Postnominal modifiers update

UUDigitalHumanitieslab · Dec 6, 2024 · bd491dd · bd491dd
1 parent 8b954c7
commit bd491dd
Show file tree

Hide file tree

Showing 14 changed files with 477 additions and 120 deletions.
diff --git a/src/sastadev/Sziplus.py b/src/sastadev/Sziplus.py
@@ -85,12 +85,12 @@ def isvcinforppart(node: SynTree) -> bool:
 
 def isrealnode(node: SynTree) -> bool:
     '''
-    The function *isrealnode* determines whether a nide is a real node, which it is if:
+    The function *isrealnode* determines whether a node is a real node, which it is if:
 
     * it is not a node for an interpunction sign
     * it is not a nonfinite complement
     * if it is not a separable particle word of a verb
-    * if it is not an index node 9as detemined by the function *isindexnode*)
+    * if it is not an index node (as determined by the function *isindexnode*)
 
     The function *isindexnode* is defined as follows:
 
@@ -102,7 +102,7 @@ def isrealnode(node: SynTree) -> bool:
         result = False
     elif isvcinforppart(node):
         result = False
-    elif rel == 'svp' and pt in node.attrib:
+    elif rel == 'svp' and 'word' in node.attrib:
         result = False
     elif isindexnode(node):
         result = False

diff --git a/src/sastadev/basicreplacements.py b/src/sastadev/basicreplacements.py
@@ -245,6 +245,8 @@ def combine(strlist: List[str]) -> str:
                                                 ('blon', 'ballon', pron, infpron, voweldel, dp),
                                                 ('ooien', 'gooien', pron, wrongpron, onsetred, dp),
                                                 ('poppe', 'pop', pron, wrongpron, emphasis, dp),
+                                                ('lus', 'lust', pron, infpron, codared, dp),
+                                                ('jou', 'jouw', pron, infpron, codared, -dp) # Td 22, 30 ik wil ook keer naar jou huis find criterion
                                                 # ('leggen', 'liggen', lexical, dial, '', dp), # moved to corrector : only if parse is illformed
                                                 # ('legt', 'ligt', lexical, dial, '', dp),  # moved to corrector : only if parse is illformed
                                                 # ('leg', 'lig', lexical, dial, '', dp) # moved to corrector : only if parse is illformed

diff --git a/src/sastadev/cleanCHILDEStokens.py b/src/sastadev/cleanCHILDEStokens.py
@@ -230,13 +230,15 @@ def removesuspecttokens(tokens: List[Token]) -> List[Token]:
 
 
 
-RobustnessTuple = Tuple[Pattern, str, str, str]
+RobustnessTuple = Tuple[Pattern, str, str, str]  # regex, instring, outstring, message
 
 robustnessrules: List[RobustnessTuple] = [(re.compile(r'\u2026'), '\u2026', '...', 'Horizontal Ellipsis (\u2026, Unicode U+2026) replaced by a sequence of three Full Stops (..., Unicode U+002E) '),
                                           (re.compile('#'), '#', '', 'Number Sign (#, Unicode U+0023) removed'),
                                           #(re.compile('#'), '#', '(.)', 'Number Sign (#, Unicode U+0023) replaced by CHAT (short) pause code: (.)'),
-                                          (re.compile(r'\[\+bch\]'), '[+bch]', '[+ bch]', 'Missing space'),
-                                          (re.compile(r'\[\+trn\]'), '[+trn]', '[+ trn]', 'Missing space'),
+                                          (re.compile(r'\[\+bch\]', re.I), '[+bch]', '[+ bch]', 'Missing space'),
+                                          (re.compile(r'\[\+trn\]', re.I), '[+trn]', '[+ trn]', 'Missing space'),
+                                          (re.compile(r'\[\+ea\]', re.I), '[+ea]', '[+ ea]', 'Missing space'),
+                                          (re.compile(r'\[%(?![\s])'), '[%', '[% ', 'Missing space'),
                                           (re.compile(r'\[:(?![:\s])'), '[:', '[: ', 'Missing space'),
                                           (re.compile(r'(?<=\w)\+\.\.\.'), '+...', ' +...', 'Missing space'),
                                           (re.compile(r'\u2018'), '\u2018', "'", "Left Single Quotation Mark (\u2018. Unicode U+2018) replaced by Apostrophe ' (Unicode U+0027)"),

diff --git a/src/sastadev/correcttreebank.py b/src/sastadev/correcttreebank.py
@@ -21,6 +21,7 @@
                                SASTA, ADULTSPELLINGCORRECTION, ALLSAMPLECORRECTIONS, BASICREPLACEMENTS, CONTEXT,
                                HISTORY, CHILDRENSPELLINGCORRECTION, THISSAMPLECORRECTIONS, replacementsubsources
                                )
+from sastadev.postnominalmodifiers import transformbwinnp, transformppinnp
 from sastadev.sastatok import sasta_tokenize
 from sastadev.sastatoken import Token, insertinflate, tokenlist2stringlist, tokenlist2string
 from sastadev.sastatypes import (AltId, CorrectionMode, ErrorDict, MetaElement,
@@ -41,7 +42,7 @@
                                         showtree, simpleshow, subclasscompatible, transplant_node,
                                         treeinflate, treewithtokenpos,
                                         updatetokenpos)
-from sastadev.treetransform import transformtreeld, transformtreenogeen, transformtreenogde
+from sastadev.treetransform import transformtagcomma, transformtreeld, transformtreenogeen, transformtreenogde
 
 ampersand = '&'
 
@@ -564,7 +565,10 @@ def correct_stree(stree: SynTree,  corr: CorrectionMode, correctionparameters: C
 
     # tree transformations
     if correctionparameters.method in ['tarsp', ' stap']:
+        stree = transformtagcomma(stree)
         stree = transformtreeld(stree)
+        stree = transformppinnp(stree)
+        stree = transformbwinnp(stree)
         stree = transformtreenogeen(stree)
         stree = transformtreenogde(stree)
 
@@ -908,7 +912,10 @@ def correct_stree(stree: SynTree,  corr: CorrectionMode, correctionparameters: C
 
     # tree transformations
     if correctionparameters.method in ['tarsp', ' stap']:
+        fulltree = transformtagcomma(fulltree)
         fulltree = transformtreeld(fulltree)
+        fulltree = transformppinnp(fulltree)
+        fulltree = transformbwinnp(fulltree)
         fulltree = transformtreenogeen(fulltree)
         fulltree = transformtreenogde(fulltree)
 

diff --git a/src/sastadev/data/childescorrections/children_samplecorrections.txt b/src/sastadev/data/childescorrections/children_samplecorrections.txt
@@ -121,6 +121,7 @@ siko	cirkel	replacement	1
 bjokje	blokje	replacement	1
 kikke	kikker	noncompletion	1
 springe	springen	noncompletion	1
+in	met	replacement	1
 in	erin	noncompletion	3
 in	deze	replacement	2
 in	op	replacement	2
@@ -211,7 +212,8 @@ ke	die	replacement	1
 dese	deze	replacement	4
 hijs	hij is	replacement	1
 eigk	eigenlijk	replacement	1
-de	het	replacement	21
+de	het	replacement	23
+de	zitten	replacement	1
 de	der	noncompletion	11
 de	dan	replacement	2
 de	deze	replacement	2
@@ -249,7 +251,7 @@ worst	borst	replacement	1
 ennu	en	replacement	1
 koeie	koeien	noncompletion	1
 o	op	noncompletion	3
-mij	mijn	noncompletion	9
+mij	mijn	noncompletion	10
 bet	bent	noncompletion	1
 bent	ben	replacement	1
 tinnen	tillen	replacement	1
@@ -378,6 +380,8 @@ deze	dit	replacement	4
 deze	waar hoort deze	explanation	1
 evallen	gevallen	noncompletion	1
 nog	ook	explanation	1
+voor	bij	replacement	1
+voor	om	replacement	2
 voor	van	explanation	1
 wat	waar	replacement	2
 wordt	hoort	replacement	2
@@ -443,6 +447,7 @@ pantoet	pannekoeken	replacement	2
 eet	eten	replacement	2
 bejo	hallo	explanation	1
 teje	gaan	replacement	2
+maar	normaal	replacement	1
 maar	en	replacement	7
 maar	want	replacement	1
 maar	ga daar maar	explanation	1
@@ -472,7 +477,7 @@ bochje	bochtje	explanation	1
 anders	andere	explanation	1
 saar	daar	explanation	1
 saal	haal	explanation	1
-saan	gaan	explanation	1
+saan	gaan	replacement	2
 bruiken	gebruiken	explanation	1
 keerd	verkeerd	explanation	1
 vrastauto	vrachtauto	replacement	1
@@ -520,6 +525,7 @@ gun	ging	replacement	2
 gin	ging	noncompletion	4
 som	soms	noncompletion	6
 witte	wit	replacement	2
+dee	deed	noncompletion	1
 dee	was	replacement	2
 dan	toen	replacement	6
 ging	gingen	replacement	2
@@ -549,12 +555,12 @@ lew	wel	replacement	2
 slin	ging	replacement	2
 stond	stonden	replacement	4
 stond	was	replacement	2
+van	bij	replacement	2
 van	om	replacement	2
 van	door	replacement	1
 van	met	replacement	1
 van	uit	replacement	1
 van	veel	replacement	1
-van	bij	replacement	1
 arreen	alleen	replacement	2
 het	er	replacement	2
 het	hij	replacement	2
@@ -614,6 +620,7 @@ zaten	hadden	replacement	1
 zo	zo'n	replacement	1
 ook	blauwe ramen zitten er ook	explanation	1
 ginnen	gingen	replacement	2
+dat	daar	replacement	1
 dat	het	replacement	1
 allegrooste	allegrootste	noncompletion	2
 en	maar	replacement	3
@@ -641,9 +648,11 @@ ta	tat	noncompletion	1
 peltje	pijltje	replacement	1
 pijwtje	pijltje	replacement	1
 sef	zeg	replacement	1
+doe	toen	replacement	1
 doe	ga	replacement	1
 zwaarden	vechten	replacement	1
 lekkes	lekkers	noncompletion	1
+was	ging	replacement	1
 was	waren	replacement	1
 was	wat	replacement	1
 suimpjes	sguimpjes	noncompletion	1
@@ -666,6 +675,8 @@ itten	eten	replacement	2
 ziekje	muziekje	replacement	1
 Bobdebouwerkattet	Bobdebouwerkwartet	replacement	1
 blazen	geblazen	replacement	1
+ik	je	replacement	1
+ik	jij	replacement	1
 ik	mij	replacement	1
 wou	wilde	replacement	3
 vashouden	vasthouden	noncompletion	2
@@ -802,6 +813,9 @@ rekele	rekenen	replacement	1
 daarzo	daar	replacement	1
 voorbeeld	bijvoorbeeld	replacement	1
 freene	free-runnen	replacement	1
+x	racen-x	replacement	1
+x	los ?	replacement	1
+x	gelaten ?	replacement	1
 x	gewoon	replacement	1
 feerunnen	freerunnen	replacement	1
 wees	geweest	replacement	1
@@ -812,3 +826,22 @@ zorgen	verzorgen	replacement	1
 kantie	vakantie	replacement	1
 zomerkantie	vakantie	replacement	1
 boekje	'n boekje voor David	explanation	1
+xxx	?	replacement	1
+koppie	kopbal	replacement	1
+weer	meer	replacement	1
+kubbelbum	bubblegum ?	replacement	1
+probreren	proberen	replacement	2
+doem	doen	replacement	1
+durf	durft	noncompletion	1
+tik	tikkertje	replacement	1
+teek	betekent	replacement	1
+palaplu	paraplu	replacement	1
+valaag	vandaag	replacement	1
+,	dat	replacement	1
+tegen	aan	replacement	1
+zeggen	vragen	replacement	1
+vliegtuig	vliegveld	replacement	1
+anneen	alleen	replacement	1
+na	dan	replacement	1
+noen	doen	replacement	1
+teeëndertig	tweeëndertig	replacement	1