diff options
| -rw-r--r-- | glcdgraphics/common.c | 35 | ||||
| -rw-r--r-- | glcdgraphics/common.h | 8 | ||||
| -rw-r--r-- | glcdskin/parser.c | 2 | ||||
| -rw-r--r-- | glcdskin/xml.c | 96 | ||||
| -rw-r--r-- | glcdskin/xml.h | 13 |
5 files changed, 106 insertions, 48 deletions
diff --git a/glcdgraphics/common.c b/glcdgraphics/common.c index 8d6e8f6..1c78e23 100644 --- a/glcdgraphics/common.c +++ b/glcdgraphics/common.c @@ -59,15 +59,12 @@ std::string trim(const std::string & s) } -// character to return when erraneous utf-8 sequence (for now: space) -#define UTF8_ERRCODE 0x0020 -// for debugging issues return '_' instead: -//#define UTF8_ERRCODE 0x005F - -void encodedCharAdjustCounter(const bool isutf8, const std::string & str, uint32_t & c, unsigned int & i) +bool encodedCharAdjustCounter(const bool isutf8, const std::string & str, uint32_t & c, unsigned int & i, const uint32_t errChar) { + bool rv = false; + if (i >= str.length()) - return; + return rv; if ( isutf8 ) { uint8_t c0,c1,c2,c3; @@ -80,41 +77,47 @@ void encodedCharAdjustCounter(const bool isutf8, const std::string & str, uint32 if ( (c0 & 0x80) == 0x00) { // one byte: 0xxxxxxx c = c0; + rv = true; } else if ( (c0 & 0xE0) == 0xC0 ) { // two byte utf8: 110yyyyy 10xxxxxx -> 00000yyy yyxxxxxx if ( (c1 & 0xC0) == 0x80 ) { c = ( (c0 & 0x1F) << 6 ) | ( (c1 & 0x3F) ); + rv = true; } else { - //syslog(LOG_INFO, "GraphLCD: illegal 2-byte UTF-8 sequence found: 0x%02x 0x%02x\n", c0, c1); - c = UTF8_ERRCODE; + //syslog(LOG_INFO, "GraphLCD: illegal 2-byte UTF-8 sequence found: %02x %02x, pos=%d, str: %s\n", c0,c1,i,str.c_str()); + c = errChar; } - i += 1; + i += 1; } else if ( (c0 & 0xF0) == 0xE0 ) { // three byte utf8: 1110zzzz 10yyyyyy 10xxxxxx -> zzzzyyyy yyxxxxxx if ( ((c1 & 0xC0) == 0x80) && ((c2 & 0xC0) == 0x80) ) { c = ( (c0 & 0x0F) << 12 ) | ( (c1 & 0x3F) << 6 ) | ( c2 & 0x3F ); + rv = true; } else { - //syslog(LOG_INFO, "GraphLCD: illegal 3-byte UTF-8 sequence found: 0x%02x 0x%02x 0x%02x\n", c0, c1, c2); - c = UTF8_ERRCODE; + //syslog(LOG_INFO, "GraphLCD: illegal 3-byte UTF-8 sequence found: %02x %02x %02x, pos=%d, str: %s\n", c0,c1,c2,i,str.c_str()); + c = errChar; } i += 2; } else if ( (c0 & 0xF8) == 0xF0 ) { // four byte utf8: 11110www 10zzzzzz 10yyyyyy 10xxxxxx -> 000wwwzz zzzzyyyy yyxxxxxx if ( ((c1 & 0xC0) == 0x80) && ((c2 & 0xC0) == 0x80) && ((c3 & 0xC0) == 0x80) ) { c = ( (c0 & 0x07) << 18 ) | ( (c1 & 0x3F) << 12 ) | ( (c2 & 0x3F) << 6 ) | (c3 & 0x3F); + rv = true; } else { - //syslog(LOG_INFO, "GraphLCD: illegal 4-byte UTF-8 sequence found: 0x%02x 0x%02x 0x%02x 0x%02x\n", c0, c1, c2, c3); - c = UTF8_ERRCODE; + //syslog(LOG_INFO, "GraphLCD: illegal 4-byte UTF-8 sequence found: %02x %02x %02x %02x, pos=%d, str: %s\n", c0,c1,c2,c3,i,str.c_str()); + c = errChar; } i += 3; } else { // 1xxxxxxx is invalid! - //syslog(LOG_INFO, "GraphLCD: illegal 1-byte UTF-8 char found: 0x%02x\n", c0); - c = UTF8_ERRCODE; + //syslog(LOG_INFO, "GraphLCD: illegal 1-byte UTF-8 char found: %02x, pos=%d, str: %s\n", c0,i,str.c_str()); + c = errChar; } } else { c = str[i]; + rv = true; } + return rv; } } // end of namespace diff --git a/glcdgraphics/common.h b/glcdgraphics/common.h index 0043390..f0376bf 100644 --- a/glcdgraphics/common.h +++ b/glcdgraphics/common.h @@ -15,13 +15,19 @@ #include <string> #include <stdint.h> +// character to return when erraneous utf-8 sequence (for now: space) +//#define UTF8_ERRCHAR 0x0020 +// for debugging issues return '_' instead: +#define UTF8_ERRCHAR 0x005F + + namespace GLCD { void clip(int & value, int min, int max); void sort(int & value1, int & value2); std::string trim(const std::string & s); -void encodedCharAdjustCounter(const bool isutf8, const std::string & str, uint32_t & c, unsigned int & i); +bool encodedCharAdjustCounter(const bool isutf8, const std::string & str, uint32_t & c, unsigned int & i, const uint32_t errChar = UTF8_ERRCHAR); } // end of namespace diff --git a/glcdskin/parser.c b/glcdskin/parser.c index 0fc9689..d64f1c6 100644 --- a/glcdskin/parser.c +++ b/glcdskin/parser.c @@ -448,7 +448,7 @@ cSkin * XmlParse(cSkinConfig & Config, const std::string & Name, const std::stri skin = new cSkin(Config, Name); context.clear(); - cXML xml(fileName); + cXML xml(fileName, skin->Config().CharSet()); xml.SetNodeStartCB(StartElem); xml.SetNodeEndCB(EndElem); xml.SetCDataCB(CharData); diff --git a/glcdskin/xml.c b/glcdskin/xml.c index ed29d11..f36d7bb 100644 --- a/glcdskin/xml.c +++ b/glcdskin/xml.c @@ -15,32 +15,14 @@ #include <iostream> #include <fstream> +#include <string.h> + #include "xml.h" +#include "../glcdgraphics/common.h" namespace GLCD { -std::string trim(const std::string & s) -{ - std::string::size_type start, end; - - start = 0; - while (start < s.length()) - { - if (!isspace(s[start])) - break; - start++; - } - end = s.length() - 1; - while (end >= 0) - { - if (!isspace(s[end])) - break; - end--; - } - return s.substr(start, end - start + 1); -} - enum { LOOK4START, // looking for first element start LOOK4TAG, // looking for element tag @@ -57,7 +39,7 @@ enum { INCLOSETAG, // reading closing tag }; -cXML::cXML(const std::string & file) +cXML::cXML(const std::string & file, const std::string sysCharset) : nodestartcb(NULL), nodeendcb(NULL), cdatacb(NULL), @@ -66,6 +48,18 @@ cXML::cXML(const std::string & file) { char * buffer; long size; + sysEncoding = sysCharset; + sysIsUTF8 = (sysEncoding == "UTF-8"); + if (!sysIsUTF8) { + // convert from utf-8 to system encoding + iconv_cd = iconv_open(sysEncoding.c_str(), "UTF-8"); + if (iconv_cd == (iconv_t) -1) { + syslog(LOG_ERR, "ERROR: system encoding %s is not supported\n", sysEncoding.c_str()); + iconv_cd = NULL; + } + } else { + iconv_cd = NULL; + } #if (__GNUC__ < 3) std::ifstream f(file.c_str(), std::ios::in | std::ios::binary | std::ios::ate); @@ -89,6 +83,7 @@ cXML::cXML(const std::string & file) delete[] buffer; } +#if 0 cXML::cXML(const char * mem, unsigned int len) : nodestartcb(NULL), nodeendcb(NULL), @@ -98,6 +93,13 @@ cXML::cXML(const char * mem, unsigned int len) { data.assign(mem, len); } +#endif + +cXML::~cXML() +{ + if (iconv_cd != NULL) + iconv_close(iconv_cd); +} void cXML::SetNodeStartCB(XML_NODE_START_CB(cb)) { @@ -129,14 +131,26 @@ int cXML::Parse(void) int percent = 0; int last = 0; std::string::size_type len; + uint32_t c, c_tmp; + unsigned int i_old; + int l, char_size; state = LOOK4START; linenr = 1; skipping = false; len = data.length(); - for (std::string::size_type i = 0; i < len; i++) + + unsigned int i = 0; + while (i < (unsigned int)len) { - if (ReadChar(data[i]) != 0) + i_old = i; + encodedCharAdjustCounter(true, data, c_tmp, i); + char_size = (i - i_old) + 1; + c = 0; + for (l = 0 ; l < char_size; l++) + c += ( (0xFF & data[i_old + l]) << ( l << 3) ); + + if (ReadChar(c /*data[i]*/, char_size) != 0) return -1; if (progresscb) { @@ -147,6 +161,7 @@ int cXML::Parse(void) last = percent; } } + i++; } return 0; } @@ -156,8 +171,15 @@ bool cXML::IsTokenChar(bool start, int c) return isalpha(c) || c == '_' || (!start && isdigit(c)); } -int cXML::ReadChar(int c) +int cXML::ReadChar(unsigned int c, int char_size) { + // buffer for conversions (when conversion from utf8 to system encoding is required) + char convbufin[5]; + char convbufout[5]; + char* convbufinp = convbufin; + char* convbufoutp = convbufout; + size_t bufin_size, bufout_size, bufconverted; + // new line? if (c == '\n') linenr++; @@ -190,7 +212,29 @@ int cXML::ReadChar(int c) state = LOOK4TAG; } else - cdata += c; + { + int i; + //cdata += c; + // convert text-data on the fly if system encoding != UTF-8 + if (iconv_cd != NULL && char_size > 1 /* ((c & 0x80) == 0x80)*/) { + for (i = 0; i < char_size; i++) + convbufin[i] = ( (char)((c >> ( i << 3) ) & 0xFF) ); + convbufin[char_size] = '\0'; + bufin_size = strlen(convbufin); + bufout_size = bufin_size; + bufconverted = iconv(iconv_cd, &convbufinp, &bufin_size, &convbufoutp, &bufout_size); + + if (bufconverted != (size_t)-1 && strlen(convbufout) != 0) { + for (i = 0; i < (int)strlen(convbufout); i++) + cdata += convbufout[i]; + } else { + cdata += "?"; + } + } else { + for (i = 0; i < char_size; i++) + cdata += ( (unsigned char)((c >> ( i << 3) ) & 0xFF) ); + } + } // silently ignore until resync break; diff --git a/glcdskin/xml.h b/glcdskin/xml.h index 009679c..30210cc 100644 --- a/glcdskin/xml.h +++ b/glcdskin/xml.h @@ -15,6 +15,7 @@ #include <string> #include <map> +#include <iconv.h> namespace GLCD { @@ -36,7 +37,10 @@ private: bool skipping; int state; int linenr; - int delim; + unsigned int delim; + std::string sysEncoding; + bool sysIsUTF8; + iconv_t iconv_cd; std::string data, cdata, tag, attrn, attrv; std::map<std::string, std::string> attr; @@ -49,11 +53,12 @@ private: protected: bool IsTokenChar(bool start, int c); - int ReadChar(int c); + int ReadChar(unsigned int c, int char_size); public: - cXML(const std::string & file); - cXML(const char * mem, unsigned int len); + cXML(const std::string & file, const std::string sysCharset = "UTF-8"); + //cXML(const char * mem, unsigned int len); + ~cXML(); void SetNodeStartCB(XML_NODE_START_CB(cb)); void SetNodeEndCB(XML_NODE_END_CB(cb)); |