Updated FSTs to correctly have multicharacter symbols for tags

UAlbertaALTLab · Nov 19, 2024 · e0ebd1b · e0ebd1b
1 parent 08dfc20
commit e0ebd1b
Show file tree

Hide file tree

Showing 4 changed files with 103 additions and 17 deletions.
diff --git a/src/stoeng/resources/dictionary/stoeng_test_db.importjson b/src/stoeng/resources/dictionary/stoeng_test_db.importjson
@@ -1,22 +1,40 @@
 [
   {
     "head": "agdaa",
+    "analysis": [
+      [],
+      "agdaa",
+      [
+        "+3Sg",
+        "+ContentInterr"
+      ]
+    ],
     "linguistInfo": {
       "inflectional_category": "wa-ya-Reg",
       "pos": "V",
-      "stem": "a_gdaa"
+      "stem": "a_gdaa",
+      "wordclass" : "VI"
     },
     "paradigm": "VI",
     "senses": [
       {
-        "definition": "for S to run",
+        "definition": "for s/he to run",
         "sources": ["McA&K&C"]
       }
     ],
     "slug": "agdaa"
   },
   {
     "head": "apa",
+    "analysis": [
+      [],
+      "apa",
+      [
+        "+3Sg",
+        "+3SgO",
+        "+ContentInterr"
+      ]
+    ],
     "linguistInfo": {
       "inflectional_category": "poly-wa-ya",
       "pos": "V",
@@ -25,14 +43,22 @@
     "paradigm": "VT",
     "senses": [
       {
-        "definition": "for S to hit O",
+        "definition": "for s/he to hit s.o.",
         "sources": ["McA&K&C"]
       }
     ],
     "slug": "apa"
   },
   {
     "head": "cîga",
+    "analysis": [
+      [],
+      "cîga",
+      [
+        "+3Sg",
+        "+ContentInterr"
+      ]
+    ],
     "linguistInfo": {
       "inflectional_category": "wa-ya-Reg",
       "pos": "V",
@@ -41,14 +67,22 @@
     "paradigm": "VI",
     "senses": [
       {
-        "definition": "for S to want him/her/it",
+        "definition": "for s/he to want s.o.",
         "sources": ["McA&K&C"]
       }
     ],
     "slug": "cîga"
   },
   {
     "head": "hnagda",
+    "analysis": [
+      [],
+      "hnagda",
+      [
+        "+3Sg",
+        "+ContentInterr"
+      ]
+    ],
     "linguistInfo": {
       "inflectional_category": "wa-ya-Reg",
       "pos": "V",
@@ -57,14 +91,22 @@
     "paradigm": "VI",
     "senses": [
       {
-        "definition": "for S to drink their own",
+        "definition": "for s/he to drink their own",
         "sources": ["McA&K&C"]
       }
     ],
     "slug": "hnagda"
   },
   {
     "head": "hnuda",
+      "analysis": [
+      [],
+      "hnuda",
+      [
+        "+3Sg",
+        "+ContentInterr"
+      ]
+    ],
     "linguistInfo": {
       "inflectional_category": "wa-ya-Reg",
       "pos": "V",
@@ -73,14 +115,22 @@
     "paradigm": "VI",
     "senses": [
       {
-        "definition": "for S to eat their own O",
+        "definition": "for s/he to eat their own s.o.",
         "sources": ["McA&K&C"]
       }
     ],
     "slug": "hnuda"
   },
   {
     "head": "ishdima",
+    "analysis": [
+      [],
+      "ishdima",
+      [
+        "+3Sg",
+        "+ContentInterr"
+      ]
+    ],
     "linguistInfo": {
       "inflectional_category": "m-n-Stems",
       "pos": "V",
@@ -89,14 +139,23 @@
     "paradigm": "VI",
     "senses": [
       {
-        "definition": "for S to be asleep",
+        "definition": "for s/he to be asleep",
         "sources": ["McA&K&C"]
       }
     ],
     "slug": "ishdima"
   },
   {
     "head": "nata",
+    "analysis": [
+      [],
+      "nata",
+      [
+        "+3Sg",
+        "+3SgO",
+        "+ContentInterr"
+      ]
+    ],
     "linguistInfo": {
       "inflectional_category": "poly-wa-ya",
       "pos": "V",
@@ -105,14 +164,23 @@
     "paradigm": "VT",
     "senses": [
       {
-        "definition": "for S to kick O",
+        "definition": "for s/he to kick s.o.",
         "sources": ["McA&K&C"]
       }
     ],
     "slug": "nata"
   },
   {
     "head": "uge",
+    "analysis": [
+      [],
+      "uge",
+      [
+        "+3Sg",
+        "+3SgO",
+        "+ContentInterr"
+      ]
+    ],
     "linguistInfo": {
       "inflectional_category": "poly-wa-ya",
       "pos": "V",
@@ -121,14 +189,22 @@
     "paradigm": "VT",
     "senses": [
       {
-        "definition": "for S to help O",
+        "definition": "for s/he to help s.o.",
         "sources": ["McA&K&C"]
       }
     ],
     "slug": "uge"
   },
   {
     "head": "yagda",
+    "analysis": [
+      [],
+      "yagda",
+      [
+        "+3Sg",
+        "+ContentInterr"
+      ]
+    ],
     "linguistInfo": {
       "inflectional_category": "m-n-Stems",
       "pos": "V",
@@ -137,23 +213,33 @@
     "paradigm": "VI",
     "senses": [
       {
-        "definition": "for S to drink it",
+        "definition": "for s/he to drink it",
         "sources": ["McA&K&C"]
       }
     ],
     "slug": "yagda"
   },
   {
     "head": "yúda",
+    "analysis": [
+      [],
+      "yúda",
+      [
+        "+2Sg",
+        "+Imp",
+        "+Imm"
+      ]
+    ],
     "linguistInfo": {
       "inflectional_category": "m-n-Stems",
       "pos": "V",
-      "stem": "0_y2uda"
+      "stem": "0_y2uda",
+      "wordclass": "VI"
     },
     "paradigm": "VI",
     "senses": [
       {
-        "definition": "for S to eat it",
+        "definition": "s/he eats s.t.",
         "sources": ["McA&K&C"]
       }
     ],

diff --git a/src/stoeng/resources/fst/analyser-gt-dict-norm.hfstol b/src/stoeng/resources/fst/analyser-gt-dict-norm.hfstol
diff --git a/src/stoeng/resources/fst/generator-gt-dict-norm.hfstol b/src/stoeng/resources/fst/generator-gt-dict-norm.hfstol
diff --git a/src/stoeng/site/settings.py b/src/stoeng/site/settings.py
@@ -49,7 +49,7 @@
 
 # Without this, importjson by default will not attempt to add inflected definitions using phrase translation.
 MORPHODICT_SUPPORTS_AUTO_DEFINITIONS = True
-MORPHODICT_ENABLE_FST_LEMMA_SUPPORT = True
+
 DEFAULT_TARGET_LANGUAGE_PHRASE_TAGS = ("+V",)
 
 MORPHODICT_ORTHOGRAPHY = {