public List <GrepSearchResult> Search(string file, string searchPattern, SearchType searchType, GrepSearchOption searchOptions, Encoding encoding)
{
string tempFolder = Path.Combine(Utils.GetTempFolder(), "dnGREP-PDF");
try
{
// Extract text
string tempFile = ExtractText(file);
if (!File.Exists(tempFile))
{
throw new ApplicationException("pdftotext failed to create text file.");
}
// GrepCore cannot check encoding of the original pdf file. If the encoding parameter is not default
// then it is the user-specified code page. If the encoding parameter *is* the default,
// then it most likely not been set, so get the encoding of the extracted text file:
if (encoding == Encoding.Default)
{
encoding = Utils.GetFileEncoding(tempFile);
}
IGrepEngine engine = GrepEngineFactory.GetSearchEngine(tempFile, initParams, FileFilter);
List <GrepSearchResult> results = engine.Search(tempFile, searchPattern, searchType, searchOptions, encoding);
if (results.Count > 0)
{
using (FileStream reader = File.Open(tempFile, FileMode.Open, FileAccess.Read, FileShare.ReadWrite))
using (StreamReader streamReader = new StreamReader(reader, encoding))
{
foreach (var result in results)
{
result.SearchResults = Utils.GetLinesEx(streamReader, result.Matches, initParams.LinesBefore, initParams.LinesAfter);
}
}
foreach (GrepSearchResult result in results)
{
result.ReadOnly = true;
if (file.Contains(tempFolder))
{
result.FileNameDisplayed = file.Substring(tempFolder.Length + 1);
}
else
{
result.FileNameDisplayed = file;
}
result.FileNameReal = file;
}
}
GrepEngineFactory.ReturnToPool(tempFile, engine);
return(results);
}
catch (Exception ex)
{
logger.Error(ex, $"Failed to search inside PDF file: {ex.Message}");
return(new List <GrepSearchResult>());
}
}
public List <GrepSearchResult> Search(string file, string searchPattern, SearchType searchType, GrepSearchOption searchOptions, Encoding encoding)
{
string tempFolder = Path.Combine(Utils.GetTempFolder(), "dnGREP-PDF");
try
{
// Extract text
string tempFile = extractText(file);
if (!File.Exists(tempFile))
{
throw new ApplicationException("pdftotext failed to create text file.");
}
IGrepEngine engine = GrepEngineFactory.GetSearchEngine(tempFile, initParams, FileFilter);
List <GrepSearchResult> results = engine.Search(tempFile, searchPattern, searchType, searchOptions, encoding);
if (results.Count > 0)
{
using (FileStream reader = File.Open(tempFile, FileMode.Open, FileAccess.Read, FileShare.ReadWrite))
using (StreamReader streamReader = new StreamReader(reader))
{
foreach (var result in results)
{
result.SearchResults = Utils.GetLinesEx(streamReader, result.Matches, initParams.LinesBefore, initParams.LinesAfter);
}
}
foreach (GrepSearchResult result in results)
{
result.ReadOnly = true;
if (file.Contains(tempFolder))
{
result.FileNameDisplayed = file.Substring(tempFolder.Length + 1);
}
else
{
result.FileNameDisplayed = file;
}
result.FileNameReal = file;
}
}
GrepEngineFactory.ReturnToPool(tempFile, engine);
return(results);
}
catch (Exception ex)
{
logger.Log <Exception>(LogLevel.Error, "Failed to search inside Pdf file", ex);
return(new List <GrepSearchResult>());
}
}
public List <GrepSearchResult> Search(string file, string searchPattern, SearchType searchType, GrepSearchOption searchOptions, Encoding encoding)
{
List <GrepSearchResult> searchResults = new List <GrepSearchResult>();
SevenZipExtractor extractor = new SevenZipExtractor(file);
GrepEnginePlainText plainTextEngine = new GrepEnginePlainText();
plainTextEngine.Initialize(new GrepEngineInitParams(showLinesInContext, linesBefore, linesAfter, fuzzyMatchThreshold));
string tempFolder = Utils.FixFolderName(Utils.GetTempFolder()) + "dnGREP-Archive\\" + Utils.GetHash(file) + "\\";
if (Directory.Exists(tempFolder))
{
Utils.DeleteFolder(tempFolder);
}
Directory.CreateDirectory(tempFolder);
try
{
extractor.ExtractArchive(tempFolder);
foreach (string archiveFileName in Directory.GetFiles(tempFolder, "*.*", SearchOption.AllDirectories))
{
IGrepEngine engine = GrepEngineFactory.GetSearchEngine(archiveFileName, new GrepEngineInitParams(showLinesInContext, linesBefore, linesAfter, fuzzyMatchThreshold));
var innerFileResults = engine.Search(archiveFileName, searchPattern, searchType, searchOptions, encoding);
using (FileStream reader = File.Open(archiveFileName, FileMode.Open, FileAccess.Read, FileShare.ReadWrite))
using (StreamReader streamReader = new StreamReader(reader))
{
foreach (var result in innerFileResults)
{
if (!result.HasSearchResults)
{
result.SearchResults = Utils.GetLinesEx(streamReader, result.Matches, linesBefore, linesAfter);
}
}
}
searchResults.AddRange(innerFileResults);
}
foreach (GrepSearchResult result in searchResults)
{
result.FileNameDisplayed = file + "\\" + result.FileNameDisplayed.Substring(tempFolder.Length);
result.FileNameReal = file;
result.ReadOnly = true;
}
}
catch (Exception ex)
{
logger.LogException(LogLevel.Error, "Failed to search inside archive.", ex);
}
return(searchResults);
}
public List <GrepSearchResult> Search(string file, string searchPattern, SearchType searchType, GrepSearchOption searchOptions, Encoding encoding)
{
try
{
// Extract text
string tempFile = extractText(file);
if (!File.Exists(tempFile))
{
throw new ApplicationException("pdftotext failed to create text file.");
}
IGrepEngine engine = GrepEngineFactory.GetSearchEngine(tempFile, new GrepEngineInitParams(showLinesInContext, linesBefore, linesAfter, fuzzyMatchThreshold));
List <GrepSearchResult> results = engine.Search(tempFile, searchPattern, searchType, searchOptions, encoding);
using (FileStream reader = File.Open(tempFile, FileMode.Open, FileAccess.Read, FileShare.ReadWrite))
using (StreamReader streamReader = new StreamReader(reader))
{
foreach (var result in results)
{
result.SearchResults = Utils.GetLinesEx(streamReader, result.Matches, linesBefore, linesAfter);
}
}
foreach (GrepSearchResult result in results)
{
result.ReadOnly = true;
result.FileNameDisplayed = file;
result.FileNameReal = file;
}
return(results);
}
catch (Exception ex)
{
logger.LogException(LogLevel.Error, "Failed to search inside Pdf file", ex);
return(new List <GrepSearchResult>());
}
}
/// <summary>
/// Searches folder for files whose content matches regex
/// </summary>
/// <param name="files">Files to search in. If one of the files does not exist or is open, it is skipped.</param>
/// <param name="searchRegex">Regex pattern</param>
/// <returns>List of results. If nothing is found returns empty list</returns>
public List <GrepSearchResult> Search(IEnumerable <string> files, SearchType searchType, string searchPattern, GrepSearchOption searchOptions, int codePage)
{
List <GrepSearchResult> searchResults = new List <GrepSearchResult>();
if (files == null)
{
return(searchResults);
}
Utils.CancelSearch = false;
if (searchPattern == null || searchPattern.Trim() == "")
{
foreach (string file in files)
{
searchResults.Add(new GrepSearchResult(file, searchPattern, null));
if ((searchOptions & GrepSearchOption.StopAfterFirstMatch) == GrepSearchOption.StopAfterFirstMatch)
{
break;
}
}
if (ProcessedFile != null)
{
ProcessedFile(this, new ProgressStatus(searchResults.Count, searchResults));
}
return(searchResults);
}
else
{
int processedFiles = 0;
try
{
foreach (string file in files)
{
try
{
IGrepEngine engine = GrepEngineFactory.GetSearchEngine(file, SearchParams);
processedFiles++;
Encoding encoding = null;
if (codePage == -1)
{
encoding = Utils.GetFileEncoding(file);
}
else
{
encoding = Encoding.GetEncoding(codePage);
}
if (Utils.CancelSearch)
{
return(searchResults);
}
List <GrepSearchResult> fileSearchResults = engine.Search(file, searchPattern, searchType, searchOptions, encoding);
if (fileSearchResults != null && fileSearchResults.Count > 0)
{
searchResults.AddRange(fileSearchResults);
}
if (ProcessedFile != null)
{
ProcessedFile(this, new ProgressStatus(processedFiles, fileSearchResults));
}
}
catch (Exception ex)
{
logger.LogException(LogLevel.Error, ex.Message, ex);
searchResults.Add(new GrepSearchResult(file, searchPattern, ex.Message, false));
if (ProcessedFile != null)
{
List <GrepSearchResult> _results = new List <GrepSearchResult>();
_results.Add(new GrepSearchResult(file, searchPattern, ex.Message, false));
ProcessedFile(this, new ProgressStatus(processedFiles, _results));
}
}
if ((searchOptions & GrepSearchOption.StopAfterFirstMatch) == GrepSearchOption.StopAfterFirstMatch && searchResults.Count > 0)
{
break;
}
}
}
finally
{
GrepEngineFactory.UnloadEngines();
}
return(searchResults);
}
}
private void Search(string file, SearchType searchType, string searchPattern, GrepSearchOption searchOptions, int codePage)
{
try
{
ProcessedFile(this, new ProgressStatus(true, processedFilesCount, foundfilesCount, null, file));
IGrepEngine engine = GrepEngineFactory.GetSearchEngine(file, SearchParams, FileFilter);
Interlocked.Increment(ref processedFilesCount);
Encoding encoding = Encoding.Default;
if (codePage > -1)
{
encoding = Encoding.GetEncoding(codePage);
}
else if (!Utils.IsBinary(file) && !Utils.IsPdfFile(file))
{
encoding = Utils.GetFileEncoding(file);
}
if (Utils.CancelSearch)
{
if (cancellationTokenSource != null)
{
cancellationTokenSource.Cancel();
}
return;
}
List <GrepSearchResult> fileSearchResults = engine.Search(file, searchPattern, searchType, searchOptions, encoding);
if (fileSearchResults != null && fileSearchResults.Count > 0)
{
AddSearchResults(fileSearchResults);
}
int hits = fileSearchResults.Where(r => r.IsSuccess).Count();
Interlocked.Add(ref foundfilesCount, hits);
ProcessedFile(this, new ProgressStatus(false, processedFilesCount, foundfilesCount, fileSearchResults, file));
GrepEngineFactory.ReturnToPool(file, engine);
}
catch (Exception ex)
{
logger.Log <Exception>(LogLevel.Error, ex.Message, ex);
AddSearchResult(new GrepSearchResult(file, searchPattern, ex.Message, false));
if (ProcessedFile != null)
{
List <GrepSearchResult> _results = new List <GrepSearchResult>();
_results.Add(new GrepSearchResult(file, searchPattern, ex.Message, false));
ProcessedFile(this, new ProgressStatus(false, processedFilesCount, foundfilesCount, _results, file));
}
}
finally
{
if ((searchOptions & GrepSearchOption.StopAfterFirstMatch) == GrepSearchOption.StopAfterFirstMatch && searchResults.Count > 0)
{
if (cancellationTokenSource != null)
{
cancellationTokenSource.Cancel();
}
}
}
}
public List <GrepSearchResult> Search(Stream input, string file, string searchPattern, SearchType searchType, GrepSearchOption searchOptions, Encoding encoding)
{
List <GrepSearchResult> searchResults = new List <GrepSearchResult>();
var filter = FileFilter.ToStandardFilter();
var includeRegexPatterns = new List <Regex>();
var excludeRegexPatterns = new List <Regex>();
Utils.PrepareFilters(filter, includeRegexPatterns, excludeRegexPatterns);
List <string> hiddenDirectories = new List <string>();
try
{
using (SevenZipExtractor extractor = new SevenZipExtractor(input))
{
foreach (var fileInfo in extractor.ArchiveFileData)
{
var attr = (FileAttributes)fileInfo.Attributes;
string innerFileName = fileInfo.FileName;
if (fileInfo.IsDirectory)
{
if (!filter.IncludeHidden && attr.HasFlag(FileAttributes.Hidden) && !hiddenDirectories.Contains(innerFileName))
{
hiddenDirectories.Add(innerFileName);
}
continue;
}
if (CheckHidden(filter, attr) &&
CheckHidden(filter, innerFileName, hiddenDirectories) &&
CheckSize(filter, fileInfo.Size) &&
CheckDate(filter, fileInfo) &&
IsPatternMatch(innerFileName, includeRegexPatterns) &&
!IsPatternMatch(innerFileName, excludeRegexPatterns))
{
using (Stream stream = new MemoryStream())
{
extractor.ExtractFile(innerFileName, stream);
stream.Seek(0, SeekOrigin.Begin);
if (CheckBinary(filter, stream))
{
// Need to check the encoding of each file in the archive. If the encoding parameter is not default
// then it is the user-specified code page. If the encoding parameter *is* the default,
// then it most likely not been set, so get the encoding of the extracted text file:
if (encoding == Encoding.Default && !Utils.IsBinary(stream))
{
stream.Seek(0, SeekOrigin.Begin);
encoding = Utils.GetFileEncoding(stream);
}
IGrepEngine engine = GrepEngineFactory.GetSearchEngine(innerFileName, initParams, filter);
var innerFileResults = engine.Search(stream, innerFileName, searchPattern, searchType, searchOptions, encoding);
if (innerFileResults.Count > 0)
{
using (Stream readStream = new MemoryStream())
{
extractor.ExtractFile(innerFileName, readStream);
readStream.Seek(0, SeekOrigin.Begin);
using (StreamReader streamReader = new StreamReader(readStream, encoding))
{
foreach (var result in innerFileResults)
{
if (Utils.CancelSearch)
{
break;
}
if (!result.HasSearchResults)
{
result.SearchResults = Utils.GetLinesEx(streamReader, result.Matches, initParams.LinesBefore, initParams.LinesAfter);
}
}
}
searchResults.AddRange(innerFileResults);
}
}
GrepEngineFactory.ReturnToPool(innerFileName, engine);
}
if (Utils.CancelSearch)
{
break;
}
}
}
}
}
foreach (GrepSearchResult result in searchResults)
{
result.FileNameDisplayed = file + "\\" + result.FileNameDisplayed;
result.FileNameReal = file;
result.ReadOnly = true;
}
}
catch (Exception ex)
{
logger.Log <Exception>(LogLevel.Error, string.Format("Failed to search inside archive '{0}'", file), ex);
}
return(searchResults);
}
public List <GrepSearchResult> Search(string file, string searchPattern, SearchType searchType, GrepSearchOption searchOptions, Encoding encoding)
{
List <GrepSearchResult> searchResults = new List <GrepSearchResult>();
SevenZipExtractor extractor = new SevenZipExtractor(file);
string tempFolder = Utils.FixFolderName(Utils.GetTempFolder()) + "dnGREP-Archive\\" + Utils.GetHash(file) + "\\";
FileFilter filter = FileFilter.ChangePath(tempFolder);
// if the search pattern(s) only match archive files, need to include an 'any' file type to search inside the archive.
// otherwise, keep the original pattern set so the user can specify what types of files to search inside the archive.
var patterns = Utils.SplitPath(FileFilter.NamePatternToInclude).ToList();
bool hasNonArchivePattern = patterns.Where(p => !Utils.IsArchiveExtension(Path.GetExtension(p))).Any();
if (!hasNonArchivePattern)
{
patterns.Add(FileFilter.IsRegex ? ".*" : "*.*");
filter = filter.ChangeIncludePattern(string.Join(";", patterns.ToArray()));
}
if (Directory.Exists(tempFolder))
{
Utils.DeleteFolder(tempFolder);
}
Directory.CreateDirectory(tempFolder);
try
{
extractor.ExtractArchive(tempFolder);
foreach (var innerFileName in Utils.GetFileListEx(filter))
{
IGrepEngine engine = GrepEngineFactory.GetSearchEngine(innerFileName, initParams, FileFilter);
var innerFileResults = engine.Search(innerFileName, searchPattern, searchType, searchOptions, encoding);
if (innerFileResults.Count > 0)
{
using (FileStream reader = File.Open(innerFileName, FileMode.Open, FileAccess.Read, FileShare.ReadWrite))
using (StreamReader streamReader = new StreamReader(reader))
{
foreach (var result in innerFileResults)
{
if (Utils.CancelSearch)
{
break;
}
if (!result.HasSearchResults)
{
result.SearchResults = Utils.GetLinesEx(streamReader, result.Matches, initParams.LinesBefore, initParams.LinesAfter);
}
}
}
searchResults.AddRange(innerFileResults);
}
if (Utils.CancelSearch)
{
break;
}
}
foreach (GrepSearchResult result in searchResults)
{
result.FileNameDisplayed = file + "\\" + result.FileNameDisplayed.Substring(tempFolder.Length);
result.FileNameReal = file;
result.ReadOnly = true;
}
}
catch (Exception ex)
{
logger.Log <Exception>(LogLevel.Error, string.Format("Failed to search inside archive '{0}'", file), ex);
}
return(searchResults);
}
private void Search(string file, SearchType searchType, string searchPattern, GrepSearchOption searchOptions, int codePage)
{
try
{
ProcessedFile(this, new ProgressStatus(true, processedFilesCount, foundfilesCount, null, file));
bool isArchive = Utils.IsArchive(file);
Encoding encoding = Encoding.Default;
if (codePage > -1)
{
encoding = Encoding.GetEncoding(codePage);
}
else if (!isArchive && !Utils.IsBinary(file) && !Utils.IsPdfFile(file))
{
encoding = Utils.GetFileEncoding(file);
}
if (Utils.CancelSearch)
{
if (cancellationTokenSource != null)
{
cancellationTokenSource.Cancel();
}
return;
}
IGrepEngine engine = GrepEngineFactory.GetSearchEngine(file, SearchParams, FileFilter, searchType);
if (isArchive && engine is ArchiveEngine archiveEngine)
{
archiveEngine.SetSearchOptions(FileFilter, SearchParams);
archiveEngine.StartingFileSearch += ArchiveEngine_StartingFileSearch;
foreach (var fileSearchResults in archiveEngine.Search(file, searchPattern, searchType, searchOptions, encoding))
{
if (fileSearchResults != null && fileSearchResults.Count > 0)
{
AddSearchResults(fileSearchResults);
}
int hits = fileSearchResults.Where(r => r.IsSuccess).Count();
Interlocked.Add(ref foundfilesCount, hits);
ProcessedFile(this, new ProgressStatus(false, processedFilesCount, foundfilesCount, fileSearchResults, file));
}
archiveEngine.StartingFileSearch -= ArchiveEngine_StartingFileSearch;
}
else
{
Interlocked.Increment(ref processedFilesCount);
var fileSearchResults = engine.Search(file, searchPattern, searchType, searchOptions, encoding).ToList();
if (fileSearchResults != null && fileSearchResults.Count > 0)
{
AddSearchResults(fileSearchResults);
}
int hits = fileSearchResults.Where(r => r.IsSuccess).Count();
Interlocked.Add(ref foundfilesCount, hits);
ProcessedFile(this, new ProgressStatus(false, processedFilesCount, foundfilesCount, fileSearchResults, file));
}
GrepEngineFactory.ReturnToPool(file, engine);
}
catch (Exception ex)
{
logger.Error(ex, "Failed in Search");
AddSearchResult(new GrepSearchResult(file, searchPattern, ex.Message, false));
if (ProcessedFile != null)
{
List <GrepSearchResult> _results = new List <GrepSearchResult>
{
new GrepSearchResult(file, searchPattern, ex.Message, false)
};
ProcessedFile(this, new ProgressStatus(false, processedFilesCount, foundfilesCount, _results, file));
}
}
finally
{
if (searchOptions.HasFlag(GrepSearchOption.StopAfterFirstMatch) && searchResults.Count > 0)
{
if (cancellationTokenSource != null)
{
cancellationTokenSource.Cancel();
}
}
}
}
public List <GrepSearchResult> Search(Stream input, string file, string searchPattern, SearchType searchType, GrepSearchOption searchOptions, Encoding encoding)
{
List <GrepSearchResult> searchResults = new List <GrepSearchResult>();
var includeRegexPatterns = new List <Regex>();
var excludeRegexPatterns = new List <Regex>();
Utils.PrepareFilters(FileFilter, includeRegexPatterns, excludeRegexPatterns);
List <string> hiddenDirectories = new List <string>();
try
{
using (SevenZipExtractor extractor = new SevenZipExtractor(input))
{
foreach (var fileInfo in extractor.ArchiveFileData)
{
var attr = (FileAttributes)fileInfo.Attributes;
string innerFileName = fileInfo.FileName;
if (fileInfo.IsDirectory)
{
if (!FileFilter.IncludeHidden && attr.HasFlag(FileAttributes.Hidden) && !hiddenDirectories.Contains(innerFileName))
{
hiddenDirectories.Add(innerFileName);
}
continue;
}
if (CheckHidden(FileFilter, attr) &&
CheckHidden(FileFilter, innerFileName, hiddenDirectories) &&
CheckSize(FileFilter, fileInfo.Size) &&
CheckDate(FileFilter, fileInfo) &&
IsPatternMatch(innerFileName, includeRegexPatterns) &&
!IsPatternMatch(innerFileName, excludeRegexPatterns))
{
using (Stream stream = new MemoryStream())
{
extractor.ExtractFile(innerFileName, stream);
stream.Seek(0, SeekOrigin.Begin);
if (CheckBinary(FileFilter, stream))
{
IGrepEngine engine = GrepEngineFactory.GetSearchEngine(innerFileName, initParams, FileFilter);
var innerFileResults = engine.Search(stream, innerFileName, searchPattern, searchType, searchOptions, encoding);
if (innerFileResults.Count > 0)
{
using (Stream readStream = new MemoryStream())
{
extractor.ExtractFile(innerFileName, readStream);
readStream.Seek(0, SeekOrigin.Begin);
using (StreamReader streamReader = new StreamReader(readStream))
{
foreach (var result in innerFileResults)
{
if (Utils.CancelSearch)
{
break;
}
if (!result.HasSearchResults)
{
result.SearchResults = Utils.GetLinesEx(streamReader, result.Matches, initParams.LinesBefore, initParams.LinesAfter);
}
}
}
searchResults.AddRange(innerFileResults);
}
}
}
if (Utils.CancelSearch)
{
break;
}
}
}
}
}
foreach (GrepSearchResult result in searchResults)
{
result.FileNameDisplayed = file + "\\" + result.FileNameDisplayed;
result.FileNameReal = file;
result.ReadOnly = true;
}
}
catch (Exception ex)
{
logger.Log <Exception>(LogLevel.Error, string.Format("Failed to search inside archive '{0}'", file), ex);
}
return(searchResults);
}