SAS dataset label

src/bin/readstat.c

@@ -424,15 +424,19 @@ size_t readstat_strftime(char *s, size_t maxsize, const char *format, time_t tim
 static int dump_metadata(readstat_metadata_t *metadata, void *ctx) {
     printf("Columns: %d\n", readstat_get_var_count(metadata));
     printf("Rows: %d\n", readstat_get_row_count(metadata));
+    const char *table_name = readstat_get_table_name(metadata);
     const char *file_label = readstat_get_file_label(metadata);
     const char *orig_encoding = readstat_get_file_encoding(metadata);
     long version = readstat_get_file_format_version(metadata);
     time_t timestamp = readstat_get_creation_time(metadata);
     readstat_compress_t compression = readstat_get_compression(metadata);
     readstat_endian_t endianness = readstat_get_endianness(metadata);
 
+    if (table_name && table_name[0]) {
+        printf("Table name: %s\n", table_name);
+    }
     if (file_label && file_label[0]) {
-        printf("File label: %s\n", file_label);
+        printf("Table label: %s\n", file_label);
     }
     if (version) {
         printf("Format version: %ld\n", version);

src/readstat.h

@@ -483,7 +483,7 @@ typedef struct readstat_writer_s {
 
     int                         row_count;
     int                         current_row;
-    char                        file_label[100];
+    char                        file_label[257];
     char                        table_name[33];
     const readstat_variable_t  *fweight_variable;
 

src/sas/readstat_sas.c

@@ -189,7 +189,7 @@ readstat_error_t sas_read_header(readstat_io_t *io, sas_header_info_t *hinfo,
         retval = READSTAT_ERROR_UNSUPPORTED_CHARSET;
         goto cleanup;
     }
-    memcpy(hinfo->file_label, header_start.file_label, sizeof(header_start.file_label));
+    memcpy(hinfo->table_name, header_start.table_name, sizeof(header_start.table_name));
     if (io->seek(hinfo->pad1, READSTAT_SEEK_CUR, io->io_ctx) == -1) {
         retval = READSTAT_ERROR_SEEK;
         goto cleanup;
@@ -317,16 +317,16 @@ readstat_error_t sas_write_header(readstat_writer_t *writer, sas_header_info_t *
     struct tm epoch_tm = { .tm_year = 60, .tm_mday = 1 };
     time_t epoch = mktime(&epoch_tm);
 
-    memset(header_start.file_label, ' ', sizeof(header_start.file_label));
+    memset(header_start.table_name, ' ', sizeof(header_start.table_name));
 
-    size_t file_label_len = strlen(writer->file_label);
-    if (file_label_len > sizeof(header_start.file_label))
-        file_label_len = sizeof(header_start.file_label);
+    size_t table_name_len = strlen(writer->table_name);
+    if (table_name_len > sizeof(header_start.table_name))
+        table_name_len = sizeof(header_start.table_name);
 
-    if (file_label_len) {
-        memcpy(header_start.file_label, writer->file_label, file_label_len);
+    if (table_name_len) {
+        memcpy(header_start.table_name, writer->table_name, table_name_len);
     } else {
-        memcpy(header_start.file_label, "DATASET", sizeof("DATASET")-1);
+        memcpy(header_start.table_name, "DATASET", sizeof("DATASET")-1);
     }
 
     retval = readstat_write_bytes(writer, &header_start, sizeof(sas_header_start_t));
@@ -379,7 +379,7 @@ readstat_error_t sas_write_header(readstat_writer_t *writer, sas_header_info_t *
         goto cleanup;
 
     sas_header_end_t header_end = {
-        .host = "W32_VSPRO"
+        .host = "9.0401M6Linux"
     };
 
     char release[sizeof(header_end.release)+1] = { 0 };

src/sas/readstat_sas.h

@@ -17,7 +17,8 @@ typedef struct sas_header_start_s {
     unsigned char encoding;
     unsigned char mystery5[13];
     char          file_type[8];
-    char          file_label[64];
+    char          table_name[32];
+    unsigned char mystery6[32];
     char          file_info[8];
 } sas_header_start_t;
 
@@ -47,7 +48,8 @@ typedef struct sas_header_info_s {
     int64_t  header_size;
     time_t   creation_time;
     time_t   modification_time;
-    char     file_label[64];
+    char     table_name[32];
+    char     file_label[256];
     char    *encoding;
 } sas_header_info_t;
 

src/sas/readstat_sas7bcat_read.c

@@ -415,7 +415,7 @@ readstat_error_t readstat_parse_sas7bcat(readstat_parser_t *parser, const char *
     }
 
     if (ctx->metadata_handler) {
-        char file_label[4*64+1];
+        char table_name[4*32+1];
         readstat_metadata_t metadata = { 
             .file_encoding = ctx->input_encoding, /* orig encoding? */
             .modified_time = hinfo->modification_time,
@@ -424,12 +424,12 @@ readstat_error_t readstat_parse_sas7bcat(readstat_parser_t *parser, const char *
             .endianness = hinfo->little_endian ? READSTAT_ENDIAN_LITTLE : READSTAT_ENDIAN_BIG,
             .is64bit = ctx->u64
         };
-        retval = readstat_convert(file_label, sizeof(file_label), 
-                hinfo->file_label, sizeof(hinfo->file_label), ctx->converter);
+        retval = readstat_convert(table_name, sizeof(table_name),
+                hinfo->table_name, sizeof(hinfo->table_name), ctx->converter);
         if (retval != READSTAT_OK)
             goto cleanup;
 
-        metadata.file_label = file_label;
+        metadata.table_name = table_name;
 
         if (ctx->metadata_handler(&metadata, ctx->user_ctx) != READSTAT_HANDLER_OK) {
             retval = READSTAT_ERROR_USER_ABORT;

src/sas/readstat_sas7bdat_read.c

@@ -85,7 +85,8 @@ typedef struct sas7bdat_ctx_s {
     time_t         ctime;
     time_t         mtime;
     int            version;
-    char           file_label[4*64+1];
+    char           table_name[4*32+1];
+    char           file_label[4*256+1];
     char           error_buf[2048];
 
     unsigned int  rdc_compression:1;
@@ -285,6 +286,7 @@ static readstat_error_t sas7bdat_parse_column_name_subheader(const char *subhead
     int i;
     const char *cnp = &subheader[signature_len+8];
     uint16_t remainder = sas_read2(&subheader[signature_len], ctx->bswap);
+    int off;
 
     if (remainder != sas_subheader_remainder(len, signature_len)) {
         retval = READSTAT_ERROR_PARSE;
@@ -298,6 +300,39 @@ static readstat_error_t sas7bdat_parse_column_name_subheader(const char *subhead
 
     for (i=ctx->col_names_count-cmax; i<ctx->col_names_count; i++) {
         ctx->col_info[i].name_ref = sas7bdat_parse_text_ref(cnp, ctx);
+        if (i == 0) {
+            if (ctx->text_blobs == NULL || ctx->text_blob_lengths == NULL) {
+                retval = READSTAT_ERROR_PARSE;
+                goto cleanup;
+            }
+            if (ctx->version < 9) {
+                off = 36;
+            } else {
+                if (ctx->text_blob_lengths[0] < 19) {
+                    retval = READSTAT_ERROR_PARSE;
+                    goto cleanup;
+                }
+                if (!memcmp(&ctx->text_blobs[0][12], "SASYZCR", 7)) {
+                        off = 44;
+                } else {
+                    off = ctx->u64 ? 36 : 12;
+                }
+            }
+            if (ctx->col_info[0].name_ref.offset >=
+                ctx->text_blob_lengths[0] ||
+                ctx->col_info[0].name_ref.offset < off) {
+                retval = READSTAT_ERROR_PARSE;
+                goto cleanup;
+            }
+            retval = readstat_convert(ctx->file_label,
+                                      sizeof(ctx->file_label),
+                                      &ctx->text_blobs[0][off],
+                                      ctx->col_info[0].name_ref.offset - off,
+                                      ctx->converter
+                                      );
+            if (retval != READSTAT_OK)
+                goto cleanup;
+        }
         cnp += 8;
     }
 
@@ -712,6 +747,7 @@ static readstat_error_t sas7bdat_submit_columns(sas7bdat_ctx_t *ctx, int compres
         readstat_metadata_t metadata = {
             .row_count = ctx->row_limit,
             .var_count = ctx->column_count,
+            .table_name = ctx->table_name,
             .file_label = ctx->file_label,
             .file_encoding = ctx->input_encoding, /* orig encoding? */
             .creation_time = ctx->ctime,
@@ -1219,8 +1255,8 @@ readstat_error_t readstat_parse_sas7bdat(readstat_parser_t *parser, const char *
         ctx->converter = converter;
     }
 
-    if ((retval = readstat_convert(ctx->file_label, sizeof(ctx->file_label),
-                hinfo->file_label, sizeof(hinfo->file_label), ctx->converter)) != READSTAT_OK) {
+    if ((retval = readstat_convert(ctx->table_name, sizeof(ctx->table_name),
+                hinfo->table_name, sizeof(hinfo->table_name), ctx->converter)) != READSTAT_OK) {
         goto cleanup;
     }
 

src/sas/readstat_sas7bdat_write.c

@@ -140,7 +140,7 @@ static readstat_error_t sas7bdat_emit_header(readstat_writer_t *writer, sas_head
         .file_format = SAS_FILE_FORMAT_UNIX,
         .encoding = 20, /* UTF-8 */
         .file_type = "SAS FILE",
-        .file_info = "DATA ~ ~"
+        .file_info = "DATA    "
     };
 
     memcpy(&header_start.magic, sas7bdat_magic_number, sizeof(header_start.magic));
@@ -218,13 +218,16 @@ static sas7bdat_subheader_t *sas7bdat_col_name_subheader_init(readstat_writer_t
     sas7bdat_subheader_t *subheader = sas7bdat_subheader_init(
             SAS_SUBHEADER_SIGNATURE_COLUMN_NAME, len);
     memcpy(&subheader->data[signature_len], &remainder, sizeof(uint16_t));
+
+    sas_text_ref_t text_ref = sas7bdat_make_text_ref(column_text_array, "READSTAT");
+    text_ref = sas7bdat_make_text_ref(column_text_array, writer->file_label);
 
     int i;
     char *ptrs = &subheader->data[signature_len+8];
     for (i=0; i<writer->variables_count; i++) {
         readstat_variable_t *variable = readstat_get_variable(writer, i);
         const char *name = readstat_variable_get_name(variable);
-        sas_text_ref_t text_ref = sas7bdat_make_text_ref(column_text_array, name);
+        text_ref = sas7bdat_make_text_ref(column_text_array, name);
         memcpy(&ptrs[0], &text_ref.index, sizeof(uint16_t));
         memcpy(&ptrs[2], &text_ref.offset, sizeof(uint16_t));
         memcpy(&ptrs[4], &text_ref.length, sizeof(uint16_t));

src/sas/readstat_xport_read.c

@@ -32,7 +32,7 @@ typedef struct xport_ctx_s {
     int            row_offset;
     size_t         row_length;
     int            parsed_row_count;
-    char           file_label[40*4+1];
+    char           file_label[256*4+1];
     char           table_name[32*4+1];
 
     readstat_variable_t **variables;

src/test/test_read.c

@@ -77,14 +77,16 @@ void parse_ctx_reset(rt_parse_ctx_t *parse_ctx, long file_format) {
     } else if ((file_format & RT_FORMAT_SAV)) {
         parse_ctx->max_file_label_len = 64;
     } else if ((file_format & RT_FORMAT_SAS7BDAT)) {
-        parse_ctx->max_file_label_len = 64;
+        parse_ctx->max_table_name_len = 32;
+        parse_ctx->max_file_label_len = 256;
     } else {
         parse_ctx->max_file_label_len = 20;
     }
     if ((file_format & RT_FORMAT_XPORT_5)) {
         parse_ctx->max_table_name_len = 8;
     } else if ((file_format & RT_FORMAT_XPORT_8)) {
         parse_ctx->max_table_name_len = 32;
+        parse_ctx->max_file_label_len = 256;
     }
     parse_ctx->var_index = -1;
     parse_ctx->obs_index = -1;

src/test/test_write.c

@@ -31,7 +31,9 @@ readstat_error_t write_file_to_buffer(rt_test_file_t *file, rt_buffer_t *buffer,
 
     readstat_writer_t *writer = readstat_writer_init();
     readstat_set_data_writer(writer, &write_data);
-    readstat_writer_set_file_label(writer, file->label);
+    if ((format & RT_FORMAT_SAS7BCAT)) {
+        strncpy(file->label, "", 1);
+    } else readstat_writer_set_file_label(writer, file->label);
     readstat_writer_set_table_name(writer, file->table_name);
     readstat_writer_set_error_handler(writer, &handle_error);
     if (file->timestamp.tm_year) {