Import RegExp classes from Ham.

Minor adjustments made by myself to fit into Haiku better.

headers/private/shared/RegExp.h

@@ -0,0 +1,78 @@
+/*
+ * Copyright 2013, Ingo Weinhold, ingo_weinhold@gmx.de.
+ * Distributed under the terms of the MIT License.
+ */
+#ifndef REG_EXP_H
+#define REG_EXP_H
+
+
+#include <stddef.h>
+
+
+class RegExp {
+public:
+			enum PatternType {
+				PATTERN_TYPE_REGULAR_EXPRESSION,
+				PATTERN_TYPE_WILDCARD
+			};
+
+			class MatchResult;
+
+public:
+								RegExp();
+								RegExp(const char* pattern,
+									PatternType patternType
+										= PATTERN_TYPE_REGULAR_EXPRESSION);
+								RegExp(const RegExp& other);
+								~RegExp();
+
+			bool				IsValid() const
+									{ return fData != NULL; }
+
+			bool				SetPattern(const char* pattern,
+									PatternType patternType
+										= PATTERN_TYPE_REGULAR_EXPRESSION);
+
+			MatchResult			Match(const char* string) const;
+
+			RegExp&				operator=(const RegExp& other);
+
+private:
+			struct Data;
+			struct MatchResultData;
+
+private:
+			Data*				fData;
+};
+
+
+class RegExp::MatchResult {
+public:
+								MatchResult();
+								MatchResult(const MatchResult& other);
+								~MatchResult();
+
+			bool				HasMatched() const;
+
+			size_t				StartOffset() const;
+			size_t				EndOffset() const;
+
+			size_t				GroupCount() const;
+			size_t				GroupStartOffsetAt(size_t index) const;
+			size_t				GroupEndOffsetAt(size_t index) const;
+
+			MatchResult&		operator=(const MatchResult& other);
+
+private:
+			friend class RegExp;
+
+private:
+								MatchResult(MatchResultData* data);
+									// takes over the data reference
+
+private:
+			MatchResultData*	fData;
+};
+
+
+#endif	// REG_EXP_H

src/kits/shared/Jamfile

@@ -28,6 +28,7 @@ StaticLibrary libshared.a :
 	NaturalCompare.cpp
 	PromptWindow.cpp
 	QueryFile.cpp
+	RegExp.cpp
 	RWLockManager.cpp
 	SHA256.cpp
 	ShakeTrackingFilter.cpp

src/kits/shared/RegExp.cpp

@@ -0,0 +1,386 @@
+/*
+ * Copyright 2013, Ingo Weinhold, ingo_weinhold@gmx.de.
+ * Copyright 2013, Rene Gollent, rene@gollent.com.
+ * Distributed under the terms of the MIT License.
+ */
+
+
+#include <RegExp.h>
+
+#include <new>
+
+#include <regex.h>
+
+#include <String.h>
+
+#include <Referenceable.h>
+
+
+// #pragma mark - RegExp::Data
+
+
+struct RegExp::Data : public BReferenceable {
+	Data(const char* pattern, PatternType patternType)
+		:
+		BReferenceable()
+	{
+		// convert the shell pattern to a regular expression
+		BString patternString;
+		if (patternType == PATTERN_TYPE_WILDCARD) {
+			while (*pattern != '\0') {
+				char c = *pattern++;
+				switch (c) {
+					case '?':
+						patternString += '.';
+						continue;
+					case '*':
+						patternString += ".*";
+						continue;
+					case '[':
+					{
+						// find the matching ']' first
+						const char* end = pattern;
+						while (*end != ']') {
+							if (*end++ == '\0') {
+								fError = REG_EBRACK;
+								return;
+							}
+						}
+
+						if (pattern == end) {
+							// Empty bracket expression. It will never match
+							// anything. Strictly speaking this is not
+							// considered an error, but we handle it like one.
+							fError = REG_EBRACK;
+							return;
+						}
+
+						patternString += '[';
+
+						// We need to avoid "[." ... ".]", "[=" ... "=]", and
+						// "[:" ... ":]" sequences, since those have special
+						// meaning in regular expressions. If we encounter
+						// a '[' followed by either of '.', '=', or ':', we
+						// replace the '[' by "[.[.]".
+						while (pattern < end) {
+							c = *pattern++;
+							if (c == '[' && pattern < end) {
+								switch (*pattern) {
+									case '.':
+									case '=':
+									case ':':
+										patternString += "[.[.]";
+										continue;
+								}
+							}
+							patternString += c;
+						}
+
+						pattern++;
+						patternString += ']';
+						break;
+					}
+
+					case '\\':
+					{
+						// Quotes the next character. Works the same way for
+						// regular expressions.
+						if (*pattern == '\0') {
+							fError = REG_EESCAPE;
+							return;
+						}
+
+						patternString += '\\';
+						patternString += *pattern++;
+						break;
+					}
+
+					case '^':
+					case '.':
+					case '$':
+					case '(':
+					case ')':
+					case '|':
+					case '+':
+					case '{':
+						// need to be quoted
+						patternString += '\\';
+						// fall through
+					default:
+						patternString += c;
+						break;
+				}
+			}
+
+			pattern = patternString.String();
+		}
+
+		fError = regcomp(&fCompiledExpression, pattern, REG_EXTENDED);
+	}
+
+	~Data()
+	{
+		if (fError == 0)
+			regfree(&fCompiledExpression);
+	}
+
+	bool IsValid() const
+	{
+		return fError == 0;
+	}
+
+	const regex_t* CompiledExpression() const
+	{
+		return &fCompiledExpression;
+	}
+
+private:
+	int		fError;
+	regex_t	fCompiledExpression;
+};
+
+
+// #pragma mark - RegExp::MatchResultData
+
+
+struct RegExp::MatchResultData : public BReferenceable {
+	MatchResultData(const regex_t* compiledExpression, const char* string)
+		:
+		BReferenceable(),
+		fMatchCount(0),
+		fMatches(NULL)
+	{
+		// Do the matching: Since we need to provide a buffer for the matches
+		// for regexec() to fill in, but don't know the number of matches
+		// beforehand, we need to guess and retry with a larger buffer, if it
+		// wasn't large enough.
+		size_t maxMatchCount = 32;
+		for (;;) {
+			fMatches = new regmatch_t[maxMatchCount];
+			if (regexec(compiledExpression, string, maxMatchCount, fMatches, 0)
+					!= 0) {
+				delete[] fMatches;
+				fMatches = NULL;
+				fMatchCount = 0;
+				break;
+			}
+
+			if (fMatches[maxMatchCount - 1].rm_so == -1) {
+				// determine the match count
+				size_t lower = 0;
+				size_t upper = maxMatchCount;
+				while (lower < upper) {
+					size_t mid = (lower + upper) / 2;
+					if (fMatches[mid].rm_so == -1)
+						upper = mid;
+					else
+						lower = mid + 1;
+				}
+				fMatchCount = lower;
+				break;
+			}
+
+			// buffer too small -- try again with larger buffer
+			delete[] fMatches;
+			fMatches = NULL;
+			maxMatchCount *= 2;
+		}
+	}
+
+	~MatchResultData()
+	{
+		delete[] fMatches;
+	}
+
+	size_t MatchCount() const
+	{
+		return fMatchCount;
+	}
+
+	const regmatch_t* Matches() const
+	{
+		return fMatches;
+	}
+
+private:
+	size_t		fMatchCount;
+	regmatch_t*	fMatches;
+};
+
+
+// #pragma mark - RegExp
+
+
+RegExp::RegExp()
+	:
+	fData(NULL)
+{
+}
+
+
+RegExp::RegExp(const char* pattern, PatternType patternType)
+	:
+	fData(NULL)
+{
+	SetPattern(pattern, patternType);
+}
+
+
+RegExp::RegExp(const RegExp& other)
+	:
+	fData(other.fData)
+{
+	if (fData != NULL)
+		fData->AcquireReference();
+}
+
+
+RegExp::~RegExp()
+{
+	if (fData != NULL)
+		fData->ReleaseReference();
+}
+
+
+bool
+RegExp::SetPattern(const char* pattern, PatternType patternType)
+{
+	if (fData != NULL) {
+		fData->ReleaseReference();
+		fData = NULL;
+	}
+
+	fData = new Data(pattern, patternType);
+	if (!fData->IsValid()) {
+		delete fData;
+		fData = NULL;
+		return false;
+	}
+
+	return true;
+}
+
+
+RegExp::MatchResult
+RegExp::Match(const char* string) const
+{
+	if (!IsValid())
+		return MatchResult();
+
+	return MatchResult(
+		new(std::nothrow) MatchResultData(fData->CompiledExpression(),
+			string));
+}
+
+
+RegExp&
+RegExp::operator=(const RegExp& other)
+{
+	if (fData != NULL)
+		fData->ReleaseReference();
+
+	fData = other.fData;
+
+	if (fData != NULL)
+		fData->AcquireReference();
+
+	return *this;
+}
+
+
+// #pragma mark - RegExp::MatchResult
+
+
+RegExp::MatchResult::MatchResult()
+	:
+	fData(NULL)
+{
+}
+
+
+RegExp::MatchResult::MatchResult(MatchResultData* data)
+	:
+	fData(data)
+{
+}
+
+
+RegExp::MatchResult::MatchResult(const MatchResult& other)
+	:
+	fData(other.fData)
+{
+	if (fData != NULL)
+		fData->AcquireReference();
+}
+
+
+RegExp::MatchResult::~MatchResult()
+{
+	if (fData != NULL)
+		fData->ReleaseReference();
+}
+
+
+bool
+RegExp::MatchResult::HasMatched() const
+{
+	return fData != NULL && fData->MatchCount() > 0;
+}
+
+
+size_t
+RegExp::MatchResult::StartOffset() const
+{
+	return fData != NULL && fData->MatchCount() > 0
+		? fData->Matches()[0].rm_so : 0;
+}
+
+
+size_t
+RegExp::MatchResult::EndOffset() const
+{
+	return fData != NULL && fData->MatchCount() > 0
+		? fData->Matches()[0].rm_eo : 0;
+}
+
+
+size_t
+RegExp::MatchResult::GroupCount() const
+{
+	if (fData == NULL)
+		return 0;
+
+	size_t matchCount = fData->MatchCount();
+	return matchCount > 0 ? matchCount - 1 : 0;
+}
+
+
+size_t
+RegExp::MatchResult::GroupStartOffsetAt(size_t index) const
+{
+	return fData != NULL && fData->MatchCount() > index + 1
+		? fData->Matches()[index + 1].rm_so : 0;
+}
+
+
+size_t
+RegExp::MatchResult::GroupEndOffsetAt(size_t index) const
+{
+	return fData != NULL && fData->MatchCount() > index + 1
+		? fData->Matches()[index + 1].rm_eo : 0;
+}
+
+
+RegExp::MatchResult&
+RegExp::MatchResult::operator=(const MatchResult& other)
+{
+	if (fData != NULL)
+		fData->ReleaseReference();
+
+	fData = other.fData;
+
+	if (fData != NULL)
+		fData->AcquireReference();
+
+	return *this;
+}