From 2659b4595d809cbd69a77e5ff7e2fc08d225f065 Mon Sep 17 00:00:00 2001
From: Andreas Stöckel <astoecke@techfak.uni-bielefeld.de>
Date: Sun, 15 Feb 2015 00:02:54 +0100
Subject: Tidied OsxmlEventParser up, implemented correct whitespace handling,
 started to write unit tests for the osxml parser

---
 src/core/common/Utils.hpp             | 21 ++++++++++--
 src/core/common/WhitespaceHandler.hpp | 60 +++++++++++++++++++++++++++++++++++
 2 files changed, 79 insertions(+), 2 deletions(-)

(limited to 'src/core')

diff --git a/src/core/common/Utils.hpp b/src/core/common/Utils.hpp
index 16a9136..8361973 100644
--- a/src/core/common/Utils.hpp
+++ b/src/core/common/Utils.hpp
@@ -119,9 +119,26 @@ public:
 	 */
 	template <class T, class Filter>
 	static std::pair<size_t, size_t> trim(const T &s, Filter f)
+	{
+		return trim(s, s.size(), f);
+	}
+
+	/**
+	 * Trims the given string or vector of chars by returning the start and end
+	 * index.
+	 *
+	 * @param s is the container that should be trimmed.
+	 * @param len is the number of elements in the container.
+	 * @param f is a function that returns true for values that should be
+	 * removed.
+	 * @return start and end index. Note that "end" points at the character
+	 * beyond the end, thus "end" minus "start"
+	 */
+	template <class T, class Filter>
+	static std::pair<size_t, size_t> trim(const T &s, size_t len, Filter f)
 	{
 		size_t start = 0;
-		for (size_t i = 0; i < s.size(); i++) {
+		for (size_t i = 0; i < len; i++) {
 			if (!f(s[i])) {
 				start = i;
 				break;
@@ -129,7 +146,7 @@ public:
 		}
 
 		size_t end = 0;
-		for (ssize_t i = s.size() - 1; i >= static_cast<ssize_t>(start); i--) {
+		for (ssize_t i = len - 1; i >= static_cast<ssize_t>(start); i--) {
 			if (!f(s[i])) {
 				end = i + 1;
 				break;
diff --git a/src/core/common/WhitespaceHandler.hpp b/src/core/common/WhitespaceHandler.hpp
index 79e0518..ed52ea3 100644
--- a/src/core/common/WhitespaceHandler.hpp
+++ b/src/core/common/WhitespaceHandler.hpp
@@ -97,6 +97,25 @@ public:
 	 * @param end is the end byte offset of the given character.
 	 */
 	void append(char c, size_t start, size_t end)
+	{
+		append(c, start, end, textBuf, textStart, textEnd);
+	}
+
+	/**
+	 * Static version of PreservingWhitespaceHandler append
+	 *
+	 * @param c is the character that should be appended to the internal buffer.
+	 * @param start is the start byte offset of the given character.
+	 * @param end is the end byte offset of the given character.
+	 * @param textBuf is a reference at the text buffer that is to be used.
+	 * @param textStart is a reference at the text start variable that is to be
+	 * used.
+	 * @param textEnd is a reference at the text end variable that is to be
+	 * used.
+	 */
+	static void append(char c, size_t start, size_t end,
+	                   std::vector<char> &textBuf, size_t &textStart,
+	                   size_t &textEnd)
 	{
 		if (textBuf.empty()) {
 			textStart = start;
@@ -129,6 +148,27 @@ public:
 	 * @param end is the end byte offset of the given character.
 	 */
 	void append(char c, size_t start, size_t end)
+	{
+		append(c, start, end, textBuf, textStart, textEnd, whitespaceBuf);
+	}
+
+	/**
+	 * Static version of TrimmingWhitespaceHandler append
+	 *
+	 * @param c is the character that should be appended to the internal buffer.
+	 * @param start is the start byte offset of the given character.
+	 * @param end is the end byte offset of the given character.
+	 * @param textBuf is a reference at the text buffer that is to be used.
+	 * @param textStart is a reference at the text start variable that is to be
+	 * used.
+	 * @param textEnd is a reference at the text end variable that is to be
+	 * used.
+	 * @param whitespaceBuf is a reference at the buffer for storing whitespace
+	 * characters.
+	 */
+	static void append(char c, size_t start, size_t end,
+	                   std::vector<char> &textBuf, size_t &textStart,
+	                   size_t &textEnd, std::vector<char> &whitespaceBuf)
 	{
 		// Handle whitespace characters
 		if (Utils::isWhitespace(c)) {
@@ -174,6 +214,26 @@ public:
 	 * @param end is the end byte offset of the given character.
 	 */
 	void append(char c, size_t start, size_t end)
+	{
+		append(c, start, end, textBuf, textStart, textEnd, hasWhitespace);
+	}
+
+	/**
+	 * Static version of CollapsingWhitespaceHandler append
+	 *
+	 * @param c is the character that should be appended to the internal buffer.
+	 * @param start is the start byte offset of the given character.
+	 * @param end is the end byte offset of the given character.
+	 * @param textBuf is a reference at the text buffer that is to be used.
+	 * @param textStart is a reference at the text start variable that is to be
+	 * used.
+	 * @param textEnd is a reference at the text end variable that is to be
+	 * used.
+	 * @param hasWhitespace is a reference at the "hasWhitespace" flag.
+	 */
+	static void append(char c, size_t start, size_t end,
+	                   std::vector<char> &textBuf, size_t &textStart,
+	                   size_t &textEnd, bool &hasWhitespace)
 	{
 		// Handle whitespace characters
 		if (Utils::isWhitespace(c)) {
-- 
cgit v1.2.3