libgsf r996 - in trunk: . gsf tools

[jody svn gnome org]

Author: jody
Date: Mon Aug 25 01:00:13 2008
New Revision: 996
URL: http://svn.gnome.org/viewvc/libgsf?rev=996&view=rev

Log:
2008-08-24  Jody Goldberg <jody gnome org>

	* gsf/gsf-infile-msvba.c (gsf_input_find_vba) : move this here from
	  tools/gsf-vba-dump.
	(gsf_infile_msvba_get_modules) : new.
	(vba_extract_module_source) : store the code rather than dumping it to stdout.

	* tools/gsf-vba-dump.c (test) : use the new utilities.
	(cb_dump_vba) : do the dumping here.


Modified:
   trunk/ChangeLog
   trunk/NEWS
   trunk/gsf/gsf-infile-ar.c
   trunk/gsf/gsf-infile-msole.c
   trunk/gsf/gsf-infile-msvba.c
   trunk/gsf/gsf-infile-msvba.h
   trunk/gsf/gsf-infile-zip.c
   trunk/gsf/gsf-libxml.c
   trunk/gsf/gsf-open-pkg-utils.c
   trunk/gsf/gsf-structured-blob.c
   trunk/gsf/gsf-zip-utils.c
   trunk/tools/gsf-vba-dump.c

Modified: trunk/NEWS
==============================================================================
--- trunk/NEWS	(original)
+++ trunk/NEWS	Mon Aug 25 01:00:13 2008
@@ -9,6 +9,7 @@
 	* Be more forgiving of corrupt ole2.
 	* Add additional MS OpenPkg support to facilitate pivots.
 	* gtk-doc improvements.
+	* Make the VBA extractor more accessible.
 
 Morten:
 	* Use g_base64_-routines from glib when available.

Modified: trunk/gsf/gsf-infile-ar.c
==============================================================================
--- trunk/gsf/gsf-infile-ar.c	(original)
+++ trunk/gsf/gsf-infile-ar.c	Mon Aug 25 01:00:13 2008
@@ -450,8 +450,8 @@
 
 /**
  * gsf_infile_ar_new :
- * @source:
- * @err:
+ * @source: #GsfInput
+ * @err: #Gerror
  *
  * Opens the root directory of a Ar file.
  * <note>This adds a reference to @source.</note>

Modified: trunk/gsf/gsf-infile-msole.c
==============================================================================
--- trunk/gsf/gsf-infile-msole.c	(original)
+++ trunk/gsf/gsf-infile-msole.c	Mon Aug 25 01:00:13 2008
@@ -425,15 +425,8 @@
 	return info;
 }
 
-/**
- * ole_dup:
- * @src:
- *
- * Utility routine to _partially_ replicate a file.  It does NOT copy the bat
- * blocks, or init the dirent.
- *
- * Returns: the partial duplicate.
- **/
+/* Utility routine to _partially_ replicate a file.  It does NOT copy the bat
+ * blocks, or init the dirent. */
 static GsfInfileMSOle *
 ole_dup (GsfInfileMSOle const *src, GError **err)
 {

Modified: trunk/gsf/gsf-infile-msvba.c
==============================================================================
--- trunk/gsf/gsf-infile-msvba.c	(original)
+++ trunk/gsf/gsf-infile-msvba.c	Mon Aug 25 01:00:13 2008
@@ -33,6 +33,9 @@
 #include <gsf/gsf-input-memory.h>
 #include <gsf/gsf-impl-utils.h>
 #include <gsf/gsf-msole-utils.h>
+#include <gsf/gsf-infile-msole.h>
+#include <gsf/gsf-infile-zip.h>
+#include <gsf/gsf-open-pkg-utils.h>
 #include <gsf/gsf-utils.h>
 
 #include <stdio.h>
@@ -45,6 +48,8 @@
 
 	GsfInfile	*source;
 	GList		*children;
+
+	GHashTable	*modules;
 };
 typedef GsfInfileClass GsfInfileMSVBAClass;
 
@@ -67,7 +72,7 @@
 vba_extract_module_source (GsfInfileMSVBA *vba, char const *name, guint32 src_offset)
 {
 	GsfInput *module;
-	guint8 *src_code;
+	guint8 *code;
 	int inflated_size;
 
 	g_return_if_fail (name != NULL);
@@ -76,15 +81,16 @@
 	if (module == NULL)
 		return;
 
-	src_code = gsf_vba_inflate (module, (gsf_off_t) src_offset, &inflated_size, TRUE);
-	if (src_code != NULL) {
-		printf ("<module name=\"%s\">\n<![CDATA[%s]]>\n</module>\n", name, src_code);
-		g_free (src_code);
+	code = gsf_vba_inflate (module, (gsf_off_t) src_offset, &inflated_size, TRUE);
+	if (code != NULL) {
+		if (NULL == vba->modules)
+			vba->modules = g_hash_table_new_full (g_str_hash, g_str_equal,
+				(GDestroyNotify)g_free, (GDestroyNotify)g_free);
+		g_hash_table_insert (vba->modules, g_strdup (name), code);
 	} else
 		g_warning ("Problems extracting the source for %s @ %u", name, src_offset);
 
 	g_object_unref (module);
-	module = NULL;
 }
 
 /**
@@ -391,6 +397,10 @@
 {
 	GsfInfileMSVBA *vba = GSF_INFILE_MSVBA (obj);
 
+	if (NULL != vba->modules) {
+		g_hash_table_destroy (vba->modules);
+		vba->modules = NULL;
+	}
 	if (vba->source != NULL) {
 		g_object_unref (G_OBJECT (vba->source));
 		vba->source = NULL;
@@ -404,6 +414,7 @@
 	GsfInfileMSVBA *vba = GSF_INFILE_MSVBA (obj);
 
 	vba->source		= NULL;
+	vba->modules		= NULL;
 	vba->children		= NULL;
 }
 
@@ -444,3 +455,87 @@
 	g_object_unref (G_OBJECT (vba));
 	return NULL;
 }
+
+/**
+ * gsf_infile_msvba_get_modules :
+ * @vba_stream : #GsfInfile
+ *
+ * a collection of names and source code.
+ *
+ * Returns: %NULL, or a hashtable of names and source code (unknown encoding).
+ **/
+GHashTable *
+gsf_infile_msvba_get_modules (GsfInfileMSVBA const *vba_stream)
+{
+	g_return_val_if_fail (GSF_IS_INFILE_MSVBA (vba_stream), NULL);
+	return vba_stream->modules;
+}
+
+/**
+ * gsf_infile_msvba_steal_modules :
+ * @vba_stream : #GsfInfile
+ *
+ * A collection of names and source code which the caller is responsible for destroying.
+ *
+ * Returns: %NULL, or a hashtable of names and source code (unknown encoding).
+ **/
+GHashTable *
+gsf_infile_msvba_steal_modules (GsfInfileMSVBA *vba_stream)
+{
+	GHashTable *res;
+	g_return_val_if_fail (GSF_IS_INFILE_MSVBA (vba_stream), NULL);
+	res = vba_stream->modules;
+	vba_stream->modules = NULL;
+	return res;
+}
+
+/**
+ * gsf_input_find_vba :
+ * @input : #GsfInput
+ * @err : #GError, optionally %NULL.
+ *
+ * A utility routine that attempts to find the VBA file withint a stream.
+ *
+ * Returns: a GsfInfileMSVBA *gsf_input_find_vba (GsfInput *input, GError *err);
+ **/
+GsfInfileMSVBA *
+gsf_input_find_vba (GsfInput *input, GError **err)
+{
+	GsfInput  *vba = NULL;
+	GsfInfile *infile;
+
+	if (NULL != (infile = gsf_infile_msole_new (input, NULL))) {
+		/* 1) Try XLS */
+		vba = gsf_infile_child_by_vname (infile, "_VBA_PROJECT_CUR", "VBA", NULL);
+		/* 2) DOC */
+		if (NULL == vba)
+			vba = gsf_infile_child_by_vname (infile, "Macros", "VBA", NULL);
+
+		/* TODO : PPT is more complex */
+
+		g_object_unref (G_OBJECT (infile));
+	} else if (NULL != (infile = gsf_infile_zip_new (input, NULL))) {
+		GsfInput *main_part = gsf_open_pkg_get_rel_by_type (GSF_INPUT (infile),
+			"http://schemas.openxmlformats.org/officeDocument/2006/relationships/officeDocument";);
+
+		if (NULL != main_part) {
+			GsfInput *vba_stream = gsf_open_pkg_get_rel_by_type (main_part,
+				"http://schemas.microsoft.com/office/2006/relationships/vbaProject";);
+			if (NULL != vba_stream) {
+				GsfInfile *ole = gsf_infile_msole_new (vba_stream, err);
+				if (NULL != ole) {
+					vba = gsf_infile_child_by_vname (ole, "VBA", NULL);
+					g_object_unref (G_OBJECT (ole));
+				}
+				g_object_unref (G_OBJECT (vba_stream));
+			}
+			g_object_unref (G_OBJECT (main_part));
+		}
+		g_object_unref (G_OBJECT (infile));
+	}
+
+	if (NULL != vba)
+	       return (GsfInfileMSVBA *)
+			gsf_infile_msvba_new (GSF_INFILE (vba), err);
+	return NULL;
+}

Modified: trunk/gsf/gsf-infile-msvba.h
==============================================================================
--- trunk/gsf/gsf-infile-msvba.h	(original)
+++ trunk/gsf/gsf-infile-msvba.h	Mon Aug 25 01:00:13 2008
@@ -36,7 +36,12 @@
 GType gsf_infile_msvba_get_type      (void) G_GNUC_CONST;
 /* void  gsf_infile_msvba_register_type (GTypeModule *module); glib dynamic types are not thread safe */
 
-GsfInfile *gsf_infile_msvba_new (GsfInfile *source, GError **err);
+GsfInfile  *gsf_infile_msvba_new (GsfInfile *source, GError **err);
+GHashTable *gsf_infile_msvba_get_modules   (GsfInfileMSVBA const *vba_stream);
+GHashTable *gsf_infile_msvba_steal_modules (GsfInfileMSVBA       *vba_stream);
+
+/* Utility */
+GsfInfileMSVBA *gsf_input_find_vba (GsfInput *input, GError **err);
 
 G_END_DECLS
 

Modified: trunk/gsf/gsf-infile-zip.c
==============================================================================
--- trunk/gsf/gsf-infile-zip.c	(original)
+++ trunk/gsf/gsf-infile-zip.c	Mon Aug 25 01:00:13 2008
@@ -261,12 +261,7 @@
 	g_free (info);
 }
 
-/**
- * zip_dup :
- * @src :
- *
- * Return value: the partial duplicate.
- **/
+/* Returns a partial duplicate. */
 static GsfInfileZip *
 zip_dup (GsfInfileZip const *src, GError **err)
 {

Modified: trunk/gsf/gsf-libxml.c
==============================================================================
--- trunk/gsf/gsf-libxml.c	(original)
+++ trunk/gsf/gsf-libxml.c	Mon Aug 25 01:00:13 2008
@@ -371,6 +371,17 @@
 	return res;
 }
 
+/**
+ * gsf_xmlDocFormatDump :
+ * @output : #GsfOutput
+ * @cur : #xmlDocPtr
+ * @encoding : The encoding to use.
+ * @format : %TRUE to reformat the output.
+ *
+ * Dumps the document @cur into @output.
+ *
+ * Returns: status from xmlSaveFormatFileTo.
+ **/
 int
 gsf_xmlDocFormatDump (GsfOutput *output, xmlDocPtr cur, char const *encoding,
 		      gboolean format)
@@ -1108,6 +1119,9 @@
  * @ns_id : The name space id to check
  * @name  : The target node name
  *
+ * Checks to see if @str is the same as @ns_id::@name with either an explicit
+ * namespace or the current default namespace.
+ *
  * Returns: %TRUE if @str == @ns_id:@name according to @state.
  **/
 gboolean
@@ -1758,6 +1772,14 @@
 	g_free (tmp);
 }
 
+/**
+ * gsf_xml_out_get_output :
+ * @xout : #GsfXMLOut
+ *
+ * Get the #GsfInput we are parsing from.
+ *
+ * Returns: #GsfInput or %NULL.
+ **/
 GsfOutput *
 gsf_xml_out_get_output (GsfXMLOut const *xout)
 {

Modified: trunk/gsf/gsf-open-pkg-utils.c
==============================================================================
--- trunk/gsf/gsf-open-pkg-utils.c	(original)
+++ trunk/gsf/gsf-open-pkg-utils.c	Mon Aug 25 01:00:13 2008
@@ -843,7 +843,7 @@
  * gsf_outfile_open_pkg_add_rel:
  * @dir : #GsfOutfile
  * @name : target name
- * @content_type :
+ * @content_type : non-%NULL content type
  * @parent : #GsfOutfile
  * @type : target type
  *

Modified: trunk/gsf/gsf-structured-blob.c
==============================================================================
--- trunk/gsf/gsf-structured-blob.c	(original)
+++ trunk/gsf/gsf-structured-blob.c	Mon Aug 25 01:00:13 2008
@@ -196,7 +196,10 @@
  * gsf_structured_blob_read :
  * @input: An input (potentially a GsfInfile) holding the blob
  *
- * Returns: a freshly created tree of blobs
+ * Create a tree of binary blobs with unknown content from a #GsfInput or
+ * #GsfInfile and store it in a newly created #GsfStructuredBlob.
+ *
+ * Returns: a new #GsfStructuredBlob object which the caller is responsible for.
  **/
 GsfStructuredBlob *
 gsf_structured_blob_read (GsfInput *input)

Modified: trunk/gsf/gsf-zip-utils.c
==============================================================================
--- trunk/gsf/gsf-zip-utils.c	(original)
+++ trunk/gsf/gsf-zip-utils.c	Mon Aug 25 01:00:13 2008
@@ -25,6 +25,14 @@
 #include <string.h>
 #include "gsf-zip-impl.h"
 
+/**
+ * SECTION:zip
+ * @Short_description: Utilities for reading and writing ZIP/JAR files
+ * @Title: Zip files
+ *
+ * #GsfInfile and #GsfOutfile support for zip files.
+ **/
+
 /* Doesn't do much, but include for symmetry */
 GsfZipDirent*
 gsf_zip_dirent_new (void)

Modified: trunk/tools/gsf-vba-dump.c
==============================================================================
--- trunk/tools/gsf-vba-dump.c	(original)
+++ trunk/tools/gsf-vba-dump.c	Mon Aug 25 01:00:13 2008
@@ -31,22 +31,15 @@
 #include <stdio.h>
 
 static void
-dump_vba (GsfInput *vba, char const *filename, GError **err)
+cb_dump_vba (char const *name, guint8 const *src_code)
 {
-	GsfInfile *vba_wrapper;
-
-	fprintf (stderr, "%s\n", filename);
-
-	vba_wrapper = gsf_infile_msvba_new (GSF_INFILE (vba), err);
-	if (vba_wrapper != NULL)
-		g_object_unref (G_OBJECT (vba_wrapper));
+	printf ("<module name=\"%s\">\n<![CDATA[%s]]>\n</module>\n", name, src_code);
 }
 
 static int
 test (int argc, char *argv[])
 {
-	GsfInput  *input, *vba;
-	GsfInfile *infile;
+	GsfInput  *input;
 	GError    *err = NULL;
 	int i;
 
@@ -54,50 +47,16 @@
 		input = gsf_input_mmap_new (argv[i], NULL);
 		if (input == NULL)	/* Only report error if stdio fails too */
 			input = gsf_input_stdio_new (argv[i], &err);
+
 		if (input != NULL) {
-			if (NULL != (infile = gsf_infile_msole_new (input, &err))) {
-				// Try XLS first
-				vba = gsf_infile_child_by_vname (infile, "_VBA_PROJECT_CUR", "VBA", NULL);
-
-				// Try DOC next
-				if (NULL == vba)
-					vba = gsf_infile_child_by_vname (infile, "Macros", "VBA", NULL);
-
-				// TODO : PPT is more complex
-
-				if (vba != NULL) {
-					dump_vba (vba, argv[1], &err);
-					g_object_unref (G_OBJECT (vba));
-				}
-				g_object_unref (G_OBJECT (infile));
-			} else if (NULL != (infile = gsf_infile_zip_new (input, NULL))) {
-				GsfInput *main_part = gsf_open_pkg_get_rel_by_type (GSF_INPUT (infile),
-					"http://schemas.openxmlformats.org/officeDocument/2006/relationships/officeDocument";);
-				if (NULL != err)  {
-					g_error_free (err);
-					err = NULL;
-				}
-
-				if (NULL != main_part) {
-					GsfInput *vba_stream = gsf_open_pkg_get_rel_by_type (main_part,
-						"http://schemas.microsoft.com/office/2006/relationships/vbaProject";);
-					if (NULL != vba_stream) {
-						GsfInfile *ole = gsf_infile_msole_new (vba_stream, &err);
-						if (NULL != ole) {
-							vba = gsf_infile_child_by_vname (ole, "VBA", NULL);
-							if (NULL != vba) {
-								dump_vba (vba, argv[1], &err);
-								g_object_unref (G_OBJECT (vba));
-							}
-							g_object_unref (G_OBJECT (ole));
-						}
-						g_object_unref (G_OBJECT (vba_stream));
-					}
-					g_object_unref (G_OBJECT (main_part));
-				}
-				g_object_unref (G_OBJECT (infile));
+			GsfInfileMSVBA *vba = gsf_input_find_vba (input, &err);
+			if (NULL != vba) {
+				GHashTable *modules = gsf_infile_msvba_get_modules (vba);
+				if (NULL != modules)
+					g_hash_table_foreach (modules,
+						(GHFunc) cb_dump_vba, NULL);
+				g_object_unref (G_OBJECT (vba));
 			}
-
 			g_object_unref (G_OBJECT (input));
 		}