introduced streaming integrity digest updater

Signed-off-by: Michael Krelin <hacker@klever.net>

2 files changed, 39 insertions, 0 deletions

diff --git a/src/eyetil.cc b/src/eyetil.cc
index cd22eea..6ccc4ae 100644
--- a/src/eyetil.cc
+++ b/src/eyetil.cc
@@ -68,48 +68,76 @@ binary_t binary_t::md5() const {
 		(unsigned char*)&(rv.front()) ))
 	throw std::runtime_error("failed to md5()");
     return rv;
 }
 
 void md5_digester::init() {
     if(!MD5_Init(&ctx)) throw std::runtime_error("failed to MD5_Init()");
 }
 void md5_digester::update(const void *d,size_t l) {
     if(!MD5_Update(&ctx,d,l)) throw std::runtime_error("failed to MD5_Update()");
 }
 binary_t md5_digester::final() {
     binary_t rv(MD5_DIGEST_LENGTH);
     if(!MD5_Final((unsigned char*)&(rv.front()), &ctx))
 	throw std::runtime_error("failed to MD5_Final()");
     return rv;
 }
 
 uint16_t block512_t::tcpcksum(block512_t& data) {
     uint32_t sum = std::accumulate(data.data,data.data+words,0);
     while(uint32_t hw = sum>>16) sum = (sum&0xffff)+hw;
     return 0xffff&~sum;
 }
 
+void integrity_digester::update(const void *d_,size_t s) {
+    uint8_t *d=(uint8_t*)d_;
+    if(data_size) {
+	int l = sizeof(data)-data_size;
+	if(l>s) {
+	    memmove(data.dptr(data_size),d,s); data_size+=s; return;
+	}
+	memmove(data.dptr(data_size),d,l); d+=l; s-=l;
+	md5.update<uint16_t>( data.tcpcksum(data) );
+    }
+    if(s<sizeof(data)) {
+	memmove(data.dptr(0),d,s); data_size=s; return;
+    }
+    size_t bb=s/sizeof(block512_t);
+    std::transform((block512_t*)d,((block512_t*)d)+bb,
+	    md5.updater<uint16_t>(),block512_t::tcpcksum);
+    size_t ss=bb*sizeof(block512_t);
+    d+=ss; s-=ss;
+    assert(s<sizeof(block512_t));
+    if(s) memmove(data.dptr(0),d,data_size=s);
+}
+
+binary_t integrity_digester::final(const std::string& ukey) {
+    assert(!data_size);
+    md5.update( binary_t(ukey) );
+    return md5.final();
+}
+
 static void make_path_for_template(const std::string& p,mode_t m) {
     struct stat st;
     std::string pp;
     for(std::string::size_type sl=p.find('/',1);
 	    sl!=std::string::npos;
 	    sl=p.find('/',sl+1)) {
 	if(stat( (pp=p.substr(0,sl)).c_str() ,&st)
 		|| !S_ISDIR(st.st_mode)) {
 	    if(mkdir(pp.c_str(),m))
 		throw std::runtime_error("failed to mkdir()");
 	}
     }
 }
 
 tmpdir_t::tmpdir_t(const std::string& dt) : dir(dt) {
     make_path_for_template(dt,0777);
     if(!mkdtemp((char*)dir.data()))
 	throw std::runtime_error("failed to mkdtmp()");
 }
 tmpdir_t::~tmpdir_t() {
     assert(!dir.empty());
     if(rmdir(dir.c_str())) {
 	syslog(LOG_WARNING,"Failed to remove '%s' directory",dir.c_str());
     }
diff --git a/src/eyetil.h b/src/eyetil.h
index 7517ba6..8784cb4 100644
--- a/src/eyetil.h
+++ b/src/eyetil.h
@@ -53,48 +53,59 @@ struct md5_digester {
 	}
     };
 
     template<typename T>
     update_iterator<T> updater() {
 	return update_iterator<T>(this);
     }
 
 };
 template<> inline void md5_digester::update<binary_t>(const binary_t& x) {
     update((const unsigned char*)&(x.front()),x.size());
 }
 
 #pragma pack(1)
 struct block512_t {
     enum { words = 512 / sizeof(uint16_t) };
     uint16_t data[words];
 
     inline uint8_t *dptr(size_t o) { return ((uint8_t*)this)+o; }
 
     static uint16_t tcpcksum(block512_t& data);
 };
 #pragma pack()
 
+struct integrity_digester {
+    md5_digester md5;
+    size_t data_size;
+    block512_t data;
+
+    integrity_digester() : data_size(0) { }
+    void update(const void *d,size_t s);
+    binary_t final(const std::string& ukey);
+};
+
+
 class tmpdir_t {
     public:
 	std::string dir;
 
 	tmpdir_t(const std::string& dt);
 	~tmpdir_t();
 
 	std::string get_file(const std::string& f);
 };
 
 class tarchive_t {
     public:
 	struct archive *a;
 	struct archive_entry *e;
 
 	tarchive_t(void *p,size_t s);
 	~tarchive_t();
 
 	bool read_next_header();
 
 	std::string entry_pathname();
 
 	bool read_data_into_fd(int fd);
 };