I want to find a way to find most suitable row from table which contains a word that is most similar to the word i'm entering. any idea? (I'm using OCR that finds words not exactly the same sometimes reads word 'specific' as 'spccific')
If you are using Oracle then you can try UTL_MATCH which uses something known as the Levenshtein Distance to calculate the minimum number of edits to transform one string into another. Other systems may have something similar or you can use the alogrithm as a starting point for your own function.
Maybe you can use the SOUNDEX functionality (SQL Server) or SOUNDS LIKE (MySQL) if it is available with the SQL engine you are using.
Related
IS there a way to find common misspellings in sql server?For example, if I had a legitimate search… like “Forclosure” instead of “Foreclosure”, is there a way to find the connection within a table without doing LIKE or CASE WHEN?
I’m not 100% sure it will work but you can try experimenting with SOUNDEX function and see if it can give you any results.
SOUNDEX converts an alphanumeric string to a four-character code that is based on how the string sounds when spoken.
For example:
SELECT SOUNDEX ('Smith'), SOUNDEX ('Smythe');
Gives same results for both
Oracle has a function called translate that can be used to replace individual characters of the string by others, in the same order that they appear. It is different than the replace function, which replaces the entire second argument occurence by the entire third argument.
translate('1tech23', '123', '456'); --would return '4tech56'
translate('222tech', '2ec', '3it'); --would return '333tith'
I need this to implement a search on a SQLite database ignoring accents (brazilian portuguese language) on my query string. The data in the table that will be queried could be with or without accents, so, depending on how the user type the query string, the results would be different.
Example:
Searching for "maçã", the user could type "maca", "maça", "macã" or "maçã", and the data in the table could also be in one of the four possibilities.
Using oracle, I would only use this:
Select Name, Id
From Fruits
Where Translate(Name, 'ãç','ac') = Translate(:QueryString, 'ãç','ac')
... and these other character substitutions:
áéíóúÁÉÍÓÚàèìòùÀÈÌÒÙãõÃÕäëïöüÄËÏÖÜâêîôûÂÊÎÔÛñÑçÇ
by:
aeiouAEIOUaeiouAEIOUaoAOaeiouAEIOUaeiouAEIOUnNcC
Of course I could nest several calls to Replace, but this wouldn't be a good choice.
Thanks in advance by some help.
Open-source Oracle functions for SQLite have been written at Kansas State University. They include translate() (full UTF-8 support, by the way) and can be found here.
I don't believe there is anything in sqlite that will translate text in a single pass as you describe.
This wouldn't be difficult to implement as a user defined function however. Here is a decent starting reference.
I used replace
REPLACE(string,pattern,replacement)
https://www.sqlitetutorial.net/sqlite-replace-function/
I have a database that has around 10k records and some of them contain HTML characters which I would like to replace.
For example I can find all occurrences:
SELECT * FROM TABLE
WHERE TEXTFIELD LIKE '%/%'
the original string example:
this is the cool mega string that contains /
how to replace all / with / ?
The end result should be:
this is the cool mega string that contains /
If you want to replace a specific string with another string or transformation of that string, you could use the "replace" function in postgresql. For instance, to replace all occurances of "cat" with "dog" in the column "myfield", you would do:
UPDATE tablename
SET myfield = replace(myfield,"cat", "dog")
You could add a WHERE clause or any other logic as you see fit.
Alternatively, if you are trying to convert HTML entities, ASCII characters, or between various encoding schemes, postgre has functions for that as well. Postgresql String Functions.
The answer given by #davesnitty will work, but you need to think very carefully about whether the text pattern you're replacing could appear embedded in a longer pattern you don't want to modify. Otherwise you'll find someone's nooking a fire, and that's just weird.
If possible, use a suitable dedicated tool for what you're un-escaping. Got URLEncoded text? use a url decoder. Got XML entities? Process them though an XSLT stylesheet in text mode output. etc. These are usually safer for your data than hacking it with find-and-replace, in that find and replace often has unfortunate side effects if not applied very carefully, as noted above.
It's possible you may want to use a regular expression. They are not a universal solution to all problems but are really handy for some jobs.
If you want to unconditionally replace all instances of "/" with "/", you don't need a regexp.
If you want to replace "/" but not "Ǘ", you might need a regexp, because you can do things like match only whole words, match various patterns, specify min/max runs of digits, etc.
In the PostgreSQL string functions and operators documentation you'll find the regexp_replace function, which will let you apply a regexp during an UPDATE statement.
To be able to say much more I'd need to know what your real data is and what you're really trying to do.
If you don't have postgres, you can export all database to a sql file, replace your string with a text editor and delete your db on your host, and re-import your new db
PS: be careful
I have a SQLite database with a word list. In a table there is a word list that includes the word "você". This word has this representation in unicode "voc\U00ea".
I've found out that the same word can have the following representation with the same visual output:
"voc\U00ea",
"voce\U0302"
When I query my db using the second representation it returns blank. Does anyone know a way for the query work using both representations without duplicating the records in the table?
Thanks,
Miguel
These two forms are known as nfc(normal form composed) and nfd("normal form decomposed"). The letter \U0302 is known as a "combining circumflex", which modifies a preceding letter.
To cope with this situation, do the following:
Pick a normalization. Usually choosing nfc is a good idea. (Although iOS/OS X file system uses nfd.)
Before putting a string into the database, always normalize. In iOS, you can use precomposedStringWithCanonicalMapping or precomosedStringWithCompatibilityMapping. To understand the difference between canonical and compatibility mappings, see this description.
Before performing a query, always normalize the query to the same normal form.
I am trying to index in Lucene a field that could have RDF literal in different languages.
Most of the approaches I have seen so far are:
Use a single index, where each document has a field per each language it uses, or
Use M indexes, M being the number of languages in the corpus.
Lucene 2.9+ has a feature called Payload that allows to attach attributes to term. Is anyone use this mechanism to store language (or other attributes such as datatypes) information ? How is performance compared to the two other approaches ? Any pointer on source code showing how it is done would help. Thanks.
It depends.
Do you want to allow something like: "Search all english text for 'foo'"? If so, then you will need one field per language.
Or do you want "Search all text for 'foo' and present the user with which language the match was found in?" If this is what you want, then either payloads or separate fields will work.
An alternative way to do it is to index all your text in one field, then have another field saying the language of the document. (Assuming each document is in a single language.) Then your search would be something like +text:foo +language:english.
In terms of efficiency: you probably want to avoid payloads, since you would have to repeat the name of the language for every term, and you can't search based on payloads (at least not easily).
so basically lucene is a ranking algorithm, it just looks at strings and compares them to other string. they can be encoded in different character encodings but their similarity is the same non the less. Just make sure you load the SnowBallAnalyzer with the supported langugage stemmer and you should get results. Like say Spanish or Chinese