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
@@ -1,165 +1,193 @@
 #include <stdlib.h>
 #include <sys/stat.h>
 #include <syslog.h>
 #include <iostream>
 #include <cassert>
 #include <stdexcept>
 #include <algorithm>
 #include <numeric>
 #include <openssl/md5.h>
 #include "eyetil.h"
 
 #include "config.h"
 #ifdef HAVE_LIBUUID
 # include <uuid/uuid.h>
 #endif
 
 binary_t& binary_t::from_hex(const std::string& h) {
     std::string::size_type hs = h.length();
     if(hs&1)
         throw std::runtime_error("odd number of characters in hexadecimal number");
     size_t rvs = hs>>1;
     resize(rvs);
     const unsigned char *hp = (const unsigned char*)h.data();
     iterator oi=begin();
     char t[3] = { 0,0,0 };
     for(size_t i=0;i<rvs;++i) {
         t[0]=*(hp++); t[1]=*(hp++);
         *(oi++) = static_cast<binary_t::value_type>(0xff&strtol(t,0,16));
     }
     return *this;
 }
 
 binary_t& binary_t::from_data(const void *d,size_t s) {
     resize(s);
     std::copy((const unsigned char*)d,(const unsigned char *)d+s,
             begin() );
     return *this;
 }
 
 binary_t& binary_t::make_nonce() {
 #ifdef HAVE_LIBUUID
     uuid_t uuid;
     uuid_generate(uuid);
     from_data((unsigned char*)uuid,sizeof(uuid));
 #else
     resize(16);
     std::generate_n(begin(),16,rand);
 #endif /* HAVE_LIBUUID */
     return *this;
 }
 
 std::string binary_t::hex() const {
     std::string rv;
     rv.reserve((size()<<1)+1);
     char t[3] = {0,0,0};
     for(const_iterator i=begin(),ie=end();i!=ie;++i) {
         size_t rc = snprintf(t,sizeof(t),"%02x",*i);
         assert(rc<sizeof(t));
         rv += t;
     }
     return rv;
 }
 
 binary_t binary_t::md5() const {
     binary_t rv(MD5_DIGEST_LENGTH);
     if(!MD5(
                 (const unsigned char*)&(front()),size(),
                 (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());
     }
 }
 
 std::string tmpdir_t::get_file(const std::string& f) {
     std::string::size_type ls = f.rfind('/');
     return dir+'/'+(
             (ls==std::string::npos)
             ? f
             : f.substr(ls+1)
         );
 }
 
 tarchive_t::tarchive_t(void *p,size_t s) : a(archive_read_new()), e(0) {
     if(!a) throw std::runtime_error("failed to archive_read_new()");
     if(archive_read_support_format_tar(a)) {
         archive_read_finish(a);
         throw std::runtime_error("failed to archive_read_support_format_tar()");
     }
     if(archive_read_open_memory(a,p,s)) {
         archive_read_finish(a);
         throw std::runtime_error("failed to archive_read_open_memory()");
     }
 }
 tarchive_t::~tarchive_t() {
     assert(a);
     archive_read_finish(a);
 }
 
 bool tarchive_t::read_next_header() {
     assert(a);
     return archive_read_next_header(a,&e)==ARCHIVE_OK;
 }
 
 std::string tarchive_t::entry_pathname() {
     assert(a); assert(e);
     return archive_entry_pathname(e);
 }
 
 bool tarchive_t::read_data_into_fd(int fd) {
     assert(a);
     return archive_read_data_into_fd(a,fd)==ARCHIVE_OK;
 }
 
 
 binary_t integrity_digest(const void *ptr,size_t size,const std::string& ukey) {
     md5_digester rv;
     std::transform( (block512_t*)ptr, ((block512_t*)ptr)+size/sizeof(block512_t),
             rv.updater<uint16_t>(), block512_t::tcpcksum );
     rv.update( binary_t(ukey) );
     return rv.final();
 }
diff --git a/src/eyetil.h b/src/eyetil.h
index 7517ba6..8784cb4 100644
--- a/src/eyetil.h
+++ b/src/eyetil.h
@@ -1,105 +1,116 @@
 #ifndef __EYETIL_H
 #define __EYETIL_H
 
 #include <vector>
 #include <string>
 #include <archive.h>
 #include <archive_entry.h>
 #include "openssl/md5.h"
 
 struct throwable_exit {
     int rc;
     throwable_exit(int rc_) : rc(rc_) { }
 };
 
 class binary_t : public std::vector<unsigned char> {
     public:
         binary_t() { }
         binary_t(size_type  n) : std::vector<unsigned char>(n) { }
         binary_t(const std::string& h) { from_hex(h); }
         binary_t(const void *d,size_t s) { from_data(d,s); }
 
         binary_t& from_hex(const std::string& h);
         binary_t& from_data(const void *d,size_t s);
         binary_t& make_nonce();
 
         std::string hex() const;
         binary_t md5() const;
 };
 
 struct md5_digester {
     MD5_CTX ctx;
     md5_digester() { init(); }
 
     void init();
     void update(const void *d,size_t l);
     binary_t final();
 
     template<typename T>
     void update(const T& x) { update(&x,sizeof(x)); }
 
     template<typename T>
     struct update_iterator : public std::iterator<std::output_iterator_tag,T,void,T*,T&> {
         md5_digester *d;
         update_iterator(md5_digester *d_) : d(d_) { }
         update_iterator(const update_iterator& x) : d(x.d) { }
 
         update_iterator& operator*() { return *this; }
         update_iterator& operator++() { return *this; }
         update_iterator& operator++(int) { return *this; }
 
         update_iterator& operator=(const T& x) {
             d->update(x); return *this;
         }
     };
 
     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);
 };
 
 binary_t integrity_digest(const void *ptr,size_t size,
         const std::string& ukey);
 
 #endif /* __EYETIL_H */