implement md5 convenience file and string wrappers. Modify readfile to support this

src/utils/Makefile

@@ -1,7 +1,8 @@
 depth = ..
 include $(depth)/mk/sysconf
 
-PROGS = trreadfile trfileudi trconftree wipedir smallut trfstreewalk trpathut \
+PROGS = trmd5 trreadfile trfileudi trconftree wipedir smallut \
+      trfstreewalk trpathut \
       transcode trbase64 \
       trmimeparse trexecmd utf8iter idfile
 
@@ -25,6 +26,13 @@ trreadfile.o : readfile.cpp readfile.h
 	$(CXX) -o trreadfile.o -c $(ALL_CXXFLAGS) \
 	       -DTEST_READFILE readfile.cpp
 
+MD5_OBJS= trmd5.o md5.o $(BIGLIB)
+trmd5 : $(MD5_OBJS)
+	$(CXX) -o trmd5 $(MD5_OBJS) $(LIBICONV) $(LIBSYS)
+trmd5.o : md5.cpp md5.h
+	$(CXX) -o trmd5.o -c $(ALL_CXXFLAGS) \
+	       -DTEST_MD5 md5.cpp
+
 PATHUT_OBJS= trpathut.o  $(BIGLIB)
 trpathut : $(PATHUT_OBJS) 
 	$(CXX) $(ALL_CXXFLAGS) -o trpathut $(PATHUT_OBJS)  $(LIBICONV)

src/utils/md5.cpp

@@ -27,7 +27,7 @@
  * This code is the same as the code published by RSA Inc.  It has been
  * edited for clarity and style only.
  */
-
+#ifndef TEST_MD5
 #include <string.h>
 
 #include "md5.h"
@@ -316,3 +316,129 @@ MD5Transform (md5uint32 state[4], const unsigned char block[64])
 	/* Zeroize sensitive information. */
 	memset ((void *)x, 0, sizeof (x));
 }
+
+/*************** Convenience  / utilities */
+void MD5Final(string &digest, MD5_CTX *context)
+{
+    unsigned char d[16];
+    MD5Final (d, context);
+    digest.assign((const char *)d, 16);
+}
+
+string& MD5String(const string& data, string& digest)
+{
+    MD5_CTX ctx;
+    MD5Init(&ctx);
+    MD5Update(&ctx, (const unsigned char*)data.c_str(), data.length());
+    MD5Final(digest, &ctx);
+    return digest;
+}
+
+string& MD5HexPrint(const string& digest, string &out)
+{
+    out.erase();
+    out.reserve(33);
+    static const char hex[]="0123456789abcdef";
+    const unsigned char *hash = (const unsigned char *)digest.c_str();
+    for (int i = 0; i < 16; i++) {
+	out.append(1, hex[hash[i] >> 4]);
+	out.append(1, hex[hash[i] & 0x0f]);
+    }
+    return out;
+}
+string& MD5HexScan(const string& xdigest, string& digest)
+{
+    digest.erase();
+    if (xdigest.length() != 32) {
+	return digest;
+    }
+    for (unsigned int i = 0; i < 16; i++) {
+	unsigned int val;
+	if (sscanf(xdigest.c_str() + 2*i, "%2x", &val) != 1) {
+	    digest.erase();
+	    return digest;
+	}
+	digest.append(1, (unsigned char)val);
+    }
+    return digest;
+}
+
+#include "readfile.h"
+class FileScanMd5 : public FileScanDo {
+public:
+    FileScanMd5(string& d) : digest(d) {}
+    virtual bool init(unsigned int size, string *reason) 
+    {
+	MD5Init(&ctx);
+	return true;
+    }
+    virtual bool data(const char *buf, int cnt, string* reason) 
+    {
+	MD5Update(&ctx, (const unsigned char*)buf, cnt);
+	return true;
+    }
+    string &digest;
+    MD5_CTX ctx;
+};
+bool MD5File(const string& filename, string &digest, string *reason)
+{
+    FileScanMd5 md5er(digest);
+    if (!file_scan(filename, &md5er, reason))
+	return false;
+    // We happen to know that digest and md5er.digest are the same object
+    MD5Final(md5er.digest, &md5er.ctx);
+    return true;
+}
+#else
+
+// Test driver
+#include <stdlib.h>
+
+#include <string>
+#include <iostream>
+#include "md5.h"
+
+using namespace std;
+
+static const char *thisprog;
+static char usage [] =
+"trmd5 filename\n\n"
+;
+static void
+Usage(void)
+{
+    fprintf(stderr, "%s: usage:\n%s", thisprog, usage);
+    exit(1);
+}
+
+int main(int argc, const char **argv)
+{
+    thisprog = argv[0];
+    argc--; argv++;
+
+  if (argc != 1)
+    Usage();
+  string filename =  *argv++;argc--;
+
+  string reason, digest;
+  if (!MD5File(filename, digest, &reason)) {
+      cerr << reason << endl;
+      exit(1);
+  } else {
+      string hex;
+      cout <<  "MD5 (" << filename << ") = " << MD5HexPrint(digest, hex) << endl;
+
+      string digest1;
+      MD5HexScan(hex, digest1);
+      if (digest1.compare(digest)) {
+	  cout << "MD5HexScan Failure" << endl;
+	  cout <<  MD5HexPrint(digest, hex) << " " << digest.length() << " -> " 
+	       << MD5HexPrint(digest1, hex) << " " << digest1.length() << endl;
+	  exit(1);
+      }
+
+  }
+  exit(0);
+}
+
+#endif

src/utils/md5.h

@@ -26,8 +26,7 @@ These notices must be retained in any copies of any part of this
 documentation and/or software.
  */
 
-extern "C" {
+/* Base functions from original file */
-
 /* MD5 context. */
 typedef struct MD5Context {
   unsigned int state[4];	/* state (ABCD) */
@@ -35,9 +34,16 @@ typedef struct MD5Context {
   unsigned char buffer[64];	/* input buffer */
 } MD5_CTX;
 
-void   MD5Init (MD5_CTX *);
+extern void   MD5Init (MD5_CTX *);
-void   MD5Update (MD5_CTX *, const unsigned char *, unsigned int);
+extern void   MD5Update (MD5_CTX *, const unsigned char *, unsigned int);
-void   MD5Final (unsigned char [16], MD5_CTX *);
+extern void   MD5Final (unsigned char [16], MD5_CTX *);
-}
 
+/* Convenience / utilities */
+#include <string>
+using std::string;
+extern void MD5Final(string& digest, MD5_CTX *);
+extern bool MD5File(const string& filename, string& digest, string *reason);
+extern string& MD5String(const string& data, string& digest);
+extern string& MD5HexPrint(const string& digest, string& xdigest);
+extern string& MD5HexScan(const string& xdigest, string& digest);
 #endif /* _MD5_H_ */

src/utils/readfile.cpp

@@ -40,27 +40,53 @@ using std::string;
 
 static void caterrno(string *reason, const char *what)
 {
-#define ERRBUFSZ 200    
-    char errbuf[ERRBUFSZ];
     if (reason) {
-	*reason += "file_to_string: ";
+	reason->append("file_to_string: ");
-	*reason += what;
+	reason->append(what);
-	*reason += ": ";
+	reason->append(": ");
 #ifdef sun
 	// Note: sun strerror is noted mt-safe ??
-	*reason += strerror(errno);
+	reason->append(strerror(errno));
 #else
+#define ERRBUFSZ 200    
+	char errbuf[ERRBUFSZ];
 	strerror_r(errno, errbuf, ERRBUFSZ);
-	*reason += errbuf;
+	reason->append(errbuf);
 #endif
     }
 }
 
-// Note: the fstat() + reserve() calls divide cpu usage almost by 2
+class FileToString : public FileScanDo {
+public:
+    FileToString(string& data) : m_data(data) {}
+    string& m_data;
+    bool init(unsigned int size, string *reason) {
+	if (size > 0)
+	    m_data.reserve(size); 
+	return true;
+    }
+    bool data(const char *buf, int cnt, string *reason) {
+	try {
+	    m_data.append(buf, cnt);
+	} catch (...) {
+	    caterrno(reason, "append");
+	    return false;
+	}
+	return true;
+    }
+};
+
+bool file_to_string(const string &fn, string &data, string *reason)
+{
+    FileToString accum(data);
+    return file_scan(fn, &accum, reason);
+}
+
+// Note: the fstat() + reserve() (in init()) calls divide cpu usage almost by 2
 // on both linux i586 and macosx (compared to just append())
 // Also tried a version with mmap, but it's actually slower on the mac and not
 // faster on linux.
-bool file_to_string(const string &fn, string &data, string *reason)
+bool file_scan(const string &fn, FileScanDo* doer, string *reason)
 {
     bool ret = false;
     bool noclosing = true;
@@ -72,19 +98,16 @@ bool file_to_string(const string &fn, string &data, string *reason)
     // If we have a file name, open it, else use stdin.
     if (!fn.empty()) {
 	fd = open(fn.c_str(), O_RDONLY|O_STREAMING);
-	if (fd < 0 
+	if (fd < 0 || fstat(fd, &st) < 0) {
-#if 1
-	    || fstat(fd, &st) < 0
-#endif
-	    ) {
 	    caterrno(reason, "open/stat");
 	    return false;
 	}
 	noclosing = false;
     }
     if (st.st_size > 0)
-	data.reserve(st.st_size+1);
+	doer->init(st.st_size+1, reason);
-
+    else 
+	doer->init(0, reason);
     char buf[4096];
     for (;;) {
 	int n = read(fd, buf, 4096);
@@ -95,10 +118,7 @@ bool file_to_string(const string &fn, string &data, string *reason)
 	if (n == 0)
 	    break;
 
-	try {
+	if (!doer->data(buf, n, reason)) {
-	    data.append(buf, n);
-	} catch (...) {
-	    caterrno(reason, "append");
 	    goto out;
 	}
     }
@@ -206,7 +226,7 @@ int main(int argc, const char **argv)
   } else if (S_ISREG(st.st_mode)) {
       string s, reason;
       if (!file_to_string(top, s, &reason)) {
-	  cerr << reason;
+	  cerr << reason << endl;
 	  exit(1);
       } else {
 	  cout << s;

src/utils/readfile.h

@@ -19,12 +19,21 @@
 /* @(#$Id: readfile.h,v 1.3 2007-06-02 08:30:42 dockes Exp $  (C) 2004 J.F.Dockes */
 
 #include <string>
+using std::string;
 
 /**
  * Read whole file into string. 
  * @return true for ok, false else
  */
-bool file_to_string(const std::string &filename, std::string &data, 
+bool file_to_string(const string &filename, string &data, string *reason = 0);
-		    std::string *reason = 0);
+
+class FileScanDo {
+public:
+    virtual ~FileScanDo() {}
+    virtual bool init(unsigned int size, string *reason) = 0;
+    virtual bool data(const char *buf, int cnt, string* reason) = 0;
+};
+bool file_scan(const std::string &filename, FileScanDo* doer,  
+	       std::string *reason = 0);
 
 #endif /* _READFILE_H_INCLUDED_ */