From fcad0d7280164e60a3e869c13ab338c104f4bb85 Mon Sep 17 00:00:00 2001
From: Jaromil <jaromil@dyne.org>
Date: Sun, 15 Dec 2024 20:18:18 +0100
Subject: [PATCH] feat: introduce muntargz extract api function

---
 src/muntar.c     | 23 ++++++++++++++++++++---
 src/muntar.h     |  4 ++++
 test/muntar.bats | 33 ++++++++++++++++++++++++++++-----
 3 files changed, 52 insertions(+), 8 deletions(-)

diff --git a/src/muntar.c b/src/muntar.c
index b4d43e5..09f89bf 100644
--- a/src/muntar.c
+++ b/src/muntar.c
@@ -158,7 +158,7 @@ static int mtar_rewind(mtar_t *tar) {
 
 
 #include <sys/stat.h> // for mkdir(2)
-#if defined(_WIN32
+#if defined(_WIN32)
 #include <windows.h>
 #define makedir(path) CreateDirectory(path, NULL)
 #else
@@ -237,13 +237,30 @@ int untar_to_path(const char *path, const uint8_t *buf,
 	return(MTAR_ESUCCESS);
 }
 
+#if !defined(NOGUNZIP)
 // gunzip and untar all in one
 #include <tinf.h>
+#define DECOMPRESSED_SIZE_RATIO 6 // raise this on errors
 int untargz_to_path(const char *path, const uint8_t *buf,
 		    const unsigned int len) {
-	return(0);
+	int res;
+	unsigned int destlen = len*6;
+	uint8_t *dest = malloc(destlen);
+	res = tinf_gzip_uncompress(dest,&destlen,buf,len);
+	// fprintf(stdout,"Compressed source length: %u\n",len);
+	// fprintf(stdout,"Uncompressed destination length: %u\n",destlen);
+	// fprintf(stdout,"Max buffer length: %u\n",len*6);
+	// fprintf(stdout,"Multiplier ratio: %u\n",destlen/len);
+	if(res != TINF_OK) {
+		fprintf(stderr,"Error in gunzip decompression (untargz_to_path)\n");
+		free(dest);
+		exit(res);
+	}
+	res = untar_to_path(path, dest, destlen);
+	free(dest);
+	return(res);
 }
-
+#endif
 
 int mtar_load(mtar_t *tar, const char *name,
 	      const uint8_t *buf, size_t size) {
diff --git a/src/muntar.h b/src/muntar.h
index f92be95..46b34e6 100644
--- a/src/muntar.h
+++ b/src/muntar.h
@@ -32,6 +32,10 @@
 // used by extract_embeddings(char *tmpdir)
 int untar_to_path(const char *path,
 		  const uint8_t *buf, const unsigned int len);
+#if !defined(NOGUNZIP)
+int untargz_to_path(const char *path,
+		    const uint8_t *buf, const unsigned int len);
+#endif
 
 enum {
 	MTAR_ESUCCESS     =  0,
diff --git a/test/muntar.bats b/test/muntar.bats
index 0a4665a..e78c344 100644
--- a/test/muntar.bats
+++ b/test/muntar.bats
@@ -40,7 +40,7 @@ int main(int argc, char **argv) {
     exit(0);
 }
 EOF
-    gcc -o muntar_list -I ${R}/src \
+    gcc -o muntar_list -DNOGUNZIP -I ${R}/src \
         ${R}/src/muntar.c ${R}/src/io.c examples.c muntar_list.c
     run ./muntar_list
     assert_success
@@ -48,7 +48,6 @@ EOF
     assert_line --partial "examples/donut.c (743 bytes)"
 }
 
-
 @test "muntar extract contents" {
       cat << EOF > muntar_extract.c
 #include <stdio.h>
@@ -63,13 +62,11 @@ int main(int argc, char **argv) {
     exit(res);
 }
 EOF
-    gcc -o muntar_extract -I ${R}/src \
+    gcc -o muntar_extract -DNOGUNZIP -I ${R}/src \
         ${R}/src/muntar.c examples.c muntar_extract.c
     run ./muntar_extract ${TMP}/extracted
     >&3 cat ${TMP}/extracted/examples/donut.c
     assert_success
-    # assert_file_size_equals ${TMP}/extracted/examples/donut.c \
-    # `ls -l ${R}/examples/donut.c | cut -d' ' -f5`
     l=`sha256sum ${TMP}/extracted/examples/donut.c | cut -d' ' -f1`
     r=`sha256sum ${R}/examples/donut.c | cut -d' ' -f1`
     assert_equal $l $r
@@ -125,3 +122,29 @@ EOF
     r=`sha256sum examples.tar | cut -d' ' -f1`
     assert_equal $l $r
 }
+
+
+@test "muntargz extract contents" {
+      cat << EOF > muntargz_extract.c
+#include <stdio.h>
+#include <stdlib.h>
+#include <muntar.h>
+extern unsigned char examples_tar_gz[];
+extern unsigned int examples_tar_gz_len;
+int main(int argc, char **argv) {
+    int res;
+    fprintf(stderr,"extract to %s\n",argv[1]);
+    res = untargz_to_path(argv[1], examples_tar_gz, examples_tar_gz_len);
+    exit(res);
+}
+EOF
+    gcc -o muntargz_extract -I ${R}/src \
+    ${R}/src/tinfgzip.c ${R}/src/tinflate.c ${R}/src/muntar.c \
+    examples_gzip.c muntargz_extract.c
+    run ./muntargz_extract ${TMP}/extractgz
+    >&3 cat ${TMP}/extractgz/examples/donut.c
+    assert_success
+    l=`sha256sum ${TMP}/extractgz/examples/donut.c | cut -d' ' -f1`
+    r=`sha256sum ${R}/examples/donut.c | cut -d' ' -f1`
+    assert_equal $l $r
+}