/// <summary>
/// Initializes a new instance of the Grep class.
/// </summary>
/// <history>
/// [Curtis_Beard] 07/12/2006 Created
/// [Andrew_Radford] 13/08/2009 Added Const. dependency on ISearchSpec, IFileFilterSpec
/// [Curtis_Beard] 05/28/2015 FIX: 69, Created for speed improvements for encoding detection
/// </history>
public Grep(ISearchSpec searchSpec, IFileFilterSpec filterSpec)
{
if (searchSpec.EncodingDetectionOptions == null)
{
throw new ArgumentNullException("ISearchSpec.EncodingDetectionOptions", "EncodingDetectionOptions must not be null");
}
SearchSpec = searchSpec;
FileFilterSpec = filterSpec;
MatchResults = new List<MatchResult>();
if (FileFilterSpec.FilterItems != null)
{
// get first file->minimum hit count filter (should only be 1)
var fileCountFilter = (from f in FileFilterSpec.FilterItems where f.FilterType.Category == FilterType.Categories.File && f.FilterType.SubCategory == FilterType.SubCategories.MinimumHitCount select f).FirstOrDefault();
if (fileCountFilter != null)
{
int.TryParse(fileCountFilter.Value, out userFilterCount);
}
}
if (SearchSpec.EncodingDetectionOptions.DetectFileEncoding && SearchSpec.EncodingDetectionOptions.UseEncodingCache)
{
EncodingCache.Instance.Load(SearchSpec.EncodingDetectionOptions.PerformanceSetting);
}
}
/// <summary>
/// Searches the given file for the given search text.
/// </summary>
/// <param name="file">FileInfo object</param>
/// <param name="searchSpec">ISearchSpec interface value</param>
/// <param name="ex">Exception holder if error occurs</param>
/// <returns>Hitobject containing grep results, null if on error</returns>
public MatchResult Grep(FileInfo file, ISearchSpec searchSpec, ref Exception ex)
{
// initialize Exception object to null
ex = null;
MatchResult match = null;
if (Parser.IsParseable(file.FullName))
{
string fileContent = Parser.Parse(file.FullName);
if (!string.IsNullOrEmpty(fileContent))
{
string[] lines = fileContent.Split(new char[] { '\n', '\r' });
for (int i = 0; i < lines.Length; i++)
{
string line = lines[i];
int posInStr = -1;
Regex reg = null;
MatchCollection regCol = null;
if (searchSpec.UseRegularExpressions)
{
string pattern = string.Format("{0}{1}{0}", searchSpec.UseWholeWordMatching ? "\\b" : string.Empty, searchSpec.SearchText);
RegexOptions options = searchSpec.UseCaseSensitivity ? RegexOptions.None : RegexOptions.IgnoreCase;
reg = new Regex(pattern, options);
regCol = reg.Matches(line);
if (regCol.Count > 0)
{
posInStr = 1;
}
}
else
{
// If we are looking for whole worlds only, perform the check.
if (searchSpec.UseWholeWordMatching)
{
reg = new Regex("\\b" + Regex.Escape(searchSpec.SearchText) + "\\b", searchSpec.UseCaseSensitivity ? RegexOptions.None : RegexOptions.IgnoreCase);
// if match is found, also check against our internal line hit count method to be sure they are in sync
Match mtc = reg.Match(line);
if (mtc != null && mtc.Success && libbSearch.Grep.RetrieveLineMatches(line, searchSpec).Count > 0)
{
posInStr = mtc.Index;
}
}
else
{
posInStr = line.IndexOf(searchSpec.SearchText, searchSpec.UseCaseSensitivity ? StringComparison.Ordinal : StringComparison.OrdinalIgnoreCase);
}
}
if (posInStr > -1)
{
if (match == null)
{
match = new MatchResult(file);
// found hit in file so just return
if (searchSpec.ReturnOnlyFileNames)
{
break;
}
}
var matchLineFound = new MatchResultLine()
{
Line = line, LineNumber = -1, HasMatch = true
};
if (searchSpec.UseRegularExpressions)
{
posInStr = regCol[0].Index;
match.SetHitCount(regCol.Count);
foreach (Match regExMatch in regCol)
{
matchLineFound.Matches.Add(new MatchResultLineMatch(regExMatch.Index, regExMatch.Length));
}
}
else
{
var lineMatches = libbSearch.Grep.RetrieveLineMatches(line, searchSpec);
match.SetHitCount(lineMatches.Count);
matchLineFound.Matches = lineMatches;
}
matchLineFound.ColumnNumber = 1;
match.Matches.Add(matchLineFound);
}
}
}
}
return(match);
}
/// <summary>
/// Searches the given file for the given search text.
/// </summary>
/// <param name="file">FileInfo object</param>
/// <param name="searchSpec">ISearchSpec interface value</param>
/// <param name="ex">Exception holder if error occurs</param>
/// <returns>Hitobject containing grep results, null if on error</returns>
/// <history>
/// [Curtis_Beard] 07/28/2006 Created
/// [Curtis_Beard] 05/25/2007 ADD: support for Exception object
/// [Curtis_Beard] 03/31/2015 CHG: rework Grep/Matches
/// </history>
public MatchResult Grep(FileInfo file, ISearchSpec searchSpec, ref Exception ex)
{
// initialize Exception object to null
ex = null;
if (IsAvailable && IsUsable)
{
try
{
if (file.Exists)
{
//const int MARGINSIZE = 4;
int count = 0;
MatchResult match = null;
int prevLine = 0;
int prevPage = 0;
//string _spacer = new string(' ', MARGINSIZE);
//string _contextSpacer = string.Empty;
//if (searchSpec.ContextLines > 0)
//{
// _contextSpacer = new string(' ', MARGINSIZE);
// _spacer = _contextSpacer.Substring(_contextSpacer.Length - MARGINSIZE - 2) + "> ";
//}
//else
// _spacer = new string(' ', MARGINSIZE);
// Open a given Word document as readonly
// Note: Word 2003+ requires write mode since reading mode doesn't allow use of home/end keys to select text)
object appversion = __WordApplication.GetType().InvokeMember("Version", BindingFlags.GetProperty, null, __WordApplication, null);
double version = 0;
double.TryParse(appversion.ToString(), out version);
bool useReadOnly = version >= 12.00 ? false : true;
object wordDocument = OpenDocument(file.FullName, useReadOnly);
// Get Selection Property
__WordSelection = __WordApplication.GetType().InvokeMember("Selection", BindingFlags.GetProperty,
null, __WordApplication, null);
// create range and find objects
object range = GetProperty(wordDocument, "Content");
object find = GetProperty(range, "Find");
// setup find
RunRoutine(find, "ClearFormatting", null);
SetProperty(find, "Forward", true);
SetProperty(find, "Text", searchSpec.SearchText);
SetProperty(find, "MatchWholeWord", searchSpec.UseWholeWordMatching);
SetProperty(find, "MatchCase", searchSpec.UseCaseSensitivity);
// start find
FindExecute(find);
// keep finding text
while ((bool)GetProperty(find, "Found") == true)
{
count += 1;
if (count == 1)
{
// create hit object
match = new MatchResult(file);
}
// since a hit was found and only displaying file names, quickly exit
if (searchSpec.ReturnOnlyFileNames)
break;
// retrieve find information
int start = (int)GetProperty(range, "Start");
int colNum = (int)Information(range, WdInformation.wdFirstCharacterColumnNumber);
int lineNum = (int)Information(range, WdInformation.wdFirstCharacterLineNumber);
int pageNum = (int)Information(range, WdInformation.wdActiveEndPageNumber);
string line = GetFindTextLine(start);
// don't add a hit if on same line
if (!(prevLine == lineNum && prevPage == pageNum))
{
// check for line numbers
//if (searchSpec.IncludeLineNumbers)
//{
// // setup line header
// _spacer = "(" + string.Format("{0},{1}", lineNum, pageNum);
// if (_spacer.Length <= 5)
// {
// _spacer = _spacer + new string(' ', 6 - _spacer.Length);
// }
// _spacer = _spacer + ") ";
// //_contextSpacer = "(" + new string(' ', _spacer.Length - 3) + ") ";
//}
// remove any odd characters from the text
line = RemoveSpecialCharacters(line);
// add context lines before
// if (__contextLines > 0){
// For i As int = __contextLines To 1 Step -1
// SetProperty(__WordSelection, "Start", start)
// SelectionMoveUp(WdUnits.wdLine, i, WdMovementType.wdMove)
// Dim cxt As string = GetFindTextLine()
// cxt = RemoveSpecialCharacters(cxt)
// if (Not HitExists(cxt, hit)){
// hit.Add(_contextSpacer & cxt & NEW_LINE, lineNum - i, 1)
// End If
// Next
// End If
// add line
MatchResultLine matchLine = new MatchResultLine()
{
HasMatch = true,
ColumnNumber = colNum,
LineNumber = lineNum,
Line = line
};
var lineMatches = libAstroGrep.Grep.RetrieveLineMatches(line, searchSpec);
match.SetHitCount(lineMatches.Count);
matchLine.Matches = lineMatches;
match.Matches.Add(matchLine);
//match.Add(_spacer, line, lineNum, colNum);
// add context lines after
// if (__contextLines > 0){
// For i As int = 1 To __contextLines
// SetProperty(__WordSelection, "Start", start)
// SelectionMoveDown(WdUnits.wdLine, i, WdMovementType.wdMove)
// Dim cxt As string = GetFindTextLine()
// cxt = RemoveSpecialCharacters(cxt)
// if (Not HitExists(cxt, hit)){
// hit.Add(_contextSpacer & cxt & NEW_LINE, lineNum + i, 1)
// End If
// Next
// End If
}
//match.SetHitCount();
prevLine = lineNum;
prevPage = pageNum;
// find again
FindExecute(find);
}
ReleaseSelection();
CloseDocument(wordDocument);
return match;
}
else
{
string msg = string.Format("File does not exist: {0}", file.FullName);
ex = new Exception(msg);
Trace(msg);
}
}
catch (Exception mainEx)
{
ex = mainEx;
Trace(mainEx.ToString());
}
}
else
{
ex = new Exception("Plugin not available or usable.");
Trace("Plugin not available or usable.");
}
return null;
}
/// <summary>
/// Retrieves the number of instances of searchText in the given line
/// </summary>
/// <param name="line">Line of text to search</param>
/// <param name="searchSpec">Current ISearchSpec interface</param>
/// <returns>Count of how many instances</returns>
/// <history>
/// [Curtis_Beard] 12/06/2005 Created
/// [Curtis_Beard] 01/12/2007 FIX: check for correct position of IndexOf
/// [Curtis_Beard] 03/05/2015 FIX: cleanup logic for whole word/case sensitive
/// </history>
public static List<MatchResultLineMatch> RetrieveLineMatches(string line, ISearchSpec searchSpec)
{
List<MatchResultLineMatch> lineMatches = new List<MatchResultLineMatch>();
int pos = line.IndexOf(searchSpec.SearchText, searchSpec.UseCaseSensitivity ? StringComparison.Ordinal : StringComparison.OrdinalIgnoreCase);
while (pos > -1)
{
// retrieve parts of text
string begin = line.Substring(0, pos);
string end = line.Substring(pos + searchSpec.SearchText.Length);
// do a check to see if begin and end are valid for wholeword searches
bool _highlight = true;
if (searchSpec.UseWholeWordMatching)
{
_highlight = WholeWordOnly(begin, end);
}
// found a hit
if (_highlight)
{
MatchResultLineMatch lineMatch = new MatchResultLineMatch(pos, searchSpec.SearchText.Length);
lineMatches.Add(lineMatch);
}
// Check remaining string for other hits in same line
pos = line.IndexOf(searchSpec.SearchText, pos + 1, searchSpec.UseCaseSensitivity ? StringComparison.Ordinal : StringComparison.OrdinalIgnoreCase);
}
return lineMatches;
}
/// <summary>
/// Logs a start search message to log file.
/// </summary>
/// <param name="searchSpec">Current search specification</param>
/// <param name="fileFilterSpec">Current file filter specification</param>
/// <history>
/// [Curtis_Beard] 05/15/2015 Initial
/// [Curtis_Beard] 05/26/2015 FIX: 69, add performance setting, cache for file encoding detection
/// </history>
private void LogStartSearchMessage(ISearchSpec searchSpec, IFileFilterSpec fileFilterSpec)
{
StringBuilder searchTextOptions = new StringBuilder();
LogSearchOptionHelper(searchTextOptions, searchSpec.UseRegularExpressions, "regex");
LogSearchOptionHelper(searchTextOptions, searchSpec.UseCaseSensitivity, "case sensitive");
LogSearchOptionHelper(searchTextOptions, searchSpec.UseWholeWordMatching, "whole word");
LogSearchOptionHelper(searchTextOptions, searchSpec.UseNegation, "negation");
LogSearchOptionHelper(searchTextOptions, searchSpec.ReturnOnlyFileNames, "only file names");
LogSearchOptionHelper(searchTextOptions, searchSpec.ContextLines > 0, string.Format("{0} context lines", searchSpec.ContextLines));
if (searchTextOptions.Length > 0)
{
searchTextOptions.Insert(0, "[");
searchTextOptions.Append("]");
}
StringBuilder fileEncoding = new StringBuilder();
if (searchSpec.EncodingDetectionOptions.DetectFileEncoding)
{
fileEncoding.Append("[");
fileEncoding.Append("detect encoding");
fileEncoding.AppendFormat(", performance set at {0}", Enum.GetName(typeof(EncodingOptions.Performance), GeneralSettings.EncodingPerformance).ToLower());
if (searchSpec.EncodingDetectionOptions.UseEncodingCache)
{
fileEncoding.Append(", cache enabled");
}
fileEncoding.Append("]");
}
LogClient.Instance.Logger.Info("Search started in '{0}'{1} against {2}{3} for {4}{5}",
searchSpec.StartFilePaths != null && searchSpec.StartFilePaths.Length > 0 ? string.Join(", ", searchSpec.StartFilePaths) : string.Join(", ", searchSpec.StartDirectories),
searchSpec.SearchInSubfolders ? "[include sub folders]" : "",
fileFilterSpec.FileFilter,
fileEncoding.ToString(),
searchSpec.SearchText,
searchTextOptions.ToString());
}
/// <summary>
/// Searches the given file for the given search text.
/// </summary>
/// <param name="file">FileInfo object</param>
/// <param name="searchSpec">ISearchSpec interface value</param>
/// <param name="ex">Exception holder if error occurs</param>
/// <returns>Hitobject containing grep results, null if on error</returns>
/// <history>
/// [Curtis_Beard] 07/28/2006 Created
/// [Curtis_Beard] 05/25/2007 ADD: support for Exception object
/// [Curtis_Beard] 03/31/2015 CHG: rework Grep/Matches
/// </history>
public MatchResult Grep(FileInfo file, ISearchSpec searchSpec, ref Exception ex)
{
// initialize Exception object to null
ex = null;
if (IsAvailable && IsUsable)
{
try
{
if (file.Exists)
{
//const int MARGINSIZE = 4;
int count = 0;
MatchResult match = null;
int prevLine = 0;
int prevPage = 0;
//string _spacer = new string(' ', MARGINSIZE);
//string _contextSpacer = string.Empty;
//if (searchSpec.ContextLines > 0)
//{
// _contextSpacer = new string(' ', MARGINSIZE);
// _spacer = _contextSpacer.Substring(_contextSpacer.Length - MARGINSIZE - 2) + "> ";
//}
//else
// _spacer = new string(' ', MARGINSIZE);
// Open a given Word document as readonly
// Note: Word 2003+ requires write mode since reading mode doesn't allow use of home/end keys to select text)
object appversion = __WordApplication.GetType().InvokeMember("Version", BindingFlags.GetProperty, null, __WordApplication, null);
double version = 0;
double.TryParse(appversion.ToString(), out version);
bool useReadOnly = version >= 12.00 ? false : true;
object wordDocument = OpenDocument(file.FullName, useReadOnly);
// Get Selection Property
__WordSelection = __WordApplication.GetType().InvokeMember("Selection", BindingFlags.GetProperty,
null, __WordApplication, null);
// create range and find objects
object range = GetProperty(wordDocument, "Content");
object find = GetProperty(range, "Find");
// setup find
RunRoutine(find, "ClearFormatting", null);
SetProperty(find, "Forward", true);
SetProperty(find, "Text", searchSpec.SearchText);
SetProperty(find, "MatchWholeWord", searchSpec.UseWholeWordMatching);
SetProperty(find, "MatchCase", searchSpec.UseCaseSensitivity);
// start find
FindExecute(find);
// keep finding text
while ((bool)GetProperty(find, "Found") == true)
{
count += 1;
if (count == 1)
{
// create hit object
match = new MatchResult(file);
}
// since a hit was found and only displaying file names, quickly exit
if (searchSpec.ReturnOnlyFileNames)
{
break;
}
// retrieve find information
int start = (int)GetProperty(range, "Start");
int colNum = (int)Information(range, WdInformation.wdFirstCharacterColumnNumber);
int lineNum = (int)Information(range, WdInformation.wdFirstCharacterLineNumber);
int pageNum = (int)Information(range, WdInformation.wdActiveEndPageNumber);
string line = GetFindTextLine(start);
// don't add a hit if on same line
if (!(prevLine == lineNum && prevPage == pageNum))
{
// check for line numbers
//if (searchSpec.IncludeLineNumbers)
//{
// // setup line header
// _spacer = "(" + string.Format("{0},{1}", lineNum, pageNum);
// if (_spacer.Length <= 5)
// {
// _spacer = _spacer + new string(' ', 6 - _spacer.Length);
// }
// _spacer = _spacer + ") ";
// //_contextSpacer = "(" + new string(' ', _spacer.Length - 3) + ") ";
//}
// remove any odd characters from the text
line = RemoveSpecialCharacters(line);
// add context lines before
// if (__contextLines > 0){
// For i As int = __contextLines To 1 Step -1
// SetProperty(__WordSelection, "Start", start)
// SelectionMoveUp(WdUnits.wdLine, i, WdMovementType.wdMove)
// Dim cxt As string = GetFindTextLine()
// cxt = RemoveSpecialCharacters(cxt)
// if (Not HitExists(cxt, hit)){
// hit.Add(_contextSpacer & cxt & NEW_LINE, lineNum - i, 1)
// End If
// Next
// End If
// add line
MatchResultLine matchLine = new MatchResultLine()
{
HasMatch = true,
ColumnNumber = colNum,
LineNumber = lineNum,
Line = line
};
var lineMatches = libAstroGrep.Grep.RetrieveLineMatches(line, searchSpec);
match.SetHitCount(lineMatches.Count);
matchLine.Matches = lineMatches;
match.Matches.Add(matchLine);
//match.Add(_spacer, line, lineNum, colNum);
// add context lines after
// if (__contextLines > 0){
// For i As int = 1 To __contextLines
// SetProperty(__WordSelection, "Start", start)
// SelectionMoveDown(WdUnits.wdLine, i, WdMovementType.wdMove)
// Dim cxt As string = GetFindTextLine()
// cxt = RemoveSpecialCharacters(cxt)
// if (Not HitExists(cxt, hit)){
// hit.Add(_contextSpacer & cxt & NEW_LINE, lineNum + i, 1)
// End If
// Next
// End If
}
//match.SetHitCount();
prevLine = lineNum;
prevPage = pageNum;
// find again
FindExecute(find);
}
ReleaseSelection();
CloseDocument(wordDocument);
return(match);
}
else
{
string msg = string.Format("File does not exist: {0}", file.FullName);
ex = new Exception(msg);
Trace(msg);
}
}
catch (Exception mainEx)
{
ex = mainEx;
Trace(mainEx.ToString());
}
}
else
{
ex = new Exception("Plugin not available or usable.");
Trace("Plugin not available or usable.");
}
return(null);
}
/// <summary>
/// Searches the given file for the given search text.
/// </summary>
/// <param name="file">FileInfo object</param>
/// <param name="searchSpec">ISearchSpec interface value</param>
/// <param name="ex">Exception holder if error occurs</param>
/// <returns>Hitobject containing grep results, null if on error</returns>
/// <history>
/// [Curtis_Beard] 10/17/2012 Created
/// [Curtis_Beard] 03/31/2015 CHG: rework Grep/Matches
/// </history>
public MatchResult Grep(FileInfo file, ISearchSpec searchSpec, ref Exception ex)
{
// initialize Exception object to null
ex = null;
MatchResult match = null;
if (Parser.IsParseable(file.FullName))
{
string fileContent = Parser.Parse(file.FullName);
if (!string.IsNullOrEmpty(fileContent))
{
string[] lines = fileContent.Split(new char[] { '\n', '\r' });
for (int i = 0; i < lines.Length; i++)
{
string line = lines[i];
int posInStr = -1;
Regex reg = null;
MatchCollection regCol = null;
if (searchSpec.UseRegularExpressions)
{
string pattern = string.Format("{0}{1}{0}", searchSpec.UseWholeWordMatching ? "\\b" : string.Empty, searchSpec.SearchText);
RegexOptions options = searchSpec.UseCaseSensitivity ? RegexOptions.None : RegexOptions.IgnoreCase;
reg = new Regex(pattern, options);
regCol = reg.Matches(line);
if (regCol.Count > 0)
{
posInStr = 1;
}
}
else
{
// If we are looking for whole worlds only, perform the check.
if (searchSpec.UseWholeWordMatching)
{
reg = new Regex("\\b" + Regex.Escape(searchSpec.SearchText) + "\\b", searchSpec.UseCaseSensitivity ? RegexOptions.None : RegexOptions.IgnoreCase);
// if match is found, also check against our internal line hit count method to be sure they are in sync
Match mtc = reg.Match(line);
if (mtc != null && mtc.Success && libAstroGrep.Grep.RetrieveLineMatches(line, searchSpec).Count > 0)
{
posInStr = mtc.Index;
}
}
else
{
posInStr = line.IndexOf(searchSpec.SearchText, searchSpec.UseCaseSensitivity ? StringComparison.Ordinal : StringComparison.OrdinalIgnoreCase);
}
}
if (posInStr > -1)
{
if (match == null)
{
match = new MatchResult(file);
// found hit in file so just return
if (searchSpec.ReturnOnlyFileNames)
{
break;
}
}
var matchLineFound = new MatchResultLine() { Line = line, LineNumber = -1, HasMatch = true };
if (searchSpec.UseRegularExpressions)
{
posInStr = regCol[0].Index;
match.SetHitCount(regCol.Count);
foreach (Match regExMatch in regCol)
{
matchLineFound.Matches.Add(new MatchResultLineMatch(regExMatch.Index, regExMatch.Length));
}
}
else
{
var lineMatches = libAstroGrep.Grep.RetrieveLineMatches(line, searchSpec);
match.SetHitCount(lineMatches.Count);
matchLineFound.Matches = lineMatches;
}
matchLineFound.ColumnNumber = 1;
match.Matches.Add(matchLineFound);
}
}
}
}
return match;
}
public Grep(ISearchSpec searchSpec)
{
SearchSpec = searchSpec;
MatchResults = new List <MatchResult>();
}
