Data update
This commit is contained in:
Ingy döt Net
parent
5150844a7d
commit
4bb20c9b71
7735 changed files with 38060 additions and 199180 deletions
|
|
@ -1,91 +0,0 @@
|
|||
with Ada.Text_IO;
|
||||
|
||||
procedure Jaro_Distances is
|
||||
|
||||
type Jaro_Measure is new Float;
|
||||
|
||||
function Jaro_Distance (Left, Right : in String) return Jaro_Measure
|
||||
is
|
||||
Left_Matches : array (Left'Range) of Boolean := (others => False);
|
||||
Right_Matches : array (Right'Range) of Boolean := (others => False);
|
||||
Matches : Natural := 0;
|
||||
Transpositions : Natural := 0;
|
||||
begin
|
||||
if Left'Length = 0 and Right'Length = 0 then
|
||||
return 1.000;
|
||||
end if;
|
||||
|
||||
declare
|
||||
Match_Distance : constant Natural := Natural'Max (Left'Length, Right'Length) / 2 - 1;
|
||||
begin
|
||||
for L in Left'Range loop
|
||||
declare
|
||||
First : constant Natural := Natural'Max (Right'First, Right'First + L - Left'First - Match_Distance);
|
||||
Last : constant Natural := Natural'Min (L - Left'First + Match_Distance + Right'First, Right'Last);
|
||||
begin
|
||||
for R in First .. Last loop
|
||||
if
|
||||
not Right_Matches (R) and
|
||||
Left (L) = Right (R)
|
||||
then
|
||||
Left_Matches (L) := True;
|
||||
Right_Matches (R) := True;
|
||||
Matches := Matches + 1;
|
||||
exit;
|
||||
end if;
|
||||
end loop;
|
||||
end;
|
||||
end loop;
|
||||
end;
|
||||
|
||||
if Matches = 0 then
|
||||
return 0.000;
|
||||
end if;
|
||||
|
||||
declare
|
||||
R : Natural := Right'First;
|
||||
begin
|
||||
for L in Left'Range loop
|
||||
if Left_Matches (L) then
|
||||
while not Right_Matches (R) loop
|
||||
R := R + 1;
|
||||
end loop;
|
||||
if Left (L) /= Right (R) then
|
||||
Transpositions := Transpositions + 1;
|
||||
end if;
|
||||
R := R + 1;
|
||||
end if;
|
||||
end loop;
|
||||
end;
|
||||
|
||||
declare
|
||||
Match : constant Float := Float (Matches);
|
||||
Term_1 : constant Float := Match / Float (Left'Length);
|
||||
Term_2 : constant Float := Match / Float (Right'Length);
|
||||
Term_3 : constant Float := (Match - Float (Transpositions) / 2.0) / Match;
|
||||
begin
|
||||
return Jaro_Measure ((Term_1 + Term_2 + Term_3) / 3.0);
|
||||
end;
|
||||
end Jaro_Distance;
|
||||
|
||||
procedure Show_Jaro (Left, Right : in String)
|
||||
is
|
||||
package Jaro_IO is
|
||||
new Ada.Text_IO.Float_IO (Jaro_Measure);
|
||||
use Ada.Text_IO;
|
||||
|
||||
Distance : constant Jaro_Measure := Jaro_Distance (Left, Right);
|
||||
begin
|
||||
Jaro_IO.Put (Distance, Fore => 1, Aft => 5, Exp => 0);
|
||||
Set_Col (10); Put (Left);
|
||||
Set_Col (22); Put (Right);
|
||||
New_Line;
|
||||
end Show_Jaro;
|
||||
|
||||
S1 : constant String := " MARTHA VS MARHTA ";
|
||||
begin
|
||||
Show_Jaro ("DWAYNE", "DUANE");
|
||||
Show_Jaro ("DIXON", "DICKSONX");
|
||||
Show_Jaro ("JELLYFISH", "SMELLYFISH");
|
||||
Show_Jaro (S1 (3 .. 8), S1 (13 .. 18));
|
||||
end Jaro_Distances;
|
||||
|
|
@ -1,98 +0,0 @@
|
|||
identification division.
|
||||
program-id. JaroDistance.
|
||||
|
||||
environment division.
|
||||
configuration section.
|
||||
repository.
|
||||
function length intrinsic
|
||||
function trim intrinsic
|
||||
function max intrinsic
|
||||
function min intrinsic
|
||||
.
|
||||
|
||||
data division.
|
||||
working-storage section.
|
||||
77 s pic x(255).
|
||||
77 t pic x(255).
|
||||
77 s-length pic 9(3).
|
||||
77 t-length pic 9(3).
|
||||
77 i pic 9(3).
|
||||
77 j pic 9(3).
|
||||
77 k pic 9(3).
|
||||
77 start-pos pic 9(3).
|
||||
77 end-pos pic 9(3).
|
||||
77 match-distance pic 9(3).
|
||||
77 matches pic 9(3).
|
||||
77 transpositions pic 9(3).
|
||||
77 distance pic 9v9(8).
|
||||
|
||||
01 jaro-table.
|
||||
05 filler occurs 255.
|
||||
10 filler pic 9(1).
|
||||
88 s-matches value 1 when set to false is 0.
|
||||
10 filler pic 9(1).
|
||||
88 t-matches value 1 when set to false is 0.
|
||||
|
||||
procedure division.
|
||||
main.
|
||||
move "MARTHA" to s
|
||||
move "MARHTA" to t
|
||||
perform jaro-calc-and-show
|
||||
move "DIXON" to s
|
||||
move "DICKSONX" to t
|
||||
perform jaro-calc-and-show
|
||||
move "JELLYFISH" to s
|
||||
move "SMELLYFISH" to t
|
||||
perform jaro-calc-and-show
|
||||
stop run
|
||||
.
|
||||
jaro-calc-and-show.
|
||||
perform jaro-distance
|
||||
display trim(s) " -> " trim(t) ", distance=" distance
|
||||
.
|
||||
jaro-distance.
|
||||
move length(trim(s)) to s-length
|
||||
move length(trim(t)) to t-length
|
||||
if s-length = zeros and t-length = zeros
|
||||
move 1 to distance
|
||||
exit paragraph
|
||||
end-if
|
||||
|
||||
compute match-distance = max(s-length, t-length) / 2 - 1
|
||||
move low-values to jaro-table
|
||||
move zeros to matches
|
||||
move zeros to transpositions
|
||||
perform varying i from 1 by 1 until i > s-length
|
||||
move max(1, i - match-distance) to start-pos
|
||||
move min(i + match-distance, t-length) to end-pos
|
||||
perform varying j from start-pos by 1 until j > end-pos
|
||||
if t-matches(j) or s(i:1) <> t(j:1)
|
||||
exit perform cycle
|
||||
end-if,
|
||||
set s-matches(i), t-matches(j) to true
|
||||
add 1 to matches
|
||||
exit perform
|
||||
end-perform
|
||||
end-perform
|
||||
if matches = zeros
|
||||
move matches to distance
|
||||
exit paragraph
|
||||
end-if
|
||||
|
||||
move 1 to k
|
||||
perform varying i from 1 by 1 until i > s-length
|
||||
if not s-matches(i)
|
||||
exit perform cycle
|
||||
end-if
|
||||
perform until t-matches(k)
|
||||
add 1 to k
|
||||
end-perform
|
||||
if s(i:1) <> t(k:1)
|
||||
add 1 to transpositions
|
||||
end-if
|
||||
add 1 to k
|
||||
end-perform
|
||||
|
||||
compute distance = ((matches / s-length) + (matches / t-length) +
|
||||
((matches - transpositions / 2) / matches)) / 3
|
||||
.
|
||||
|
|
@ -1,55 +0,0 @@
|
|||
(let ()
|
||||
(defun jaro (s1 s2)
|
||||
(let (mw mflags1 mflags2 fn-reset-mflags fn-reset-all-mflags fn-cnt-trans)
|
||||
(setq mflags1 (make-vector (length s1) nil))
|
||||
(setq mflags2 (make-vector (length s2) nil))
|
||||
(setq mw (1- (/ (max (length s1) (length s2)) 2)))
|
||||
(setq fn-reset-mflags
|
||||
(lambda (idx)
|
||||
(let ((start (max 0 (- idx mw)))
|
||||
(end (min (1- (length s2)) (+ idx mw))))
|
||||
(cl-loop for i from start to end do
|
||||
(when (and (not (elt mflags1 idx))
|
||||
(not (elt mflags2 i)))
|
||||
(when (equal (elt s1 idx) (elt s2 i))
|
||||
(aset mflags1 idx 't)
|
||||
(aset mflags2 i 't) ) ) ) ) ) )
|
||||
(setq fn-reset-all-mflags
|
||||
(lambda ()
|
||||
(dotimes (idx (length s1))
|
||||
(funcall fn-reset-mflags idx) ) ) )
|
||||
|
||||
(setq fn-cnt-trans
|
||||
(lambda ()
|
||||
(let ((cur2 0) (transposition 0))
|
||||
(dotimes (cur1 (length s1))
|
||||
(when (aref mflags1 cur1)
|
||||
(while (not (aref mflags2 cur2))
|
||||
(setq cur2 (1+ cur2)) )
|
||||
(when (not (equal (aref s1 cur1)
|
||||
(aref s2 cur2)))
|
||||
(setq transposition (1+ transposition)) )
|
||||
(setq cur2 (1+ cur2))
|
||||
)
|
||||
)
|
||||
transposition ) ) )
|
||||
|
||||
(funcall fn-reset-all-mflags)
|
||||
(let ((m (seq-count (lambda (f) f) mflags1))
|
||||
(tr (funcall fn-cnt-trans)))
|
||||
;;(message "matches: %s, transposition: %s, |s1|: %d |s2|: %d" m tr (length s1) (length s2))
|
||||
(if (= m 0)
|
||||
0
|
||||
(progn (/ (+ (/ (float m) (length s1)) (/ (float m) (length s2)) (/ (float (- m (/ (float tr) 2))) m) ) 3))
|
||||
) ) ) )
|
||||
|
||||
(let ((params '(("MARTHA" "MARHTA")
|
||||
("DIXON" "DICKSONX")
|
||||
("JELLYFISH" "SMELLYFISH"))))
|
||||
(dolist (p params)
|
||||
(message "jaro(%s, %s) = %f"
|
||||
(nth 0 p) (nth 1 p)
|
||||
(jaro (nth 0 p) (nth 1 p)))
|
||||
)
|
||||
)
|
||||
)
|
||||
|
|
@ -1,39 +0,0 @@
|
|||
class Jaro {
|
||||
private static function jaro(s1: String, s2: String): Float {
|
||||
var s1_len = s1.length;
|
||||
var s2_len = s2.length;
|
||||
if (s1_len == 0 && s2_len == 0) return 1;
|
||||
|
||||
var match_distance = Std.int(Math.max(s1_len, s2_len)) / 2 - 1;
|
||||
var matches = { s1: [for(n in 0...s1_len) false], s2: [for(n in 0...s2_len) false] };
|
||||
var m = 0;
|
||||
for (i in 0...s1_len) {
|
||||
var start = Std.int(Math.max(0, i - match_distance));
|
||||
var end = Std.int(Math.min(i + match_distance + 1, s2_len));
|
||||
for (j in start...end)
|
||||
if (!matches.s2[j] && s1.charAt(i) == s2.charAt(j)) {
|
||||
matches.s1[i] = true;
|
||||
matches.s2[j] = true;
|
||||
m++;
|
||||
break;
|
||||
}
|
||||
}
|
||||
if (m == 0) return 0;
|
||||
|
||||
var k = 0;
|
||||
var t = 0.;
|
||||
for (i in 0...s1_len)
|
||||
if (matches.s1[i]) {
|
||||
while (!matches.s2[k]) k++;
|
||||
if (s1.charAt(i) != s2.charAt(k++)) t += 0.5;
|
||||
}
|
||||
|
||||
return (m / s1_len + m / s2_len + (m - t) / m) / 3.0;
|
||||
}
|
||||
|
||||
public static function main() {
|
||||
Sys.println(jaro( "MARTHA", "MARHTA"));
|
||||
Sys.println(jaro( "DIXON", "DICKSONX"));
|
||||
Sys.println(jaro("JELLYFISH", "SMELLYFISH"));
|
||||
}
|
||||
}
|
||||
Loading…
Add table
Add a link
Reference in a new issue