Page 1 of 1
Optimizing findstring
Posted: Wed Dec 15, 2010 4:02 pm
by buzzqw
Hi all!
first of all i must admin that i am not a programmer in any way! but i enjoying using purebasic for automatizing simply task
i have to check 2 cvs of several thousand hundred lines, i must read a line in file A and check in file B if i found/not found some parameters
this is a snippet of code
Code: Select all
ReadFile(888,firstfile.s)
While Eof(888)=0
line.s=ReadString(888)
If StringField(line.s,2,";")="S"
a.s=StringField(line.s,3,";")
b.s=StringField(line.s,4,";")
c.s=StringField(line.s,6,";")
ReadFile(889,secondfile.s)
While Eof(889)=0
secondline.s=ReadString(889)
If FindString(secondline.s,a.s,0) And FindString(secondline.s,b.s,0) And FindString(secondline.s,c.s,0)
WriteStringN(890,line.s+" >>> "+secondline.s)
EndIf
Wend
CloseFile(889)
EndIf
Wend
CloseFile(888)
CloseFile(890)
i have a very strong feeling that i wrote a very bad routine (but working.. slow)
any help from professional developers ?
Thanks!
BHH
Re: Optimizing findstring
Posted: Wed Dec 15, 2010 4:05 pm
by IdeasVacuum
Re: Optimizing findstring
Posted: Wed Dec 15, 2010 4:31 pm
by buzzqw
very fascinating ... but over my understanding
Code: Select all
Procedure TUNEDBM(Start.l,Array Source.a(1), Array Pattern.a(1))
Protected.i i, j, m, n, k, match, shift
Protected.i Dim bmBc(255)
m = ArraySize(Pattern())
n = ArraySize(Source())
; Preprocessing
preBmBc(Pattern(), bmBc())
shift = Max(bmBc(Pattern(m - 1)),1)
bmBc(Pattern(m - 1)) = 0
; Searching
j = Max(0,Start-1)
While j < n
While j + m < n And bmBc(Source(j + m -1)) <> 0
j + bmBc(Source(j + m -1))
If j + m >= n : Break : EndIf
j + bmBc(Source(j + m -1))
If j + m >= n : Break : EndIf
j + bmBc(Source(j + m -1))
If j + m >= n : Break : EndIf
Wend
If CompareMemory(@Source(j), @Pattern(), ArraySize(Pattern()))
ProcedureReturn j
EndIf
j + shift
Wend
ProcedureReturn 0
EndProcedure
how this code is to use in my situation ?
BHH
Re: Optimizing findstring
Posted: Wed Dec 15, 2010 4:49 pm
by IdeasVacuum
...that is the search algorithm. On the same page is some code to load a whole file into an array (Array Source). Write your own code that calls TUNEDBM with a pattern (string) to search for (Array Pattern). You can use the same process for both files.
Re: Optimizing findstring
Posted: Wed Dec 15, 2010 4:50 pm
by Trond
You will get more help if you post a working code. Code snippets that don't work as they are posted are annoying to work with.
Re: Optimizing findstring
Posted: Wed Dec 15, 2010 5:02 pm
by buzzqw
here the full code
http://www.64k.it/andres/data/Varie/findstring.rar (3kb)
sorry pal but code is in italian
in fist string field you put the first csv
in second string the second csv
"salva log" mean "save log" where analysis file is done
"avvia analisi" mean start analisys
BHH
Re: Optimizing findstring
Posted: Wed Dec 15, 2010 5:36 pm
by Trond
First problem: Procedures that are run as a thread must have exactly one integer parameter:
Else your program may crash after a while.
Edit:
The next thing I notice is that you read in your file twice. Performance will be better if you read the files to be compared into two arrays and then use the arrays twice instead of reading the file twice.