#10 hxltmcli: better tolerance to the soon to be deprecated format

EticaAI · Nov 28, 2021 · 4a0d186 · 4a0d186
1 parent 9ad50d1
commit 4a0d186
Show file tree

Hide file tree

Showing 9 changed files with 80 additions and 23 deletions.
diff --git a/bin/hxltmcli.py b/bin/hxltmcli.py
@@ -180,6 +180,11 @@
 
 [eng-Latn]_
 """
+
+# TODO: since this commit an error with syntax is breaking using on python 3.7
+#       https://github.com/EticaAI/hxltm/runs/4343555339?check_suite_focus=true
+#       This needs better check later if still persists or is false positive
+
 import sys
 import os
 import logging
@@ -248,7 +253,7 @@
 from liquid.context import Context as LiquidContext
 # from liquid.token import TOKEN_EXPRESSION as LIQUID_TOKEN_EXPRESSION
 
-__VERSION__ = "v0.8.9"
+__VERSION__ = "v0.9.0-rc1"
 
 # _[eng-Latn]
 # Note: If you are doing a fork and making it public, please customize
@@ -271,6 +276,7 @@
 
 __ATTRIBUTUM_OPTIONEM__ = {
     # No annotationem (text notes) at concept level: please use meta_conceptum
+    'accuratum': ['#item+terminum+__linguam__+accuratum'],
     'annotationem': [
         '#meta+linguam+__linguam__+annotationem',
         '#meta+terminum+__linguam__+annotationem',
@@ -348,11 +354,7 @@
 
 
 class HXLTMCLI:  # pylint: disable=too-many-instance-attributes
-    """
-    _[eng-Latn] hxltmcli is an working draft of a tool to
-                convert prototype of translation memory stored with HXL to
-                XLIFF v2.1
-    [eng-Latn]_
+    """HXLTMCLI
     """
 
     def __init__(self):
@@ -454,9 +456,6 @@ def _initiale_hxltm_asa(self, archivum: str) -> bool:
             bool: If okay.
         """
 
-        # with open(archivum, 'r') as arch:
-        #     hxltm_crudum = arch.read().splitlines()
-
         self.hxltm_asa = HXLTMASA(
             archivum,
             ontologia=self._ontologia,
@@ -1172,6 +1171,7 @@ def in_noop(self, hxlated_input, tab_output, is_stdout):
 
         Requires that the input must be a valid HXLated file
         """
+        # pylint: disable=no-self-use
 
         with open(hxlated_input, 'r') as csv_file:
             csv_reader = csv.reader(csv_file)
@@ -1632,7 +1632,7 @@ class HXLTMArgumentum:  # pylint: disable=too-many-instance-attributes
     tmeta: InitVar[dict] = None
     agendum_linguam: InitVar[List[Type['HXLTMLinguam']]] = []
     auxilium_linguam: InitVar[List[Type['HXLTMLinguam']]] = []
-    agendum_attributum: InitVar[List[Type[str]]] = []
+    agendum_attributum: InitVar[List[str]] = []
     fontem_linguam: InitVar[Type['HXLTMLinguam']] = None
     objectivum_linguam: InitVar[Type['HXLTMLinguam']] = None
     objectivum_formatum: InitVar[str] = 'HXLTM'
@@ -1651,7 +1651,6 @@ class HXLTMArgumentum:  # pylint: disable=too-many-instance-attributes
     versionem: InitVar[bool] = False
     # crudum_argparse: InitVar[Dict] = {}
 
-    # def de_argparse(self, args_rem: Type['ArgumentParser']):
     def de_argparse(self, args_rem: Dict = None):
         """Argūmentum de Python argparse
 
@@ -1662,7 +1661,8 @@ def de_argparse(self, args_rem: Dict = None):
         Returns:
             [HXLTMArgumentum]: Ego HXLTMArgumentum
         """
-        # print(args_rem)
+        # pylint: disable=too-many-branches,too-many-statements
+
         if args_rem is not None:
             if hasattr(args_rem, 'outfile'):
                 self.objectivum_archivum_nomen = args_rem.outfile
@@ -1956,7 +1956,7 @@ def est_venandum_insectum(self, rem: bool):
 
         return self
 
-    def v(self, _verbosum: bool = None):  # pylint: disable=invalid-name
+    def v(self, _verbosum: bool = None):
         """Ego python Dict
 
         Trivia:
@@ -1969,6 +1969,8 @@ def v(self, _verbosum: bool = None):  # pylint: disable=invalid-name
         Returns:
             [Dict]: Python objectīvum
         """
+        # pylint: disable=invalid-name,too-many-branches
+
         # TODO: add a commom helper of this for all other .v()
         # TODO: make it at least one level more deep (or recursive)
 
@@ -2066,6 +2068,7 @@ class HXLTMDatum:
 
 #>>> crudum_datum
     """
+    # pylint: disable=too-many-instance-attributes
 
     # crudum: InitVar[List] = []
     crudum_caput: InitVar[List] = []
@@ -4982,6 +4985,9 @@ class HXLTMOntologia:
 
     """
 
+    # dēprecātum, https://en.wiktionary.org/wiki/deprecatus#Latin
+    _deprecatum: Type[set] = set()
+
     def __init__(self, ontologia: Dict, vacuum: bool = False):
         """
         _[eng-Latn] Constructs all the necessary attributes for the
@@ -5421,6 +5427,15 @@ def quod_nomen_breve_de_hxl(self, hxl_hashtag: str) -> str:
             nomen_breve = 'conceptum_typum'
 
         elif hxl_hashtag.startswith('#status+rem+accuratum+i_'):
+            # Deprecated
+            self._deprecatum.add(
+                'deprecatum [{0}]: #item+terminum+__linguam__+accuratum'.format(
+                    hxl_hashtag)
+            )
+            nomen_breve = 'accuratum__L__'
+
+        elif hxl_hashtag.startswith('#item+terminum+i_') and \
+                hxl_hashtag.endswith('+accuratum'):
             nomen_breve = 'accuratum__L__'
 
         elif hxl_hashtag.startswith('#status+rem+textum+i_'):
@@ -5430,6 +5445,12 @@ def quod_nomen_breve_de_hxl(self, hxl_hashtag: str) -> str:
             nomen_breve = 'statum_rem_json__L__'
 
         elif hxl_hashtag.startswith('#item+rem+i_'):
+            # Deprecated
+            self._deprecatum.add('deprecatum [{0}]'.format(hxl_hashtag))
+            nomen_breve = 'rem__L__'
+
+        elif hxl_hashtag.startswith('#item+terminum+i_') and \
+                hxl_hashtag.endswith('+rem'):
             nomen_breve = 'rem__L__'
 
         return nomen_breve
@@ -5691,6 +5712,12 @@ def initialle(self, strictum: bool):  # pylint: disable=too-many-branches
             if len(parts[0]) == 2:
                 self.iso6391a2 = parts[0].lower()
 
+        # Improved message over
+        #     ValueError: not enough values to unpack (expected 2, got 1)
+        if self.linguam.find('-') == -1:
+            raise ValueError(
+                'HXLTMLinguam zzz-Zzzz? [{0}]'.format(str(self.linguam)))
+
         self.iso6393, self.iso115924 = \
             list(self.linguam.split('-'))
 

diff --git a/bin/hxltmdexml.py b/bin/hxltmdexml.py
@@ -152,6 +152,10 @@
 [eng-Latn]_
 """
 
+# TODO: since this commit an error with syntax is breaking using on python 3.7
+#       https://github.com/EticaAI/hxltm/runs/4343555339?check_suite_focus=true
+#       This needs better check later if still persists or is false positive
+
 # import xml.etree.ElementTree as et
 import sys
 import os
@@ -191,7 +195,7 @@
 # pip3 install langcodes
 # import langcodes
 
-__VERSION__ = "v0.8.9"
+__VERSION__ = "v0.9.0-rc1"
 
 # _[eng-Latn]
 # Note: If you are doing a fork and making it public, please customize
@@ -235,6 +239,7 @@
     'meta_conceptum': ['#meta+conceptum'],
     'meta_linguam': ['#meta+linguam+__linguam__'],
     'meta_terminum': ['#meta+terminum+__linguam__'],
+    # __nomen_breve: 'rem__L__'
     'rem': ['#item+terminum+__linguam__+rem'],
 }
 __ATTRIBUTUM_DEFALLO__ = [
@@ -3252,6 +3257,7 @@ def __init__(
         ontologia: Type['HXLTMOntologia'],
         agendum_linguam: Type[List['HXLTMLinguam']] = [],
         agendum_attributum: Type[List[str]] = [],
+        # agendum_attributum: Type[List['str']] = [],
         fontem_linguam: Type['HXLTMLinguam'] = None,
         objectivum_linguam: Type['HXLTMLinguam'] = None
     ):

diff --git a/docs/eng-Latn/hxltm.adoc b/docs/eng-Latn/hxltm.adoc
@@ -56,6 +56,9 @@ include::../testum/hxltm-salve-mundi.hxltm.xml[]
 
 ////
 hxltmdexml --agendum-linguam lat-Latn,arb-Arab testum/hxltm-salve-mundi.hxltm.xml > testum/resultatum/hxltm-salve-mundi.tm.hxl.csv
+
+hxltmcli --agendum-linguam eng-Latn@en,por-Latn@pt hxl-non-hxltm-exemplum.hxl.csv
+
 ////
 [source,bash]
 ----
@@ -68,6 +71,9 @@ hxltmdexml --agendum-linguam lat-Latn,arb-Arab testum/hxltm-salve-mundi.hxltm.xm
 include::../testum/resultatum/hxltm-salve-mundi.tm.hxl.csv[]
 ----
 
+> TODO: make it work with new format
+>       `hxltmcli hxltm-exemplum-glossarium-minimum.tm.hxl.csv --objectivum-TMX`
+
 ////
 == Drafts
 

diff --git a/ontologia/cor.hxltm.215.yml b/ontologia/cor.hxltm.215.yml
@@ -1711,7 +1711,9 @@ ontologia:
       # Trivia: linguam, https://en.wiktionary.org/wiki/lingua#Latin
       __linguam__:
          # Exemplum: '#item +rem +i_en +i_eng +is_Latn'
+         # TODO: invert __HXL with __HXL_deprecatum
         __HXL: '#item +rem __linguam__'
+        __HXL_deprecatum: ['#item +terminum +__linguam__ +rem']
         __nomen_breve: 'rem__L__'  # rem.rem__L__
         __id: ontologia.commune.rem.__linguam__
         __libellam: rem

diff --git a/testum/hxl-non-hxltm-exemplum.hxl.csv b/testum/hxl-non-hxltm-exemplum.hxl.csv
@@ -0,0 +1,7 @@
+#vocab+id,#vocab+term+i_en,#vocab+term+i_pt
+1,UN,ONU
+2,United Nations,Organização das Nações Unidas
+3,IFRC,FICV/CV
+4,International Federation of Red Cross and Red Crescent Societies ,Federação Internacional das Sociedades da Cruz Vermelha e do Crescente Vermelho
+5,AI,AI
+6,Amnesty International,Amnistia Internacional
diff --git a/testum/hxltm-exemplum-glossarium-minimum.tm.hxl.csv b/testum/hxltm-exemplum-glossarium-minimum.tm.hxl.csv
@@ -1,7 +1,7 @@
-#item +rem +i_en +i_eng +is_Latn,#item +rem +i_pt +i_por +is_Latn
-UN,ONU
-United Nations,Organização das Nações Unidas
-IFRC,FICV/CV
-International Federation of Red Cross and Red Crescent Societies ,Federação Internacional das Sociedades da Cruz Vermelha e do Crescente Vermelho
-AI,AI
-Amnesty International,Amnistia Internacional
+#item+conceptum+codicem,#item+terminum+i_en+i_eng+is_Latn+rem,#item+terminum+i_pt+i_por+is_Latn+rem
+1,UN,ONU
+2,United Nations,Organização das Nações Unidas
+3,IFRC,FICV/CV
+4,International Federation of Red Cross and Red Crescent Societies ,Federação Internacional das Sociedades da Cruz Vermelha e do Crescente Vermelho
+5,AI,AI
+6,Amnesty International,Amnistia Internacional
diff --git a/testum/hxltmcli--help_eng-Latn.txt b/testum/hxltmcli--help_eng-Latn.txt
@@ -35,7 +35,7 @@ usage: hxltmcli [-h] [--sheet [number]] [--selector [path]]
                 [--venandum-insectum-est] [--versionem]
                 [infile] [outfile]
 
-_[eng-Latn] hxltmcli v0.8.9 is an example of implementation of the
+_[eng-Latn] hxltmcli v0.9.0-rc1 is an example of implementation of the
 Multilingual Terminology in Humanitarian Language Exchange (HXLTM).
 For XML processing, use hxltmdexml.
 [eng-Latn]_"

diff --git a/testum/hxltmdexml--help_eng-Latn.txt b/testum/hxltmdexml--help_eng-Latn.txt
@@ -7,7 +7,7 @@ usage: hxltmdexml [-h] [--agendum-linguam [agendum_linguam]]
                   [infile] [outfile]
 
 _[eng-Latn]
-hxltmdexml v0.8.9 is an example of implementation of the
+hxltmdexml v0.9.0-rc1 is an example of implementation of the
 Multilingual Terminology in Humanitarian Language Exchange (HXLTM)
 focused on convert XML files to the pivot format HXLTM.
 [eng-Latn]_"

diff --git a/tox.ini b/tox.ini
@@ -30,6 +30,15 @@
 # conda, etc, etc etc
 # @see https://github.com/tox-dev
 
+#### lint
+# pylint --disable=W0511,C0302 bin/hxltmcli.py
+# pylint --disable=W0511,C0302 bin/hxltmdexml.py
+# pylint --disable=W0511,C0302 bin/
+#
+# See also https://pypi.org/project/autopep8/
+# autopep8 --list-fixes bin/hxltmcli.py
+# autopep8 --diff bin/hxltmcli.py
+
 # ___________________________________________________________________________ #
 # Here is where thing start
 [tox]