#include "gedcom_internal.h"
#include "multilex.h"
#include "encoding.h"
+#include "encoding_state.h"
#include "xref.h"
int line_no = 0;
#define NEW_MODEL_FILE "new.ged"
-int lexer_init(ENCODING enc, FILE* f)
+int lexer_init(Encoding enc, FILE* f)
{
if (enc == ONE_BYTE) {
lf = &gedcom_1byte_lex;
gedcom_1byte_myinit(f);
- set_encoding_width(enc);
+ set_read_encoding_width(enc);
return open_conv_to_internal("ASCII");
}
else if (enc == TWO_BYTE_HILO) {
lf = &gedcom_hilo_lex;
gedcom_hilo_myinit(f);
- set_encoding_width(enc);
+ set_read_encoding_width(enc);
return open_conv_to_internal("UNICODE");
}
else if (enc == TWO_BYTE_LOHI) {
lf = &gedcom_lohi_lex;
gedcom_lohi_myinit(f);
- set_encoding_width(enc);
+ set_read_encoding_width(enc);
return open_conv_to_internal("UNICODE");
}
else {
return (*lf)();
}
+void rewind_file(FILE* f)
+{
+ if (fseek(f, 0, 0) != 0)
+ gedcom_warning(_("Error positioning input file: %s"), strerror(errno));
+}
+
int determine_encoding(FILE* f)
{
char first[2];
int read;
+ set_read_encoding_bom(WITHOUT_BOM);
read = fread(first, 1, 2, f);
if (read != 2) {
gedcom_warning(_("Error reading from input file: %s"), strerror(errno));
+ rewind_file(f);
return ONE_BYTE;
}
else if ((first[0] == '0') && (first[1] == ' ')) {
- gedcom_debug_print(_("One-byte encoding"));
- if (fseek(f, 0, 0) != 0)
- gedcom_warning(_("Error positioning input file: %s"), strerror(errno));
+ gedcom_debug_print("One-byte encoding");
+ rewind_file(f);
return ONE_BYTE;
}
- else if ((first[0] == '\0') && (first[1] == '0'))
- {
- gedcom_debug_print(_("Two-byte encoding, high-low"));
- if (fseek(f, 0, 0) != 0)
- gedcom_warning(_("Error positioning input file: %s"), strerror(errno));
+ else if ((first[0] == '\0') && (first[1] == '0')) {
+ gedcom_debug_print("Two-byte encoding, high-low");
+ rewind_file(f);
return TWO_BYTE_HILO;
}
- else if ((first[0] == '\xFE') && (first[1] == '\xFF'))
- {
- gedcom_debug_print(_("Two-byte encoding, high-low, with BOM"));
+ else if ((first[0] == '\xFE') && (first[1] == '\xFF')) {
+ gedcom_debug_print("Two-byte encoding, high-low, with BOM");
+ set_read_encoding_bom(WITH_BOM);
return TWO_BYTE_HILO;
}
- else if ((first[0] == '0') && (first[1] == '\0'))
- {
- gedcom_debug_print(_("Two-byte encoding, low-high"));
- if (fseek(f, 0, 0) != 0)
- gedcom_warning(_("Error positioning input file: %s"), strerror(errno));
+ else if ((first[0] == '0') && (first[1] == '\0')) {
+ gedcom_debug_print("Two-byte encoding, low-high");
+ rewind_file(f);
return TWO_BYTE_LOHI;
}
- else if ((first[0] == '\xFF') && (first[1] == '\xFE'))
- {
- gedcom_debug_print(_("Two-byte encoding, low-high, with BOM"));
+ else if ((first[0] == '\xFF') && (first[1] == '\xFE')) {
+ gedcom_debug_print("Two-byte encoding, low-high, with BOM");
+ set_read_encoding_bom(WITH_BOM);
return TWO_BYTE_LOHI;
}
+ else if ((first[0] == '\xEF') && (first[1] == '\xBB')) {
+ read = fread(first, 1, 1, f);
+ if (read != 1) {
+ gedcom_warning(_("Error reading from input file: %s"), strerror(errno));
+ rewind_file(f);
+ }
+ else if (first[0] == '\xBF') {
+ set_read_encoding_bom(WITH_BOM);
+ gedcom_debug_print("UTF-8 encoding, with BOM");
+ }
+ else {
+ gedcom_warning(_("Unknown encoding, falling back to one-byte"));
+ rewind_file(f);
+ }
+ return ONE_BYTE;
+ }
else {
gedcom_warning(_("Unknown encoding, falling back to one-byte"));
- if (fseek(f, 0, 0) != 0)
- gedcom_warning(_("Error positioning input file: %s"), strerror(errno));
+ rewind_file(f);
return ONE_BYTE;
}
}
int init_called = 0;
-void gedcom_init()
+int gedcom_init()
{
init_called = 1;
update_gconv_search_path();
- setlocale(LC_ALL, "");
+ init_encodings();
+ if (!setlocale(LC_ALL, "")
+ || ! bindtextdomain(PACKAGE, LOCALEDIR)
+ || ! bind_textdomain_codeset(PACKAGE, INTERNAL_ENCODING))
+ return 1;
+ else
+ return 0;
}
int gedcom_parse_file(const char* file_name)
{
- ENCODING enc;
+ Encoding enc;
int result = 1;
FILE* file;
- char *textdom, *save_textdom;
-
- textdom = textdomain(NULL);
- if (!textdom) {
- gedcom_error(_("Could not retrieve text domain: %s"), strerror(errno));
- return result;
- }
-
- save_textdom = strdup(textdom);
- if (! save_textdom) {
- MEMORY_ERROR;
- return result;
- }
-
- if (! bindtextdomain(PACKAGE, LOCALEDIR)
- || ! bind_textdomain_codeset(PACKAGE, INTERNAL_ENCODING)
- || ! textdomain(PACKAGE)) {
- gedcom_error(_("Could not set text domain: %s"), strerror(errno));
- return result;
- }
if (!init_called) {
gedcom_error(_("Internal error: GEDCOM parser not initialized"));
file_name, strerror(errno));
}
else {
- init_encodings();
+ line_no = 1;
enc = determine_encoding(file);
if (lexer_init(enc, file)) {
}
}
- if (! textdomain(save_textdom)) {
- gedcom_error(_("Could not restore text domain: %s"), strerror(errno));
- return result;
- }
return result;
}