More checks added to MD5.

library/Process-linux.cpp

@@ -40,6 +40,7 @@ using namespace std;
 #include "dfhack/VersionInfoFactory.h"
 #include "dfhack/VersionInfo.h"
 #include "dfhack/Error.h"
+#include <string.h>
 using namespace DFHack;
 
 Process::Process(VersionInfoFactory * known_versions)
@@ -54,8 +55,11 @@ Process::Process(VersionInfoFactory * known_versions)
     my_descriptor = 0;
 
     md5wrapper md5;
+    uint32_t length;
+    uint8_t first_kb [1024];
+    memset(first_kb, 0, sizeof(first_kb));
     // get hash of the running DF process
-    string hash = md5.getHashFromFile(exe_link_name);
+    string hash = md5.getHashFromFile(exe_link_name, length, (char *) first_kb);
     // create linux process, add it to the vector
     VersionInfo * vinfo = known_versions->getVersionInfoByMD5(hash);
     if(vinfo)
@@ -71,6 +75,29 @@ Process::Process(VersionInfoFactory * known_versions)
         cerr << "File: " << exe_link_name << endl;
         cerr << "MD5: " << hash << endl;
         cerr << "working dir: " << wd << endl;
+        cerr << "length:" << length << endl;
+        cerr << "1KB hexdump follows:" << endl;
+        for(int i = 0; i < 64; i++)
+        {
+            fprintf(stderr, "%02x %02x %02x %02x  %02x %02x %02x %02x  %02x %02x %02x %02x  %02x %02x %02x %02x\n",
+                    first_kb[i*16],
+                    first_kb[i*16+1],
+                    first_kb[i*16+2],
+                    first_kb[i*16+3],
+                    first_kb[i*16+4],
+                    first_kb[i*16+5],
+                    first_kb[i*16+6],
+                    first_kb[i*16+7],
+                    first_kb[i*16+8],
+                    first_kb[i*16+9],
+                    first_kb[i*16+10],
+                    first_kb[i*16+11],
+                    first_kb[i*16+12],
+                    first_kb[i*16+13],
+                    first_kb[i*16+14],
+                    first_kb[i*16+15]
+                   );
+        }
         free(wd);
     }
 }

library/depends/md5/md5wrapper.cpp

@@ -108,12 +108,13 @@ std::string md5wrapper::getHashFromString(std::string text)
  * (based on Ronald L. Rivest's code
  * from RFC1321 "The MD5 Message-Digest Algorithm")
  */
-std::string md5wrapper::getHashFromFile(std::string filename)
+std::string md5wrapper::getHashFromFile(std::string filename, uint32_t & length, char * first_kb)
 {
     FILE *file;
     MD5_CTX context;
 
     int len;
+    int saved = 0;
     unsigned char buffer[1024], digest[16];
 
     //open file
@@ -121,7 +122,7 @@ std::string md5wrapper::getHashFromFile(std::string filename)
     {
         return "file unreadable.";
     }
-
+    length = 0;
     //init md5
     md5->MD5Init (&context);
 
@@ -130,6 +131,12 @@ std::string md5wrapper::getHashFromFile(std::string filename)
     {
         errno = 0;
         len = fread (buffer, 1, 1024, file);
+        if(saved < 1024 && first_kb)
+        {
+            memcpy(first_kb + saved, buffer, std::min (len, 1024 - saved));
+            saved += len;
+        }
+        length += len;
         if(len != 1024)
         {
             int err = ferror(file);

library/depends/md5/md5wrapper.h

@@ -61,7 +61,7 @@ class md5wrapper
 		 * a file specified in "filename" and
 		 * returns it as string
 		 */
-		std::string getHashFromFile(std::string filename);
+		std::string getHashFromFile(const std::string filename, uint32_t & length, char * first_kb = NULL);
 };