mdbx-tools: добавление опции -c (concise) в mdbx_dump.

src/man1/mdbx_dump.1

@@ -12,6 +12,8 @@ mdbx_dump \- MDBX environment export tool
 [\c
 .BR \-q ]
 [\c
+.BR \-c ]
+[\c
 .BI \-f \ file\fR]
 [\c
 .BR \-l ]
@@ -41,6 +43,9 @@ Write the library version number to the standard output, and exit.
 .BR \-q
 Be quiet.
 .TP
+.BR \-c
+Concise mode without repeating keys in a dump, but incompatible with Berkeley DB and LMDB.
+.TP
 .BR \-f \ file
 Write to the specified file instead of to the standard output.
 .TP

src/tools/dump.c

@@ -20,6 +20,7 @@
 
 #define PRINT 1
 #define GLOBAL 2
+#define CONCISE 4
 static int mode = GLOBAL;
 
 typedef struct flagbit {
@@ -55,42 +56,23 @@ static void signal_handler(int sig) {
 
 #endif /* !WINDOWS */
 
-static const char hexc[] = "0123456789abcdef";
-
-static void dumpbyte(unsigned char c) {
-  putchar(hexc[c >> 4]);
-  putchar(hexc[c & 15]);
-}
-
-static void text(MDBX_val *v) {
-  unsigned char *c, *end;
-
+static void dumpval(const MDBX_val *v) {
+  static const char digits[] = "0123456789abcdef";
   putchar(' ');
-  c = v->iov_base;
-  end = c + v->iov_len;
-  while (c < end) {
+  for (const unsigned char *c = v->iov_base, *end = c + v->iov_len; c < end; ++c) {
+    if (mode & PRINT) {
       if (isprint(*c) && *c != '\\') {
         putchar(*c);
-    } else {
+        continue;
+      } else
         putchar('\\');
-      dumpbyte(*c);
     }
-    c++;
+    putchar(digits[*c >> 4]);
+    putchar(digits[*c & 15]);
   }
   putchar('\n');
 }
 
-static void dumpval(MDBX_val *v) {
-  unsigned char *c, *end;
-
-  putchar(' ');
-  c = v->iov_base;
-  end = c + v->iov_len;
-  while (c < end)
-    dumpbyte(*c++);
-  putchar('\n');
-}
-
 bool quiet = false, rescue = false;
 const char *prog;
 static void error(const char *func, int rc) {
@@ -185,12 +167,19 @@ static int dump_tbl(MDBX_txn *txn, MDBX_dbi dbi, char *name) {
       rc = MDBX_EINTR;
       break;
     }
-    if (mode & PRINT) {
-      text(&key);
-      text(&data);
-    } else {
     dumpval(&key);
     dumpval(&data);
+    if ((flags & MDBX_DUPSORT) && (mode & CONCISE)) {
+      while ((rc = mdbx_cursor_get(cursor, &key, &data, MDBX_NEXT_DUP)) == MDBX_SUCCESS) {
+        if (user_break) {
+          rc = MDBX_EINTR;
+          break;
+        }
+        putchar(' ');
+        dumpval(&data);
+      }
+      if (rc != MDBX_NOTFOUND)
+        break;
     }
   }
   printf("DATA=END\n");
@@ -206,10 +195,12 @@ static int dump_tbl(MDBX_txn *txn, MDBX_dbi dbi, char *name) {
 static void usage(void) {
   fprintf(stderr,
           "usage: %s "
-          "[-V] [-q] [-f file] [-l] [-p] [-r] [-a|-s table] [-u|U] "
+          "[-V] [-q] [-c] [-f file] [-l] [-p] [-r] [-a|-s table] [-u|U] "
           "dbpath\n"
           "  -V\t\tprint version and exit\n"
           "  -q\t\tbe quiet\n"
+          "  -c\t\tconcise mode without repeating keys,\n"
+          "  \t\tbut incompatible with Berkeley DB and LMDB\n"
           "  -f\t\twrite to file instead of stdout\n"
           "  -l\t\tlist tables and exit\n"
           "  -p\t\tuse printable characters\n"
@@ -268,6 +259,7 @@ int main(int argc, char *argv[]) {
                      "s:"
                      "V"
                      "r"
+                     "c"
                      "q")) != EOF) {
     switch (i) {
     case 'V':
@@ -298,6 +290,9 @@ int main(int argc, char *argv[]) {
       break;
     case 'n':
       break;
+    case 'c':
+      mode |= CONCISE;
+      break;
     case 'p':
       mode |= PRINT;
       break;

src/tools/load.c

@@ -380,7 +380,16 @@ __hot static int readline(MDBX_val *out, MDBX_val *buf) {
       return badend();
     }
   }
-  if (fgets(buf->iov_base, (int)buf->iov_len, stdin) == nullptr)
+
+  /* modern concise mode, where space in second position mean the same (previously) value */
+  c = fgetc(stdin);
+  if (c == EOF)
+    return errno ? errno : EOF;
+  if (c == ' ')
+    return (ungetc(c, stdin) == c) ? MDBX_SUCCESS : (errno ? errno : EOF);
+
+  *(char *)buf->iov_base = c;
+  if (fgets((char *)buf->iov_base + 1, (int)buf->iov_len - 1, stdin) == nullptr)
     return errno ? errno : EOF;
   lineno++;
 
@@ -721,8 +730,8 @@ int main(int argc, char *argv[]) {
     }
 
     int batch = 0;
+    MDBX_val key = {.iov_base = nullptr, .iov_len = 0}, data = {.iov_base = nullptr, .iov_len = 0};
     while (err == MDBX_SUCCESS) {
-      MDBX_val key, data;
       err = readline(&key, &kbuf);
       if (err == EOF)
         break;