• TWEAK: Changed orthography status names to todo, draft, preliminary, verified
• TWEAK: Improved Language.get_orthography to return better default picks and allow getting orthographies of specific script or status

0.2.4 (18.03.2021)

• First pip release :)

0.2.3 (10.03.2021)

• FEATURE: Implemented --include-all-orthographies to check all but deprecated orthographies and changed default behaviour to only list primary orthographies
• TWEAK: Implemented treating orthographies with preferred_as_group as one for checks
• TWEAK: Languages with multiple primary orthographies will match if one is supported
• TWEAK: Languages can be initiated with pruneRetainDecomposed to keep any precomposed characters from the database when using prune (which decomposes them to base + mark)
• TWEAK: Improved tests for CLI and improved and fixed some parsing tests
• FIX: Marginal cases fixed where using parse_chars and already parsed lists would merge a mark with a predeceding base glpyh and result in a erraneous list of base/aux characters
• DATA: Added uppercase to bicameral scripts
• DATA: All languages now have a primary orthography
• DATA: Introduced preferred_as_group orthography attribute
• TESTS: Config to ignore other library's warnings

0.2.2 (02.03.2021)

• TWEAK: Languages() now takes a validity argument to filter by validity ('weak' or better by default)
• TWEAK: parse_chars now will put decomposition components on in the input list to the end of the list
• TWEAK: Languages require an orthography that has status primary

0.2.1 (03.02.2021)

• DATA: Updated and added many scripts and languages and their speaker counts

0.2.0 (28.01.2021)

• FEATURE: Added --decomposed flag that determines if a font is required to have all glyphs of a language as code points, or if supporting all combining marks is sufficient
• TWEAK: Renamed module and database to hyperglot
• TWEAK: --strict-support refactored to --validity with default weak to pick the level of required validity on the languages that should get matched
• TWEAK: Saving and validating enforces removal of superflous mark characters that are getting implicitly extracted via glyph decomposition
• TWEAK: Detection automatically extracts all required mark glyphs for languages and the database has been pruned of any no longer required mark glyphs listed. Using the hyperglot-save will apply this pruning and save the database in its cleaned up state
• TESTS: Added tests for the Language and Languages class
• TESTS: Added test for the CLI options running against actual font files
• DOCS: Overhauled and updated the README to all latest changes

0.1.14 (27.5.2020)

• FIX: Refined character parsing to also include the encoded form of any decomposable glyphs

0.1.13 (15.5.2020)

• FIX: Improved character set parsing from database properly decomposing any combining characters into their parts and checking against those
• TESTS: Added first pytest for above case

0.1.12 (14.5.2020)

• FEATURE: Added --strict-support flag (default False) to explicitly trigger warning about languages with unconfirmed status. Since those languages have well researched charset information but just have not been confirmed by several expert sources we still want to include them in the count. Using --strict-support excludes (but lists separately) all those languages which we have not been able to confirm
• TWEAK: Renamed --strict flag to --strict-iso to be more discriptive
• TWEAK: Database file linking, one more time... as per 0.1.10
• TWEAK: Added validation check to prevent non-space separators in character list data

0.1.11 (10.4.2020)

• FEATURE: Implemented fontlang-export CLI script to export the rosetta.yaml with expanded inherits to a file, usage: $ fontlang-export thefile.yaml

0.1.10 (17.3.2020)

• FIX: Refactored setup.py to include the databased file relative to the package

0.1.9 (19.2.2020)

• FIX: "Inverted" the preferred_as_individual outcome, e.g. those languages should suppress any included languages from being listed and be listed as one language instead

0.1.7 & 0.1.8 (18.02.2020)

• FIX: Made sure preferred_as_individual in fact also removes the language that is being inheritted from the matches
• TWEAK: Update fontlang-save and sorting to not include inheritted attributes
• TWEAK: Updated and fixed validation for status attributes

0.1.6 (07.02.2020)

• FEATURE: Implemented support for the preferred_as_individual attribute on macro languages
• FEATURE: Added --strict flag to display language names and macrolanguages as per ISO data
• TWEAK: Implemented orthography attribute inherit to inherit another language's orthography for that script (if one exists)
• FIX: Language names with countries in brackets no longer have their closing parenthesis cut off
• TWEAK: Updated fontlang-validate to spec

0.1.5 (30.01.2020)

• FIX: More robust relative file path loading for database file
• TWEAK: -o output is now of same structure for single file input, and indexed by file name for several file input
• TWEAK: -o filters the languages' orthographies to only supported ones
• TWEAK: Added validation check to confirm orthographies have a 'script'
• TWEAK: Refactored validation script to fontlang-validate CLI command
• TWEAK: Languages without orthographies that are included in macrolanguages that do have orthographies silently inherit the macrolanguage's orthographies
• FEATURE: Added fontlang-save CLI command to re-save the rosetta.yaml sorted alphabetically
• FEATURE: Added --include-historical and --include-constructed flags to include those languages in results
• FEATURE: Added --version and --verbose flags

0.1.4 (28.01.2020)

• FEATURE: Added -m option ('individual', 'union', 'intersection') to compute a support comparison of several passed in fonts
• TWEAK: You can now pass in any number of font paths. By default one after the other is analyzed
• TWEAK: Make sure to print preferred_name if available

0.1.3 (23.01.2020)

• TWEAK: Merged fontlang with Rosetta Language DB repo
• TWEAK: Updated data structure in YAML and added Language class for convenience

0.1.2

• FEATURE: -o flag to specify an output yaml file path
• FEATURE: -n flag to display language names in native spelling (where available)
• FEATURE: -u flag to display language users (if available)
• TWEAK: Updated rosetta.yaml language database

0.1.1

• FEATURE: -s flag with "base" or "aux" values to set support level to check
• TWEAK: Support output sorted by scripts and language DB status (informs about not "done" langs being match)
• TWEAK: Added basic font validation for the passed in file path
• TWEAK: Fixed relative imports and cli usage for dev
• FIX: Language database typo fix

0.1.0

• FEATURE: MVP with basic $ fontlang path/to/font command