chunk/param: add a maximum chunk size param

emory/chunk/find.hpp

@@ -9,6 +9,7 @@
 #include "params.hpp"
 
 #include <cruft/util/hash/buzhash.hpp>
+#include <cruft/util/debug/assert.hpp>
 
 
 ///////////////////////////////////////////////////////////////////////////////
@@ -37,14 +38,18 @@ namespace emory::chunk {
                 hash_state = cruft::rotatel (hash_state, 1) ^ *cursor++;
 
             for ( ; cursor < src.end () - p.window; ++cursor) {
+                if (cursor < start + p.maximum) {
                     if (likely (hash_state & mask)) {
                         hash_state = cruft::rotatel (hash_state, 1)
                                      ^ cruft::rotatel (u64 (*(cursor - p.window)), p.window)
                                      ^ *cursor;
                         continue;
                     }
+                }
 
                 cruft::view<u08 const*> const region { start, cursor };
+                CHECK_GE (cursor - start, p.minimum);
+                CHECK_LE (cursor - start, p.maximum);
 
                 *dst = {
                     .offset = {

emory/chunk/ostream.cpp

@@ -23,6 +23,7 @@ emory::chunk::operator<< (std::ostream &os, params const &val)
     return os << "{ bits: "    << val.bits
               << ", window: "  << val.window
               << ", minimum: " << val.minimum
+              << ", maximum: " << val.maximum
               << " }";
 }
 

emory/chunk/params.hpp

@@ -18,11 +18,14 @@ namespace emory::chunk {
         std::size_t window;
         /// The minimum number of bytes for a matching region.
         std::ptrdiff_t minimum;
+        /// The maximum number of bytes for a matching region
+        std::ptrdiff_t maximum;
     };
 
     constexpr params DEFAULT_PARAMS {
         .bits    =   12,
         .window  =    8,
         .minimum = 4096,
+        .maximum = 4 * 1024 * 1024,
     };
 }

tools/analyse.cpp

@@ -154,6 +154,7 @@ enum {
     ARG_BITS,
     ARG_WINDOW,
     ARG_MINIMUM,
+    ARG_MAXIMUM,
 
     NUM_ARGS,
     NUM_ARGS_REQUIRED = 3,
@@ -164,10 +165,11 @@ enum {
 int main (int argc, char const **argv)
 {
     if (argc < NUM_ARGS_REQUIRED) {
-        std::cerr << "usage: " << argv[ARG_SELF] << " <input> <output> [bits] [window] [minimum]\n"
+        std::cerr << "usage: " << argv[ARG_SELF] << " <input> <output> [bits] [window] [minimum] [maximum]\n"
                   << "default bits = "    << emory::chunk::DEFAULT_PARAMS.bits    << '\n'
                   << "default window = "  << emory::chunk::DEFAULT_PARAMS.window  << '\n'
-                  << "default minimum = " << emory::chunk::DEFAULT_PARAMS.minimum << '\n';
+                  << "default minimum = " << emory::chunk::DEFAULT_PARAMS.minimum << '\n'
+                  << "default maximum = " << emory::chunk::DEFAULT_PARAMS.maximum << '\n';
         return EXIT_FAILURE;
     }
 
@@ -178,6 +180,8 @@ int main (int argc, char const **argv)
         p.window = cruft::parse::from_string<std::size_t> (argv[ARG_WINDOW]);
     if (argc > ARG_BITS + 1)
         p.minimum = cruft::parse::from_string<std::size_t> (argv[ARG_MINIMUM]);
+    if (argc > ARG_MAXIMUM + 1)
+        p.maximum = cruft::parse::from_string<std::size_t> (argv[ARG_MAXIMUM]);
 
     std::cerr << p << '\n';
 

tools/compare.cpp

@@ -24,6 +24,7 @@ enum {
     ARG_BITS,
     ARG_WINDOW,
     ARGS_MINIMUM,
+    ARGS_MAXIMUM,
     ARGS_TARGET,
     ARGS_SOURCE,
 
@@ -34,7 +35,7 @@ enum {
 int main (int argc, char const **argv)
 {
     if (argc < NUM_ARGS) {
-        std::cerr << "usage: " << argv[ARG_SELF] << " <bits> <window> <minimum> <target> <source> [...]\n";
+        std::cerr << "usage: " << argv[ARG_SELF] << " <bits> <window> <minimum> <maximum> <target> <source> [...]\n";
         return EXIT_FAILURE;
     }
 
@@ -42,6 +43,7 @@ int main (int argc, char const **argv)
         .bits    = cruft::parse::from_string<std::size_t> (argv[ARG_BITS  ]),
         .window  = cruft::parse::from_string<std::size_t> (argv[ARG_WINDOW]),
         .minimum = cruft::parse::from_string<std::ptrdiff_t> (argv[ARGS_MINIMUM]),
+        .maximum = cruft::parse::from_string<std::ptrdiff_t> (argv[ARGS_MAXIMUM]),
     };
 
     std::clog << "Hashing target\n";