From 8c34b2d9748311840a93f64494f56a5b1e81015f Mon Sep 17 00:00:00 2001 From: narroo Date: Sat, 15 Feb 2014 19:56:36 -0500 Subject: Addressed Issue #402. cIniFile can now process UTF-8 files that have a Byte Order Marker, BOM. --- lib/inifile/iniFile.cpp | 156 +++++++++++++++++++++++++++--------------------- lib/inifile/iniFile.h | 2 + 2 files changed, 91 insertions(+), 67 deletions(-) diff --git a/lib/inifile/iniFile.cpp b/lib/inifile/iniFile.cpp index afa1c110d..7b0df3d68 100644 --- a/lib/inifile/iniFile.cpp +++ b/lib/inifile/iniFile.cpp @@ -83,91 +83,97 @@ bool cIniFile::ReadFile(const AString & a_FileName, bool a_AllowExampleRedirect) } } - while (getline(f, line)) - { - // To be compatible with Win32, check for existence of '\r'. - // Win32 files have the '\r' and Unix files don't at the end of a line. - // Note that the '\r' will be written to INI files from - // Unix so that the created INI file can be read under Win32 - // without change. - size_t lineLength = line.length(); - if (lineLength == 0) - { - continue; - } - if (line[lineLength - 1] == '\r') - { - line = line.substr(0, lineLength - 1); - } + if (getline(f, line)) + { + // Removes UTF-8 Byte Order Markers (BOM) if, present. + RemoveBom(line); - if (line.length() == 0) + do { - continue; - } + // To be compatible with Win32, check for existence of '\r'. + // Win32 files have the '\r' and Unix files don't at the end of a line. + // Note that the '\r' will be written to INI files from + // Unix so that the created INI file can be read under Win32 + // without change. + size_t lineLength = line.length(); + if (lineLength == 0) + { + continue; + } + if (line[lineLength - 1] == '\r') + { + line = line.substr(0, lineLength - 1); + } - // Check that the user hasn't opened a binary file by checking the first - // character of each line! - if (!isprint(line[0])) - { - printf("%s: Binary-check failed on char %d\n", __FUNCTION__, line[0]); - f.close(); - return false; - } - if ((pLeft = line.find_first_of(";#[=")) == AString::npos) - { - continue; - } + if (line.length() == 0) + { + continue; + } - switch (line[pLeft]) - { + // Check that the user hasn't opened a binary file by checking the first + // character of each line! + if (!isprint(line[0])) + { + printf("%s: Binary-check failed on char %d\n", __FUNCTION__, line[0]); + f.close(); + return false; + } + if ((pLeft = line.find_first_of(";#[=")) == AString::npos) + { + continue; + } + + switch (line[pLeft]) + { case '[': { - if ( - ((pRight = line.find_last_of("]")) != AString::npos) && - (pRight > pLeft) - ) - { - keyname = line.substr(pLeft + 1, pRight - pLeft - 1); - AddKeyName(keyname); - } - break; + if ( + ((pRight = line.find_last_of("]")) != AString::npos) && + (pRight > pLeft) + ) + { + keyname = line.substr(pLeft + 1, pRight - pLeft - 1); + AddKeyName(keyname); + } + break; } case '=': { - valuename = line.substr(0, pLeft); - value = line.substr(pLeft + 1); - AddValue(keyname, valuename, value); - break; + valuename = line.substr(0, pLeft); + value = line.substr(pLeft + 1); + AddValue(keyname, valuename, value); + break; } case ';': case '#': { - if (names.size() == 0) - { - AddHeaderComment(line.substr(pLeft + 1)); - } - else - { - AddKeyComment(keyname, line.substr(pLeft + 1)); - } - break; + if (names.size() == 0) + { + AddHeaderComment(line.substr(pLeft + 1)); + } + else + { + AddKeyComment(keyname, line.substr(pLeft + 1)); + } + break; } - } // switch (line[pLeft]) - } // while (getline()) + } // switch (line[pLeft]) + } while (getline(f, line)); // do - f.close(); - if (names.size() == 0) - { - return false; - } - - if (IsFromExampleRedirect) - { - WriteFile(FILE_IO_PREFIX + a_FileName); + f.close(); + if (names.size() == 0) + { + return false; + } + + if (IsFromExampleRedirect) + { + WriteFile(FILE_IO_PREFIX + a_FileName); + } + return true; } - return true; } @@ -824,3 +830,19 @@ AString cIniFile::CheckCase(const AString & s) const + +void cIniFile::RemoveBom(AString & a_line) const +{ + // The BOM sequence for UTF-8 is 0xEF,0xBB,0xBF ( In Unicode Latin I:  ) + static char BOM[] = { 0xEF, 0xBB, 0xBF }; + + // The BOM sequence, if present, is always the first three characters of the input. + if (a_line.compare(0, 3, BOM) == 0) + { + a_line.erase(0, 3); + } +} + + + + diff --git a/lib/inifile/iniFile.h b/lib/inifile/iniFile.h index 40af618dc..0bf1d917e 100644 --- a/lib/inifile/iniFile.h +++ b/lib/inifile/iniFile.h @@ -51,6 +51,8 @@ private: /// If the object is case-insensitive, returns s as lowercase; otherwise returns s as-is AString CheckCase(const AString & s) const; + /// Removes the UTF-8 BOMs (Byte order makers), if present. + void RemoveBom(AString & a_line) const; public: enum errors { -- cgit v1.2.3