getting ready for Glottolog 5.0; closes #75

setup.cfg

@@ -24,6 +24,7 @@ classifiers =
     Programming Language :: Python :: 3.9
     Programming Language :: Python :: 3.10
     Programming Language :: Python :: 3.11
+    Programming Language :: Python :: 3.12
     Programming Language :: Python :: Implementation :: CPython
     Programming Language :: Python :: Implementation :: PyPy
     License :: OSI Approved :: Apache Software License
@@ -123,7 +124,7 @@ max-line-length = 100
 exclude = .tox
 
 [tox:tox]
-envlist = py38, py39, py310, py311
+envlist = py38, py39, py310, py311, py312
 isolated_build = true
 skip_missing_interpreter = true
 

src/pyglottolog/admin_commands/elpubbib.py

@@ -5,4 +5,4 @@
 
 
 def run(args):  # pragma: no cover
-    download(args.repos.bibfiles['elpub.bib'], args.log)
+    download(args.repos.bibfiles['elpub.bib'], args.log, args.repos)

src/pyglottolog/admin_commands/updatelinks.py

@@ -18,6 +18,8 @@ def run(args):
     langs = list(args.repos.languoids())
     updated = set()
     for cls in LinkProvider.__subclasses__():
+        if getattr(cls, '__inactive__', False):
+            continue
         name = cls.__name__.lower()
         if (not getattr(args, 'provider', None)) or (name in args.provider):
             args.log.info('updating {0} links ...'.format(name))

src/pyglottolog/api.py

@@ -4,13 +4,13 @@
 import re
 import typing
 import pathlib
+import functools
 import contextlib
 import collections
 
 import pycldf.util
 from csvw import TableGroup, Column
 from clldutils.path import walk, git_describe
-from clldutils.misc import lazyproperty
 from clldutils.apilib import API
 from clldutils.jsonlib import load
 import clldutils.iso_639_3
@@ -110,56 +110,56 @@ def _cfg(self, name, cls=None):
         return config.Config.from_ini(
             self.path('config', name + '.ini'), object_class=cls or config.Generic)
 
-    @lazyproperty
+    @functools.cached_property
     def aes_status(self) -> typing.Dict[str, config.AES]:
         """
         :rtype: mapping with :class:`config.AES` values.
         """
         return self._cfg('aes_status', cls=config.AES)
 
-    @lazyproperty
+    @functools.cached_property
     def aes_sources(self) -> typing.Dict[str, config.AESSource]:
         """
         :rtype: mapping with :class:`config.AESSource` values
         """
         return self._cfg('aes_sources', cls=config.AESSource)
 
-    @lazyproperty
+    @functools.cached_property
     def document_types(self) -> typing.Dict[str, config.DocumentType]:
         """
         :rtype: mapping with :class:`config.DocumentType` values
         """
         return self._cfg('document_types', cls=config.DocumentType)
 
-    @lazyproperty
+    @functools.cached_property
     def med_types(self) -> typing.Dict[str, config.MEDType]:
         """
         :rtype: mapping with :class:`config.MEDType` values
         """
         return self._cfg('med_types', cls=config.MEDType)
 
-    @lazyproperty
+    @functools.cached_property
     def macroareas(self) -> typing.Dict[str, config.Macroarea]:
         """
         :rtype: mapping with :class:`config.Macroarea` values
         """
         return self._cfg('macroareas', cls=config.Macroarea)
 
-    @lazyproperty
+    @functools.cached_property
     def language_types(self) -> typing.Dict[str, config.LanguageType]:
         """
         :rtype: mapping with :class:`config.LanguageType` values
         """
         return self._cfg('language_types', cls=config.LanguageType)
 
-    @lazyproperty
+    @functools.cached_property
     def languoid_levels(self) -> typing.Dict[str, config.LanguoidLevel]:
         """
         :rtype: mapping with :class:`config.LanguoidLevel` values
         """
         return self._cfg('languoid_levels', cls=config.LanguoidLevel)
 
-    @lazyproperty
+    @functools.cached_property
     def editors(self) -> typing.Dict[str, config.Generic]:
         """
         Metadata about editors of Glottolog
@@ -168,7 +168,7 @@ def editors(self) -> typing.Dict[str, config.Generic]:
         """
         return self._cfg('editors')
 
-    @lazyproperty
+    @functools.cached_property
     def publication(self) -> typing.Dict[str, config.Generic]:
         """
         Metadata about the Glottolog publication
@@ -177,7 +177,7 @@ def publication(self) -> typing.Dict[str, config.Generic]:
         """
         return self._cfg('publication')
 
-    @lazyproperty
+    @functools.cached_property
     def iso(self) -> clldutils.iso_639_3.ISO:
         """
         :return: `clldutils.iso_639_3.ISO` instance, fed with the data of the latest \
@@ -192,7 +192,7 @@ def ftsindex(self) -> pathlib.Path:
         """
         return self.build_path('whoosh')
 
-    @lazyproperty
+    @functools.cached_property
     def _tree_dirs(self):
         return list(walk(self.tree, mode='dirs'))
 
@@ -327,7 +327,7 @@ def newick_tree(
                 trees.append('{0};'.format(ns))
         return '\n'.join(trees)
 
-    @lazyproperty
+    @functools.cached_property
     def bibfiles(self) -> references.BibFiles:
         """
         Access reference data by BibFile.
@@ -355,13 +355,13 @@ def refs_by_languoid(self, *bibfiles, **kw):
                     res[lang.id].append(entry)
         return res, all_
 
-    @lazyproperty
+    @functools.cached_property
     def hhtypes(self):
         # Note: The file `hhtype.ini` does not exist anymore. This is fixed in HHTypes, when
         # calling `config.get_ini`. Only used when compiling monster.bib.
         return references.HHTypes(self.references_path('hhtype.ini'))
 
-    @lazyproperty
+    @functools.cached_property
     def triggers(self):
         res = {'inlg': [], 'lgcode': []}
         for lang in self.languoids():
@@ -372,7 +372,7 @@ def triggers(self):
                                        for text in lang.cfg.getlist('triggers', type_)])
         return res
 
-    @lazyproperty
+    @functools.cached_property
     def macroarea_map(self):
         res = {}
         for lang in self.languoids():

src/pyglottolog/config/__init__.py

@@ -5,6 +5,7 @@
 import attr
 from clldutils.misc import nfilter
 from clldutils.inifile import INI
+from clldutils.jsonlib import load
 
 __all__ = [
     'AES', 'AESSource', 'Macroarea', 'DocumentType', 'LanguageType', 'LanguoidLevel',
@@ -16,15 +17,17 @@ class ConfigObject(object):
     Factory to turn INI file sections into instances of `@attr.s` classes.
     """
     @classmethod
-    def from_section(cls, cfg, section):
+    def from_section(cls, cfg, section, fname):
         try:
             fields = set(f.name for f in attr.fields(cls))
         except attr.exceptions.NotAnAttrsClassError:
             fields = None
 
         kw = {'name' if 'id' in cfg[section] else 'id': section}
         kw.update(cfg[section].items())
-        return cls(**{k: v for k, v in kw.items() if fields is None or k in fields})
+        res = cls(**{k: v for k, v in kw.items() if fields is None or k in fields})
+        res._fname = fname
+        return res
 
 
 class Generic(ConfigObject):
@@ -95,6 +98,12 @@ class Macroarea(ConfigObject):
     #: Glottolog reference ID linking to further information
     reference_id = attr.ib()
 
+    @property
+    def geojson(self):
+        fname = self._fname.parent / 'macroareas' / 'voronoi' / '{}.geojson'.format(
+            self.name.lower().replace(' ', '_'))
+        return load(fname) if fname.exists() else None
+
 
 @attr.s
 class DocumentType(ConfigObject):
@@ -178,7 +187,7 @@ def from_ini(cls, fname, object_class):
         ini = get_ini(fname)
         d = collections.OrderedDict()
         for sec in ini.sections():
-            obj = object_class.from_section(ini, sec)
+            obj = object_class.from_section(ini, sec, fname)
             d[obj.id] = obj
         res = cls(**d)
         res.__defaults__ = ini['DEFAULT']

src/pyglottolog/iso.py

@@ -125,7 +125,9 @@ def iter(cls, max_year=None, cache_dir=None, log=None):
                     print('no crs for {}, page {}'.format(year, page))
                 else:
                     for i, cr in enumerate(tables[0]):
-                        yield cls(**{k.replace(' ', '_'): v for k, v in cr.items()})
+                        d = {k.replace(' ', '_'): v for k, v in cr.items()}
+                        if 'Effective_Date' in d:
+                            yield cls(**{k.replace(' ', '_'): v for k, v in cr.items()})
                 if i < 99:
                     break
                 page += 1  # pragma: no cover

src/pyglottolog/languoids/models.py

@@ -118,8 +118,9 @@ class Reference(object):
     key = attr.ib()  #:
     pages = attr.ib(default=None)  #:
     trigger = attr.ib(default=None)
+    endtag = attr.ib(default='**')
     pattern = re.compile(
-        r"\*\*(?P<key>[a-z0-9\-_]+:[a-zA-Z.?\-;*'/()\[\]!_:0-9\u2014]+?)\*\*"
+        r"\*\*(?P<key>[a-z0-9\-_]+:[a-zA-Z.?\-;*'/()\[\]!_:0-9\u2014]+?)(?P<endtag>\*\*|\(\*\*\))"
         r"(:(?P<pages>[0-9\-f]+))?"
         r'(<trigger "(?P<trigger>[^\"]+)">)?')
     old_pattern = re.compile(r'[^\[]+\[(?P<pages>[^\]]*)\]\s*\([0-9]+\s+(?P<key>[^\)]+)\)')

src/pyglottolog/links/elcat.py

@@ -41,6 +41,7 @@ def format_date() -> str:
 
 
 class ElCat(LinkProvider):
+    __inactive__ = True
     status_map = {}
 
     def __init__(self, repos=None):

src/pyglottolog/links/util.py

@@ -6,7 +6,7 @@
 
 
 def read_grouped_cldf_languages(doi):  # pragma: no cover
-    rec = cldfzenodo.Record.from_doi(doi)
+    rec = cldfzenodo.Record.from_concept_doi(doi)
     with TemporaryDirectory() as tmp:
         ds = rec.download_dataset(tmp)
         langs = sorted(
@@ -77,3 +77,17 @@ class WALS(LinkProvider):  # pragma: no cover
     __doi__ = '10.5281/zenodo.3606197'
     __url_template__ = 'https://' + __domain__ + '/languoid/lect/wals_code_{0[id]}'
     __label_template__ = '{0[name]}'
+
+
+class Grambank(LinkProvider):  # pragma: no cover
+    __domain__ = "grambank.clld.org"
+    __doi__ = '10.5281/zenodo.7740139'
+    __url_template__ = 'https://' + __domain__ + '/languages/{0[id]}'
+    __label_template__ = '{0[name]}'
+
+
+class Lexibank(LinkProvider):  # pragma: no cover
+    __domain__ = "lexibank.clld.org"
+    __doi__ = '10.5281/zenodo.5227817'
+    __url_template__ = 'https://' + __domain__ + '/languages/{0[id]}'
+    __label_template__ = '{0[name]}'