X-Git-Url: https://git.dlugolecki.net.pl/?a=blobdiff_plain;f=gedcom%2Fencoding.c;h=b2c3c416955e65d4465f4dc7eff2e71e12d57ee7;hb=f7b7f4752136dc40686676194a2895d82fe6018e;hp=7e00f767b9f85cdc546fceba9ff36e1d0318fb24;hpb=ba123c650193eb37ab152b5083b67364e8bacf30;p=gedcom-parse.git diff --git a/gedcom/encoding.c b/gedcom/encoding.c index 7e00f76..b2c3c41 100644 --- a/gedcom/encoding.c +++ b/gedcom/encoding.c @@ -27,13 +27,13 @@ #include #include #include "gedcom_internal.h" +#include "gedcom.h" #include "encoding.h" #include "hash.h" #define ENCODING_CONF_FILE "gedcom.enc" #define GCONV_SEARCH_PATH "GCONV_PATH" #define MAXBUF 255 -#define INIT_NR_ENCODINGS 10 static iconv_t cd_to_internal = (iconv_t) -1; static ENCODING the_enc = ONE_BYTE; @@ -43,7 +43,7 @@ char* charwidth_string[] = { "1", "2_HILO", "2_LOHI" }; hnode_t *node_alloc(void *c __attribute__((unused))) { - return malloc(sizeof *node_alloc(NULL)); + return (hnode_t *)malloc(sizeof *node_alloc(NULL)); } void node_free(hnode_t *n, void *c __attribute__((unused))) @@ -60,17 +60,23 @@ void add_encoding(char *gedcom_n, char* charwidth, char *iconv_n) key = (char *) malloc(strlen(gedcom_n) + strlen(charwidth) + 3); val = (char *) malloc(strlen(iconv_n) + 1); - /* sprintf is safe here (malloc'ed before) */ - sprintf(key, "%s(%s)", gedcom_n, charwidth); - strcpy(val, iconv_n); - - if (hash_lookup(encodings, key)) { - gedcom_warning(_("Duplicate entry found for encoding '%s', ignoring"), - gedcom_n); - } - else { - hash_alloc_insert(encodings, key, val); + if (key && val) { + /* sprintf is safe here (malloc'ed before) */ + sprintf(key, "%s(%s)", gedcom_n, charwidth); + strcpy(val, iconv_n); + + if (hash_lookup(encodings, key)) { + gedcom_warning(_("Duplicate entry found for encoding '%s', ignoring"), + gedcom_n); + free(key); + free(val); + } + else { + hash_alloc_insert(encodings, key, val); + } } + else + MEMORY_ERROR; } char* get_encoding(char* gedcom_n, ENCODING enc) @@ -79,16 +85,23 @@ char* get_encoding(char* gedcom_n, ENCODING enc) hnode_t *node; key = (char*)malloc(strlen(gedcom_n) + strlen(charwidth_string[enc]) + 3); - /* sprintf is safe here (malloc'ed before) */ - sprintf(key, "%s(%s)", gedcom_n, charwidth_string[enc]); - node = hash_lookup(encodings, key); - free(key); - if (node) { - return hnode_get(node); + if (key) { + /* sprintf is safe here (malloc'ed before) */ + sprintf(key, "%s(%s)", gedcom_n, charwidth_string[enc]); + + node = hash_lookup(encodings, key); + free(key); + if (node) { + return hnode_get(node); + } + else { + gedcom_error(_("No encoding defined for '%s'"), gedcom_n); + return NULL; + } } else { - gedcom_error(_("No encoding defined for '%s'"), gedcom_n); + MEMORY_ERROR; return NULL; } } @@ -98,6 +111,62 @@ void cleanup_encodings() hash_free(encodings); } +/* Let function be called before main() */ +void update_gconv_search_path() __attribute__ ((constructor)); + +/* Note: + + The environment variable GCONV_PATH has to be adjusted before the very + first call of iconv_open. For the most general case, it means that we + have to make our own constructor here (in case some of the other library + constructors would use iconv_open). + + However, it looks like a change of an environment variable in a constructor + doesn't always survive until the main() function. This is the case if + the environment variable is a new one, for which there was no room yet + in the initial environment. The initial environment is located on the + stack, but when variables are added, it is moved to the heap (to be able + to grow). Now, the main function takes again the one from the stack, not + from the heap, so changes are lost. + + For this, the function below will also be called in gedcom_init(), which + needs to be called as early as possible in the program. + */ + +void update_gconv_search_path() +{ + char *gconv_path; + /* Add gedcom data directory to gconv search path */ + gconv_path = getenv(GCONV_SEARCH_PATH); + if (gconv_path == NULL || strstr(gconv_path, PKGDATADIR) == NULL) { + char *new_gconv_path; + if (gconv_path == NULL) { + new_gconv_path = (char *)malloc(strlen(GCONV_SEARCH_PATH) + + strlen(PKGDATADIR) + + 2); + if (new_gconv_path) + sprintf(new_gconv_path, "%s=%s", GCONV_SEARCH_PATH, PKGDATADIR); + } + else { + new_gconv_path = (char *)malloc(strlen(GCONV_SEARCH_PATH) + + strlen(gconv_path) + + strlen(PKGDATADIR) + + 3); + if (new_gconv_path) + sprintf(new_gconv_path, "%s=%s:%s", + GCONV_SEARCH_PATH, gconv_path, PKGDATADIR); + } + if (new_gconv_path) + /* Ignore failures of putenv (can't do anything about it anyway) */ + putenv(new_gconv_path); + else { + fprintf(stderr, "Could not allocate memory at %s, %d\n", + __FILE__, __LINE__); + abort(); + } + } +} + void init_encodings() { if (encodings == NULL) { @@ -106,34 +175,12 @@ void init_encodings() char gedcom_n[MAXBUF + 1]; char charwidth[MAXBUF + 1]; char iconv_n[MAXBUF + 1]; - char *gconv_path; - atexit(cleanup_encodings); - - /* Add gedcom data directory to gconv search path */ - gconv_path = getenv(GCONV_SEARCH_PATH); - if (gconv_path == NULL || strstr(gconv_path, PKGDATADIR) == NULL) { - char *new_gconv_path; - if (gconv_path == NULL) { - new_gconv_path = (char *)malloc(strlen(GCONV_SEARCH_PATH) - + strlen(PKGDATADIR) - + 2); - sprintf(new_gconv_path, "%s=%s", GCONV_SEARCH_PATH, PKGDATADIR); - } - else { - new_gconv_path = (char *)malloc(strlen(GCONV_SEARCH_PATH) - + strlen(gconv_path) - + strlen(PKGDATADIR) - + 3); - sprintf(new_gconv_path, "%s=%s:%s", - GCONV_SEARCH_PATH, gconv_path, PKGDATADIR); - } - if (putenv(new_gconv_path) != 0) { - gedcom_warning(_("Failed updating conversion module path")); - } + if (atexit(cleanup_encodings) != 0) { + gedcom_warning(_("Could not register encoding cleanup function")); } - - encodings = hash_create(INIT_NR_ENCODINGS, NULL, NULL); + + encodings = hash_create(HASHCOUNT_T_MAX, NULL, NULL); hash_set_allocator(encodings, node_alloc, node_free, NULL); /* Open gedcom configuration file and read */ @@ -144,8 +191,8 @@ void init_encodings() in = fopen(path, "r"); } if (in == NULL) { - gedcom_warning(_("Could not open encoding configuration file '%s'"), - ENCODING_CONF_FILE); + gedcom_warning(_("Could not open encoding configuration file '%s': %s"), + ENCODING_CONF_FILE, strerror(errno)); } else { line_no = 1; @@ -166,7 +213,10 @@ void init_encodings() } } } - fclose(in); + if (fclose(in) != 0) { + gedcom_warning(_("Error closing file '%s': %s"), + ENCODING_CONF_FILE, strerror(errno)); + } } } } @@ -201,16 +251,20 @@ int open_conv_to_internal(char* fromcode) void close_conv_to_internal() { - iconv_close(cd_to_internal); + if (iconv_close(cd_to_internal) != 0) { + gedcom_warning(_("Error closing conversion context: %s"), strerror(errno)); + } cd_to_internal = (iconv_t) -1; } char* to_internal(char* str, size_t len, char* output_buffer, size_t out_len) { + size_t res; size_t outsize = out_len; char *wrptr = output_buffer; char *rdptr = conv_buf; + char *retval = output_buffer; /* set up input buffer (concatenate to what was left previous time) */ /* can't use strcpy, because possible null bytes from unicode */ memcpy(conv_buf + conv_buf_size, str, len); @@ -218,9 +272,24 @@ char* to_internal(char* str, size_t len, /* set up output buffer (empty it) */ memset(output_buffer, 0, out_len); /* do the conversion */ - iconv(cd_to_internal, &rdptr, &conv_buf_size, &wrptr, &outsize); + res = iconv(cd_to_internal, &rdptr, &conv_buf_size, &wrptr, &outsize); + if (res == (size_t)-1) { + if (errno == EILSEQ) { + /* restart from an empty state and return NULL */ + iconv(cd_to_internal, NULL, NULL, NULL, NULL); + retval = NULL; + rdptr++; + conv_buf_size--; + } + else if (errno == EINVAL) { + /* Do nothing, leave it to next iteration */ + } + else { + gedcom_error(_("Error in converting characters: %s"), strerror(errno)); + } + } /* then shift what is left over to the head of the input buffer */ memmove(conv_buf, rdptr, conv_buf_size); memset(conv_buf + conv_buf_size, 0, sizeof(conv_buf) - conv_buf_size); - return output_buffer; + return retval; }