Add support for --hashes=sha512_256 as another algorithm

14 files changed, 81 insertions, 28 deletions

diff --git a/doc/ledger.1 b/doc/ledger.1
index febfaf1e..0e5ac7e7 100644
--- a/doc/ledger.1
+++ b/doc/ledger.1
@@ -1,4 +1,4 @@
-.Dd November 23, 2023
+.Dd November 30, 2023
 .Dt LEDGER 1
 .Os
 .Sh NAME
@@ -639,7 +639,14 @@ Record the hash of each transaction in a
 .Ar Hash
 metadata value, according to the hashing algorithm given by the
 .Ar ALGO
-argument. Note that if a
+argument. Support algorithms are
+.Ar sha512
+or
+.Ar sh512_256
+where both use SHA512, but the latter only stores the first half of
+the hash value.
+.Pp
+If a
 .Ar Hash
 metadata value is explicitly provided and does not match what would
 have been generated, an error is reported. Hashes depend on previous
diff --git a/doc/ledger3.texi b/doc/ledger3.texi
index 494847ca..13219758 100644
--- a/doc/ledger3.texi
+++ b/doc/ledger3.texi
@@ -6765,6 +6765,15 @@ they do not match, an error is reported. You may also write just a
 prefix of the @var{Hash}, which is less verbose but still gives quite
 good assurance.
 
+The support algorithms are:
+
+@table @code
+@item sha512
+Use the SHA512 hashing algorithm.
+@item sha512_256
+Same as SHA512, but record only the first 256 bits.
+@end table
+
 Somewhat like balance assertions, which give assurance that previous
 posting amounts are correct, these @var{Hash} tags give assurance that
 all previous journal entries (in parse order) are unchanged (or at
diff --git a/src/generate.cc b/src/generate.cc
index 98986c67..ee820c98 100644
--- a/src/generate.cc
+++ b/src/generate.cc
@@ -365,7 +365,7 @@ void generate_posts_iterator::increment()
       parsing_context.get_current().journal = session.journal.get();
       parsing_context.get_current().scope   = &session;
 
-      if (session.journal->read(parsing_context, false) != 0) {
+      if (session.journal->read(parsing_context, NO_HASHES) != 0) {
         VERIFY(session.journal->xacts.back()->valid());
         posts.reset(*session.journal->xacts.back());
         post = *posts++;
diff --git a/src/global.cc b/src/global.cc
index ad1001ef..050a1d51 100644
--- a/src/global.cc
+++ b/src/global.cc
@@ -116,7 +116,7 @@ void global_scope_t::parse_init(path init_file)
   parsing_context.get_current().journal = session().journal.get();
   parsing_context.get_current().scope   = &report();
 
-  if (session().journal->read(parsing_context, false) > 0 ||
+  if (session().journal->read(parsing_context, NO_HASHES) > 0 ||
       session().journal->auto_xacts.size() > 0 ||
       session().journal->period_xacts.size() > 0) {
     throw_(parse_error, _f("Transactions found in initialization file '%1%'")
diff --git a/src/journal.cc b/src/journal.cc
index b5ae3f1b..62e33440 100644
--- a/src/journal.cc
+++ b/src/journal.cc
@@ -466,7 +466,8 @@ bool journal_t::remove_xact(xact_t * xact)
   return true;
 }
 
-std::size_t journal_t::read(parse_context_stack_t& context, bool store_hashes)
+std::size_t journal_t::read(parse_context_stack_t& context,
+                            hash_type_t hash_type)
 {
   std::size_t count = 0;
   try {
@@ -485,7 +486,7 @@ std::size_t journal_t::read(parse_context_stack_t& context, bool store_hashes)
     if (! current.master)
       current.master = master;
 
-    count = read_textual(context, store_hashes);
+    count = read_textual(context, hash_type);
     if (count > 0) {
       if (! current.pathname.empty())
         sources.push_back(fileinfo_t(current.pathname));
diff --git a/src/journal.h b/src/journal.h
index d68618bf..cfc93c0a 100644
--- a/src/journal.h
+++ b/src/journal.h
@@ -184,7 +184,8 @@ public:
     return period_xacts.end();
   }
 
-  std::size_t read(parse_context_stack_t& context, bool store_hashes);
+  std::size_t read(parse_context_stack_t& context,
+                   hash_type_t hash_type);
 
   bool has_xdata();
   void clear_xdata();
@@ -193,7 +194,8 @@ public:
 
 private:
 
-  std::size_t read_textual(parse_context_stack_t& context, bool store_hashes);
+  std::size_t read_textual(parse_context_stack_t& context,
+                           hash_type_t hash_type);
 
   bool        should_check_payees();
   bool        payee_not_registered(const string& name);
diff --git a/src/precmd.cc b/src/precmd.cc
index 25c28fc8..f90e8ced 100644
--- a/src/precmd.cc
+++ b/src/precmd.cc
@@ -74,7 +74,7 @@ namespace {
         parsing_context.get_current().journal = report.session.journal.get();
         parsing_context.get_current().scope   = &report.session;
 
-        report.session.journal->read(parsing_context, false);
+        report.session.journal->read(parsing_context, NO_HASHES);
         report.session.journal->clear_xdata();
       }
     }
diff --git a/src/session.cc b/src/session.cc
index 4587f24a..f02ee9e9 100644
--- a/src/session.cc
+++ b/src/session.cc
@@ -134,7 +134,7 @@ std::size_t session_t::read_data(const string& master_account)
       parsing_context.push(*price_db_path);
       parsing_context.get_current().journal = journal.get();
       try {
-        if (journal->read(parsing_context, HANDLED(hashes_)) > 0)
+        if (journal->read(parsing_context, HANDLER(hashes_).hash_type) > 0)
           throw_(parse_error, _("Transactions not allowed in price history file"));
       }
       catch (...) {
@@ -169,7 +169,7 @@ std::size_t session_t::read_data(const string& master_account)
     parsing_context.get_current().journal = journal.get();
     parsing_context.get_current().master  = acct;
     try {
-      xact_count += journal->read(parsing_context, HANDLED(hashes_));
+      xact_count += journal->read(parsing_context, HANDLER(hashes_).hash_type);
     }
     catch (...) {
       parsing_context.pop();
@@ -230,7 +230,7 @@ journal_t * session_t::read_journal_from_string(const string& data)
   parsing_context.get_current().journal = journal.get();
   parsing_context.get_current().master  = journal->master;
   try {
-    journal->read(parsing_context, HANDLED(hashes_));
+    journal->read(parsing_context, HANDLER(hashes_).hash_type);
   }
   catch (...) {
     parsing_context.pop();
diff --git a/src/session.h b/src/session.h
index ce1b69a3..c83be644 100644
--- a/src/session.h
+++ b/src/session.h
@@ -163,11 +163,6 @@ public:
      data_files.push_back(str);
    });
 
-  enum hash_type_t {
-    NO_HASHES = 0,
-    HASH_SHA512 = 1
-  };
-
   OPTION__
   (session_t, hashes_,
    hash_type_t hash_type = NO_HASHES;
@@ -175,6 +170,8 @@ public:
    DO_(str) {
      if (str == "sha512" || str == "SHA512") {
        hash_type = HASH_SHA512;
+     } else if (str == "sha512_256" || str == "SHA512_256") {
+       hash_type = HASH_SHA512_256;
      } else {
         throw_(std::invalid_argument, _f("Unrecognized hash type"));
      }
diff --git a/src/textual.cc b/src/textual.cc
index af208232..5488c6a3 100644
--- a/src/textual.cc
+++ b/src/textual.cc
@@ -40,6 +40,7 @@
 #include "query.h"
 #include "pstream.h"
 #include "pool.h"
+#include "session.h"
 #include <algorithm>
 #if HAVE_BOOST_PYTHON
 #include "pyinterp.h"
@@ -79,7 +80,7 @@ namespace {
     instance_t *             parent;
     std::list<application_t> apply_stack;
     bool                     no_assertions;
-    bool                     store_hashes;
+    hash_type_t              hash_type;
 #if TIMELOG_SUPPORT
     time_log_t               timelog;
 #endif
@@ -88,10 +89,10 @@ namespace {
                parse_context_t&       _context,
                instance_t *           _parent = NULL,
                const bool             _no_assertions = false,
-               const bool             _store_hashes = false)
+               const hash_type_t      _hash_type = NO_HASHES)
       : context_stack(_context_stack), context(_context),
         in(*context.stream.get()), parent(_parent),
-        no_assertions(_no_assertions), store_hashes(_store_hashes),
+        no_assertions(_no_assertions), hash_type(_hash_type),
         timelog(context) {}
 
     virtual string description() {
@@ -805,7 +806,7 @@ void instance_t::include_directive(char * line)
           context_stack.get_current().scope   = scope;
           try {
             instance_t instance(context_stack, context_stack.get_current(),
-                                this, no_assertions, store_hashes);
+                                this, no_assertions, hash_type);
             instance.apply_stack.push_front(application_t("account", master));
             instance.parse();
           }
@@ -2020,11 +2021,12 @@ xact_t * instance_t::parse_xact(char *          line,
 
   TRACE_STOP(xact_details, 1);
 
-  if (store_hashes) {
+  if (hash_type != NO_HASHES) {
     string expected_hash =
       xact->hash(previous_xact &&
                  previous_xact->has_tag("Hash") ?
-                 previous_xact->get_tag("Hash")->to_string() : "");
+                 previous_xact->get_tag("Hash")->to_string() : "",
+                 hash_type);
     if (xact->has_tag("Hash")) {
       string current_hash = xact->get_tag("Hash")->to_string();
       if (! std::equal(expected_hash.begin(),
@@ -2059,13 +2061,13 @@ expr_t::ptr_op_t instance_t::lookup(const symbol_t::kind_t kind,
 }
 
 std::size_t journal_t::read_textual(parse_context_stack_t& context_stack,
-                                    bool store_hashes)
+                                    hash_type_t hash_type)
 {
   TRACE_START(parsing_total, 1, "Total time spent parsing text:");
   {
     instance_t instance(context_stack, context_stack.get_current(), NULL,
                         checking_style == journal_t::CHECK_PERMISSIVE,
-                        store_hashes);
+                        hash_type);
     instance.apply_stack.push_front
       (application_t("account", context_stack.get_current().master));
     instance.parse();
diff --git a/src/utils.h b/src/utils.h
index c17c8fb1..34308807 100644
--- a/src/utils.h
+++ b/src/utils.h
@@ -605,6 +605,12 @@ inline string sha1sum(const string& str)
 
 extern const string version;
 
+enum hash_type_t {
+  NO_HASHES = 0,
+  HASH_SHA512 = 1,
+  HASH_SHA512_256 = 2
+};
+
 } // namespace ledger
 
 /*@}*/
diff --git a/src/xact.cc b/src/xact.cc
index 5c3c7042..ab53794a 100644
--- a/src/xact.cc
+++ b/src/xact.cc
@@ -594,7 +594,7 @@ namespace {
   }
 }
 
-string xact_t::hash(string nonce) const {
+string xact_t::hash(string nonce, hash_type_t hash_type) const {
   std::ostringstream repr;
   repr << nonce;
   repr << date();
@@ -621,7 +621,8 @@ string xact_t::hash(string nonce) const {
   unsigned char data[128];
   string repr_str(repr.str());
   SHA512((void *)repr_str.c_str(), repr_str.length(), data);
-  return bufferToHex(data, 64 /*SHA512_DIGEST_LENGTH*/);
+  return bufferToHex(
+    data, hash_type == HASH_SHA512 ? 64 : 32 /*SHA512_DIGEST_LENGTH*/);
 }
 
 namespace {
diff --git a/src/xact.h b/src/xact.h
index 2118c390..379ce778 100644
--- a/src/xact.h
+++ b/src/xact.h
@@ -128,7 +128,7 @@ public:
 
   virtual bool valid() const;
 
-  string hash(string nonce) const;
+  string hash(string nonce, hash_type_t hash_type) const;
 };
 
 class auto_xact_t : public xact_base_t
diff --git a/test/baseline/opt-hashes-sha512_256.test b/test/baseline/opt-hashes-sha512_256.test
new file mode 100644
index 00000000..f269652d
--- /dev/null
+++ b/test/baseline/opt-hashes-sha512_256.test
@@ -0,0 +1,28 @@
+2008/01/01 January
+    Expenses:Books          $10.00
+    Assets:Cash
+
+2008/01/31 End of January
+    Expenses:Books          $10.00
+    Assets:Cash
+
+2008/02/01 February
+    Assets:Cash
+    Expenses:Books          $20.00
+
+test print --hashes=sha512_256
+2008/01/01 January
+    ; Hash: c0f3322857def0d32bd2c9398e0785aacc8206f9ba64ef923978cbdcea13f59c
+    Expenses:Books                            $10.00
+    Assets:Cash
+
+2008/01/31 End of January
+    ; Hash: 151aa236cc3c632e492a94739d92fa76f321939dac04102b4c5ef893384ac839
+    Expenses:Books                            $10.00
+    Assets:Cash
+
+2008/02/01 February
+    ; Hash: 0336590b40e1fc83d6ade4a8c8e8f1c75af10deaac2be80580fbb73b0086324d
+    Assets:Cash
+    Expenses:Books                            $20.00
+end test