readstat_error_t readstat_parse_sas7bcat(readstat_parser_t *parser, const char *path, void *user_ctx) {
readstat_error_t retval = READSTAT_OK;
readstat_io_t *io = parser->io;
int64_t i;
char *page = NULL;
char *buffer = NULL;
sas7bcat_ctx_t *ctx = calloc(1, sizeof(sas7bcat_ctx_t));
sas_header_info_t *hinfo = calloc(1, sizeof(sas_header_info_t));
ctx->block_pointers = malloc((ctx->block_pointers_capacity = 200) * sizeof(uint64_t));
ctx->value_label_handler = parser->value_label_handler;
ctx->metadata_handler = parser->metadata_handler;
ctx->input_encoding = parser->input_encoding;
ctx->output_encoding = parser->output_encoding;
ctx->user_ctx = user_ctx;
ctx->io = io;
if (io->open(path, io->io_ctx) == -1) {
retval = READSTAT_ERROR_OPEN;
goto cleanup;
}
if ((retval = sas_read_header(io, hinfo, parser->error_handler, user_ctx)) != READSTAT_OK) {
goto cleanup;
}
ctx->u64 = hinfo->u64;
ctx->pad1 = hinfo->pad1;
ctx->bswap = machine_is_little_endian() ^ hinfo->little_endian;
ctx->header_size = hinfo->header_size;
ctx->page_count = hinfo->page_count;
ctx->page_size = hinfo->page_size;
if (ctx->input_encoding == NULL) {
ctx->input_encoding = hinfo->encoding;
}
if (ctx->u64) {
retval = READSTAT_ERROR_PARSE;
goto cleanup;
}
if (ctx->input_encoding && ctx->output_encoding && strcmp(ctx->input_encoding, ctx->output_encoding) != 0) {
iconv_t converter = iconv_open(ctx->output_encoding, ctx->input_encoding);
if (converter == (iconv_t)-1) {
retval = READSTAT_ERROR_UNSUPPORTED_CHARSET;
goto cleanup;
}
ctx->converter = converter;
}
if (parser->metadata_handler) {
char file_label[4*64+1];
retval = readstat_convert(file_label, sizeof(file_label),
hinfo->file_label, sizeof(hinfo->file_label), ctx->converter);
if (retval != READSTAT_OK)
goto cleanup;
if (ctx->metadata_handler(file_label, hinfo->modification_time,
10000 * hinfo->major_version + hinfo->minor_version, ctx->user_ctx) != READSTAT_HANDLER_OK) {
retval = READSTAT_ERROR_USER_ABORT;
goto cleanup;
}
}
if ((page = malloc(ctx->page_size)) == NULL) {
retval = READSTAT_ERROR_MALLOC;
goto cleanup;
}
if (io->seek(ctx->header_size+SAS_CATALOG_FIRST_INDEX_PAGE*ctx->page_size, READSTAT_SEEK_SET, io->io_ctx) == -1) {
retval = READSTAT_ERROR_SEEK;
goto cleanup;
}
if (io->read(page, ctx->page_size, io->io_ctx) < ctx->page_size) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
sas7bcat_augment_index(&page[856+2*ctx->pad1], ctx->page_size - 856 - 2*ctx->pad1, ctx);
// Pass 1 -- find the XLSR entries
for (i=SAS_CATALOG_USELESS_PAGES; i<ctx->page_count; i++) {
if (io->seek(ctx->header_size+i*ctx->page_size, READSTAT_SEEK_SET, io->io_ctx) == -1) {
retval = READSTAT_ERROR_SEEK;
goto cleanup;
}
if (io->read(page, ctx->page_size, io->io_ctx) < ctx->page_size) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
if (memcmp(&page[16], "XLSR", sizeof("XLSR")-1) == 0) {
sas7bcat_augment_index(&page[16], ctx->page_size - 16, ctx);
}
}
sas7bcat_sort_index(ctx);
sas7bcat_uniq_index(ctx);
// Pass 2 -- look up the individual block pointers
for (i=0; i<ctx->block_pointers_used; i++) {
int start_page = ctx->block_pointers[i] >> 32;
int start_page_pos = (ctx->block_pointers[i]) & 0xFFFF;
int buffer_len = sas7bcat_block_size(start_page, start_page_pos, ctx, &retval);
if (buffer_len == -1) {
goto cleanup;
} else if (buffer_len == 0) {
continue;
}
buffer = realloc(buffer, buffer_len);
if ((retval = sas7bcat_read_block(buffer, buffer_len, start_page, start_page_pos, ctx)) != READSTAT_OK)
goto cleanup;
if ((retval = sas7bcat_parse_block(buffer, buffer_len, ctx)) != READSTAT_OK)
goto cleanup;
}
cleanup:
io->close(io->io_ctx);
if (page)
free(page);
if (buffer)
free(buffer);
if (ctx)
sas7bcat_ctx_free(ctx);
if (hinfo)
free(hinfo);
return retval;
}
readstat_error_t readstat_parse_sas7bcat(readstat_parser_t *parser, const char *filename, void *user_ctx) {
readstat_error_t retval = READSTAT_OK;
int64_t i;
char *page = NULL;
char *buffer = NULL;
sas_catalog_ctx_t *ctx = calloc(1, sizeof(sas_catalog_ctx_t));
sas_header_info_t *hinfo = calloc(1, sizeof(sas_header_info_t));
ctx->block_pointers = malloc((ctx->block_pointers_capacity = 200) * sizeof(uint64_t));
ctx->value_label_handler = parser->value_label_handler;
ctx->user_ctx = user_ctx;
if ((ctx->fd = readstat_open(filename)) == -1) {
retval = READSTAT_ERROR_OPEN;
goto cleanup;
}
if ((retval = sas_read_header(ctx->fd, hinfo, parser->error_handler, user_ctx)) != READSTAT_OK) {
goto cleanup;
}
ctx->u64 = hinfo->u64;
ctx->pad1 = hinfo->pad1;
ctx->bswap = machine_is_little_endian() ^ hinfo->little_endian;
ctx->header_size = hinfo->header_size;
ctx->page_count = hinfo->page_count;
ctx->page_size = hinfo->page_size;
if (ctx->u64) {
retval = READSTAT_ERROR_PARSE;
goto cleanup;
}
if (strcmp(hinfo->encoding, "UTF-8") != 0 &&
strcmp(hinfo->encoding, "US-ASCII") != 0) {
iconv_t converter = iconv_open("UTF-8", hinfo->encoding);
if (converter == (iconv_t)-1) {
retval = READSTAT_ERROR_UNSUPPORTED_CHARSET;
goto cleanup;
}
ctx->converter = converter;
}
if ((page = malloc(ctx->page_size)) == NULL) {
retval = READSTAT_ERROR_MALLOC;
goto cleanup;
}
if (readstat_lseek(ctx->fd, ctx->header_size+SAS_CATALOG_FIRST_INDEX_PAGE*ctx->page_size, SEEK_SET) == -1) {
retval = READSTAT_ERROR_SEEK;
goto cleanup;
}
if (read(ctx->fd, page, ctx->page_size) < ctx->page_size) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
sas_catalog_augment_index(&page[856+2*ctx->pad1], ctx->page_size - 856 - 2*ctx->pad1, ctx);
// Pass 1 -- find the XLSR entries
for (i=SAS_CATALOG_USELESS_PAGES; i<ctx->page_count; i++) {
if (readstat_lseek(ctx->fd, ctx->header_size+i*ctx->page_size, SEEK_SET) == -1) {
retval = READSTAT_ERROR_SEEK;
goto cleanup;
}
if (read(ctx->fd, page, ctx->page_size) < ctx->page_size) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
if (memcmp(&page[16], "XLSR", sizeof("XLSR")-1) == 0) {
sas_catalog_augment_index(&page[16], ctx->page_size - 16, ctx);
}
}
// Pass 2 -- look up the individual block pointers
for (i=0; i<ctx->block_pointers_used; i++) {
int start_page = ctx->block_pointers[i] >> 32;
int start_page_pos = (ctx->block_pointers[i]) & 0xFFFF;
int buffer_len = sas_catalog_block_size(start_page, start_page_pos, ctx, &retval);
if (buffer_len == -1) {
goto cleanup;
} else if (buffer_len == 0) {
continue;
}
buffer = realloc(buffer, buffer_len);
if ((retval = sas_catalog_read_block(buffer, buffer_len, start_page, start_page_pos, ctx)) != READSTAT_OK)
goto cleanup;
if ((retval = sas_catalog_parse_block(buffer, buffer_len, ctx)) != READSTAT_OK)
goto cleanup;
}
cleanup:
if (page)
free(page);
if (buffer)
free(buffer);
if (ctx)
sas_catalog_ctx_free(ctx);
if (hinfo)
free(hinfo);
return retval;
}