TextSearch: use BString-based searching instead of Grep-based.

Huge performance improvement. Fixes #9529.

src/apps/text_search/GrepWindow.cpp

@@ -90,7 +90,6 @@ GrepWindow::GrepWindow(BMessage* message)
 	fRecurseDirs(NULL),
 	fSkipDotDirs(NULL),
 	fCaseSensitive(NULL),
-	fEscapeText(NULL),
 	fTextOnly(NULL),
 	fInvokePe(NULL),
 	fHistoryMenu(NULL),
@@ -211,10 +210,6 @@ void GrepWindow::MessageReceived(BMessage* message)
 			_OnCaseSensitive();
 			break;
 
-		case MSG_ESCAPE_TEXT:
-			_OnEscapeText();
-			break;
-
 		case MSG_TEXT_ONLY:
 			_OnTextOnly();
 			break;
@@ -259,10 +254,6 @@ void GrepWindow::MessageReceived(BMessage* message)
 			_OnReportResult(message);
 			break;
 
-		case MSG_REPORT_ERROR:
-			_OnReportError(message);
-			break;
-
 		case MSG_SELECT_ALL:
 			_OnSelectAll(message);
 			break;
@@ -452,9 +443,6 @@ GrepWindow::_CreateMenus()
 	fCaseSensitive = new BMenuItem(
 		B_TRANSLATE("Case-sensitive"), new BMessage(MSG_CASE_SENSITIVE));
 
-	fEscapeText = new BMenuItem(
-		B_TRANSLATE("Escape search text"), new BMessage(MSG_ESCAPE_TEXT));
-
 	fTextOnly = new BMenuItem(
 		B_TRANSLATE("Text files only"), new BMessage(MSG_TEXT_ONLY));
 
@@ -486,7 +474,6 @@ GrepWindow::_CreateMenus()
 	fPreferencesMenu->AddItem(fRecurseDirs);
 	fPreferencesMenu->AddItem(fSkipDotDirs);
 	fPreferencesMenu->AddItem(fCaseSensitive);
-	fPreferencesMenu->AddItem(fEscapeText);
 	fPreferencesMenu->AddItem(fTextOnly);
 	fPreferencesMenu->AddItem(fInvokePe);
 
@@ -607,7 +594,6 @@ GrepWindow::_LoadPrefs()
 	fRecurseLinks->SetMarked(fModel->fRecurseLinks);
 	fSkipDotDirs->SetMarked(fModel->fSkipDotDirs);
 	fCaseSensitive->SetMarked(fModel->fCaseSensitive);
-	fEscapeText->SetMarked(fModel->fEscapeText);
 	fTextOnly->SetMarked(fModel->fTextOnly);
 	fInvokePe->SetMarked(fModel->fInvokePe);
 
@@ -1019,15 +1005,6 @@ GrepWindow::_OnReportResult(BMessage* message)
 }
 
 
-void
-GrepWindow::_OnReportError(BMessage* message)
-{
-	const char* buf;
-	if (message->FindString("error", &buf) == B_OK)
-		fSearchResults->AddItem(new BStringItem(buf));
-}
-
-
 void
 GrepWindow::_OnRecurseLinks()
 {
@@ -1055,15 +1032,6 @@ GrepWindow::_OnSkipDotDirs()
 }
 
 
-void
-GrepWindow::_OnEscapeText()
-{
-	fModel->fEscapeText = !fModel->fEscapeText;
-	fEscapeText->SetMarked(fModel->fEscapeText);
-	_ModelChanged();
-}
-
-
 void
 GrepWindow::_OnCaseSensitive()
 {

src/apps/text_search/GrepWindow.h

@@ -50,11 +50,9 @@ private:
 			void				_OnNodeMonitorPulse();
 			void				_OnReportFileName(BMessage* message);
 			void				_OnReportResult(BMessage* message);
-			void				_OnReportError(BMessage* message);
 			void				_OnRecurseLinks();
 			void				_OnRecurseDirs();
 			void				_OnSkipDotDirs();
-			void				_OnEscapeText();
 			void				_OnCaseSensitive();
 			void				_OnTextOnly();
 			void				_OnInvokePe();
@@ -103,7 +101,6 @@ private:
 			BMenuItem*			fRecurseDirs;
 			BMenuItem*			fSkipDotDirs;
 			BMenuItem*			fCaseSensitive;
-			BMenuItem*			fEscapeText;
 			BMenuItem*			fTextOnly;
 			BMenuItem*			fInvokePe;
 			BMenu*				fHistoryMenu;

src/apps/text_search/Grepper.cpp

@@ -17,6 +17,7 @@
 #include <Locale.h>
 #include <NodeInfo.h>
 #include <Path.h>
+#include <String.h>
 #include <UTF8.h>
 
 #include "FileIterator.h"
@@ -59,33 +60,10 @@ strdup_to_utf8(uint32 encode, const char* src, int32 length)
 }
 
 
-char*
+Grepper::Grepper(BString pattern, const Model* model,
-strdup_from_utf8(uint32 encode, const char* src, int32 length)
-{
-	int32 srcLen = length;
-	int32 dstLen = srcLen;
-	char* dst = new (nothrow) char[dstLen + 1];
-	if (dst == NULL)
-		return NULL;
-	int32 cookie = 0;
-	convert_from_utf8(encode, src, &srcLen, dst, &dstLen, &cookie);
-	// TODO: See above.
-	dst[dstLen] = '\0';
-	char* dup = strdup(dst);
-	delete[] dst;
-	if (srcLen != length) {
-		fprintf(stderr, "strdup_from_utf8(%" B_PRId32 ", %" B_PRId32
-			") dst allocate smalled(%" B_PRId32 ")\n", encode, length, dstLen);
-	}
-	return dup;
-}
-
-
-Grepper::Grepper(const char* pattern, const Model* model,
 		const BHandler* target, FileIterator* iterator)
-	: fPattern(NULL),
+	: fPattern(pattern),
 	  fTarget(target),
-	  fEscapeText(model->fEscapeText),
 	  fCaseSensitive(model->fCaseSensitive),
 	  fEncoding(model->fEncoding),
 
@@ -93,19 +71,12 @@ Grepper::Grepper(const char* pattern, const Model* model,
 	  fThreadId(-1),
 	  fMustQuit(false)
 {
-	if (fEncoding > 0) {
-		char* src = strdup_from_utf8(fEncoding, pattern, strlen(pattern));
-		_SetPattern(src);
-		free(src);
-	} else
-		_SetPattern(pattern);
 }
 
 
 Grepper::~Grepper()
 {
 	Cancel();
-	free(fPattern);
 	delete fIterator;
 }
 
@@ -160,17 +131,9 @@ int32
 Grepper::_GrepperThread()
 {
 	BMessage message;
-
 	char fileName[B_PATH_NAME_LENGTH];
-	char tempString[B_PATH_NAME_LENGTH];
-	char command[B_PATH_NAME_LENGTH + 32];
-
-	BPath tempFile;
-	sprintf(fileName, "/tmp/SearchText%" B_PRId32, fThreadId);
-	tempFile.SetTo(fileName);
 
 	while (!fMustQuit && fIterator->GetNextName(fileName)) {
-
 		message.MakeEmpty();
 		message.what = MSG_REPORT_FILE_NAME;
 		message.AddString("filename", fileName);
@@ -191,58 +154,44 @@ Grepper::_GrepperThread()
 			continue;
 		}
 
-		if (!_EscapeSpecialChars(fileName, B_PATH_NAME_LENGTH)) {
+		BString contents;
-			sprintf(tempString, B_TRANSLATE("%s: Not enough room to escape "
-				"the filename."), fileName);
 
-			message.MakeEmpty();
+		BFile file(&ref, B_READ_WRITE);
-			message.what = MSG_REPORT_ERROR;
+		off_t size;
-			message.AddString("error", tempString);
+		file.GetSize(&size);
+		file.Seek(0, SEEK_SET);
+		char* buffer = new char[size];
+		file.Read(buffer, size);
+		if (fEncoding > 0) {
+			char* temp = strdup_to_utf8(fEncoding, buffer, size);
+			contents.SetTo(temp, size);
+			free(temp);
+		} else
+			contents.SetTo(buffer, size);
+		delete[] buffer;
+
+		int32 index = 0, lines = 1; // First line is 1 not 0
+		while (true) {
+			int32 newPos;
+			if (fCaseSensitive)
+				newPos = contents.FindFirst(fPattern, index);
+			else
+				newPos = contents.IFindFirst(fPattern, index);
+			if (newPos == B_ERROR)
+				break;
+
+			lines += _CountLines(contents, index, newPos);
+			BString linenoAndLine;
+			linenoAndLine.SetToFormat("%" B_PRId32 ":%s", lines, _GetLine(contents, newPos).String());
+			message.AddString("text", linenoAndLine);
+
+			index = newPos + 1;
+		}
+
+		if (message.HasString("text") || fIterator->NotifyNegatives())
 			fTarget.SendMessage(&message);
-			continue;
-		}
-
-		sprintf(command, "grep -hn %s %s \"%s\" > \"%s\"",
-			fCaseSensitive ? "" : "-i", fPattern, fileName, tempFile.Path());
-
-		int res = system(command);
-
-		if (res == 0 || res == 1) {
-			FILE *results = fopen(tempFile.Path(), "r");
-
-			if (results != NULL) {
-				while (fgets(tempString, B_PATH_NAME_LENGTH, results) != 0) {
-					if (fEncoding > 0) {
-						char* tempdup = strdup_to_utf8(fEncoding, tempString,
-							strlen(tempString));
-						message.AddString("text", tempdup);
-						free(tempdup);
-					} else
-						message.AddString("text", tempString);
-				}
-
-				if (message.HasString("text") || fIterator->NotifyNegatives())
-					fTarget.SendMessage(&message);
-
-				fclose(results);
-				continue;
-			}
-		}
-
-		sprintf(tempString, B_TRANSLATE("%s: There was a problem running grep."), fileName);
-
-		message.MakeEmpty();
-		message.what = MSG_REPORT_ERROR;
-		message.AddString("error", tempString);
-		fTarget.SendMessage(&message);
 	}
 
-	// We wait with removing the temporary file until after the
-	// entire search has finished, to prevent a lot of flickering
-	// if the Tracker window for /tmp/ might be open.
-
-	remove(tempFile.Path());
-
 	message.MakeEmpty();
 	message.what = MSG_SEARCH_FINISHED;
 	fTarget.SendMessage(&message);
@@ -251,79 +200,36 @@ Grepper::_GrepperThread()
 }
 
 
-void
+int32
-Grepper::_SetPattern(const char* src)
+Grepper::_CountLines(BString& str, int32 startPos, int32 endPos)
 {
-	if (src == NULL)
+	int32 ret = 0;
-		return;
+	for (int32 i = startPos; i < endPos; i++) {
-
+		if (str[i] == '\n')
-	if (!fEscapeText) {
+			ret++;
-		fPattern = strdup(src);
-		return;
 	}
-
+	return ret;
-	// We will simply guess the size of the memory buffer
-	// that we need. This should always be large enough.
-	fPattern = (char*)malloc((strlen(src) + 1) * 3 * sizeof(char));
-	if (fPattern == NULL)
-		return;
-
-	const char* srcPtr = src;
-	char* dstPtr = fPattern;
-
-	// Put double quotes around the pattern, so separate
-	// words are considered to be part of a single string.
-	*dstPtr++ = '"';
-
-	while (*srcPtr != '\0') {
-		char c = *srcPtr++;
-
-		// Put a backslash in front of characters
-		// that should be escaped.
-		if ((c == '.')  || (c == ',')
-			||  (c == '[')  || (c == ']')
-			||  (c == '?')  || (c == '*')
-			||  (c == '+')  || (c == '-')
-			||  (c == ':')  || (c == '^')
-			||  (c == '"')	|| (c == '`')) {
-			*dstPtr++ = '\\';
-		} else if ((c == '\\') || (c == '$')) {
-			// Some characters need to be escaped
-			// with *three* backslashes in a row.
-			*dstPtr++ = '\\';
-			*dstPtr++ = '\\';
-			*dstPtr++ = '\\';
-		}
-
-		// Note: we do not have to escape the
-		// { } ( ) < > and | characters.
-
-		*dstPtr++ = c;
-	}
-
-	*dstPtr++ = '"';
-	*dstPtr = '\0';
 }
 
 
-bool
+BString
-Grepper::_EscapeSpecialChars(char* buffer, ssize_t bufferSize)
+Grepper::_GetLine(BString& str, int32 pos)
 {
-	char* copy = strdup(buffer);
+	int32 startPos = 0;
-	char* start = buffer;
+	int32 endPos = str.Length();
-	uint32 len = strlen(copy);
+	for (int32 i = pos; i > 0; i--) {
-	bool result = true;
+		if (str[i] == '\n') {
-	for (uint32 count = 0; count < len; ++count) {
+			startPos = i + 1;
-		if (copy[count] == '"' || copy[count] == '$')
+			break;
-			*buffer++ = '\\';
+		}
-		if (buffer - start == bufferSize - 1) {
+	}
-			result = false;
+	for (int32 i = pos; i < endPos; i++) {
+		if (str[i] == '\n') {
+			endPos = i;
 			break;
 		}
-		*buffer++ = copy[count];
 	}
-	*buffer = '\0';
-	free(copy);
-	return result;
-}
 
+	BString ret;
+	return str.CopyInto(ret, startPos, endPos - startPos);
+}

src/apps/text_search/Grepper.h

@@ -6,6 +6,7 @@
 #define GREPPER_H
 
 #include <Messenger.h>
+#include <String.h>
 
 class FileIterator;
 class Model;
@@ -13,7 +14,7 @@ class Model;
 // Executes "grep" in a background thread.
 class Grepper {
 public:
-								Grepper(const char* pattern, const Model* model,
+								Grepper(BString pattern, const Model* model,
 									const BHandler* target,
 									FileIterator* iterator);
 	virtual						~Grepper();
@@ -30,16 +31,14 @@ private:
 	// The thread function that does the actual grepping.
 			int32				_GrepperThread();
 
-	// Remembers, and possibly escapes, the search pattern.
+	// Counts the number of linebreaks from startPos to endPos.
-			void				_SetPattern(const char* source);
+			int32				_CountLines(BString& str, int32 startPos,
-
+											int32 endPos);
-	// Prepends all quotes, dollars and backslashes with at backslash
+	// Gets the complete line that "pos" is on.
-	// to prevent the shell from misinterpreting them.
+			BString				_GetLine(BString& str, int32 pos);
-			bool				_EscapeSpecialChars(char* buffer,
-									ssize_t bufferSize);
 	private:
-	// The (escaped) search pattern.
+	// The search pattern.
-			char*				fPattern;
+			BString				fPattern;
 
 	// The settings from the model.
 			BMessenger			fTarget;

src/apps/text_search/Model.cpp

@@ -35,7 +35,6 @@ Model::Model()
 	fRecurseLinks(false),
 	fSkipDotDirs(true),
 	fCaseSensitive(false),
-	fEscapeText(true),
 	fTextOnly(true),
 	fInvokePe(false),
 
@@ -91,9 +90,6 @@ Model::LoadPrefs()
 			sizeof(int32)) > 0)
 		fCaseSensitive = (value != 0);
 
-	if (file.ReadAttr("EscapeText", B_INT32_TYPE, 0, &value, sizeof(int32)) > 0)
-		fEscapeText = (value != 0);
-
 	if (file.ReadAttr("TextOnly", B_INT32_TYPE, 0, &value, sizeof(int32)) > 0)
 		fTextOnly = (value != 0);
 
@@ -147,9 +143,6 @@ Model::SavePrefs()
 	value = fCaseSensitive ? 1 : 0;
 	file.WriteAttr("CaseSensitive", B_INT32_TYPE, 0, &value, sizeof(int32));
 
-	value = fEscapeText ? 1 : 0;
-	file.WriteAttr("EscapeText", B_INT32_TYPE, 0, &value, sizeof(int32));
-
 	value = fTextOnly ? 1 : 0;
 	file.WriteAttr("TextOnly", B_INT32_TYPE, 0, &value, sizeof(int32));
 

src/apps/text_search/Model.h

@@ -23,7 +23,6 @@ enum {
 	MSG_RECURSE_DIRS,
 	MSG_SKIP_DOT_DIRS,
 	MSG_CASE_SENSITIVE,
-	MSG_ESCAPE_TEXT,
 	MSG_TEXT_ONLY,
 	MSG_INVOKE_PE,
 	MSG_CHECKBOX_SHOW_LINES,
@@ -35,7 +34,6 @@ enum {
 
 	MSG_REPORT_FILE_NAME,
 	MSG_REPORT_RESULT,
-	MSG_REPORT_ERROR,
 	MSG_SEARCH_FINISHED,
 
 	MSG_NEW_WINDOW,
@@ -87,9 +85,6 @@ public:
 			// Whether the search is case sensitive.
 			bool				fCaseSensitive;
 
-			// Whether the search pattern will be escaped.
-			bool				fEscapeText;
-
 			// Whether we look at text files only.
 			bool				fTextOnly;