Added option to choose how to export blobs

Supported modes are: - strip (don't export) - raw (compatibility) - octal (export as \ooo) Note that gmdb2 dialog where updated, but not the code. Many fixes to do there first :/
samlowe · Dec 2, 2012 · a66531c · a66531c
1 parent 473d820
commit a66531c
Show file tree

Hide file tree

Showing 6 changed files with 110 additions and 30 deletions.
diff --git a/doc/mdb-export.txt b/doc/mdb-export.txt
@@ -2,23 +2,24 @@ NAME
   mdb-export - Export data in an MDB database table to CSV format.
 
 SYNOPSIS
-  mdb-export [-H] [-d delimiter] [-R row_delim] [[-Q] | [-q quote [-X escape]]] [-I backend] [-D format] [-N namespace] database table
+  mdb-export [-H] [-d delimiter] [-R row_delim] [[-Q] | [-q quote [-X escape]]] [-I backend] [-D format] [-N namespace] [-b strip|raw|octal] database table
 
 DESCRIPTION
   mdb-export is a utility program distributed with MDB Tools. 
 
   It produces a CSV (comma separated value) output for the given table. Such output is suitable for importation into databases or spreadsheets.
 
 OPTIONS
-  -H            Supress header row
-  -Q            Don't wrap text-like fields (text, memo, date) in quotes.  If not specified text fiels will be surrounded by " (double quote) characters.
-  -d            Specify an alternative column delimiter If no delimiter is specified, table names will be delimited by a , (comma) character.
-  -R            Specify a row delimiter
-  -I backend    INSERT statements (instead of CSV). You must specify which SQL backend dialect to use. Allowed values are: access, sybase, oracle, postgres and mysql.
-  -D            Set the date format (see strftime(3) for details)
-  -q            Use to wrap text-like fields. Default is ".
-  -X            Use to escape quoted characters within a field.  Default is doubling.
-  -N namespace  Prefix identifiers with namespace.
+  -H                  Supress header row
+  -Q                  Don't wrap text-like fields (text, memo, date) in quotes.  If not specified text fiels will be surrounded by " (double quote) characters.
+  -d                  Specify an alternative column delimiter If no delimiter is specified, table names will be delimited by a , (comma) character.
+  -R                  Specify a row delimiter
+  -I backend          INSERT statements (instead of CSV). You must specify which SQL backend dialect to use. Allowed values are: access, sybase, oracle, postgres and mysql.
+  -D                  Set the date format (see strftime(3) for details)
+  -q                  Use to wrap text-like fields. Default is ".
+  -X                  Use to escape quoted characters within a field.  Default is doubling.
+  -N namespace        Prefix identifiers with namespace.
+  -b strip|raw|octal  Binary export mode: strip binaries, export as-is, or output \ooo style octal data.
 
 
 NOTES 

diff --git a/src/gmdb2/gladefiles/gmdb-export.glade b/src/gmdb2/gladefiles/gmdb-export.glade
@@ -13,7 +13,7 @@
           <widget class="GtkTable" id="table1">
             <property name="visible">True</property>
             <property name="border_width">30</property>
-            <property name="n_rows">6</property>
+            <property name="n_rows">7</property>
             <property name="n_columns">2</property>
             <property name="column_spacing">30</property>
             <property name="row_spacing">8</property>
@@ -93,8 +93,8 @@
                 <property name="use_markup">True</property>
               </widget>
               <packing>
-                <property name="top_attach">5</property>
-                <property name="bottom_attach">6</property>
+                <property name="top_attach">6</property>
+                <property name="bottom_attach">7</property>
                 <property name="x_options">GTK_FILL</property>
                 <property name="y_options"></property>
               </packing>
@@ -112,8 +112,8 @@
               <packing>
                 <property name="left_attach">1</property>
                 <property name="right_attach">2</property>
-                <property name="top_attach">5</property>
-                <property name="bottom_attach">6</property>
+                <property name="top_attach">6</property>
+                <property name="bottom_attach">7</property>
                 <property name="x_options">GTK_FILL</property>
                 <property name="y_options"></property>
               </packing>
@@ -176,6 +176,32 @@
                 <property name="bottom_attach">5</property>
               </packing>
             </child>
+            <child>
+              <widget class="GtkLabel" id="label8">
+                <property name="visible">True</property>
+                <property name="xalign">0</property>
+                <property name="label" translatable="yes">&lt;b&gt;Binaries:&lt;/b&gt;</property>
+                <property name="use_markup">True</property>
+              </widget>
+              <packing>
+                <property name="top_attach">5</property>
+                <property name="bottom_attach">6</property>
+                <property name="x_options">GTK_FILL</property>
+                <property name="y_options"></property>
+              </packing>
+            </child>
+            <child>
+              <widget class="GtkComboBox" id="bin_combo">
+                <property name="visible">True</property>
+                <property name="items" translatable="yes"></property>
+              </widget>
+              <packing>
+                <property name="left_attach">1</property>
+                <property name="right_attach">2</property>
+                <property name="top_attach">5</property>
+                <property name="bottom_attach">6</property>
+              </packing>
+            </child>
           </widget>
           <packing>
             <property name="position">2</property>

diff --git a/src/gmdb2/gmdb.h b/src/gmdb2/gmdb.h
@@ -88,6 +88,7 @@ void gmdb_export_get_delimiter(GladeXML *xml, gchar *delimiter, int max_buf);
 void gmdb_export_get_lineterm(GladeXML *xml, gchar *lineterm, int max_buf);
 int gmdb_export_get_quote(GladeXML *xml);
 char gmdb_export_get_quotechar(GladeXML *xml);
+int gmdb_export_get_binmode(GladeXML *xml);
 int gmdb_export_get_headers(GladeXML *xml);
 gchar *gmdb_export_get_filepath(GladeXML *xml);
 

diff --git a/src/gmdb2/sql.c b/src/gmdb2/sql.c
@@ -60,6 +60,7 @@ gmdb_sql_write_rslt_cb(GtkWidget *w, GladeXML *xml)
 	int need_quote = 0;
 	gchar delimiter[11];
 	gchar quotechar;
+	//TODO int binmode;
 	gchar lineterm[5];
 	gchar *str;
 	int rows=0, n_columns;
@@ -78,6 +79,7 @@ gmdb_sql_write_rslt_cb(GtkWidget *w, GladeXML *xml)
 	gmdb_export_get_lineterm(xml, lineterm, 5);
 	need_quote = gmdb_export_get_quote(xml);
 	quotechar = gmdb_export_get_quotechar(xml);
+	//TODO binmode = gmdb_export_get_binmode(xml);
 	need_headers = gmdb_export_get_headers(xml);
 	file_path = gmdb_export_get_filepath(xml);
 

diff --git a/src/gmdb2/table_export.c b/src/gmdb2/table_export.c
@@ -42,6 +42,10 @@ MdbCatalogEntry *cat_entry;
 #define NEVER "Never"
 #define AUTOMAT "Automatic (where necessary)"
 
+#define BIN_STRIP "Strip"
+#define BIN_RAW "Raw"
+#define BIN_OCTAL "Octal"
+
 void
 gmdb_print_quote(FILE *outfile, int need_quote, char quotechar, char *colsep, char *str)
 {
@@ -116,6 +120,22 @@ gmdb_export_get_quotechar(GladeXML *xml)
 
 	return quotechar;
 }
+
+int
+gmdb_export_get_binmode(GladeXML *xml)
+{
+	GtkComboBox *combobox;
+	gchar *str;
+
+	combobox = GTK_COMBO_BOX(glade_xml_get_widget(xml, "bin_combo"));
+	str = gtk_combo_box_get_active_text (combobox);
+
+	if (!strcmp(str,BIN_STRIP)) return 1;
+	else if (!strcmp(str,BIN_OCTAL)) return 2;
+	else return 0;
+}
+
+
 int
 gmdb_export_get_headers(GladeXML *xml)
 {
@@ -161,6 +181,7 @@ int need_quote = 0;
 gchar delimiter[11];
 gchar quotechar;
 gchar lineterm[5];
+int binmode = 1;
 int rows=0;
 
 	GtkWidget *exportwin, *dlg;
@@ -170,6 +191,7 @@ int rows=0;
 	need_quote = gmdb_export_get_quote(exportwin_xml);
 	quotechar = gmdb_export_get_quotechar(exportwin_xml);
 	need_headers = gmdb_export_get_headers(exportwin_xml);
+	binmode = gmdb_export_get_binmode(exportwin_xml);
 	file_path = gmdb_export_get_filepath(exportwin_xml);
 
 	// printf("file path %s\n",file_path);
@@ -271,4 +293,10 @@ gmdb_table_export_populate_dialog(GladeXML *xml)
 	gtk_combo_box_append_text(combobox, "'");
 	gtk_combo_box_append_text(combobox, "`");
 	gtk_combo_box_set_active(combobox, 0);
+
+	combobox = GTK_COMBO_BOX(glade_xml_get_widget(xml, "bin_combo"));
+	gtk_combo_box_append_text(combobox, BIN_STRIP);
+	gtk_combo_box_append_text(combobox, BIN_RAW);
+	gtk_combo_box_append_text(combobox, BIN_OCTAL);
+	gtk_combo_box_set_active(combobox, 1);
 }
diff --git a/src/util/mdb-export.c b/src/util/mdb-export.c
@@ -28,11 +28,15 @@
 #define is_quote_type(x) (x==MDB_TEXT || x==MDB_OLE || x==MDB_MEMO || x==MDB_DATETIME || x==MDB_BINARY || x==MDB_REPID)
 #define is_binary_type(x) (x==MDB_OLE || x==MDB_BINARY || x==MDB_REPID)
 
+#define BIN_MODE_STRIP   0
+#define BIN_MODE_RAW     1
+#define BIN_MODE_OCTAL   2
+
 static char *escapes(char *s);
 
 //#define DONT_ESCAPE_ESCAPE
 static void
-print_col(gchar *col_val, int quote_text, int col_type, int bin_len, char *quote_char, char *escape_char)
+print_col(gchar *col_val, int quote_text, int col_type, int bin_len, char *quote_char, char *escape_char, int bin_mode)
 {
 	size_t quote_len = strlen(quote_char); /* multibyte */
 
@@ -46,6 +50,8 @@ print_col(gchar *col_val, int quote_text, int col_type, int bin_len, char *quote
 		fputs(quote_char,stdout);
 		while (1) {
 			if (is_binary_type(col_type)) {
+				if (bin_mode == BIN_MODE_STRIP)
+					break;
 				if (!bin_len--)
 					break;
 			} else /* use \0 sentry */
@@ -60,12 +66,14 @@ print_col(gchar *col_val, int quote_text, int col_type, int bin_len, char *quote
 				fprintf(stdout, "%s%s", escape_char, escape_char);
 				col_val += orig_escape_len;
 #endif
-			} else
+			} else if (is_binary_type(col_type) && *col_val < 0 && bin_mode == BIN_MODE_OCTAL)
+				fprintf(stdout, "\\%03o", *(unsigned char*)col_val++);
+			else
 				putc(*col_val++, stdout);
 		}
-		fputs(quote_char,stdout);
+		fputs(quote_char, stdout);
 	} else
-		fputs(col_val,stdout);
+		fputs(col_val, stdout);
 }
 int
 main(int argc, char **argv)
@@ -84,11 +92,12 @@ main(int argc, char **argv)
 	char quote_text = 1;
 	char *insert_dialect = NULL;
 	char *namespace = NULL;
+	int bin_mode = BIN_MODE_RAW;
 	int  opt;
 	char *value;
 	size_t length;
 
-	while ((opt=getopt(argc, argv, "HQq:X:d:D:R:I:N:"))!=-1) {
+	while ((opt=getopt(argc, argv, "HQq:X:d:D:R:I:N:b:"))!=-1) {
 		switch (opt) {
 		case 'H':
 			header_row = 0;
@@ -118,6 +127,18 @@ main(int argc, char **argv)
 		case 'N':
 			namespace = (char *) g_strdup(optarg);
 		break;
+		case 'b':
+			if (!strcmp(optarg, "strip"))
+				bin_mode = BIN_MODE_STRIP;
+			else if (!strcmp(optarg, "raw"))
+				bin_mode = BIN_MODE_RAW;
+			else if (!strcmp(optarg, "octal"))
+				bin_mode = BIN_MODE_OCTAL;
+			else {
+				fprintf(stderr, "Invalid binary mode\n");
+				exit(1);
+			}
+		break;
 		default:
 		break;
 		}
@@ -139,15 +160,16 @@ main(int argc, char **argv)
 	if (argc-optind < 2) {
 		fprintf(stderr,"Usage: %s [options] <file> <table>\n",argv[0]);
 		fprintf(stderr,"where options are:\n");
-		fprintf(stderr,"  -H             supress header row\n");
-		fprintf(stderr,"  -Q             don't wrap text-like fields in quotes\n");
-		fprintf(stderr,"  -d <delimiter> specify a column delimiter\n");
-		fprintf(stderr,"  -R <delimiter> specify a row delimiter\n");
-		fprintf(stderr,"  -I <backend>   INSERT statements (instead of CSV)\n");
-		fprintf(stderr,"  -D <format>    set the date format (see strftime(3) for details)\n");
-		fprintf(stderr,"  -q <char>      Use <char> to wrap text-like fields. Default is \".\n");
-		fprintf(stderr,"  -X <char>      Use <char> to escape quoted characters within a field. Default is doubling.\n");
-		fprintf(stderr,"  -N <namespace> Prefix identifiers with namespace\n");
+		fprintf(stderr,"  -H                   supress header row\n");
+		fprintf(stderr,"  -Q                   don't wrap text-like fields in quotes\n");
+		fprintf(stderr,"  -d <delimiter>       specify a column delimiter\n");
+		fprintf(stderr,"  -R <delimiter>       specify a row delimiter\n");
+		fprintf(stderr,"  -I <backend>         INSERT statements (instead of CSV)\n");
+		fprintf(stderr,"  -D <format>          set the date format (see strftime(3) for details)\n");
+		fprintf(stderr,"  -q <char>            Use <char> to wrap text-like fields. Default is \".\n");
+		fprintf(stderr,"  -X <char>            Use <char> to escape quoted characters within a field. Default is doubling.\n");
+		fprintf(stderr,"  -N <namespace>       Prefix identifiers with namespace\n");
+		fprintf(stderr,"  -b strip|raw|octal   Binary export mode.\n");
 		g_free (delimiter);
 		g_free (row_delimiter);
 		g_free (quote_char);
@@ -231,7 +253,7 @@ main(int argc, char **argv)
 					value = bound_values[j];
 					length = bound_lens[j];
 				}
-				print_col(value, quote_text, col->col_type, length, quote_char, escape_char);
+				print_col(value, quote_text, col->col_type, length, quote_char, escape_char, bin_mode);
 				if (col->col_type == MDB_OLE)
 					free(value);
 			}