Advanced text searching

• Download source - 3 Kb

Introduction

When dealing with large amounts of data (either a website or in other specialized programs) it is very good to offer the user the possibility to narrow down the subset of data he needs to look at. The only way to avoid this and still not confuse the user is to partition the data in small chunks (such as presenting an address book by the first letter of each name). This isn't always possible so a search system can be helpful.

Description of the solution

I've developed a class which implements a simple query language. The syntax is presented below by some elementary examples and their interpretation:

abc cde fgh	The character sequences "abc", "cde" and "fgh" must be all present in the string, however their order doesn't matter (so cdeabcfgh is a valid sequence for this rule). Every blank character is considered a separator (actually I use the Char.IsSeparator method to determine if a character is separator)
"abc cde"""	The sequence abc cde" must be present in the string. Observe that in order to include the quote character ("), you must double it. Besides of grouping words the quote character is also useful in inserting special characters ( (, ), [, ]).
[a b c]	The string must contain at least one of the elements (OR).
(a b c)	The string must contain all the elements (the order is not important!)
[(a b) (c "[")]	The parenthesizes can be nested. The expression means: the string either contains a AND b (the order is not important) OR it contains c and [ (observe how the special character [ needs to be between quotes)

A more formal description of the syntax would be:

queryString ::= <queryPart>[<empty characters><queryString>]
queryPart ::= <keyword>|<orComposition>|<andComposition>
orComposition ::= [<queryPart>{<empty characters><queryPart>}]
andComposition ::= (<queryPart>{<empty characters><queryPart>})
emptyCharacters ::= as defined by the Char.IsSeparator method
keyword ::= "<characters>"|<non empty, non special characters>

This syntax may appear a little strange and has the drawback that it's non-standard. Because of this it might require a little getting used to from the users perspective. I don't recommend using it as the default search language on a website, however it can be presented as an alternative (for advanced searching for example). Also, I found it to be a little more obvious than the normal approach (if you write "Anna and Betty", what do you want to search for: for the exact string or for a string containing both words?) and it's simpler to implement.

To search using this code, you must do the following simple steps:

1. Create a search tree and remember its root
   Dim x As SearchTreeNode = SearchTreeNode.ContructSearchTree("[a b] [c d]")
2. Now call the .MatchesString(string to match, root) function for each element you want to verify. It returns true or false depending whether the supplied string matches the rule or not.

Implementation details

The SearchTreeNode is a node of a multi-way tree which remembers it's first child and next sibling as shown below:

The nodes of the tree can be of three types: AND nodes, OR nodes and KEYWORD nodes (or leafs). An AND node matches a string if all of its children match the string. An OR node matches a string if at least one of its children match the string. A keyword node matches a string if it is contained within the string. For example the following rule is translated to the tree seen below: "[(a b) (c d)] e"

You probably observed the extra level with an AND node. This is done to avoid treating a particular case in the routine and has no impact on the run. The above tree is constructed with the ConstructSearchTree static (shared) method. This method throws an exception if it's unable to construct the tree. The message of the exception tries to hint to a possible cause of the syntax error.

The MatchesString static method needs a string which it should compare to the rule and the root of the tree. It then performs a recursive walk of the tree using the MatchString private method and returns true or false.

Some notes on the performance

I'm actually quite satisfied with the performance of the code. It can search 10000 strings containing 100 "x" characters with the rule "[a b] [c d]" in 16 milliseconds. This represents the worst case scenario for the leafs (where the actual searching is done using the indexOf method), because no match is found. Also in this case the full tree must be walked, so with a string that actually matches the rule the search could be aborted much sooner. The tests were done on an AMD Sempron 2200+ machine with 256 MB RAM running WinXP SP2.

One possibility to speed up the matching would be to implement a faster string matching algorithm like Boyer-Moore or TurboSearch. The helper tables for these could be generated during the creation of the tree. The problem is that the strings are Unicode encoded and thus a helper table (for the TurboSearch algorithm for example) could reach very big sizes (64K elements). Also there is the fact that the keywords will probably be rather small (5-10 charactes) and the indexOf method performs quite well.

Additional features

There are two methods: one for building the filter part of an SQL query from the tree and one to build a regular expression from it. They are both static (shared) methods and their syntax is as follows: BuildSQLQuery(field name, root element) returns a string that looks like: (<field name> LIKE "%keyword%" AND ... In the keyword all occurrences of the quote character (") are replaced with backslash quote (\").

The second method builds a regular expression from the search tree: BuildRegEx(root element) and returns the regular expression as string. One limitation of this method is that it doesn't generate a regular expression that handles all possible orders or appearances of the supplied argument, just the one defined in the tree. I used this approach because given the limitations of regular expressions a full implementation would generate very long and complicated expressions (one would have to explicitly specify all possible permutations in the expression).

Final notes

I hope that you found this article useful. If you found this article stupid, annoying, incorrect, etc. express this fact by rating the article as you see fit. However I would appreciate if you would leave a comment explaining your reasons so that I can (hopefully) learn from my mistakes.

You must Sign In to use this message board.

FAQ  Noise Tolerance Very HighHighMediumLowVery Low  Search Messages    Layout Expand All MessagesThread ViewNo Javascript (slow)No Javascript Preview    Per page 102550

Msgs 1 to 17 of 17 (Total in Forum: 17) (Refresh) FirstPrevNext

Subject   Author  Date  advanced search aisyahhumairah 18:10 8 Oct '07   hi.. can give me a source code using vb .net to create advanced search like all words, anywords, and exact word ? plsss humairah Sign In·View Thread·PermaLink Seacrching document or speific word in windows Drives kk_upadhyay 22:08 7 Oct '07   hi , i am making a window application in vb.net.I want to search a prticular word or a document through this application? how can i do this .suggest me the appropiate solution for this..... hhhhhhhhhh Sign In·View Thread·PermaLink Binding Data to DataGrid from database? khmays2012 4:10 28 Mar '07   I want bind Data from Table in Database by using code or New class How am I doing That in VB.net Or C#? Please Guide me? Sign In·View Thread·PermaLink 1.50/5 (2 votes) Binding Data to DataGrid from database? khmays2012 4:09 28 Mar '07   I want bind Data from Table in Database by using code or New class How am I doing That? Please Guide me? Sign In·View Thread·PermaLink 2.00/5 (1 vote) i am a chinese mqcan0116 1:09 23 Jan '07   there is two reasons coming here for me . one,i want to practise my english the other is i want to prove my program level and make some foreign friends! thanks chinese! Sign In·View Thread·PermaLink 1.00/5 (1 vote) .net frame lock Hiran Rajapaksha 10:19 19 Dec '06   hwo i lock the Form frame in vb.net 2003 Sign In·View Thread·PermaLink Can the approach be used for normal (OR/AND) syntax? Michael Freidgeim 19:43 20 Jun '06   Can you give any ideas, what to change in this code to implement normal(Google/MSDN help style) syntax? E.g. queries like "Anna AND Betty" ,"Michael OR Peter" etc. Michael Freidgeim. Blog: http://geekswithblogs.net/mnf/ Sign In·View Thread·PermaLink Re: Can the approach be used for normal (OR/AND) syntax? Cd-MaN 21:32 20 Jun '06   Sorry, this was written as a module for a project which has been finished in the meantime and I don't have time to return to it. Sign In·View Thread·PermaLink Nicely Done computerguru92382 17:36 1 Mar '06   Very nicely done article. Paul Sign In·View Thread·PermaLink C# Version Neo Fount 0:50 17 Jan '06   using System.Text; using System.Diagnostics; using System.Collections; public enum SearchTreeNodeType { And_Node = 0, Or_Node = 1, Keyword_Node = 2 } public class SearchTreeNode { #region varible private SearchTreeNodeType NodeType; private SearchTreeNode FirstChild; private SearchTreeNode NextSybling; private string Keyword; #endregion public static bool MatchesString(string query, SearchTreeNode root) { if (root == null) { return false; } return root.MatchString(ref query); } /// ///Build a SQL query for the given node and a given file name. The result will look like: ///(fieldName="a" AND fieldName="b") OR fieldName="c" /// /// /// /// public static string BuildSQLQuery(string FieldName, SearchTreeNode root) { if (root == null) { return ""; } return root.CreateSQLQuery(FieldName); } /// ///Build a regular expression from the tree. One limitation of this approach is that ///the generated regex string doesn't take in account all the possible order of the query ///parts (or else the query would get very long) /// /// /// public static string BuildRegEx(SearchTreeNode root) { if (root == null) { return ""; } return root.CreateRegEx(); } /// /// The function which construct the tree for a given query or throws an exception with the error /// /// /// public static SearchTreeNode ContructSearchTree(string keywords) { if (keywords == null) { return null; } keywords = keywords.Trim(); if (keywords == "") { return null; } ArrayList words = SplitQueryString(keywords); return BuildSearchList(words, SearchTreeNodeType.And_Node); } /// /// For debugging. dump the tree to the tracer window /// public void Dump() { if (NodeType == SearchTreeNodeType.And_Node) { Trace.WriteLine("Node type: AND"); } else if (NodeType == SearchTreeNodeType.Keyword_Node) { Trace.WriteLine("Node type: Keyword Text: " + Keyword); } else if (NodeType == SearchTreeNodeType.Or_Node) { Trace.WriteLine("Node type: OR"); } Trace.Indent(); SearchTreeNode s = this.FirstChild; while (!(s == null)) { s.Dump(); s = s.NextSybling; } Trace.Unindent(); } #region Private Method private bool MatchString(ref string query) { // 'see what type of node we are in and behave accordingly if (NodeType == SearchTreeNodeType.And_Node) { SearchTreeNode sNode = FirstChild; while (!(sNode == null)) { //'if at least one child is false, return false if (sNode.MatchString(ref query) == false) { return false; } sNode = sNode.NextSybling; } //if all the children are true, return true return true; } else if (NodeType == SearchTreeNodeType.Or_Node) { SearchTreeNode sNode = FirstChild; while (!(sNode == null)) { //'if at least one child is true, return true if (sNode.MatchString(ref query) == true) { return true; } sNode = sNode.NextSybling; } //if all the children are false, return false return false; } else if (NodeType == SearchTreeNodeType.Keyword_Node) { //we are in a leaf to match the keyword return query.IndexOf(Keyword) > -1; } else { throw new System.Exception("Unknow SearchTreeNode type!"); } } private string CreateSQLQuery(string fieldName) { if (NodeType == SearchTreeNodeType.And_Node) { //if we have a single child node, return the string generated by it, //as the final value depends only by it if (FirstChild == null) { throw new System.Exception("Illegal SearchTreeNode!"); } if (FirstChild.NextSybling == null) { return FirstChild.CreateSQLQuery(fieldName); } StringBuilder res = new StringBuilder(); SearchTreeNode currChild = FirstChild; while (!(currChild == null)) { if (res.Length > 0) { res.Append(" AND "); } res.Append(currChild.CreateSQLQuery(fieldName)); currChild = currChild.NextSybling; } return "(" + res.ToString() + ")"; } else if (NodeType == SearchTreeNodeType.Or_Node) { //if we have a single child node, return the string generated by it, //as the final value depends only by it if (FirstChild == null) { throw new System.Exception("Illegal SearchTreeNode!"); } if (FirstChild.NextSybling == null) { return FirstChild.CreateSQLQuery(fieldName); } StringBuilder res = new StringBuilder(); SearchTreeNode currChild = FirstChild; while (!(currChild == null)) { if (res.Length > 0) { res.Append(" OR "); } res.Append(currChild.CreateSQLQuery(fieldName)); currChild = currChild.NextSybling; } return "(" + res.ToString() + ")"; } else if (NodeType == SearchTreeNodeType.Keyword_Node) { string res = Keyword; //replace " with \" and construct the string res = "(" + fieldName + " LIKE \"%" + res.Replace("\"", "\\\"") + "%\")"; return res; } else { throw new System.Exception("Unknow SearchTreeNode type!"); } } private string CreateRegEx() { if (NodeType == SearchTreeNodeType.And_Node) { //if we have a single child node, return the string generated by it, //as the final value depends only by it if (FirstChild == null) { throw new System.Exception("Illegal SearchTreeNode!"); } if (FirstChild.NextSybling == null) { return FirstChild.CreateRegEx(); } StringBuilder res = new StringBuilder(); SearchTreeNode currChild = FirstChild; while (!(currChild == null)) { //to optimize the expression a little and avoid the superflous .*.* pattern if (res.Length > 0) { if (res[res.Length - 1] == '*') { res.Length = res.Length - 2; } } res.Append(currChild.CreateRegEx()); currChild = currChild.NextSybling; } return res.ToString(); } else if (NodeType == SearchTreeNodeType.Or_Node) { //if we have a single child node, return the string generated by it, //as the final value depends only by it if (FirstChild == null) { throw new System.Exception("Illegal SearchTreeNode!"); } if (FirstChild.NextSybling == null) { return FirstChild.CreateRegEx(); } StringBuilder res = new StringBuilder(); SearchTreeNode currChild = FirstChild; while (!(currChild == null)) { if (res.Length > 0) { res.Append("|"); } res.Append(currChild.CreateRegEx()); currChild = currChild.NextSybling; } return "(" + res.ToString() + ")"; } else if (NodeType == SearchTreeNodeType.Keyword_Node) { //replace every special character by \ //it is important to start with the \ character! //The special characters are: \ | ( ) [ { ^ $ * + ? . string res = Keyword; res = res.Replace("\\", "\\\\"); res = res.Replace("|", "\\|"); res = res.Replace("(", "\\("); res = res.Replace(")", "\\)"); res = res.Replace("[", "\\["); res = res.Replace("]", "\\]"); res = res.Replace("{", "\\{"); res = res.Replace("}", "\\}"); res = res.Replace("^", "\\^"); res = res.Replace("$", "\\$"); res = res.Replace("*", "\\*"); res = res.Replace("+", "\\+"); res = res.Replace("?", "\\?"); res = res.Replace(".", "\\."); //the final partial regex is like this: // return ".*" + res + ".*"; } else { throw new System.Exception("Unknow SearchTreeNode type!"); } } /// /// This function splits the query string in parts and throws an exception if it's not possible /// /// /// private static ArrayList SplitQueryString(string keywords) { StringBuilder buffer = new StringBuilder(); bool inQuote = false; int parantCount = 0; int straightParantCount = 0; ArrayList words = new ArrayList(); int i = 0; while (i < keywords.Length) { if (keywords[i] == '(') { buffer.Append("("); if (!(inQuote)) { parantCount = parantCount + 1; } } else if (keywords[i] == ')') { buffer.Append(")"); if (!(inQuote)) { parantCount = parantCount - 1; } if (parantCount < 0) { throw new System.Exception("Closing ) without opened"); } if ((!(inQuote)) & (parantCount == 0) & (straightParantCount == 0)) { if ((buffer.Length > 0) & (buffer.ToString() != "()")) { words.Add(buffer.ToString().Trim()); } buffer.Length = 0; } } else if (keywords[i] == '[') { buffer.Append("["); if (!(inQuote)) { straightParantCount = straightParantCount + 1; } } else if (keywords[i] == ']') { buffer.Append("]"); if (!(inQuote)) { straightParantCount = straightParantCount - 1; } if (straightParantCount < 0) { throw new System.Exception("Closing ] without opened"); } if ((!(inQuote)) & (parantCount == 0) & (straightParantCount == 0)) { if ((buffer.Length > 0) & (buffer.ToString() != "[]")) { words.Add(buffer.ToString().Trim()); } buffer.Length = 0; } } else if (keywords[i] == '"') { bool doubleQuote = false; if (i < keywords.Length - 1) { if (keywords[i+1] == '"') { doubleQuote = true; } } if (doubleQuote) { buffer.Append("\""); i = i + 1; } else { inQuote = !(inQuote); if ((!(inQuote)) & (parantCount == 0) & (straightParantCount == 0)) { if ((buffer.Length > 0) & (buffer.ToString() != "[]")) { words.Add(buffer.ToString().Trim()); } buffer.Length = 0; } } } else if (char.IsSeparator(keywords[i])) { if (inQuote | (parantCount > 0) | (straightParantCount > 0)) { buffer.Append(keywords[i]); } else if (buffer.Length > 0) { words.Add(buffer.ToString()); buffer.Length = 0; } } else { buffer.Append(keywords[i]); } i = i + 1; } if ((!(inQuote)) & (parantCount == 0) & (straightParantCount == 0)) { if (buffer.Length > 0) { words.Add(buffer.ToString().Trim()); } buffer.Length = 0; } else { throw new System.Exception("Error in the query string. Possible paranthesis or quote mismatch"); } return words; } private static SearchTreeNode BuildSearchList(ArrayList keywords, SearchTreeNodeType ParentKind) { SearchTreeNode res = new SearchTreeNode(); if ((keywords.Count == 0)) { throw new System.Exception("Unknown error during building the query tree"); } //Handle the final item if ((keywords.Count == 1) & (((string)(keywords[0]))[0] != '(') & (((string)(keywords[0]))[0] != '[')) { res.NodeType = SearchTreeNodeType.Keyword_Node; res.Keyword = (string)keywords[0]; return res; } //Add all the elements res.NodeType = ParentKind; SearchTreeNode lastChild = null; foreach (string _Word in keywords) { SearchTreeNode newNode; ArrayList subChildWords; if (_Word.Length > 2) { if (_Word[0] == '(') { //_Word = _Word.Substring(1, _Word.Length - 2); subChildWords = SplitQueryString(_Word.Substring(1, _Word.Length - 2)); newNode = BuildSearchList(subChildWords, SearchTreeNodeType.And_Node); } else if (_Word[0] == '[') { //_Word = _Word.Substring(1, _Word.Length - 2); subChildWords = SplitQueryString(_Word.Substring(1, _Word.Length - 2)); newNode = BuildSearchList(subChildWords, SearchTreeNodeType.Or_Node); } else { subChildWords = SplitQueryString(_Word); newNode = BuildSearchList(subChildWords, SearchTreeNodeType.And_Node); } } else { subChildWords = SplitQueryString(_Word); newNode = BuildSearchList(subChildWords, SearchTreeNodeType.And_Node); } if (lastChild == null) { res.FirstChild = newNode; lastChild = newNode; } else { lastChild.NextSybling = newNode; lastChild = newNode; } } if (res.FirstChild == null) { throw new System.Exception("Unknown error during building the query tree"); } //if we only created one subitem, return it instead if (res.FirstChild.NextSybling == null) { return res.FirstChild; } return res; } #endregion } http://hardrock.cnblogs.com Sign In·View Thread·PermaLink C++ re-writing + NOT operator + optimisation christophe.carpentier@alogie.fr 2:29 17 May '05   I have re-written the same program using C++ instruction with these changes. -a ! (NOT) operator added (apply to both words and sub-expression) -no extra level in the tree(the code detect naturally extra level and remove them) -the search string is not reallocated in every node but instead nodes point directly to the string -the program is used in a database application where it is supposed to resolve millions lines with great performance -possible improvement : reorder the nodes in each AND and OR nodes to place first the child nodes whose cost are minimal class AnaWord { public: AnaWord(); ~AnaWord(); struct AnaNode { int type; char*c; int c_len; struct AnaNode*next; struct AnaNode*up; }; enum { ANA_AND=1, ANA_OR=2, ANA_WORD=3, ANA_WORD_NOT=4 }; private: struct AnaNode*alloc_node(); struct AnaNode*alloc_node_copy(struct AnaNode*e); void apply_not(AnaNode*root); int build_analsys(struct AnaNode*root); public: void free_node(void*root); void * build_analsys(char*c,int c_len); BOOL is_matching(void*root,char*c); void dump(void*root,int indent); }; AnaWord::AnaWord() { } AnaWord::~AnaWord() { } void AnaWord::free_node(void*root) { struct AnaNode*e=((struct AnaNode*)root)->up; while(e) { struct AnaNode*tmp=e->next; free_node(e); e=tmp; } my_free(root); } AnaWord::AnaNode*AnaWord::alloc_node() { struct AnaNode* e= (struct AnaNode*)my_malloc(sizeof(struct AnaNode)); e->next=e->up=0; return e; } AnaWord::AnaNode*AnaWord::alloc_node_copy(struct AnaNode*e) { struct AnaNode*a=alloc_node(); memcpy(a,e,sizeof(struct AnaNode)); return a; } void * AnaWord::build_analsys(char*c,int c_len) { struct AnaNode root; root.c=c; root.c_len=c_len; root.up=0; root.next=0; build_analsys(&root); if(!root.up)return 0; if(!root.up->next)return root.up; root.type=ANA_AND; return alloc_node_copy(&root); } int AnaWord::build_analsys(AnaNode*root) { BOOL b_not=FALSE; char*c=root->c; int c_len=root->c_len; AnaNode*head=0; for(; { while(c_len&&*c==' '){c++;c_len--;} if(!c_len) { break; } if(*c=='!') { c++;c_len--; b_not=!b_not; continue; } if(*c=='('||*c=='[') { AnaNode tmp; tmp.up=0; tmp.next=0; if(*c=='(')tmp.type=ANA_AND;else tmp.type=ANA_OR; c++;c_len--; tmp.c=c; tmp.c_len=c_len; c_len=build_analsys(&tmp); c+=tmp.c_len-c_len; if(!tmp.up){b_not=FALSE;continue;} AnaNode*e; if(tmp.up->next)e=alloc_node_copy(&tmp);else e=tmp.up; if(b_not){apply_not(e);b_not=FALSE;} if(head){head->next=e;head=e;}else root->up=head=e; continue; } if(*c==')'||*c==']') { c++;c_len--; break; } if(*c=='"') { c++;c_len--; struct AnaNode*e=alloc_node(); if(head){head->next=e;head=e;}else root->up=head=e; if(b_not){e->type=ANA_WORD_NOT;b_not=FALSE;}else e->type=ANA_WORD; e->c=c; while(c_len) { if(*c=='\\'&&c_len>1){c+=2;c_len-=2;} if(*c=='"')break; c++;c_len--; } e->c_len=c-e->c; if(c_len&&*c=='"'){c++;c_len--;} continue; } struct AnaNode*e=alloc_node(); if(head){head->next=e;head=e;}else root->up=head=e; if(b_not){e->type=ANA_WORD_NOT;b_not=FALSE;}else e->type=ANA_WORD; e->c=c; while(c_len&&*c!='('&&*c!='['&&*c!=')'&&*c!=']'&&*c!=' '&&*c!='"'){c++;c_len--;} e->c_len=c-e->c; } return c_len; } void AnaWord::apply_not(AnaNode*root) { switch(root->type) { case ANA_WORD: root->type=ANA_WORD_NOT; return; case ANA_WORD_NOT: root->type=ANA_WORD; return; case ANA_AND: root->type=ANA_OR; { struct AnaNode*e=root->up; while(e){apply_not(e);e=e->next;} } return; case ANA_OR: root->type=ANA_AND; { struct AnaNode*e=root->up; while(e){apply_not(e);e=e->next;} } return; } } void AnaWord::dump(void*p,int indent) { if(!p) { OutputDebugString("NULL NODE\n"); return; } struct AnaNode*root=(struct AnaNode*)p; CString s1; CString s2; char*c=s1.GetBufferSetLength(indent); int i; for(i=0;i if(root->type==ANA_WORD||root->type==ANA_WORD_NOT) { CString s; char*c=s.GetBufferSetLength(root->c_len); memcpy(c,root->c,root->c_len); s+="\n"; if(root->type==ANA_WORD)OutputDebugString(s1+s); else OutputDebugString(s1+"!"+s); return; } switch(root->type) { case ANA_AND: s2="AND"; break; case ANA_OR: s2="OR"; break; } OutputDebugString(s1+s2+"\n"); AnaNode*e=root->up; while(e) { dump(e,indent+1); e=e->next; } } BOOL AnaWord::is_matching(void*p,char*text) { if(!p) { return TRUE; } if(((AnaNode*)p)->type==ANA_WORD||((AnaNode*)p)->type==ANA_WORD_NOT) { int c_len=((AnaNode*)p)->c_len; char*c=((AnaNode*)p)->c; char*d=text; for(; { int len=c_len; char*e=c; char*f=d; for(; { if(!len) { if(((AnaNode*)p)->type==ANA_WORD)return TRUE; return FALSE; } if(!*f) { if(((AnaNode*)p)->type==ANA_WORD)return FALSE; return TRUE; } char a=*e; char b=*f; if(a=='\\'&&len>1) { len--;e++; a=*e; } if(a>='A'&&a<='Z')a+=0x20; if(b>='A'&&b<='Z')b+=0x20; if(a!=b)break; e++;f++;len--; } d++; } } int type=((AnaNode*)p)->type; AnaNode*e=((AnaNode*)p)->up; while(e) { BOOL b=check(e,text); e=e->next; if(!b&&type==ANA_AND)return FALSE; if(b&&type==ANA_OR)return TRUE; } if(type==ANA_AND)return TRUE; if(type==ANA_OR)return FALSE; return FALSE; } Sign In·View Thread·PermaLink 5.00/5 (2 votes) Using this to search NTEXT column in SQL Server 2000 Ravidhu 12:10 3 Apr '05   Hi, I am trying to use your approach in my application where I have to search data stored in NTEXT database columns using keywords, AND, OR etc. For example the Admin can search for 1) CRM and Oracle 2) CRM OR PEOPLESOFT The data is stored in 5 seperate tables and each table has their work experience, technology skills, resume etc. Since this does not have a sample project. I'm not sure how to use this. I have created a View in the database that has all the columns where the data is stored in NTEXT columns. Can you explain with the above example how I can proceed?? I would really appreciate your help. Regards -Raman Sign In·View Thread·PermaLink Why not just use Regular Expressions? TLWallace.NET 6:22 12 Mar '05   Why not use regular expressions? Regular expressions are widely becoming the defacto standard. And there is plenty of information on how to use them. I'm just curious, are you introducing an alternative to regular expressions or a quantative improvement to such? TL Wallace Sign In·View Thread·PermaLink Re: Why not just use Regular Expressions? Cd-MaN 9:58 12 Mar '05   Regular expressions are very nice things indeed and I use them extensively. However they have a major drawback: AFAIK they don't offer the possibility to find the strings in any order. For example A.*B will match the string ACB but not the string BCA. My method uses the .IndexOf separately on each token, so there is no restriction placed on the order of the strings. I find this more usefull in the case when I have to search a website for example. P.S. Thank you for reading my article. Sign In·View Thread·PermaLink 2.75/5 (4 votes) Re: Why not just use Regular Expressions? danielcomer 7:43 19 Jan '06   - A.*B will match the string ACB but not the string BCA. AFAIK, there are VERY few languages (collections of words) that a regular expression could not define. Sign In·View Thread·PermaLink 1.00/5 (1 vote) Re: Why not just use Regular Expressions? Polymorpher 4:52 3 Feb '06   You have a good point but this does seem to be easyer for the end user who may not understand regular expresions....smaller lerning curve Pablo Sign In·View Thread·PermaLink 2.00/5 (1 vote) Re: Why not just use Regular Expressions? Shane Story 9:44 27 Dec '06   Wouldn't the solution be to translate English or whatever into regular expressions for the user? Shane Sign In·View Thread·PermaLink

Last Visit: Wednesday, October 15, 2008 12:32 PM     Last Update:Wednesday, October 15, 2008 12:32 PM