Index: /trunk/include/iprt/cpp/ministring.h =================================================================== --- /trunk/include/iprt/cpp/ministring.h (revision 35566) +++ /trunk/include/iprt/cpp/ministring.h (revision 35567) @@ -185,7 +185,7 @@ * String length in bytes. * - * Returns the length of the member string, which is equal to strlen(c_str()). - * In other words, this does not count unicode codepoints but returns the number - * of bytes. This is always cached so calling this is cheap and requires no + * Returns the length of the member string in bytes, which is equal to strlen(c_str()). + * In other words, this does not count unicode codepoints; use utf8length() for that. + * The byte length is always cached so calling this is cheap and requires no * strlen() invocation. * @@ -195,4 +195,17 @@ { return m_cch; + } + + /** + * String length in UTF-8 codepoints. + * + * As opposed to length(), which returns the length in bytes, this counts the number + * of UTF-8 codepoints. This is *not* cached so calling this is expensive. + * + * @returns Number of codepoints in the member string. + */ + size_t utf8length() const + { + return m_psz ? RTStrUniLen(m_psz) : 0; } @@ -652,6 +665,6 @@ * Find the given substring. * - * Looks for pcszFind in "this" starting at "pos" and returns its position, - * counting from the beginning of "this" at 0. + * Looks for pcszFind in "this" starting at "pos" and returns its position + * as a byte (not codepoint) offset, counting from the beginning of "this" at 0. * * @param pcszFind The substring to find. @@ -676,7 +689,23 @@ * Returns a substring of "this" as a new Utf8Str. * - * Works exactly like its equivalent in std::string except that this interprets - * pos and n as unicode codepoints instead of bytes. With the default - * parameters "0" and "npos", this always copies the entire string. + * Works exactly like its equivalent in std::string. With the default + * parameters "0" and "npos", this always copies the entire string. The + * "pos" and "n" arguments represent bytes; it is the caller's responsibility + * to ensure that the offsets do not copy invalid UTF-8 sequences. When + * used in conjunction with find() and length(), this will work. + * + * @param pos Index of first byte offset to copy from "this", counting from 0. + * @param n Number of bytes to copy, starting with the one at "pos". + * The copying will stop if the null terminator is encountered before + * n bytes have been copied. + */ + iprt::MiniString substr(size_t pos = 0, size_t n = npos) const + { + return MiniString(*this, pos, n); + } + + /** + * Returns a substring of "this" as a new Utf8Str. As opposed to substr(), + * this variant takes codepoint offsets instead of byte offsets. * * @param pos Index of first unicode codepoint to copy from @@ -686,8 +715,6 @@ * terminator is encountered before n codepoints have * been copied. - * - * @remarks This works on code points, not bytes! - */ - iprt::MiniString substr(size_t pos = 0, size_t n = npos) const; + */ + iprt::MiniString substrCP(size_t pos = 0, size_t n = npos) const; /** Index: /trunk/src/VBox/Runtime/common/string/ministring.cpp =================================================================== --- /trunk/src/VBox/Runtime/common/string/ministring.cpp (revision 35566) +++ /trunk/src/VBox/Runtime/common/string/ministring.cpp (revision 35567) @@ -224,5 +224,5 @@ } -MiniString MiniString::substr(size_t pos /*= 0*/, size_t n /*= npos*/) +MiniString MiniString::substrCP(size_t pos /*= 0*/, size_t n /*= npos*/) const { @@ -256,11 +256,14 @@ size_t cbCopy = psz - pFirst; - ret.reserve(cbCopy + 1); // may throw bad_alloc -#ifndef RT_EXCEPTIONS_ENABLED - AssertRelease(capacity() >= cbCopy + 1); -#endif - memcpy(ret.m_psz, pFirst, cbCopy); - ret.m_cch = cbCopy; - ret.m_psz[cbCopy] = '\0'; + if (cbCopy) + { + ret.reserve(cbCopy + 1); // may throw bad_alloc +#ifndef RT_EXCEPTIONS_ENABLED + AssertRelease(capacity() >= cbCopy + 1); +#endif + memcpy(ret.m_psz, pFirst, cbCopy); + ret.m_cch = cbCopy; + ret.m_psz[cbCopy] = '\0'; + } } } Index: /trunk/src/VBox/Runtime/testcase/tstIprtMiniString.cpp =================================================================== --- /trunk/src/VBox/Runtime/testcase/tstIprtMiniString.cpp (revision 35566) +++ /trunk/src/VBox/Runtime/testcase/tstIprtMiniString.cpp (revision 35567) @@ -211,4 +211,44 @@ CHECK_EQUAL(SubStr15, "cdef"); + /* substr() and substrCP() functions */ + iprt::MiniString strTest(""); + CHECK_EQUAL(strTest.substr(0), ""); + CHECK_EQUAL(strTest.substrCP(0), ""); + CHECK_EQUAL(strTest.substr(1), ""); + CHECK_EQUAL(strTest.substrCP(1), ""); + + /* now let's have some non-ASCII to chew on */ + strTest = "abcdefßäbcdef"; + // 13 codepoints, but 15 bytes (excluding null terminator); + // "ß" and "ä" consume two bytes each + CHECK_EQUAL(strTest.substr(0), strTest.c_str()); + CHECK_EQUAL(strTest.substrCP(0), strTest.c_str()); + + CHECK_EQUAL(strTest.substr(2), "cdefßäbcdef"); + CHECK_EQUAL(strTest.substrCP(2), "cdefßäbcdef"); + + CHECK_EQUAL(strTest.substr(2, 2), "cd"); + CHECK_EQUAL(strTest.substrCP(2, 2), "cd"); + + CHECK_EQUAL(strTest.substr(6), "ßäbcdef"); + CHECK_EQUAL(strTest.substrCP(6), "ßäbcdef"); + + CHECK_EQUAL(strTest.substr(6, 2), "ß"); // UTF-8 "ß" consumes two bytes + CHECK_EQUAL(strTest.substrCP(6, 1), "ß"); + + CHECK_EQUAL(strTest.substr(8), "äbcdef"); // UTF-8 "ß" consumes two bytes + CHECK_EQUAL(strTest.substrCP(7), "äbcdef"); + + CHECK_EQUAL(strTest.substr(8, 3), "äb"); // UTF-8 "ä" consumes two bytes + CHECK_EQUAL(strTest.substrCP(7, 2), "äb"); + + CHECK_EQUAL(strTest.substr(14, 1), "f"); + CHECK_EQUAL(strTest.substrCP(12, 1), "f"); + + CHECK_EQUAL(strTest.substr(15, 1), ""); + CHECK_EQUAL(strTest.substrCP(13, 1), ""); + + CHECK_EQUAL(strTest.substr(16, 1), ""); + CHECK_EQUAL(strTest.substrCP(15, 1), ""); /* special constructor and assignment arguments */