Merge pull request #48 from rafal-c/feature/header_handler

Implement a handler for file header

src/rdata.h

@@ -42,6 +42,22 @@ typedef enum rdata_file_format_e {
     RDATA_SINGLE_OBJECT
 } rdata_file_format_t;
 
+typedef enum rdata_compression_e {
+	RDATA_COMPRESSION_NONE,
+	RDATA_COMPRESSION_GZIP,
+	RDATA_COMPRESSION_BZIP2,
+	RDATA_COMPRESSION_LZMA
+} rdata_compression_t;
+
+#pragma pack(push, 1)
+typedef struct rdata_header_s {
+  char       header[2];
+  uint32_t   format_version;
+  uint32_t   writer_version;
+  uint32_t   reader_version;
+} rdata_header_t;
+#pragma pack(pop)
+
 const char *rdata_error_message(rdata_error_t error_code);
 
 typedef int (*rdata_column_handler)(const char *name, rdata_type_t type,
@@ -52,6 +68,15 @@ typedef int (*rdata_column_name_handler)(const char *value, int index, void *ctx
 typedef void (*rdata_error_handler)(const char *error_message, void *ctx);
 typedef int (*rdata_progress_handler)(double progress, void *ctx);
 
+/* Callback type for passing meta-information from file header to the caller. Arguments:
+ * - compression - compression algorithm used on the file,
+ * - header_line - first 5 bytes of the file containing information about file type (RData vs RDS, ascii vs binary, etc.),
+ * - header - structure with information about file version, writer R version and minimal R version required to read the file back,
+ * - ctx - user context, same as in all other handlers
+ * header_line and header must not be free'd by the user.
+ */
+typedef int (*rdata_header_handler)(rdata_compression_t compression, const char *header_line, const rdata_header_t* header, void *ctx);
+
 #if defined(_MSC_VER)
 #include <BaseTsd.h>
 typedef SSIZE_T ssize_t;
@@ -96,6 +121,7 @@ typedef struct rdata_parser_s {
     rdata_column_handler        dim_handler;
     rdata_text_value_handler    dim_name_handler;
     rdata_error_handler         error_handler;
+    rdata_header_handler        header_handler;
     rdata_io_t                 *io;
 } rdata_parser_t;
 
@@ -111,6 +137,7 @@ rdata_error_t rdata_set_value_label_handler(rdata_parser_t *parser, rdata_text_v
 rdata_error_t rdata_set_dim_handler(rdata_parser_t *parser, rdata_column_handler dim_handler);
 rdata_error_t rdata_set_dim_name_handler(rdata_parser_t *parser, rdata_text_value_handler dim_name_handler);
 rdata_error_t rdata_set_error_handler(rdata_parser_t *parser, rdata_error_handler error_handler);
+rdata_error_t rdata_set_header_handler(rdata_parser_t *parser, rdata_header_handler header_handler);
 rdata_error_t rdata_set_open_handler(rdata_parser_t *parser, rdata_open_handler open_handler);
 rdata_error_t rdata_set_close_handler(rdata_parser_t *parser, rdata_close_handler close_handler);
 rdata_error_t rdata_set_seek_handler(rdata_parser_t *parser, rdata_seek_handler seek_handler);

src/rdata_internal.h

@@ -6,13 +6,6 @@
 
 #pragma pack(push, 1)
 
-typedef struct rdata_v2_header_s {
-    char       header[2];
-    uint32_t   format_version;
-    uint32_t   writer_version;
-    uint32_t   reader_version;
-} rdata_v2_header_t;
-
 typedef struct rdata_sexptype_header_s {
     unsigned int   type:8;
     unsigned int   object:1;
@@ -78,3 +71,6 @@ typedef struct rdata_sexptype_info_s {
 #define RDATA_SEXPTYPE_LANGUAGE_OBJECT_ATTR    240
 #define RDATA_SEXPTYPE_PAIRLIST_ATTR           239
 #define RDATA_PSEUDO_SXP_ALTREP                238
+
+/* we read this many characters from the beginning of the file to determine file format */
+#define RDATA_HEADER_LENGTH 5

src/rdata_parser.c

@@ -63,6 +63,11 @@ rdata_error_t rdata_set_error_handler(rdata_parser_t *parser, rdata_error_handle
     return RDATA_OK;
 }
 
+rdata_error_t rdata_set_header_handler(rdata_parser_t *parser, rdata_header_handler header_handler) {
+    parser->header_handler = header_handler;
+    return RDATA_OK;
+}
+
 rdata_error_t rdata_set_open_handler(rdata_parser_t *parser, rdata_open_handler open_handler) {
     parser->io->open = open_handler;
     return RDATA_OK;

src/rdata_read.c

@@ -60,7 +60,8 @@ typedef struct rdata_ctx_s {
     rdata_text_value_handler     value_label_handler;
     rdata_column_handler         dim_handler;
     rdata_text_value_handler     dim_name_handler;
-    rdata_error_handler       error_handler;
+    rdata_error_handler          error_handler;
+    rdata_header_handler         header_handler;
     void                        *user_ctx;
 #if HAVE_BZIP2
     bz_stream                   *bz_strm;
@@ -656,9 +657,10 @@ void free_rdata_ctx(rdata_ctx_t *ctx) {
 rdata_error_t rdata_parse(rdata_parser_t *parser, const char *filename, void *user_ctx) {
     int is_rdata = 0;
     rdata_error_t retval = RDATA_OK;
-    rdata_v2_header_t v2_header;
+    rdata_header_t v2_header;
     rdata_ctx_t *ctx = rdata_ctx_init(parser->io, filename);
     char *encoding = NULL;
+    rdata_compression_t compression = RDATA_COMPRESSION_NONE;
 
     if (ctx == NULL) {
         retval = RDATA_ERROR_OPEN;
@@ -675,15 +677,16 @@ rdata_error_t rdata_parse(rdata_parser_t *parser, const char *filename, void *us
     ctx->dim_handler = parser->dim_handler;
     ctx->dim_name_handler = parser->dim_name_handler;
     ctx->error_handler = parser->error_handler;
+    ctx->header_handler = parser->header_handler;
 
     ctx->is_dimnames = false;
 
     if ((retval = init_stream(ctx)) != RDATA_OK) {
         goto cleanup;
     }
 
-    char header_line[5];
-    if (read_st(ctx, &header_line, sizeof(header_line)) != sizeof(header_line)) {
+    char header_line[RDATA_HEADER_LENGTH] = "";
+    if (read_st(ctx, &header_line, RDATA_HEADER_LENGTH) != RDATA_HEADER_LENGTH) {
         retval = RDATA_ERROR_READ;
         goto cleanup;
     }
@@ -704,6 +707,33 @@ rdata_error_t rdata_parse(rdata_parser_t *parser, const char *filename, void *us
         v2_header.reader_version = byteswap4(v2_header.reader_version);
     }
 
+    if (ctx->header_handler) {
+#if HAVE_BZIP2
+        if (ctx->bz_strm) {
+            compression = RDATA_COMPRESSION_BZIP2;
+        }
+#endif
+#if HAVE_APPLE_COMPRESSION
+        if (ctx->compression_strm) {
+            compression = RDATA_COMPRESSION_LZMA;
+        }
+#endif
+#if HAVE_ZLIB
+        if (ctx->z_strm) {
+            compression = RDATA_COMPRESSION_GZIP;
+        }
+#endif
+#if HAVE_LZMA
+        if (ctx->lzma_strm) {
+            compression = RDATA_COMPRESSION_LZMA;
+        }
+#endif
+        if(ctx->header_handler(compression, header_line, &v2_header, ctx->user_ctx)) {
+            retval = RDATA_ERROR_USER_ABORT;
+            goto cleanup;
+        }
+    }
+
     if (is_rdata && v2_header.format_version != header_line[3] - '0') {
         retval = RDATA_ERROR_PARSE;
         goto cleanup;

src/rdata_write.c

@@ -243,7 +243,7 @@ rdata_error_t rdata_begin_file(rdata_writer_t *writer, void *user_ctx) {
             goto cleanup;
     }
 
-    rdata_v2_header_t v2_header;
+    rdata_header_t v2_header;
     memcpy(v2_header.header, "X\n", sizeof("X\n")-1);
     v2_header.format_version = 2;
     v2_header.reader_version = 131840;