Reorganise archive extraction code

3 files changed, 94 insertions, 68 deletions

diff --git a/libbb/deb_extract.c b/libbb/deb_extract.c
index 36cebfa51..f64193970 100644
--- a/libbb/deb_extract.c
+++ b/libbb/deb_extract.c
@@ -27,71 +27,97 @@
 #include <unistd.h>
 #include <stdlib.h>
 #include <string.h>
-#include <signal.h>
 #include "libbb.h"
 
+int seek_sub_file(FILE *in_file, file_headers_t *headers, const char *tar_gz_file)
+{
+	/* find the headers for the specified .tar.gz file */
+	while (headers != NULL) {
+		if (strcmp(headers->name, tar_gz_file) == 0) {
+			fseek(in_file, headers->offset, SEEK_SET);
+			return(EXIT_SUCCESS);
+		}
+		headers = headers->next;
+	}
+
+	return(EXIT_FAILURE);
+}
+
 /*
  * The contents of argument depend on the value of function.
  * It is either a dir name or a control file or field name(see dpkg_deb.c)
  */
-extern char *deb_extract(const char *package_filename, const int function, const char *argument, const char *argument2)
+char *deb_extract(const char *package_filename, FILE *out_stream, const int extract_function,
+	const char *prefix, const char *filename)
 {
+	FILE *deb_stream, *uncompressed_stream;
+	file_headers_t *ar_headers = NULL;
+	file_headers_t *tar_headers = NULL;
+	file_headers_t *list_ptr;
+	file_headers_t *deb_extract_list = (file_headers_t *) calloc(1, sizeof(file_headers_t));
 
-	FILE *deb_file, *uncompressed_file;
-	ar_headers_t *headers = NULL;
 	char *ared_file = NULL;
 	char *output_buffer = NULL;
 	int gunzip_pid;
 
-	switch (function) {
-		case (extract_info):
-		case (extract_control):
-		case (extract_field):
-			ared_file = xstrdup("control.tar.gz");
-			break;
-		default:
-			ared_file = xstrdup("data.tar.gz");
-			break;
+	if (extract_function & extract_control_tar_gz) {
+		ared_file = xstrdup("control.tar.gz");
+	}
+	else if (extract_function & extract_data_tar_gz) {		
+		ared_file = xstrdup("data.tar.gz");
 	}
 
 	/* open the debian package to be worked on */
-	deb_file = wfopen(package_filename, "r");
+	deb_stream = wfopen(package_filename, "r");
 
-	headers = (ar_headers_t *) xmalloc(sizeof(ar_headers_t));	
+	ar_headers = (file_headers_t *) xmalloc(sizeof(file_headers_t));	
 	
 	/* get a linked list of all ar entries */
-	if ((headers = get_ar_headers(deb_file)) == NULL) {
+	ar_headers = get_ar_headers(deb_stream);
+	if (ar_headers == NULL) {
 		error_msg("Couldnt get ar headers\n");
 		return(NULL);
 	}
 
 	/* seek to the start of the .tar.gz file within the ar file*/
-	if (seek_ared_file(deb_file, headers, ared_file) == EXIT_FAILURE) {
-		error_msg("Couldnt seek to ar file");
+	fseek(deb_stream, 0, SEEK_SET);
+	if (seek_sub_file(deb_stream, ar_headers, ared_file) == EXIT_FAILURE) {
+		error_msg("Couldnt seek to file %s", ared_file);
 	}
 
-	/* open a stream of decompressed data */
-	uncompressed_file = fdopen(gz_open(deb_file, &gunzip_pid), "r");
-
-	if (function & extract_fsys_tarfile) {
-		copy_file_chunk(uncompressed_file, stdout, -1);
-	} else {
-		FILE *output;
+	/* get a linked list of all tar entries */
+	tar_headers = get_tar_gz_headers(deb_stream);
+	if (tar_headers == NULL) {
+		error_msg("Couldnt get tar headers\n");
+		return(NULL);
+	}
 
-		if (function & extract_contents_to_file) {
-			output = wfopen(argument, "w");
-		} else {
-			output = stdout;
+	if (extract_function & extract_one_to_buffer) {
+		list_ptr = tar_headers;
+		while (list_ptr != NULL) {
+			if (strcmp(filename, list_ptr->name) == 0) {
+				deb_extract_list = append_archive_list(deb_extract_list, list_ptr);
+				break;
+			}
+			list_ptr = list_ptr->next;
 		}
+	} else {
+		deb_extract_list = tar_headers;
+	}
 
-		output_buffer = untar(uncompressed_file, output, function, argument, argument2);
-		if (output != stdout) {
-			fclose(output);
-		}
+	/* seek to the start of the .tar.gz file within the ar file*/
+	if (seek_sub_file(deb_stream, ar_headers, ared_file) == EXIT_FAILURE) {
+		error_msg("Couldnt seek to ar file");
 	}
+
+	/* open a stream of decompressed data */
+	uncompressed_stream = fdopen(gz_open(deb_stream, &gunzip_pid), "r");
+
+	output_buffer = extract_archive(uncompressed_stream, out_stream, deb_extract_list, extract_function, prefix);
+
 	gz_close(gunzip_pid);
-	fclose(deb_file);
-	fclose(uncompressed_file);
+	fclose(deb_stream);
+	fclose(uncompressed_stream);
 	free(ared_file);
 
 	return(output_buffer);
diff --git a/libbb/get_ar_headers.c b/libbb/get_ar_headers.c
index 151caafe1..414498b4f 100644
--- a/libbb/get_ar_headers.c
+++ b/libbb/get_ar_headers.c
@@ -30,7 +30,7 @@
 #include <unistd.h>
 #include "libbb.h"
 
-extern ar_headers_t *get_ar_headers(FILE *in_file)
+file_headers_t *get_ar_headers(FILE *in_file)
 {
 	typedef struct raw_ar_header_s {	/* Byte Offset */
 		char name[16];	/*  0-15 */
@@ -44,7 +44,7 @@ extern ar_headers_t *get_ar_headers(FILE *in_file)
 
 	raw_ar_header_t raw_ar_header;
 
-	ar_headers_t *ar_list, *ar_tmp, *ar_entry;
+	file_headers_t *ar_list, *ar_entry;
 	char ar_magic[8];
 	char *long_name=NULL;
 	
@@ -59,10 +59,10 @@ extern ar_headers_t *get_ar_headers(FILE *in_file)
 		return(NULL);
 	}
 	
-	ar_list = (ar_headers_t *) xcalloc(1, sizeof(ar_headers_t));
+	ar_list = (file_headers_t *) xcalloc(1, sizeof(file_headers_t));
 
 	while (fread((char *) &raw_ar_header, 1, 60, in_file) == 60) {
-		ar_entry = (ar_headers_t *) xcalloc(1, sizeof(ar_headers_t));
+		ar_entry = (file_headers_t *) xcalloc(1, sizeof(file_headers_t));
 		/* check the end of header markers are valid */
 		if ((raw_ar_header.fmag[0] != '`') || (raw_ar_header.fmag[1] != '\n')) {
 			char newline;
@@ -113,12 +113,7 @@ extern ar_headers_t *get_ar_headers(FILE *in_file)
 
 		fseek(in_file, (off_t) ar_entry->size, SEEK_CUR);
 
-		ar_tmp = (ar_headers_t *) xcalloc(1, sizeof(ar_headers_t));
-		*ar_tmp = *ar_list;
-		*ar_list = *ar_entry;
-		free(ar_entry);
-		ar_list->next = ar_tmp;
+		ar_list = append_archive_list(ar_list, ar_entry);
 	}
-
 	return(ar_list);
 }
diff --git a/libbb/libbb.h b/libbb/libbb.h
index f99d25780..f24a38109 100644
--- a/libbb/libbb.h
+++ b/libbb/libbb.h
@@ -213,36 +213,41 @@ char *xreadlink(const char *path);
 char *concat_path_file(const char *path, const char *filename);
 char *last_char_is(const char *s, int c);
 
-typedef struct ar_headers_s {
+typedef struct file_headers_s {
 	char *name;
+	char *link_name;
 	off_t size;
 	uid_t uid;
 	gid_t gid;
 	mode_t mode;
 	time_t mtime;
 	off_t offset;
-	struct ar_headers_s *next;
-} ar_headers_t;
-extern ar_headers_t *get_ar_headers(FILE *in_file);
-extern int seek_ared_file(FILE *in_file, ar_headers_t *headers, const char *tar_gz_file);
-
-typedef enum extract_function_e {
-	extract_contents = 1,
-	extract_control = 2,
-	extract_info = 4,
-	extract_extract = 8,
-	extract_verbose_extract = 16,
-	extract_list = 32,
-	extract_fsys_tarfile = 64,
-	extract_field = 128,
-	extract_contents_to_file = 256
-} extract_function_t;
-extern char *deb_extract(const char *package_filename, int function,
-	const char *argument, const char *argument2);
-extern char *untar(FILE *src_tar_file, FILE *output, int untar_function,
-	const char *argument, const char *file_prefix);
-extern char *read_text_file_to_buffer(FILE *src_file);
-extern char *read_package_field(const char *package_buffer);
+	dev_t device;
+	struct file_headers_s *next;
+} file_headers_t;
+file_headers_t *get_ar_headers(FILE *in_file);
+file_headers_t *get_tar_headers(FILE *tar_stream);
+file_headers_t *get_tar_gz_headers(FILE *compressed_stream);
+file_headers_t *append_archive_list(file_headers_t *head, file_headers_t *tail_entry);
+file_headers_t *add_from_archive_list(file_headers_t *master_list, file_headers_t *new_list, const char *name);
+
+enum extract_functions_e {
+	extract_verbose_list = 1,
+	extract_list = 2,
+	extract_one_to_buffer = 4,
+	extract_to_stdout = 8,
+	extract_all_to_fs = 16,
+	extract_preserve_date = 32,
+	extract_data_tar_gz = 64,
+	extract_control_tar_gz = 128,
+	extract_unzip_only = 256
+};
+char *extract_archive(FILE *src_stream, FILE *out_stream, file_headers_t *extract_headers, int function, const char *prefix);
+char *deb_extract(const char *package_filename, FILE *out_stream, const int function,
+	const char *prefix, const char *filename);
+char *read_package_field(const char *package_buffer);
+int seek_sub_file(FILE *in_file, file_headers_t *headers, const char *tar_gz_file);
+char *fgets_str(FILE *file, const char *terminating_string);
 
 extern int unzip(FILE *l_in_file, FILE *l_out_file);
 extern void gz_close(int gunzip_pid);