Changeset: fd44dd6cb5fc for MonetDB
URL: https://dev.monetdb.org/hg/MonetDB/rev/fd44dd6cb5fc
Modified Files:
clients/Tests/MAL-signatures-hge.test
clients/Tests/MAL-signatures.test
clients/Tests/exports.stable.out
monetdb5/modules/mal/txtsim.c
Branch: txtsim
Log Message:
approved output
(tiny fix on interface mismatch (dbl vs double))
diffs (254 lines):
diff --git a/clients/Tests/MAL-signatures-hge.test
b/clients/Tests/MAL-signatures-hge.test
--- a/clients/Tests/MAL-signatures-hge.test
+++ b/clients/Tests/MAL-signatures-hge.test
@@ -35447,7 +35447,7 @@ battxtsim
similarity
command battxtsim.similarity(X_0:bat[:str], X_1:bat[:str]):bat[:dbl]
fstrcmp0_impl_bulk;
-Normalized edit distance between two strings
+(Deprecated) Normalized edit distance between two strings
baturl
extractURLHost
command baturl.extractURLHost(X_0:bat[:str], X_1:bit):bat[:str]
@@ -50879,60 +50879,80 @@ pattern tokenizer.take(X_0:oid):str
TKNZRtakeOid;
reconstruct and returns the i-th string
txtsim
+damerau_levenshtein
+pattern txtsim.damerau_levenshtein(X_0:str, X_1:str):int
+damerau_levenshtein_distance;
+Calculates Damerau-Levenshtein distance between two strings, operation costs
(ins/del = 1, replacement = 1, transposition = 2)
+txtsim
+damerau_levenshtein
+pattern txtsim.damerau_levenshtein(X_0:str, X_1:str, X_2:int, X_3:int,
X_4:int):int
+damerau_levenshtein_distance;
+Calculates Damerau-Levenshtein distance between two strings, variable
operation costs (ins/del, replacement, transposition)
+txtsim
editdistance
command txtsim.editdistance(X_0:str, X_1:str):int
-levenshteinbasic_impl;
-Alias for Levenshtein(str,str)
+damerau_levenshtein1;
+Alias for Damerau-Levenshtein(str,str), insdel cost = 1, replace cost = 1 and
transpose = 2
txtsim
editdistance2
command txtsim.editdistance2(X_0:str, X_1:str):int
-levenshteinbasic2_impl;
-Calculates Levenshtein distance (edit distance) between two strings. Cost of
transposition is 1 instead of 2
+damerau_levenshtein2;
+Alias for Damerau-Levenshtein(str,str), insdel cost = 1, replace cost = 1 and
transpose = 1
+txtsim
+jaro_winkler_similarity
+command txtsim.jaro_winkler_similarity(X_0:str, X_1:str):dbl
+jaro_winkler_similarity;
+Calculate Jaro Winkler similarity
txtsim
levenshtein
-command txtsim.levenshtein(X_0:str, X_1:str, X_2:int, X_3:int, X_4:int):int
-levenshtein_impl;
-Calculates Levenshtein distance (edit distance) between two strings, variable
operation costs (ins/del, replacement, transposition)
+pattern txtsim.levenshtein(X_0:str, X_1:str):int
+levenshtein_distance;
+Calculates Levenshtein distance between two strings, operation costs (ins/del
= 1, replacement = 1)
txtsim
levenshtein
-command txtsim.levenshtein(X_0:str, X_1:str):int
-levenshteinbasic_impl;
-Calculates Levenshtein distance (edit distance) between two strings
+pattern txtsim.levenshtein(X_0:str, X_1:str, X_2:int, X_3:int):int
+levenshtein_distance;
+Calculates Levenshtein distance between two strings, variable operation costs
(ins/del, replacement)
+txtsim
+levenshtein
+pattern txtsim.levenshtein(X_0:str, X_1:str, X_2:int, X_3:int, X_4:int):int
+levenshtein_distance;
+(Backwards compatibility purposes) Calculates Damerau-Levenshtein distance
between two strings, variable operation costs (ins/del, replacement,
transposition)
txtsim
qgramnormalize
command txtsim.qgramnormalize(X_0:str):str
-CMDqgramnormalize;
+qgram_normalize;
'Normalizes' strings (eg. toUpper and replaces non-alphanumerics with one space
txtsim
qgramselfjoin
command txtsim.qgramselfjoin(X_0:bat[:oid], X_1:bat[:oid], X_2:bat[:int],
X_3:bat[:int], X_4:flt, X_5:int) (X_6:bat[:int], X_7:bat[:int])
-CMDqgramselfjoin;
+qgram_selfjoin;
QGram self-join on ordered(!) qgram tables and sub-ordered q-gram positions
txtsim
similarity
command txtsim.similarity(X_0:str, X_1:str):dbl
fstrcmp0_impl;
-Normalized edit distance between two strings
+(Deprecated) Normalized edit distance between two strings
txtsim
similarity
command txtsim.similarity(X_0:str, X_1:str, X_2:dbl):dbl
fstrcmp_impl;
-Normalized edit distance between two strings
+(Deprecated) Normalized edit distance between two strings
txtsim
soundex
command txtsim.soundex(X_0:str):str
-soundex_impl;
+soundex;
Soundex function for phonetic matching
txtsim
str2qgrams
command txtsim.str2qgrams(X_0:str):bat[:str]
-CMDstr2qgrams;
+str_2_qgrams;
Break the string into 4-grams
txtsim
stringdiff
command txtsim.stringdiff(X_0:str, X_1:str):int
-stringdiff_impl;
-calculate the soundexed editdistance
+stringdiff;
+Calculate the soundexed editdistance
url
extractURLHost
command url.extractURLHost(X_0:str, X_1:bit):str
diff --git a/clients/Tests/MAL-signatures.test
b/clients/Tests/MAL-signatures.test
--- a/clients/Tests/MAL-signatures.test
+++ b/clients/Tests/MAL-signatures.test
@@ -26497,7 +26497,7 @@ battxtsim
similarity
command battxtsim.similarity(X_0:bat[:str], X_1:bat[:str]):bat[:dbl]
fstrcmp0_impl_bulk;
-Normalized edit distance between two strings
+(Deprecated) Normalized edit distance between two strings
baturl
extractURLHost
command baturl.extractURLHost(X_0:bat[:str], X_1:bit):bat[:str]
@@ -39204,60 +39204,80 @@ pattern tokenizer.take(X_0:oid):str
TKNZRtakeOid;
reconstruct and returns the i-th string
txtsim
+damerau_levenshtein
+pattern txtsim.damerau_levenshtein(X_0:str, X_1:str):int
+damerau_levenshtein_distance;
+Calculates Damerau-Levenshtein distance between two strings, operation costs
(ins/del = 1, replacement = 1, transposition = 2)
+txtsim
+damerau_levenshtein
+pattern txtsim.damerau_levenshtein(X_0:str, X_1:str, X_2:int, X_3:int,
X_4:int):int
+damerau_levenshtein_distance;
+Calculates Damerau-Levenshtein distance between two strings, variable
operation costs (ins/del, replacement, transposition)
+txtsim
editdistance
command txtsim.editdistance(X_0:str, X_1:str):int
-levenshteinbasic_impl;
-Alias for Levenshtein(str,str)
+damerau_levenshtein1;
+Alias for Damerau-Levenshtein(str,str), insdel cost = 1, replace cost = 1 and
transpose = 2
txtsim
editdistance2
command txtsim.editdistance2(X_0:str, X_1:str):int
-levenshteinbasic2_impl;
-Calculates Levenshtein distance (edit distance) between two strings. Cost of
transposition is 1 instead of 2
+damerau_levenshtein2;
+Alias for Damerau-Levenshtein(str,str), insdel cost = 1, replace cost = 1 and
transpose = 1
+txtsim
+jaro_winkler_similarity
+command txtsim.jaro_winkler_similarity(X_0:str, X_1:str):dbl
+jaro_winkler_similarity;
+Calculate Jaro Winkler similarity
txtsim
levenshtein
-command txtsim.levenshtein(X_0:str, X_1:str, X_2:int, X_3:int, X_4:int):int
-levenshtein_impl;
-Calculates Levenshtein distance (edit distance) between two strings, variable
operation costs (ins/del, replacement, transposition)
+pattern txtsim.levenshtein(X_0:str, X_1:str):int
+levenshtein_distance;
+Calculates Levenshtein distance between two strings, operation costs (ins/del
= 1, replacement = 1)
txtsim
levenshtein
-command txtsim.levenshtein(X_0:str, X_1:str):int
-levenshteinbasic_impl;
-Calculates Levenshtein distance (edit distance) between two strings
+pattern txtsim.levenshtein(X_0:str, X_1:str, X_2:int, X_3:int):int
+levenshtein_distance;
+Calculates Levenshtein distance between two strings, variable operation costs
(ins/del, replacement)
+txtsim
+levenshtein
+pattern txtsim.levenshtein(X_0:str, X_1:str, X_2:int, X_3:int, X_4:int):int
+levenshtein_distance;
+(Backwards compatibility purposes) Calculates Damerau-Levenshtein distance
between two strings, variable operation costs (ins/del, replacement,
transposition)
txtsim
qgramnormalize
command txtsim.qgramnormalize(X_0:str):str
-CMDqgramnormalize;
+qgram_normalize;
'Normalizes' strings (eg. toUpper and replaces non-alphanumerics with one space
txtsim
qgramselfjoin
command txtsim.qgramselfjoin(X_0:bat[:oid], X_1:bat[:oid], X_2:bat[:int],
X_3:bat[:int], X_4:flt, X_5:int) (X_6:bat[:int], X_7:bat[:int])
-CMDqgramselfjoin;
+qgram_selfjoin;
QGram self-join on ordered(!) qgram tables and sub-ordered q-gram positions
txtsim
similarity
command txtsim.similarity(X_0:str, X_1:str):dbl
fstrcmp0_impl;
-Normalized edit distance between two strings
+(Deprecated) Normalized edit distance between two strings
txtsim
similarity
command txtsim.similarity(X_0:str, X_1:str, X_2:dbl):dbl
fstrcmp_impl;
-Normalized edit distance between two strings
+(Deprecated) Normalized edit distance between two strings
txtsim
soundex
command txtsim.soundex(X_0:str):str
-soundex_impl;
+soundex;
Soundex function for phonetic matching
txtsim
str2qgrams
command txtsim.str2qgrams(X_0:str):bat[:str]
-CMDstr2qgrams;
+str_2_qgrams;
Break the string into 4-grams
txtsim
stringdiff
command txtsim.stringdiff(X_0:str, X_1:str):int
-stringdiff_impl;
-calculate the soundexed editdistance
+stringdiff;
+Calculate the soundexed editdistance
url
extractURLHost
command url.extractURLHost(X_0:str, X_1:bit):str
diff --git a/clients/Tests/exports.stable.out b/clients/Tests/exports.stable.out
--- a/clients/Tests/exports.stable.out
+++ b/clients/Tests/exports.stable.out
@@ -784,6 +784,8 @@ void TABLETdestroy_format(Tablet *as);
int TABLEToutput_file(Tablet *as, BAT *order, stream *s);
int TRACEtable(Client cntxt, BAT **r);
int TYPE_xml;
+int UTF8_strlen(const char *restrict s);
+int UTF8_strwidth(const char *restrict s);
void addMalException(MalBlkPtr mb, str msg);
str addOptimizerPipe(Client cntxt, MalBlkPtr mb, const char *name);
str addPipeDefinition(Client cntxt, const char *name, const char *pipe);
@@ -1232,6 +1234,7 @@ const char *stoptraceRef;
void strAfterCall(ValPtr v, ValPtr bak);
void strBeforeCall(ValPtr v, ValPtr bak);
const char *strRef;
+int str_strlen(const char *restrict s);
const char *streamsRef;
const char *strimpsRef;
const char *subavgRef;
diff --git a/monetdb5/modules/mal/txtsim.c b/monetdb5/modules/mal/txtsim.c
--- a/monetdb5/modules/mal/txtsim.c
+++ b/monetdb5/modules/mal/txtsim.c
@@ -394,7 +394,7 @@ jaro_winkler(const str_item *x, const st
}
static str
-jaro_winkler_similarity(double *ret, str *x, str *y)
+jaro_winkler_similarity(dbl *ret, str *x, str *y)
{
int *x_flags = NULL, *y_flags = NULL;
str_item xi = { 0 }, yi = { 0 };
_______________________________________________
checkin-list mailing list -- checkin-list@monetdb.org
To unsubscribe send an email to checkin-list-le...@monetdb.org
