Clean up and unify encoding conversions

[odcread.git] / odcread.cc

diff --git a/odcread.cc b/odcread.cc
index 858028e6babf5e252ede8ade0c0f71d2946ceb4e..d2aa025e12713d516fa5355d7d4751a03c5a1a50 100644 (file)
--- a/odcread.cc
+++ b/odcread.cc
@@ -9,6 +9,13 @@
 #include <textmodel.h>
 #include <visitor.h>
 
+// Character encoding conversions
+#include <langinfo.h> // determine the current charset
+#include <locale.h> // locale support
+#include <iconv.h> // charset conversions
+#include <errno.h> // error codes
+#include <string.h> // string descriptions of error codes
+
 namespace odc {
        class Context {
                public:
@@ -80,14 +87,61 @@ namespace odc {
                virtual void foldRight() {
                        terminateContext();
                }
+               char *getCharSet() {
+                       return nl_langinfo(CODESET);
+               }
                virtual void textShortPiece(const ShortPiece *piece) {
-                       std::string text = piece->getText();
-                       d_context.top()->addPiece(text);
+                       std::string str = convert((char *)piece->getBuffer(), piece->size() + 1, (char *)"ISO-8859-1", 1);
+                       d_context.top()->addPiece(str);
                }
                virtual void textLongPiece(const LongPiece *piece) {
-                       throw "Long Piece not handled";
-                       //std::string text = piece->getText();
-                       //d_context.top()->addPiece(text);
+                       std::string str = convert((char *)piece->getBuffer(), piece->size() + 2, (char *)"UCS-2", 2);
+                       d_context.top()->addPiece(str);
+               }
+
+               /**
+                * Convert an input character buffer in the given encoding to the
+                * locale's encoding.
+                */
+               std::string convert(char *in, size_t bytesIn, char *encodingIn, size_t inBytesPerChar) {
+                       // Convert from the input encoding to the locale's encoding
+                       iconv_t conv = iconv_open(getCharSet(), encodingIn);
+
+                       // Handle errors by throwing a readable message
+                       if (conv == (iconv_t)-1) {
+                               std::string str("iconv initialization error: ");
+                               str += strerror(errno);
+                               throw str.c_str();
+                       }
+
+                       // Assume at most 4 bytes per character are needed
+                       size_t bytesOut = 4 * bytesIn / inBytesPerChar;
+
+                       // Allocate the output buffer
+                       char *out = new char[bytesOut];
+                       char *outPtr = out;
+
+                       // Perform conversion
+                       size_t rval = iconv(conv, &in, &bytesIn, &outPtr, &bytesOut);
+                       if (rval == (size_t)-1) {
+                               std::string str("iconv error: ");
+                               str += strerror(errno);
+                               throw str.c_str();
+                       }
+
+                       // Free the iconv state
+                       iconv_close(conv);
+
+                       // Copy result into a std::string
+                       std::string str(out);
+                       delete out;
+
+                       // Convert newlines
+                       for (std::string::iterator it = str.begin(); it < str.end(); ++it) {
+                               if (*it == '\r') *it = '\n';
+                       }
+
+                       return str;
                }
        };
 
@@ -112,6 +166,10 @@ int main(int argc, char *argv[]) {
        if (argc < 2) {
                return 1;
        }
+
+       // Set the locale according to the terminal's environment
+       setlocale(LC_ALL, "");
+
        std::ifstream in(argv[1], std::ios::in | std::ios::binary);
 
        odc::Store* s;