Merge remote-tracking branch 'tq/master' into LL such that it compiles

Conflicts:
	src/Makefile
	src/backends/packrat.c
	src/compile.c
	src/hammer.h
	src/internal.h
	src/parsers/action.c
	src/parsers/and.c
	src/parsers/attr_bool.c
	src/parsers/bits.c
	src/parsers/butnot.c
	src/parsers/ch.c
	src/parsers/charset.c
	src/parsers/choice.c
	src/parsers/difference.c
	src/parsers/end.c
	src/parsers/epsilon.c
	src/parsers/ignore.c
	src/parsers/ignoreseq.c
	src/parsers/indirect.c
	src/parsers/int_range.c
	src/parsers/many.c
	src/parsers/not.c
	src/parsers/nothing.c
	src/parsers/optional.c
	src/parsers/sequence.c
	src/parsers/token.c
	src/parsers/unimplemented.c
	src/parsers/whitespace.c
	src/parsers/xor.c

.gitignore

@@ -8,3 +8,7 @@ examples/base64
 TAGS
 *.swp
 *.swo
+\#*
+.*
+docs/milestone2.dot.pdf
+*.dot.pdf

Makefile

@@ -17,6 +17,9 @@ CONFIG_VARS= INCLUDE_TESTS
 test: src/test_suite
 	$<
 
+examples/all: src/all
+examples/compile: src/compile
+
 define SUBDIR_TEMPLATE
 $(1)/%:
 	$$(MAKE) -C $(1) $$*

common.mk

@@ -6,7 +6,7 @@ endif
 include $(TOPLEVEL)/config.mk
 
 TEST_CFLAGS = $(shell pkg-config --cflags glib-2.0) -DINCLUDE_TESTS
-TEST_LDFLAGS = $(shell pkg-config --libs glib-2.0)
+TEST_LDFLAGS = $(shell pkg-config --libs glib-2.0) -lrt
 
 CFLAGS := -std=gnu99 -Wall -Wextra -Werror -Wno-unused-parameter -Wno-attributes
 LDFLAGS :=

docs/milestone2.dot

@@ -0,0 +1,36 @@
+digraph {
+	graph [rankdir=LR];
+	subgraph complete {
+		node [color="gray",fontcolor="gray"];
+		glue;
+		regex_svm;
+		regex_rvm;
+		desugaring; // Needs merged.
+	}
+	/* The end result of the milestone, along with the subtasks listed */
+	milestone2 [color="green",style="filled"];
+	llk -> milestone2;
+	lr -> milestone2;
+	lalr8_gen -> lr; // Generate parse tables for LALR(8)
+	glr_gen -> lr; // Generate parse tables for GLR
+	llk_gen -> llk; // Generate parse tables for LL(k)
+	lr_driver -> lr; // Write driver for all LR-type algs; analagous to SVM and RVM implementations
+	llk_driver -> llk; // Write driver for LL(k)
+	regex -> milestone2;
+	glue -> milestone2;
+	tests -> milestone2;
+	
+	regex_gen -> regex; // should be mostly done; the rest is concurrent with regex_svm_actions
+	regex_driver -> regex;
+	regex_svm -> regex_driver;
+	regex_rvm -> regex_driver;
+	regex_svm_actions -> regex_driver; // 1 for each way that an HParsedToken can be extracted from the stack.
+
+
+	/*
+	* 
+	*/
+	desugaring -> llk_gen;
+	desugaring -> lalr8_gen;
+	desugaring -> glr_gen;
+}

docs/milestone3.dot

@@ -0,0 +1,65 @@
+digraph {
+	graph [rankdir=LR];
+
+	subgraph complete {
+		node [color="gray",fontcolor="gray"];
+	}
+
+	subgraph groups {
+		node [color="blue",fontcolor="blue"];
+		cpp;
+		python;
+		ruby;
+		go;
+		php;
+		dotnet;
+	}
+	
+	milestone3 [color="green",style="filled"];
+
+
+	function_desc_fmt -> function_descs;
+	function_desc_fmt -> binding_generator;
+
+	binding_generator -> cpp_gen;
+	binding_generator -> python_gen;
+	binding_generator -> ruby_gen;
+	binding_generator -> go_gen;
+	binding_generator -> php_gen;
+	binding_generator -> dotnet_gen;
+
+	function_descs -> cpp_gen;
+	function_descs -> python_gen;
+	function_descs -> ruby_gen;
+	function_descs -> go_gen;
+	function_descs -> php_gen;
+	function_descs -> dotnet_gen;
+
+
+	// Plugins to generate a type of code
+	cpp_gen -> cpp;
+	python_gen -> python;
+	ruby_gen -> ruby;
+	go_gen -> go;
+	php_gen -> php;
+	dotnet_gen -> dotnet;
+	
+	// base code... developed concurrently with _gen's
+	cpp_base -> cpp;
+	python_base -> python;
+	ruby_base -> ruby;
+	go_base -> go;
+	php_base -> php;
+	dotnet_base -> dotnet;
+	
+	// Bindings for various languages. These are just groupings.
+	cpp -> milestone3;
+	python -> milestone3;
+	ruby -> milestone3;
+	go -> milestone3;
+	php -> milestone3;
+	dotnet -> milestone3;
+
+	
+
+}

docs/rvm_sample_input.rvm

@@ -0,0 +1,17 @@
++C
+int foo() {
+    return 42;
+}
+
++SVM /svm/simple
+@input ""
+@output "()"
+0 ACCEPT
+
++SVM /svm/string
+@input "quux"
+@outut "(<5555>)"
+1 MARK
+2 CAPTURE
+2 ACCEPT
+

src/Makefile

@@ -27,7 +27,8 @@ PARSERS := \
 
 BACKENDS := \
 	packrat \
-	ll
+	llk \
+	regex
 
 HAMMER_PARTS := \
 	bitreader.o \
@@ -39,7 +40,6 @@ HAMMER_PARTS := \
 	datastructures.o \
 	system_allocator.o \
 	benchmark.o \
-	compile.o \
 	cfgrammar.o \
 	$(PARSERS:%=parsers/%.o) \
 	$(BACKENDS:%=backends/%.o)

src/allocator.h

@@ -19,6 +19,7 @@
 #define HAMMER_ALLOCATOR__H__
 #include <sys/types.h>
 
+// TODO(thequux): Turn this into an "HAllocatorVtable", and add a wrapper that also takes an environment pointer.
 typedef struct HAllocator_ {
   void* (*alloc)(struct HAllocator_* allocator, size_t size);
   void* (*realloc)(struct HAllocator_* allocator, void* ptr, size_t size);

src/backends/llk.c

@@ -3,22 +3,23 @@
 #include "../cfgrammar.h"
 #include "../parsers/parser_internal.h"
 
+// XXX despite the names, this is all LL(1) right now. TODO
 
 
-/* Generating the LL parse table */
+/* Generating the LL(k) parse table */
 
 /* Maps each nonterminal (HCFChoice) of the grammar to another hash table that
  * maps lookahead tokens (HCFToken) to productions (HCFSequence).
  */
-typedef struct HLLTable_ {
+typedef struct HLLkTable_ {
   HHashTable *rows;
   HCFChoice  *start;    // start symbol
   HArena     *arena;
   HAllocator *mm__;
-} HLLTable;
+} HLLkTable;
 
 /* Interface to look up an entry in the parse table. */
-const HCFSequence *h_ll_lookup(const HLLTable *table, const HCFChoice *x, HCFToken tok)
+const HCFSequence *h_llk_lookup(const HLLkTable *table, const HCFChoice *x, HCFToken tok)
 {
   const HHashTable *row = h_hashtable_get(table->rows, x);
   assert(row != NULL);  // the table should have one row for each nonterminal
@@ -28,7 +29,7 @@ const HCFSequence *h_ll_lookup(const HLLTable *table, const HCFChoice *x, HCFTok
 }
 
 /* Allocate a new parse table. */
-HLLTable *h_lltable_new(HAllocator *mm__)
+HLLkTable *h_llktable_new(HAllocator *mm__)
 {
   // NB the parse table gets an arena separate from the grammar so we can free
   //    the latter after table generation.
@@ -37,7 +38,7 @@ HLLTable *h_lltable_new(HAllocator *mm__)
   HHashTable *rows = h_hashtable_new(arena, h_eq_ptr, h_hash_ptr);
   assert(rows != NULL);
 
-  HLLTable *table = h_new(HLLTable, 1);
+  HLLkTable *table = h_new(HLLkTable, 1);
   assert(table != NULL);
   table->mm__  = mm__;
   table->arena = arena;
@@ -46,7 +47,7 @@ HLLTable *h_lltable_new(HAllocator *mm__)
   return table;
 }
 
-void h_lltable_free(HLLTable *table)
+void h_llktable_free(HLLkTable *table)
 {
   HAllocator *mm__ = table->mm__;
   h_delete_arena(table->arena);
@@ -95,10 +96,10 @@ int fill_table_row(HCFGrammar *g, HHashTable *row,
   return 0;
 }
 
-/* Generate the LL parse table from the given grammar.
+/* Generate the LL(k) parse table from the given grammar.
  * Returns -1 on error, 0 on success.
  */
-static int fill_table(HCFGrammar *g, HLLTable *table)
+static int fill_table(HCFGrammar *g, HLLkTable *table)
 {
   table->start = g->start;
 
@@ -120,7 +121,7 @@ static int fill_table(HCFGrammar *g, HLLTable *table)
       for(s = a->seq; *s; s++) {
         // record this production in row as appropriate
         // this can signal an ambiguity conflict.
-        // NB we don't worry about deallocating anything, h_ll_compile will
+        // NB we don't worry about deallocating anything, h_llk_compile will
         //    delete the whole arena for us.
         if(fill_table_row(g, row, a, *s) < 0)
           return -1;
@@ -131,7 +132,7 @@ static int fill_table(HCFGrammar *g, HLLTable *table)
   return 0;
 }
 
-int h_ll_compile(HAllocator* mm__, HParser* parser, const void* params)
+int h_llk_compile(HAllocator* mm__, HParser* parser, const void* params)
 {
   // Convert parser to a CFG. This can fail as indicated by a NULL return.
   HCFGrammar *grammar = h_cfgrammar(mm__, parser);
@@ -143,11 +144,11 @@ int h_ll_compile(HAllocator* mm__, HParser* parser, const void* params)
   // TODO: avoid conflicts by splitting occurances?
 
   // generate table and store in parser->data.
-  HLLTable *table = h_lltable_new(mm__);
+  HLLkTable *table = h_llktable_new(mm__);
   if(fill_table(grammar, table) < 0) {
     // the table was ambiguous
     h_cfgrammar_free(grammar);
-    h_lltable_free(table);
+    h_llktable_free(table);
     return -1;
   }
   parser->data = table;
@@ -161,13 +162,14 @@ int h_ll_compile(HAllocator* mm__, HParser* parser, const void* params)
 
 
 
-/* LL driver */
+/* LL(k) driver */
 
-HParseResult *h_ll_parse(HAllocator* mm__, const HParser* parser, HParseState* state)
+HParseResult *h_llk_parse(HAllocator* mm__, const HParser* parser, HInputStream* stream)
 {
-  const HLLTable *table = parser->data;
+  const HLLkTable *table = parser->data;
-  HArena *arena = state->arena;
+  HArena *arena  = h_new_arena(mm__, 0);    // will hold the results
-  HSlist *stack = h_slist_new(arena);
+  HArena *tarena = h_new_arena(mm__, 0);    // tmp, deleted after parse
+  HSlist *stack  = h_slist_new(tarena);
   HCountedArray *seq = h_carray_new(arena); // accumulates current parse result
 
   // in order to construct the parse tree, we delimit the symbol stack into
@@ -177,7 +179,7 @@ HParseResult *h_ll_parse(HAllocator* mm__, const HParser* parser, HParseState* s
   // frame delimiter.
   // also on the stack below the mark, we store the previously accumulated
   // value for the surrounding production.
-  void *mark = h_arena_malloc(arena, 1);
+  void *mark = h_arena_malloc(tarena, 1);
 
   // initialize with the start symbol on the stack.
   h_slist_push(stack, table->start);
@@ -188,8 +190,8 @@ HParseResult *h_ll_parse(HAllocator* mm__, const HParser* parser, HParseState* s
   while(!h_slist_empty(stack)) {
     // fill up lookahead buffer as required
     if(lookahead == 0) {
-        uint8_t c = h_read_bits(&state->input_stream, 8, false);
+        uint8_t c = h_read_bits(stream, 8, false);
-        if(state->input_stream.overrun)
+        if(stream->overrun)
           lookahead = end_token;
         else
           lookahead = char_token(c);
@@ -203,16 +205,16 @@ HParseResult *h_ll_parse(HAllocator* mm__, const HParser* parser, HParseState* s
       // hit stack frame boundary
 
       // wrap the accumulated parse result, this sequence is finished
-      HParsedToken *tok = a_new(HParsedToken, 1);
+      HParsedToken *tok = h_arena_malloc(arena, sizeof(HParsedToken));
       tok->token_type = TT_SEQUENCE;
       tok->seq = seq;
       // XXX tok->index and tok->bit_offset (don't take directly from stream, cuz peek!)
 
       // call validation and semantic action, if present
-      if(x->pred && !x->pred(make_result(state, tok)))
+      if(x->pred && !x->pred(make_result(tarena, tok)))
-        return NULL;    // validation failed -> no parse
+        goto no_parse;    // validation failed -> no parse
       if(x->action)
-        tok = (HParsedToken *)x->action(make_result(state, tok));
+        tok = (HParsedToken *)x->action(make_result(arena, tok));
 
       // result becomes next left-most element of higher-level sequence
       seq = h_slist_pop(stack);
@@ -230,7 +232,7 @@ HParseResult *h_ll_parse(HAllocator* mm__, const HParser* parser, HParseState* s
       seq = h_carray_new(arena);
 
       // look up applicable production in parse table
-      const HCFSequence *p = h_ll_lookup(table, x, lookahead);
+      const HCFSequence *p = h_llk_lookup(table, x, lookahead);
 
       // push production's rhs onto the stack (in reverse order)
       HCFChoice **s;
@@ -250,40 +252,40 @@ HParseResult *h_ll_parse(HAllocator* mm__, const HParser* parser, HParseState* s
       switch(x->type) {
       case HCF_END:
         if(input != end_token)
-          return NULL;
+          goto no_parse;
         tok = NULL;
         break;
 
       case HCF_CHAR:
         if(input != char_token(x->chr))
-          return NULL;
+          goto no_parse;
-        tok = a_new(HParsedToken, 1);
+        tok = h_arena_malloc(arena, sizeof(HParsedToken));
         tok->token_type = TT_UINT;
         tok->uint = x->chr;
         break;
 
       case HCF_CHARSET:
         if(input == end_token)
-          return NULL;
+          goto no_parse;
         if(!charset_isset(x->charset, token_char(input)))
-          return NULL;
+          goto no_parse;
-        tok = a_new(HParsedToken, 1);
+        tok = h_arena_malloc(arena, sizeof(HParsedToken));
         tok->token_type = TT_UINT;
         tok->uint = token_char(input);
         break;
 
       default: // should not be reached
         assert_message(0, "unknown HCFChoice type");
-        return NULL;
+        goto no_parse;
       }
 
       // XXX tok->index and tok->bit_offset (don't take directly from stream, cuz peek!)
 
       // call validation and semantic action, if present
-      if(x->pred && !x->pred(make_result(state, tok)))
+      if(x->pred && !x->pred(make_result(tarena, tok)))
-        return NULL;    // validation failed -> no parse
+        goto no_parse;  // validation failed -> no parse
       if(x->action)
-        tok = (HParsedToken *)x->action(make_result(state, tok));
+        tok = (HParsedToken *)x->action(make_result(arena, tok));
 
       // append to result sequence
       h_carray_append(seq, tok);
@@ -293,25 +295,31 @@ HParseResult *h_ll_parse(HAllocator* mm__, const HParser* parser, HParseState* s
   // since we started with a single nonterminal on the stack, seq should
   // contain exactly the parse result.
   assert(seq->used == 1);
-  return make_result(state, seq->elements[0]);
+  h_delete_arena(tarena);
+  return make_result(arena, seq->elements[0]);
+
+  no_parse:
+    h_delete_arena(tarena);
+    h_delete_arena(arena);
+    return NULL;
 }
 
 
 
-HParserBackendVTable h__ll_backend_vtable = {
+HParserBackendVTable h__llk_backend_vtable = {
-  .compile = h_ll_compile,
+  .compile = h_llk_compile,
-  .parse = h_ll_parse
+  .parse = h_llk_parse
 };
 
 
 
 
 // dummy!
-int test_ll(void)
+int test_llk(void)
 {
-  const HParser *c = h_many(h_ch('x'));
+  HParser *c = h_many(h_ch('x'));
-  const HParser *q = h_sequence(c, h_ch('y'), NULL);
+  HParser *q = h_sequence(c, h_ch('y'), NULL);
-  const HParser *p = h_choice(q, h_end_p(), NULL);
+  HParser *p = h_choice(q, h_end_p(), NULL);
 
   HCFGrammar *g = h_cfgrammar(&system_allocator, p);
 

src/backends/packrat.c

@@ -1,7 +1,16 @@
 #include <assert.h>
+#include <string.h>
 #include "../internal.h"
 #include "../parsers/parser_internal.h"
 
+static uint32_t djbhash(const uint8_t *buf, size_t len) {
+  uint32_t hash = 5381;
+  while (len--) {
+    hash = hash * 33 + *buf++;
+  }
+  return hash;
+}
+
 // short-hand for constructing HCachedResult's
 static HCachedResult *cached_result(const HParseState *state, HParseResult *result) {
   HCachedResult *ret = a_new(HCachedResult, 1);
@@ -191,12 +200,37 @@ HParseResult* h_do_parse(const HParser* parser, HParseState *state) {
 }
 
 int h_packrat_compile(HAllocator* mm__, HParser* parser, const void* params) {
+  parser->backend = PB_PACKRAT;
   return 0; // No compilation necessary, and everything should work
 	    // out of the box.
 }
 
-HParseResult *h_packrat_parse(HAllocator* mm__, const HParser* parser, HParseState* parse_state) {
+static uint32_t cache_key_hash(const void* key) {
-  return h_do_parse(parser, parse_state);
+  return djbhash(key, sizeof(HParserCacheKey));
+}
+static bool cache_key_equal(const void* key1, const void* key2) {
+  return memcmp(key1, key2, sizeof(HParserCacheKey)) == 0;
+}
+
+HParseResult *h_packrat_parse(HAllocator* mm__, const HParser* parser, HInputStream *input_stream) {
+  HArena * arena = h_new_arena(mm__, 0);
+  HParseState *parse_state = a_new_(arena, HParseState, 1);
+  parse_state->cache = h_hashtable_new(arena, cache_key_equal, // key_equal_func
+				       cache_key_hash); // hash_func
+  parse_state->input_stream = *input_stream;
+  parse_state->lr_stack = h_slist_new(arena);
+  parse_state->recursion_heads = h_hashtable_new(arena, cache_key_equal,
+						 cache_key_hash);
+  parse_state->arena = arena;
+  HParseResult *res = h_do_parse(parser, parse_state);
+  h_slist_free(parse_state->lr_stack);
+  h_hashtable_free(parse_state->recursion_heads);
+  // tear down the parse state
+  h_hashtable_free(parse_state->cache);
+  if (!res)
+    h_delete_arena(parse_state->arena);
+
+  return res;
 }
 
 HParserBackendVTable h__packrat_backend_vtable = {

src/backends/regex.c

@@ -0,0 +1,366 @@
+#include <string.h>
+#include <assert.h>
+#include "../internal.h"
+#include "../parsers/parser_internal.h"
+#include "regex.h"
+
+#undef a_new
+#define a_new(typ, count) a_new_(arena, typ, count)
+// Stack VM
+typedef enum HSVMOp_ {
+  SVM_PUSH, // Push a mark. There is no VM insn to push an object.
+  SVM_NOP, // Used to start the chain, and possibly elsewhere. Does nothing.
+  SVM_ACTION, // Same meaning as RVM_ACTION
+  SVM_CAPTURE, // Same meaning as RVM_CAPTURE
+  SVM_ACCEPT,
+} HSVMOp;
+
+typedef struct HRVMTrace_ {
+  struct HRVMTrace_ *next; // When parsing, these are
+			   // reverse-threaded. There is a postproc
+			   // step that inverts all the pointers.
+  size_t input_pos;
+  uint16_t arg;
+  uint8_t opcode;
+} HRVMTrace;
+
+typedef struct HRVMThread_ {
+  HRVMTrace *trace;
+  uint16_t ip;
+} HRVMThread;
+
+HParseResult *run_trace(HAllocator *mm__, HRVMProg *orig_prog, HRVMTrace *trace, const uint8_t *input, int len);
+
+HRVMTrace *invert_trace(HRVMTrace *trace) {
+  HRVMTrace *last = NULL;
+  if (!trace)
+    return NULL;
+  if (!trace->next)
+    return trace;
+  do {
+    HRVMTrace *next = trace->next;
+    trace->next = last;
+    last = trace;
+    trace = next;
+  } while (trace->next);
+  return trace;
+}
+
+void* h_rvm_run__m(HAllocator *mm__, HRVMProg *prog, const uint8_t* input, size_t len) {
+  HArena *arena = h_new_arena(mm__, 0);
+  HRVMTrace **heads_p = a_new(HRVMTrace*, prog->length),
+    **heads_n = a_new(HRVMTrace*, prog->length);
+
+  HRVMTrace *ret_trace;
+  
+  uint8_t *insn_seen = a_new(uint8_t, prog->length); // 0 -> not seen, 1->processed, 2->queued
+  HRVMThread *ip_queue = a_new(HRVMThread, prog->length);
+  size_t ipq_top;
+
+  
+  
+
+#define THREAD ip_queue[ipq_top-1]
+#define PUSH_SVM(op_, arg_) do { \
+	  HRVMTrace *nt = a_new(HRVMTrace, 1); \
+	  nt->arg = (arg_);		       \
+	  nt->opcode = (op_);		       \
+	  nt->next = THREAD.trace;	       \
+	  nt->input_pos = off;		       \
+	  THREAD.trace = nt;		       \
+  } while(0)
+    
+  heads_n[0] = a_new(HRVMTrace, 1); // zeroing
+  heads_n[0]->opcode = SVM_NOP;
+
+  size_t off = 0;
+  int live_threads = 1;
+  for (off = 0; off <= len; off++) {
+    uint8_t ch = ((off == len) ? 0 : input[off]);
+    size_t ip_s; // BUG: there was an unused variable ip. Not sure if
+		 // I intended to use it somewhere.
+    /* scope */ {
+      HRVMTrace **heads_t;
+      heads_t = heads_n;
+      heads_n = heads_p;
+      heads_p = heads_t;
+      memset(heads_n, 0, prog->length * sizeof(*heads_n));
+    }
+    memset(insn_seen, 0, prog->length); // no insns seen yet
+    if (!live_threads)
+      goto match_fail;
+    live_threads = 0;
+    for (ip_s = 0; ip_s < prog->length; ip_s++) {
+      ipq_top = 1;
+      // TODO: Write this as a threaded VM
+      if (!heads_p[ip_s])
+	continue;
+      THREAD.ip = ip_s;
+
+      uint8_t hi, lo;
+      uint16_t arg;
+      while(ipq_top > 0) {
+	if (insn_seen[THREAD.ip] == 1)
+	  continue;
+	insn_seen[THREAD.ip] = 1;
+	arg = prog->insns[THREAD.ip].arg;
+	switch(prog->insns[THREAD.ip].op) {
+	case RVM_ACCEPT:
+	  PUSH_SVM(SVM_ACCEPT, 0);
+	  ret_trace = THREAD.trace;
+	  goto run_trace;
+	case RVM_MATCH:
+	  // Doesn't actually validate the "must be followed by MATCH
+	  // or STEP. It should. Preproc perhaps?
+	  hi = (arg >> 8) & 0xff;
+	  lo = arg & 0xff;
+	  THREAD.ip++;
+	  if (ch < lo || ch > hi)
+	    ipq_top--; // terminate thread
+	  goto next_insn;
+	case RVM_GOTO:
+	  THREAD.ip = arg;
+	  goto next_insn;
+	case RVM_FORK:
+	  THREAD.ip++;
+	  if (!insn_seen[arg]) {
+	    insn_seen[THREAD.ip] = 2;
+	    HRVMTrace* tr = THREAD.trace;
+	    ipq_top++;
+	    THREAD.ip = arg;
+	    THREAD.trace = tr;
+	  }
+	  goto next_insn;
+	case RVM_PUSH:
+	  PUSH_SVM(SVM_PUSH, 0);
+	  THREAD.ip++;
+	  goto next_insn;
+	case RVM_ACTION:
+	  PUSH_SVM(SVM_ACTION, arg);
+	  THREAD.ip++;
+	  goto next_insn;
+	case RVM_CAPTURE:
+	  PUSH_SVM(SVM_CAPTURE, 0);
+	  THREAD.ip++;
+	  goto next_insn;
+	case RVM_EOF:
+	  THREAD.ip++;
+	  if (off != len)
+	    ipq_top--; // Terminate thread
+	  goto next_insn;
+	case RVM_STEP:
+	  // save thread
+	  live_threads++;
+	  heads_n[THREAD.ip++] = THREAD.trace;
+	  ipq_top--;
+	  goto next_insn;
+	}
+      next_insn:
+	;
+	
+      }
+    }
+  }
+  // No accept was reached.
+ match_fail:
+  h_delete_arena(arena);
+  return NULL;
+  
+ run_trace:
+  // Invert the direction of the trace linked list.
+
+  
+  ret_trace = invert_trace(ret_trace);
+  HParseResult *ret = run_trace(mm__, prog, ret_trace, input, len);
+  // ret is in its own arena
+  h_delete_arena(arena);
+  return ret;
+}
+#undef PUSH_SVM
+#undef THREAD
+
+
+
+
+void svm_stack_ensure_cap(HAllocator *mm__, HSVMContext *ctx, size_t addl) {
+  if (ctx->stack_count + addl >= ctx->stack_capacity) {
+    ctx->stack = mm__->realloc(mm__, ctx->stack, sizeof(*ctx->stack) * (ctx->stack_capacity *= 2));
+    // TODO: check for realloc failure
+  }
+}
+
+HParseResult *run_trace(HAllocator *mm__, HRVMProg *orig_prog, HRVMTrace *trace, const uint8_t *input, int len) {
+  // orig_prog is only used for the action table
+  HSVMContext ctx;
+  HArena *arena = h_new_arena(mm__, 0);
+  ctx.stack_count = 0;
+  ctx.stack_capacity = 16;
+  ctx.stack = h_new(HParsedToken*, ctx.stack_capacity);
+
+  HParsedToken *tmp_res;
+  HRVMTrace *cur;
+  for (cur = trace; cur; cur = cur->next) {
+    switch (cur->opcode) {
+    case SVM_PUSH:
+      svm_stack_ensure_cap(mm__, &ctx, 1);
+      tmp_res = a_new(HParsedToken, 1);
+      tmp_res->token_type = TT_MARK;
+      tmp_res->index = cur->input_pos;
+      tmp_res->bit_offset = 0;
+      ctx.stack[ctx.stack_count++] = tmp_res;
+      break;
+    case SVM_NOP:
+      break;
+    case SVM_ACTION:
+      // Action should modify stack appropriately
+      if (!orig_prog->actions[cur->arg].action(arena, &ctx, orig_prog->actions[cur->arg].env)) {
+	// action failed... abort somehow
+	// TODO: Actually abort
+      }
+      break;
+    case SVM_CAPTURE: 
+      // Top of stack must be a mark
+      // This replaces said mark in-place with a TT_BYTES.
+      assert(ctx.stack[ctx.stack_count]->token_type == TT_MARK);
+      
+      tmp_res = ctx.stack[ctx.stack_count];
+      tmp_res->token_type = TT_BYTES;
+      // TODO: Will need to copy if bit_offset is nonzero
+      assert(tmp_res->bit_offset == 0);
+	
+      tmp_res->bytes.token = input + tmp_res->index;
+      tmp_res->bytes.len = cur->input_pos - tmp_res->index + 1; // inclusive
+      break;
+    case SVM_ACCEPT:
+      assert(ctx.stack_count == 1);
+      HParseResult *res = a_new(HParseResult, 1);
+      res->ast = ctx.stack[0];
+      res->bit_length = cur->input_pos * 8;
+      res->arena = arena;
+      return res;
+    }
+  }
+
+  h_delete_arena(arena);
+  return NULL;
+}
+
+uint16_t h_rvm_create_action(HRVMProg *prog, HSVMActionFunc action_func, void* env) {
+  for (uint16_t i = 0; i < prog->action_count; i++) {
+    if (prog->actions[i].action == action_func && prog->actions[i].env == env)
+      return i;
+  }
+  // Ensure that there's room in the action array...
+  if (!(prog->action_count & (prog->action_count + 1))) {
+    // needs to be scaled up.
+    size_t array_size = (prog->action_count + 1) * 2; // action_count+1 is a
+					       // power of two
+    prog->actions = prog->allocator->realloc(prog->allocator, prog->actions, array_size * sizeof(*prog->actions));
+    // TODO: Handle the allocation failed case nicely.
+  }
+
+  HSVMAction *action = &prog->actions[prog->action_count];
+  action->action = action_func;
+  action->env = env;
+  return prog->action_count++;
+}
+
+uint16_t h_rvm_insert_insn(HRVMProg *prog, HRVMOp op, uint16_t arg) {
+  // Ensure that there's room in the insn array...
+  if (!(prog->length & (prog->length + 1))) {
+    // needs to be scaled up.
+    size_t array_size = (prog->length + 1) * 2; // action_count+1 is a
+						// power of two
+    prog->insns = prog->allocator->realloc(prog->allocator, prog->insns, array_size * sizeof(*prog->insns));
+    // TODO: Handle the allocation failed case nicely.
+  }
+
+  prog->insns[prog->length].op = op;
+  prog->insns[prog->length].arg = arg;
+  return prog->length++;
+}
+
+uint16_t h_rvm_get_ip(HRVMProg *prog) {
+  return prog->length;
+}
+
+void h_rvm_patch_arg(HRVMProg *prog, uint16_t ip, uint16_t new_val) {
+  assert(prog->length > ip);
+  prog->insns[ip].arg = new_val;
+}
+
+size_t h_svm_count_to_mark(HSVMContext *ctx) {
+  size_t ctm;
+  for (ctm = 0; ctm < ctx->stack_count-1; ctm++) {
+    if (ctx->stack[ctx->stack_count - 1 - ctm]->token_type == TT_MARK)
+      return ctm;
+  }
+  return ctx->stack_count;
+}
+
+// TODO: Implement the primitive actions
+bool h_svm_action_make_sequence(HArena *arena, HSVMContext *ctx, void* env) {
+  size_t n_items = h_svm_count_to_mark(ctx);
+  assert (n_items < ctx->stack_count);
+  HParsedToken *res = ctx->stack[ctx->stack_count - 1 - n_items];
+  assert (res->token_type == TT_MARK);
+  res->token_type = TT_SEQUENCE;
+  
+  HCountedArray *ret_carray = h_carray_new_sized(arena, n_items);
+  res->seq = ret_carray;
+  // res index and bit offset are the same as the mark.
+  for (size_t i = 0; i < n_items; i++) {
+    ret_carray->elements[i] = ctx->stack[ctx->stack_count - n_items + i];
+  }
+  ctx->stack_count -= n_items;
+  return true;
+}
+
+bool h_svm_action_clear_to_mark(HArena *arena, HSVMContext *ctx, void* env) {
+  while (ctx->stack_count > 0) {
+    if (ctx->stack[--ctx->stack_count]->token_type == TT_MARK)
+      return true;
+  }
+  return false; // no mark found.
+}
+
+// Glue regex backend to rest of system
+
+bool h_compile_regex(HRVMProg *prog, const HParser *parser) {
+  return parser->vtable->compile_to_rvm(prog, parser->env);
+}
+
+static void h_regex_free(HParser *parser) {
+  HRVMProg *prog = (HRVMProg*)parser->backend_data;
+  HAllocator *mm__ = prog->allocator;
+  h_free(prog->insns);
+  h_free(prog->actions);
+  h_free(prog);
+  parser->backend_data = NULL;
+  parser->backend = PB_PACKRAT;
+}
+
+static int h_regex_compile(HAllocator *mm__, HParser* parser, const void* params) {
+  if (!parser->vtable->isValidRegular(parser->env))
+    return 1;
+  HRVMProg *prog = h_new(HRVMProg, 1);
+  prog->allocator = mm__;
+  if (!h_compile_regex(prog, parser)) {
+    h_free(prog->insns);
+    h_free(prog->actions);
+    h_free(prog);
+    return 2;
+  }
+  parser->backend_data = prog;
+  return 0;
+}
+
+static HParseResult *h_regex_parse(HAllocator* mm__, const HParser* parser, HInputStream *input_stream) {
+  return h_rvm_run__m(mm__, (HRVMProg*)parser->backend_data, input_stream->input, input_stream->length);
+}
+
+HParserBackendVTable h__regex_backend_vtable = {
+  .compile = h_regex_compile,
+  .parse = h_regex_parse,
+  .free = h_regex_free
+};

src/backends/regex.h

@@ -0,0 +1,80 @@
+// Internal defs
+#ifndef HAMMER_BACKEND_REGEX__H
+#define HAMMER_BACKEND_REGEX__H
+
+// each insn is an 8-bit opcode and a 16-bit parameter
+// [a] are actions; they add an instruction to the stackvm that is being output.
+// [m] are match ops; they can either succeed or fail, depending on the current character
+// [c] are control ops. They affect the pc non-linearly.
+typedef enum HRVMOp_ {
+  RVM_ACCEPT,  // [a]
+  RVM_GOTO,    // [c] parameter is an offset into the instruction table
+  RVM_FORK,    // [c] parameter is an offset into the instruction table
+  RVM_PUSH,    // [a] No arguments, just pushes a mark (pointer to some
+               //     character in the input string) onto the stack
+  RVM_ACTION,  // [a] argument is an action ID
+  RVM_CAPTURE, // [a] Capture the last string (up to the current
+	       //     position, non-inclusive), and push it on the
+	       //     stack. No arg.
+  RVM_EOF,     // [m] Succeeds only if at EOF.
+  RVM_MATCH,   // [m] The high byte of the parameter is an upper bound
+	       //     and the low byte is a lower bound, both
+	       //     inclusive. An inverted match should be handled
+	       //     as two ranges.
+  RVM_STEP,    // [a] Step to the next byte of input
+  RVM_OPCOUNT
+} HRVMOp;
+
+typedef struct HRVMInsn_{
+  uint8_t op;
+  uint16_t arg;
+} HRVMInsn;
+
+#define TT_MARK TT_RESERVED_1
+
+typedef struct HSVMContext_ {
+  HParsedToken **stack;
+  size_t stack_count; // number of items on the stack. Thus stack[stack_count] is the first unused item on the stack.
+  size_t stack_capacity;
+} HSVMContext;
+
+// These actions all assume that the items on the stack are not
+// aliased anywhere.
+typedef bool (*HSVMActionFunc)(HArena *arena, HSVMContext *ctx, void* env);
+typedef struct HSVMAction_ {
+  HSVMActionFunc action;
+  void* env;
+} HSVMAction;
+
+struct HRVMProg_ {
+  HAllocator *allocator;
+  size_t length;
+  size_t action_count;
+  HRVMInsn *insns;
+  HSVMAction *actions;
+};
+
+// Returns true IFF the provided parser could be compiled.
+bool h_compile_regex(HRVMProg *prog, const HParser* parser);
+
+// These functions are used by the compile_to_rvm method of HParser
+uint16_t h_rvm_create_action(HRVMProg *prog, HSVMActionFunc action_func, void* env);
+
+// returns the address of the instruction just created
+uint16_t h_rvm_insert_insn(HRVMProg *prog, HRVMOp op, uint16_t arg);
+
+// returns the address of the next insn to be created.
+uint16_t h_rvm_get_ip(HRVMProg *prog);
+
+// Used to insert forward references; the idea is to generate a JUMP
+// or FORK instruction with a target of 0, then update it once the
+// correct target is known.
+void h_rvm_patch_arg(HRVMProg *prog, uint16_t ip, uint16_t new_val);
+
+// Common SVM action funcs...
+bool h_svm_action_make_sequence(HArena *arena, HSVMContext *ctx, void* env);
+bool h_svm_action_clear_to_mark(HArena *arena, HSVMContext *ctx, void* env);
+
+extern HParserBackendVTable h__regex_backend_vtable;
+
+#endif

src/backends/regexvm_asm.pl

@@ -0,0 +1,112 @@
+#!/usr/bin/perl -w
+
+use strict;
+# The input file consists of a sequence of blocks, which can be parsed
+# as SVM test cases, RVM test cases, or C functions. Each block starts
+# with a header line, then a sequence of options, and finally text in
+# a format defined by the block type.
+#
+# Header lines start with "+TYPE", optionally followed by a name. This
+# name is semantically meaningful for SVM and RVM blocks; it
+# determines the name of the test case.
+
+# A C block's name is not used, and it takes no options. The body
+# (which continues until the first line that looks like a header), is
+# just passed straight through into the C source.
+
+# SVM blocks' names are the GLib test case name. The underlying
+# function's name is derived by substituting invalid characters with
+# '_'. Note that this can result in collisions (eg, /foo_bar/baz
+# collides with /foo/bar_baz). If this happens, it's your own damn
+# fault; rename the blocks. SVM blocks take three different options:
+# @input, @output, and @pre. The @input pragma's argument is a
+# C-quoted string that gets passed into the VM as the input string,
+# and @output is a C-quoted string that is compared against
+# h_write_result_unamb.  @pre lines are prepended verbatim to the
+# function body (with the @pre stripped, of course); they can be used
+# to initialize environment values.
+#
+# SVM instructions consist of either two or four fields:
+#
+#     input_pos opcode [arg env]
+#
+# input_pos and opcode correspond to the fields in HRVMTrace.  arg and
+# env are used to populate an HSVMAction; arg is the function, and env
+# is the object whose address should be used as the env.
+
+# RVM blocks are very similar to SVM blocks; the name and options are
+# handled exactly the same way. The assembly text is handled slightly
+# differently; the format is:
+#
+#     [label:] opcode [arg ...]
+#
+# For FORK and GOTO, the arg should be a label that is defined
+# elsewhere.
+#
+# For ACTION, the arguments are handled the same way as with SVM.
+#
+# MATCH takes two arguments, each of which can be any C integer
+# constant (not including character constants), which form the lower
+# and upper bounds of the matched character, respectively.
+#
+# No other RVM instructions take an argument.
+
+# At the beginning of any line, comments preceeded by '#' are allowed;
+# they are replaced by C++ comments and inserted in the nearest valid
+# location in the output.
+
+my $mode == "TOP";
+
+# common regexes:
+my $re_ident = qr/[A-Za-z_][A-Za-z0-9_]*/;
+my $re_cstr = qr/"(?:[^\\"]|\\["'abefnrtv0\\]|\\x[0-9a-fA-F]{2}|\\[0-7]{3})*"/;
+
+
+my %svm = (
+    name => sub {
+	my ($env, $name) = @_;
+	$env->{name} = $name;
+    },
+    pragma => sub {
+	my ($env, $name, $val) = @_;
+	if ($name eq "input") {
+	    chomp($env->{input} = $val);
+	} elsif ($name eq "output") {
+	    chomp($env->{output} = $val);
+	} elsif ($name eq "pre") {
+	    # Do I have the ref precedence right here?
+	    push(@$env->{pre}, $val);
+	} else {
+	    warn "Invalid SVM pragma";
+	}
+    },
+    body => sub {
+	my ($env, $line) = @_;
+	my ($ipos, $op, $arg, $argenv);
+	if ($line =~ /^\s*(\d+)\s+(PUSH|NOP|ACTION|CAPTURE|ACCEPT)(?:\s+($re_ident)\s+($re_ident))?/) {
+	    if ($2 eq "PUSH") {
+		# TODO: implement all the opcodes
+	    }
+	}
+    }
+    );
+
+
+while (<>) {
+    if (/^+(C|RVM|SVM)/) {
+	$mode = $1;
+    }
+    
+    if ($mode eq "TOP") {
+	if (/^#(.*)/) {
+	    print "// $1";
+	    next;
+	}
+    } elsif ($mode eq "SVM") {
+    } elsif ($mode eq "RVM") {
+    } elsif ($mode eq "C") {
+    }
+    
+}
+
+

src/benchmark.c

@@ -21,11 +21,11 @@
 
 */
 
-HBenchmarkResults *h_benchmark(const HParser* parser, HParserTestcase* testcases) {
+HBenchmarkResults *h_benchmark(HParser* parser, HParserTestcase* testcases) {
   return h_benchmark__m(&system_allocator, parser, testcases);
 }
 
-HBenchmarkResults *h_benchmark__m(HAllocator* mm__, const HParser* parser, HParserTestcase* testcases) {
+HBenchmarkResults *h_benchmark__m(HAllocator* mm__, HParser* parser, HParserTestcase* testcases) {
   // For now, just output the results to stderr
   HParserTestcase* tc = testcases;
   HParserBackend backend = PB_MIN;
@@ -33,7 +33,7 @@ HBenchmarkResults *h_benchmark__m(HAllocator* mm__, const HParser* parser, HPars
   ret->len = PB_MAX-PB_MIN;
   ret->results = h_new(HBackendResults, ret->len);
 
-  for (backend = PB_MIN; backend < PB_MAX; backend++) {
+  for (backend = PB_MIN; backend <= PB_MAX; backend++) {
     ret->results[backend].backend = backend;
     // Step 1: Compile grammar for given parser...
     if (h_compile(parser, backend, NULL) == -1) {

src/compile.c

@@ -1,17 +0,0 @@
-// This file contains functions related to managing multiple parse backends
-#include "hammer.h"
-#include "internal.h"
-
-static HParserBackendVTable *backends[PB_MAX] = {
-  &h__packrat_backend_vtable,
-  &h__ll_backend_vtable,
-};
-
-int h_compile(const HParser* parser, HParserBackend backend, const void* params) {
-  return h_compile__m(&system_allocator, parser, backend, params);
-}
-
-int h_compile__m(HAllocator* mm__, const HParser* parser, HParserBackend backend, const void* params) {
-  // be naughty and cast off the const
-  return backends[backend]->compile(mm__, (HParser *)parser, params);
-}

src/hammer.c

@@ -26,13 +26,12 @@
 #include "allocator.h"
 #include "parsers/parser_internal.h"
 
-static uint32_t djbhash(const uint8_t *buf, size_t len) {
+static HParserBackendVTable *backends[PB_MAX + 1] = {
-  uint32_t hash = 5381;
+  &h__packrat_backend_vtable,
-  while (len--) {
+  &h__regex_backend_vtable,
-    hash = hash * 33 + *buf++;
+  &h__llk_backend_vtable,
-  }
+};
-  return hash;
+
-}
 
 /* Helper function, since these lines appear in every parser */
 
@@ -42,46 +41,52 @@ typedef struct {
 } HTwoParsers;
 
 
-static uint32_t cache_key_hash(const void* key) {
-  return djbhash(key, sizeof(HParserCacheKey));
-}
-static bool cache_key_equal(const void* key1, const void* key2) {
-  return memcmp(key1, key2, sizeof(HParserCacheKey)) == 0;
-}
 
 
 HParseResult* h_parse(const HParser* parser, const uint8_t* input, size_t length) {
   return h_parse__m(&system_allocator, parser, input, length);
 }
 HParseResult* h_parse__m(HAllocator* mm__, const HParser* parser, const uint8_t* input, size_t length) {
+  // TODO: split the creation of the parse state into h_packrat_parse
   // Set up a parse state...
-  HArena * arena = h_new_arena(mm__, 0);
+  HInputStream input_stream = {
-  HParseState *parse_state = a_new_(arena, HParseState, 1);
+    .index = 0,
-  parse_state->cache = h_hashtable_new(arena, cache_key_equal, // key_equal_func
+    .bit_offset = 8,
-					      cache_key_hash); // hash_func
+    .overrun = 0,
-  parse_state->input_stream.input = input;
+    .endianness = BIT_BIG_ENDIAN | BYTE_BIG_ENDIAN,
-  parse_state->input_stream.index = 0;
+    .length = length,
-  parse_state->input_stream.bit_offset = 8; // bit big endian
+    .input = input
-  parse_state->input_stream.overrun = 0;
+  };
-  parse_state->input_stream.endianness = BIT_BIG_ENDIAN | BYTE_BIG_ENDIAN;
-  parse_state->input_stream.length = length;
-  parse_state->lr_stack = h_slist_new(arena);
-  parse_state->recursion_heads = h_hashtable_new(arena, cache_key_equal,
-						  cache_key_hash);
-  parse_state->arena = arena;
-  HParseResult *res = h_do_parse(parser, parse_state);
-  h_slist_free(parse_state->lr_stack);
-  h_hashtable_free(parse_state->recursion_heads);
-  // tear down the parse state
-  h_hashtable_free(parse_state->cache);
-  if (!res)
-    h_delete_arena(parse_state->arena);
   
-  return res;
+  return backends[parser->backend]->parse(mm__, parser, &input_stream);
 }
 
 void h_parse_result_free(HParseResult *result) {
   h_delete_arena(result->arena);
 }
 
+bool h_false(void* env) {
+  (void)env;
+  return false;
+}
 
+bool h_true(void* env) {
+  (void)env;
+  return true;
+}
+
+bool h_not_regular(HRVMProg *prog, void *env) {
+  (void)env;
+  return false;
+}
+
+int h_compile(HParser* parser, HParserBackend backend, const void* params) {
+  return h_compile__m(&system_allocator, parser, backend, params);
+}
+
+int h_compile__m(HAllocator* mm__, HParser* parser, HParserBackend backend, const void* params) {
+  int ret = backends[backend]->compile(mm__, parser, params);
+  if (!ret)
+    parser->backend = backend;
+  return ret;
+}

src/hammer.h

@@ -34,8 +34,11 @@ typedef struct HParseState_ HParseState;
 typedef enum HParserBackend_ {
   PB_MIN = 0,
   PB_PACKRAT = PB_MIN, // PB_MIN is always the default.
-  PB_LL,
+  PB_REGULAR,	// 
-  PB_MAX
+  PB_LLk,	//
+  PB_LALR,	// Not Implemented
+  PB_GLR,	// Not Implemented
+  PB_MAX = PB_LLk
 } HParserBackend;
 
 typedef enum HTokenType_ {
@@ -44,6 +47,7 @@ typedef enum HTokenType_ {
   TT_SINT,
   TT_UINT,
   TT_SEQUENCE,
+  TT_RESERVED_1, // reserved for backend-specific internal use
   TT_USER = 64,
   TT_ERR,
   TT_MAX
@@ -75,7 +79,9 @@ typedef struct HParsedToken_ {
 } HParsedToken;
 
 /**
- * The result of a successful parse.
+ * The result of a successful parse. Note that this may reference the
+ * input string.
+ *
  * If a parse fails, the parse result will be NULL.
  * If a parse is successful but there's nothing there (i.e., if end_p 
  * succeeds) then there's a parse result but its ast is NULL.
@@ -111,12 +117,14 @@ typedef const HParsedToken* (*HAction)(const HParseResult *p);
  */
 typedef bool (*HPredicate)(HParseResult *p);
 
-typedef struct HParserVtable_ HParserVtable;
-
 typedef struct HCFChoice_ HCFChoice;
+typedef struct HRVMProg_ HRVMProg;
+typedef struct HParserVtable_ HParserVtable;
 
 typedef struct HParser_ {
   const HParserVtable *vtable;
+  HParserBackend backend;
+  void* backend_data;
   void *env;
   void *data; /* e.g., parse tables */
   HCFChoice *desugared; /* if the parser can be desugared, its desugared form */
@@ -191,7 +199,7 @@ HAMMER_FN_DECL(HParseResult*, h_parse, const HParser* parser, const uint8_t* inp
  * 
  * Result token type: TT_BYTES
  */
-HAMMER_FN_DECL(const HParser*, h_token, const uint8_t *str, const size_t len);
+HAMMER_FN_DECL(HParser*, h_token, const uint8_t *str, const size_t len);
 
 /**
  * Given a single character, returns a parser that parses that 
@@ -199,7 +207,7 @@ HAMMER_FN_DECL(const HParser*, h_token, const uint8_t *str, const size_t len);
  * 
  * Result token type: TT_UINT
  */
-HAMMER_FN_DECL(const HParser*, h_ch, const uint8_t c);
+HAMMER_FN_DECL(HParser*, h_ch, const uint8_t c);
 
 /**
  * Given two single-character bounds, lower and upper, returns a parser
@@ -208,14 +216,14 @@ HAMMER_FN_DECL(const HParser*, h_ch, const uint8_t c);
  * 
  * Result token type: TT_UINT
  */
-HAMMER_FN_DECL(const HParser*, h_ch_range, const uint8_t lower, const uint8_t upper);
+HAMMER_FN_DECL(HParser*, h_ch_range, const uint8_t lower, const uint8_t upper);
 
 /**
  * Given an integer parser, p, and two integer bounds, lower and upper,
  * returns a parser that parses an integral value within the range 
  * [lower, upper] (inclusive).
  */
-HAMMER_FN_DECL(const HParser*, h_int_range, const HParser *p, const int64_t lower, const int64_t upper);
+HAMMER_FN_DECL(HParser*, h_int_range, const HParser *p, const int64_t lower, const int64_t upper);
 
 /**
  * Returns a parser that parses the specified number of bits. sign == 
@@ -223,63 +231,63 @@ HAMMER_FN_DECL(const HParser*, h_int_range, const HParser *p, const int64_t lowe
  *
  * Result token type: TT_SINT if sign == true, TT_UINT if sign == false
  */
-HAMMER_FN_DECL(const HParser*, h_bits, size_t len, bool sign);
+HAMMER_FN_DECL(HParser*, h_bits, size_t len, bool sign);
 
 /**
  * Returns a parser that parses a signed 8-byte integer value. 
  *
  * Result token type: TT_SINT
  */
-HAMMER_FN_DECL_NOARG(const HParser*, h_int64);
+HAMMER_FN_DECL_NOARG(HParser*, h_int64);
 
 /**
  * Returns a parser that parses a signed 4-byte integer value. 
  *
  * Result token type: TT_SINT
  */
-HAMMER_FN_DECL_NOARG(const HParser*, h_int32);
+HAMMER_FN_DECL_NOARG(HParser*, h_int32);
 
 /**
  * Returns a parser that parses a signed 2-byte integer value. 
  *
  * Result token type: TT_SINT
  */
-HAMMER_FN_DECL_NOARG(const HParser*, h_int16);
+HAMMER_FN_DECL_NOARG(HParser*, h_int16);
 
 /**
  * Returns a parser that parses a signed 1-byte integer value. 
  *
  * Result token type: TT_SINT
  */
-HAMMER_FN_DECL_NOARG(const HParser*, h_int8);
+HAMMER_FN_DECL_NOARG(HParser*, h_int8);
 
 /**
  * Returns a parser that parses an unsigned 8-byte integer value. 
  *
  * Result token type: TT_UINT
  */
-HAMMER_FN_DECL_NOARG(const HParser*, h_uint64);
+HAMMER_FN_DECL_NOARG(HParser*, h_uint64);
 
 /**
  * Returns a parser that parses an unsigned 4-byte integer value. 
  *
  * Result token type: TT_UINT
  */
-HAMMER_FN_DECL_NOARG(const HParser*, h_uint32);
+HAMMER_FN_DECL_NOARG(HParser*, h_uint32);
 
 /**
  * Returns a parser that parses an unsigned 2-byte integer value. 
  *
  * Result token type: TT_UINT
  */
-HAMMER_FN_DECL_NOARG(const HParser*, h_uint16);
+HAMMER_FN_DECL_NOARG(HParser*, h_uint16);
 
 /**
  * Returns a parser that parses an unsigned 1-byte integer value. 
  *
  * Result token type: TT_UINT
  */
-HAMMER_FN_DECL_NOARG(const HParser*, h_uint8);
+HAMMER_FN_DECL_NOARG(HParser*, h_uint8);
 
 /**
  * Given another parser, p, returns a parser that skips any whitespace 
@@ -287,7 +295,7 @@ HAMMER_FN_DECL_NOARG(const HParser*, h_uint8);
  *
  * Result token type: p's result type
  */
-HAMMER_FN_DECL(const HParser*, h_whitespace, const HParser* p);
+HAMMER_FN_DECL(HParser*, h_whitespace, const HParser* p);
 
 /**
  * Given two parsers, p and q, returns a parser that parses them in
@@ -295,7 +303,7 @@ HAMMER_FN_DECL(const HParser*, h_whitespace, const HParser* p);
  *
  * Result token type: p's result type
  */
-HAMMER_FN_DECL(const HParser*, h_left, const HParser* p, const HParser* q);
+HAMMER_FN_DECL(HParser*, h_left, const HParser* p, const HParser* q);
 
 /**
  * Given two parsers, p and q, returns a parser that parses them in
@@ -303,7 +311,7 @@ HAMMER_FN_DECL(const HParser*, h_left, const HParser* p, const HParser* q);
  *
  * Result token type: q's result type
  */
-HAMMER_FN_DECL(const HParser*, h_right, const HParser* p, const HParser* q);
+HAMMER_FN_DECL(HParser*, h_right, const HParser* p, const HParser* q);
 
 /**
  * Given three parsers, p, x, and q, returns a parser that parses them in
@@ -311,7 +319,7 @@ HAMMER_FN_DECL(const HParser*, h_right, const HParser* p, const HParser* q);
  *
  * Result token type: x's result type
  */
-HAMMER_FN_DECL(const HParser*, h_middle, const HParser* p, const HParser* x, const HParser* q);
+HAMMER_FN_DECL(HParser*, h_middle, const HParser* p, const HParser* x, const HParser* q);
 
 /**
  * Given another parser, p, and a function f, returns a parser that 
@@ -319,21 +327,21 @@ HAMMER_FN_DECL(const HParser*, h_middle, const HParser* p, const HParser* x, con
  *
  * Result token type: any
  */
-HAMMER_FN_DECL(const HParser*, h_action, const HParser* p, const HAction a);
+HAMMER_FN_DECL(HParser*, h_action, const HParser* p, const HAction a);
 
 /**
  * Parse a single character in the given charset. 
  *
  * Result token type: TT_UINT
  */
-HAMMER_FN_DECL(const HParser*, h_in, const uint8_t *charset, size_t length);
+HAMMER_FN_DECL(HParser*, h_in, const uint8_t *charset, size_t length);
 
 /**
  * Parse a single character *NOT* in the given charset. 
  *
  * Result token type: TT_UINT
  */
-HAMMER_FN_DECL(const HParser*, h_not_in, const uint8_t *charset, size_t length);
+HAMMER_FN_DECL(HParser*, h_not_in, const uint8_t *charset, size_t length);
 
 /**
  * A no-argument parser that succeeds if there is no more input to 
@@ -341,14 +349,14 @@ HAMMER_FN_DECL(const HParser*, h_not_in, const uint8_t *charset, size_t length);
  *
  * Result token type: None. The HParseResult exists but its AST is NULL.
  */
-HAMMER_FN_DECL_NOARG(const HParser*, h_end_p);
+HAMMER_FN_DECL_NOARG(HParser*, h_end_p);
 
 /**
  * This parser always fails. 
  *
  * Result token type: NULL. Always.
  */
-HAMMER_FN_DECL_NOARG(const HParser*, h_nothing_p);
+HAMMER_FN_DECL_NOARG(HParser*, h_nothing_p);
 
 /**
  * Given a null-terminated list of parsers, apply each parser in order.
@@ -356,7 +364,7 @@ HAMMER_FN_DECL_NOARG(const HParser*, h_nothing_p);
  *
  * Result token type: TT_SEQUENCE
  */
-HAMMER_FN_DECL_VARARGS_ATTR(__attribute__((sentinel)), const HParser*, h_sequence, const HParser* p);
+HAMMER_FN_DECL_VARARGS_ATTR(__attribute__((sentinel)), HParser*, h_sequence, const HParser* p);
 
 /**
  * Given an array of parsers, p_array, apply each parser in order. The 
@@ -365,7 +373,7 @@ HAMMER_FN_DECL_VARARGS_ATTR(__attribute__((sentinel)), const HParser*, h_sequenc
  *
  * Result token type: The type of the first successful parser's result.
  */
-HAMMER_FN_DECL_VARARGS_ATTR(__attribute__((sentinel)), const HParser*, h_choice, const HParser* p);
+HAMMER_FN_DECL_VARARGS_ATTR(__attribute__((sentinel)), HParser*, h_choice, const HParser* p);
 
 /**
  * Given two parsers, p1 and p2, this parser succeeds in the following 
@@ -375,7 +383,7 @@ HAMMER_FN_DECL_VARARGS_ATTR(__attribute__((sentinel)), const HParser*, h_choice,
  *
  * Result token type: p1's result type.
  */
-HAMMER_FN_DECL(const HParser*, h_butnot, const HParser* p1, const HParser* p2);
+HAMMER_FN_DECL(HParser*, h_butnot, const HParser* p1, const HParser* p2);
 
 /**
  * Given two parsers, p1 and p2, this parser succeeds in the following 
@@ -385,7 +393,7 @@ HAMMER_FN_DECL(const HParser*, h_butnot, const HParser* p1, const HParser* p2);
  *
  * Result token type: p1's result type.
  */
-HAMMER_FN_DECL(const HParser*, h_difference, const HParser* p1, const HParser* p2);
+HAMMER_FN_DECL(HParser*, h_difference, const HParser* p1, const HParser* p2);
 
 /**
  * Given two parsers, p1 and p2, this parser succeeds if *either* p1 or
@@ -393,7 +401,7 @@ HAMMER_FN_DECL(const HParser*, h_difference, const HParser* p1, const HParser* p
  *
  * Result token type: The type of the result of whichever parser succeeded.
  */
-HAMMER_FN_DECL(const HParser*, h_xor, const HParser* p1, const HParser* p2);
+HAMMER_FN_DECL(HParser*, h_xor, const HParser* p1, const HParser* p2);
 
 /**
  * Given a parser, p, this parser succeeds for zero or more repetitions
@@ -401,7 +409,7 @@ HAMMER_FN_DECL(const HParser*, h_xor, const HParser* p1, const HParser* p2);
  *
  * Result token type: TT_SEQUENCE
  */
-HAMMER_FN_DECL(const HParser*, h_many, const HParser* p);
+HAMMER_FN_DECL(HParser*, h_many, const HParser* p);
 
 /**
  * Given a parser, p, this parser succeeds for one or more repetitions 
@@ -409,7 +417,7 @@ HAMMER_FN_DECL(const HParser*, h_many, const HParser* p);
  *
  * Result token type: TT_SEQUENCE
  */
-HAMMER_FN_DECL(const HParser*, h_many1, const HParser* p);
+HAMMER_FN_DECL(HParser*, h_many1, const HParser* p);
 
 /**
  * Given a parser, p, this parser succeeds for exactly N repetitions 
@@ -417,7 +425,7 @@ HAMMER_FN_DECL(const HParser*, h_many1, const HParser* p);
  *
  * Result token type: TT_SEQUENCE
  */
-HAMMER_FN_DECL(const HParser*, h_repeat_n, const HParser* p, const size_t n);
+HAMMER_FN_DECL(HParser*, h_repeat_n, const HParser* p, const size_t n);
 
 /**
  * Given a parser, p, this parser succeeds with the value p parsed or 
@@ -425,7 +433,7 @@ HAMMER_FN_DECL(const HParser*, h_repeat_n, const HParser* p, const size_t n);
  *
  * Result token type: If p succeeded, the type of its result; if not, TT_NONE.
  */
-HAMMER_FN_DECL(const HParser*, h_optional, const HParser* p);
+HAMMER_FN_DECL(HParser*, h_optional, const HParser* p);
 
 /**
  * Given a parser, p, this parser succeeds if p succeeds, but doesn't 
@@ -433,7 +441,7 @@ HAMMER_FN_DECL(const HParser*, h_optional, const HParser* p);
  *
  * Result token type: None. The HParseResult exists but its AST is NULL.
  */
-HAMMER_FN_DECL(const HParser*, h_ignore, const HParser* p);
+HAMMER_FN_DECL(HParser*, h_ignore, const HParser* p);
 
 /**
  * Given a parser, p, and a parser for a separator, sep, this parser 
@@ -444,7 +452,7 @@ HAMMER_FN_DECL(const HParser*, h_ignore, const HParser* p);
  *
  * Result token type: TT_SEQUENCE
  */
-HAMMER_FN_DECL(const HParser*, h_sepBy, const HParser* p, const HParser* sep);
+HAMMER_FN_DECL(HParser*, h_sepBy, const HParser* p, const HParser* sep);
 
 /**
  * Given a parser, p, and a parser for a separator, sep, this parser matches a list of things that p can parse, separated by sep. Unlike sepBy, this ensures that the result has at least one element.
@@ -452,14 +460,14 @@ HAMMER_FN_DECL(const HParser*, h_sepBy, const HParser* p, const HParser* sep);
  *
  * Result token type: TT_SEQUENCE
  */
-HAMMER_FN_DECL(const HParser*, h_sepBy1, const HParser* p, const HParser* sep);
+HAMMER_FN_DECL(HParser*, h_sepBy1, const HParser* p, const HParser* sep);
 
 /**
  * This parser always returns a zero length match, i.e., empty string. 
  *
  * Result token type: None. The HParseResult exists but its AST is NULL.
  */
-HAMMER_FN_DECL_NOARG(const HParser*, h_epsilon_p);
+HAMMER_FN_DECL_NOARG(HParser*, h_epsilon_p);
 
 /**
  * This parser applies its first argument to read an unsigned integer
@@ -470,7 +478,7 @@ HAMMER_FN_DECL_NOARG(const HParser*, h_epsilon_p);
  *
  * Result token type: TT_SEQUENCE
  */
-HAMMER_FN_DECL(const HParser*, h_length_value, const HParser* length, const HParser* value);
+HAMMER_FN_DECL(HParser*, h_length_value, const HParser* length, const HParser* value);
 
 /**
  * This parser attaches a predicate function, which returns true or 
@@ -485,7 +493,7 @@ HAMMER_FN_DECL(const HParser*, h_length_value, const HParser* length, const HPar
  * 
  * Result token type: p's result type if pred succeeded, NULL otherwise.
  */
-HAMMER_FN_DECL(const HParser*, h_attr_bool, const HParser* p, HPredicate pred);
+HAMMER_FN_DECL(HParser*, h_attr_bool, const HParser* p, HPredicate pred);
 
 /**
  * The 'and' parser asserts that a conditional syntax is satisfied, 
@@ -502,7 +510,7 @@ HAMMER_FN_DECL(const HParser*, h_attr_bool, const HParser* p, HPredicate pred);
  *
  * Result token type: None. The HParseResult exists but its AST is NULL.
  */
-HAMMER_FN_DECL(const HParser*, h_and, const HParser* p);
+HAMMER_FN_DECL(HParser*, h_and, const HParser* p);
 
 /**
  * The 'not' parser asserts that a conditional syntax is *not* 
@@ -522,7 +530,7 @@ HAMMER_FN_DECL(const HParser*, h_and, const HParser* p);
  * 
  * Result token type: None. The HParseResult exists but its AST is NULL.
  */
-HAMMER_FN_DECL(const HParser*, h_not, const HParser* p);
+HAMMER_FN_DECL(HParser*, h_not, const HParser* p);
 
 /**
  * Create a parser that just calls out to another, as yet unknown, 
@@ -565,7 +573,7 @@ HAMMER_FN_DECL(void, h_pprint, FILE* stream, const HParsedToken* tok, int indent
  *
  * Returns -1 if grammar cannot be compiled with the specified options; 0 otherwise.
  */
-HAMMER_FN_DECL(int, h_compile, const HParser* parser, HParserBackend backend, const void* params);
+HAMMER_FN_DECL(int, h_compile, HParser* parser, HParserBackend backend, const void* params);
 
 /**
  * TODO: Document me
@@ -590,7 +598,7 @@ const uint8_t* h_bit_writer_get_buffer(HBitWriter* w, size_t *len);
 void h_bit_writer_free(HBitWriter* w);
 
 // {{{ Benchmark functions
-HAMMER_FN_DECL(HBenchmarkResults *, h_benchmark, const HParser* parser, HParserTestcase* testcases);
+HAMMER_FN_DECL(HBenchmarkResults *, h_benchmark, HParser* parser, HParserTestcase* testcases);
 void h_benchmark_report(FILE* stream, HBenchmarkResults* results);
 void h_benchmark_dump_optimized_code(FILE* stream, HBenchmarkResults* results);
 // }}}

src/internal.h

@@ -48,7 +48,7 @@ static inline void h_generic_free(HAllocator *allocator, void* ptr) {
   allocator->free(allocator, ptr);
 }
 
-HAllocator system_allocator;
+extern HAllocator system_allocator;
 
 
 typedef struct HInputStream_ {
@@ -131,7 +131,8 @@ struct HParseState_ {
 
 typedef struct HParserBackendVTable_ {
   int (*compile)(HAllocator *mm__, HParser* parser, const void* params);
-  HParseResult* (*parse)(HAllocator *mm__, const HParser* parser, HParseState* parse_state);
+  HParseResult* (*parse)(HAllocator *mm__, const HParser* parser, HInputStream* parse_state);
+  void (*free)(HParser* parser);
 } HParserBackendVTable;
 
 
@@ -213,9 +214,10 @@ struct HBitWriter_ {
 
 // }}}
 
+
 // Backends {{{
 extern HParserBackendVTable h__packrat_backend_vtable;
-extern HParserBackendVTable h__ll_backend_vtable;
+extern HParserBackendVTable h__llk_backend_vtable;
 // }}}
 
 // TODO(thequux): Set symbol visibility for these functions so that they aren't exported.
@@ -295,9 +297,14 @@ struct HParserVtable_ {
   HParseResult* (*parse)(void *env, HParseState *state);
   bool (*isValidRegular)(void *env);
   bool (*isValidCF)(void *env);
+  bool (*compile_to_rvm)(HRVMProg *prog, void* env); // FIXME: forgot what the bool return value was supposed to mean.
   HCFChoice* (*desugar)(HAllocator *mm__, void *env);
 };
 
+bool h_false(void*);
+bool h_true(void*);
+bool h_not_regular(HRVMProg*, void*);
+
 #if 0
 #include <stdlib.h>
 #define h_arena_malloc(a, s) malloc(s)

src/parsers/action.c

@@ -12,7 +12,7 @@ static HParseResult* parse_action(void *env, HParseState *state) {
     //HParsedToken *tok = a->action(h_do_parse(a->p, state));
     if(tmp) {
         const HParsedToken *tok = a->action(tmp);
-        return make_result(state, (HParsedToken*)tok);
+        return make_result(state->arena, (HParsedToken*)tok);
     } else
       return NULL;
   } else // either the parser's missing or the action's missing
@@ -44,18 +44,24 @@ static bool action_isValidCF(void *env) {
   return a->p->vtable->isValidCF(a->p->env);
 }
 
+static bool action_ctrvm(HRVMProg *prog, void* env) {
+  HParseAction *a = (HParseAction*)env;
+  return a->p->vtable->compile_to_rvm(prog, a->p->env);
+}
+
 static const HParserVtable action_vt = {
   .parse = parse_action,
   .isValidRegular = action_isValidRegular,
   .isValidCF = action_isValidCF,
   .desugar = desugar_action,
+  .compile_to_rvm = action_ctrvm,
 };
 
-const HParser* h_action(const HParser* p, const HAction a) {
+HParser* h_action(const HParser* p, const HAction a) {
   return h_action__m(&system_allocator, p, a);
 }
 
-const HParser* h_action__m(HAllocator* mm__, const HParser* p, const HAction a) {
+HParser* h_action__m(HAllocator* mm__, const HParser* p, const HAction a) {
   HParseAction *env = h_new(HParseAction, 1);
   env->p = p;
   env->action = a;

src/parsers/and.c

@@ -5,7 +5,7 @@ static HParseResult *parse_and(void* env, HParseState* state) {
   HParseResult *res = h_do_parse((HParser*)env, state);
   state->input_stream = bak;
   if (res)
-    return make_result(state, NULL);
+    return make_result(state->arena, NULL);
   return NULL;
 }
 
@@ -22,13 +22,14 @@ static const HParserVtable and_vt = {
 				revision. --mlp, 18/12/12 */
   .isValidCF = h_false,      /* despite TODO above, this remains false. */
   .desugar = desugar_and,
+  .compile_to_rvm = h_not_regular,
 };
 
 
-const HParser* h_and(const HParser* p) {
+HParser* h_and(const HParser* p) {
   return h_and__m(&system_allocator, p);
 }
-const HParser* h_and__m(HAllocator* mm__, const HParser* p) {
+HParser* h_and__m(HAllocator* mm__, const HParser* p) {
   // zero-width postive lookahead
   return h_new_parser(mm__, &and_vt, (void *)p);
 }

src/parsers/attr_bool.c

@@ -47,18 +47,24 @@ static HCFChoice* desugar_ab(HAllocator *mm__, void *env) {
   return ret;
 }
 
+static bool ab_ctrvm(HRVMProg *prog, void *env) {
+  HAttrBool *ab = (HAttrBool*)env;
+  return h_compile_regex(prog, ab->p);
+}
+
 static const HParserVtable attr_bool_vt = {
   .parse = parse_attr_bool,
   .isValidRegular = ab_isValidRegular,
   .isValidCF = ab_isValidCF,
   .desugar = desugar_ab,
+  .compile_to_rvm = ab_ctrvm,
 };
 
 
-const HParser* h_attr_bool(const HParser* p, HPredicate pred) {
+HParser* h_attr_bool(const HParser* p, HPredicate pred) {
   return h_attr_bool__m(&system_allocator, p, pred);
 }
-const HParser* h_attr_bool__m(HAllocator* mm__, const HParser* p, HPredicate pred) { 
+HParser* h_attr_bool__m(HAllocator* mm__, const HParser* p, HPredicate pred) { 
   HAttrBool *env = h_new(HAttrBool, 1);
   env->p = p;
   env->pred = pred;

src/parsers/bits.c

@@ -1,3 +1,4 @@
+#include <assert.h>
 #include "parser_internal.h"
 
 struct bits_env {
@@ -13,7 +14,7 @@ static HParseResult* parse_bits(void* env, HParseState *state) {
     result->sint = h_read_bits(&state->input_stream, env_->length, true);
   else
     result->uint = h_read_bits(&state->input_stream, env_->length, false);
-  return make_result(state, result);
+  return make_result(state->arena, result);
 }
 
 static HCFChoice* desugar_bits(HAllocator *mm__, void *env) {
@@ -41,16 +42,43 @@ static HCFChoice* desugar_bits(HAllocator *mm__, void *env) {
   return ret;
 }
 
+static bool h_svm_action_bits(HArena *arena, HSVMContext *ctx, void* env) {
+  // BUG: relies un undefined behaviour: int64_t is a signed uint64_t; not necessarily true on 32-bit
+  struct bits_env *env_ = env;
+  HParsedToken *top = ctx->stack[ctx->stack_count-1];
+  assert(top->token_type == TT_BYTES);
+  uint64_t res = 0;
+  for (size_t i = 0; i < top->bytes.len; i++)
+    res = (res << 8) | top->bytes.token[i];   // TODO: Handle other endiannesses.
+  top->uint = res; // possibly cast to signed through union
+  top->token_type = (env_->signedp ? TT_SINT : TT_UINT);
+  return true;
+}
+
+static bool bits_ctrvm(HRVMProg *prog, void* env) {
+  struct bits_env *env_ = (struct bits_env*)env;
+  h_rvm_insert_insn(prog, RVM_PUSH, 0);
+  for (size_t i=0; (i < env_->length)/8; ++i) { // FUTURE: when we can handle non-byte-aligned, the env_->length/8 part will be different
+    h_rvm_insert_insn(prog, RVM_MATCH, 0xFF00);
+    h_rvm_insert_insn(prog, RVM_STEP, 0);
+  }
+  h_rvm_insert_insn(prog, RVM_CAPTURE, 0);
+  h_rvm_insert_insn(prog, RVM_ACTION, h_rvm_create_action(prog, h_svm_action_bits, env));
+  return true;
+}
+
 static const HParserVtable bits_vt = {
   .parse = parse_bits,
   .isValidRegular = h_true,
   .isValidCF = h_true,
   .desugar = desugar_bits,
+  .compile_to_rvm = bits_ctrvm,
 };
-const HParser* h_bits(size_t len, bool sign) {
+
+HParser* h_bits(size_t len, bool sign) {
   return h_bits__m(&system_allocator, len, sign);
 }
-const HParser* h_bits__m(HAllocator* mm__, size_t len, bool sign) {
+HParser* h_bits__m(HAllocator* mm__, size_t len, bool sign) {
   struct bits_env *env = h_new(struct bits_env, 1);
   env->length = len;
   env->signedp = sign;
@@ -58,10 +86,10 @@ const HParser* h_bits__m(HAllocator* mm__, size_t len, bool sign) {
 }
 
 #define SIZED_BITS(name_pre, len, signedp) \
-  const HParser* h_##name_pre##len () {				\
+  HParser* h_##name_pre##len () {				\
     return h_bits__m(&system_allocator, len, signedp);		\
   }								\
-  const HParser* h_##name_pre##len##__m(HAllocator* mm__) {	\
+  HParser* h_##name_pre##len##__m(HAllocator* mm__) {	\
     return h_bits__m(mm__, len, signedp);			\
   }
 SIZED_BITS(int, 8, true)

src/parsers/butnot.c

@@ -43,14 +43,15 @@ static HCFChoice* desugar_butnot(HAllocator *mm__, void *env) {
 static const HParserVtable butnot_vt = {
   .parse = parse_butnot,
   .isValidRegular = h_false,
-  .isValidCF = h_false,
+  .isValidCF = h_false, // XXX should this be true if both p1 and p2 are CF?
   .desugar = desugar_butnot,
+  .compile_to_rvm = h_not_regular,
 };
 
-const HParser* h_butnot(const HParser* p1, const HParser* p2) {
+HParser* h_butnot(const HParser* p1, const HParser* p2) {
   return h_butnot__m(&system_allocator, p1, p2);
 }
-const HParser* h_butnot__m(HAllocator* mm__, const HParser* p1, const HParser* p2) {
+HParser* h_butnot__m(HAllocator* mm__, const HParser* p1, const HParser* p2) {
   HTwoParsers *env = h_new(HTwoParsers, 1);
   env->p1 = p1;
   env->p2 = p2;

src/parsers/ch.c

@@ -6,7 +6,7 @@ static HParseResult* parse_ch(void* env, HParseState *state) {
   if (c == r) {
     HParsedToken *tok = a_new(HParsedToken, 1);    
     tok->token_type = TT_UINT; tok->uint = r;
-    return make_result(state, tok);
+    return make_result(state->arena, tok);
   } else {
     return NULL;
   }
@@ -20,16 +20,25 @@ static HCFChoice* desugar_ch(HAllocator *mm__, void *env) {
   return ret;
 }
 
+static bool ch_ctrvm(HRVMProg *prog, void* env) {
+  uint8_t c = (uint8_t)(unsigned long)(env);
+  // TODO: Does this capture anything?
+  h_rvm_insert_insn(prog, RVM_MATCH, c & c << 8);
+  h_rvm_insert_insn(prog, RVM_STEP, 0);
+  return true;
+}
+
 static const HParserVtable ch_vt = {
   .parse = parse_ch,
   .isValidRegular = h_true,
   .isValidCF = h_true,
   .desugar = desugar_ch,
+  .compile_to_rvm = ch_ctrvm,
 };
 
-const HParser* h_ch(const uint8_t c) {
+HParser* h_ch(const uint8_t c) {
   return h_ch__m(&system_allocator, c);
 }
-const HParser* h_ch__m(HAllocator* mm__, const uint8_t c) {  
+HParser* h_ch__m(HAllocator* mm__, const uint8_t c) {  
   return h_new_parser(mm__, &ch_vt, (void *)(uintptr_t)c);
 }

src/parsers/charset.c

@@ -9,7 +9,7 @@ static HParseResult* parse_charset(void *env, HParseState *state) {
   if (charset_isset(cs, in)) {
     HParsedToken *tok = a_new(HParsedToken, 1);
     tok->token_type = TT_UINT; tok->uint = in;
-    return make_result(state, tok);    
+    return make_result(state->arena, tok);    
   } else
     return NULL;
 }
@@ -22,17 +22,38 @@ static HCFChoice* desugar_charset(HAllocator *mm__, void *env) {
   return ret;
 }
 
+// FUTURE: this is horribly inefficient
+static bool cs_ctrvm(HRVMProg *prog, void *env) {
+  HCharset cs = (HCharset)env;
+  uint16_t start = h_rvm_get_ip(prog);
+  for (size_t i=0; i<256; ++i) {
+    if (charset_isset(cs, i)) {
+      uint16_t insn = h_rvm_insert_insn(prog, RVM_FORK, 0);
+      h_rvm_insert_insn(prog, RVM_MATCH, i & i << 8);
+      h_rvm_insert_insn(prog, RVM_GOTO, 0);
+      h_rvm_patch_arg(prog, insn, h_rvm_get_ip(prog));
+    }
+  }
+  uint16_t jump = h_rvm_insert_insn(prog, RVM_STEP, 0);
+  for (size_t i=start; i<jump; ++i) {
+    if (RVM_GOTO == prog->insns[i].op)
+      h_rvm_patch_arg(prog, i, jump);
+  }
+  return true;
+}
+
 static const HParserVtable charset_vt = {
   .parse = parse_charset,
   .isValidRegular = h_true,
   .isValidCF = h_true,
   .desugar = desugar_charset,
+  .compile_to_rvm = cs_ctrvm,
 };
 
-const HParser* h_ch_range(const uint8_t lower, const uint8_t upper) {
+HParser* h_ch_range(const uint8_t lower, const uint8_t upper) {
   return h_ch_range__m(&system_allocator, lower, upper);
 }
-const HParser* h_ch_range__m(HAllocator* mm__, const uint8_t lower, const uint8_t upper) {
+HParser* h_ch_range__m(HAllocator* mm__, const uint8_t lower, const uint8_t upper) {
   HCharset cs = new_charset(mm__);
   for (int i = 0; i < 256; i++)
     charset_set(cs, i, (lower <= i) && (i <= upper));
@@ -40,7 +61,7 @@ const HParser* h_ch_range__m(HAllocator* mm__, const uint8_t lower, const uint8_
 }
 
 
-static const HParser* h_in_or_not__m(HAllocator* mm__, const uint8_t *options, size_t count, int val) {
+static HParser* h_in_or_not__m(HAllocator* mm__, const uint8_t *options, size_t count, int val) {
   HCharset cs = new_charset(mm__);
   for (size_t i = 0; i < 256; i++)
     charset_set(cs, i, 1-val);
@@ -50,19 +71,19 @@ static const HParser* h_in_or_not__m(HAllocator* mm__, const uint8_t *options, s
   return h_new_parser(mm__, &charset_vt, cs);
 }
 
-const HParser* h_in(const uint8_t *options, size_t count) {
+HParser* h_in(const uint8_t *options, size_t count) {
   return h_in_or_not__m(&system_allocator, options, count, 1);
 }
 
-const HParser* h_in__m(HAllocator* mm__, const uint8_t *options, size_t count) {
+HParser* h_in__m(HAllocator* mm__, const uint8_t *options, size_t count) {
   return h_in_or_not__m(mm__, options, count, 1);
 }
 
-const HParser* h_not_in(const uint8_t *options, size_t count) {
+HParser* h_not_in(const uint8_t *options, size_t count) {
   return h_in_or_not__m(&system_allocator, options, count, 0);
 }
 
-const HParser* h_not_in__m(HAllocator* mm__, const uint8_t *options, size_t count) {
+HParser* h_not_in__m(HAllocator* mm__, const uint8_t *options, size_t count) {
   return h_in_or_not__m(mm__, options, count, 0);
 }
 

src/parsers/choice.c

@@ -54,34 +54,53 @@ static HCFChoice* desugar_choice(HAllocator *mm__, void *env) {
   return ret;
 }
 
+static bool choice_ctrvm(HRVMProg *prog, void* env) {
+  HSequence *s = (HSequence*)env;
+  uint16_t gotos[s->len];
+  uint16_t start = h_rvm_get_ip(prog);
+  for (size_t i=0; i<s->len; ++i) {
+    uint16_t insn = h_rvm_insert_insn(prog, RVM_FORK, 0);
+    if (!h_compile_regex(prog, s->p_array[i]->env))
+      return false;
+    gotos[i] = h_rvm_insert_insn(prog, RVM_GOTO, 0);
+    h_rvm_patch_arg(prog, insn, h_rvm_get_ip(prog));
+  }
+  uint16_t jump = h_rvm_insert_insn(prog, RVM_STEP, 0);
+  for (size_t i=start; i<s->len; ++i) {
+      h_rvm_patch_arg(prog, gotos[i], jump);
+  }
+  return true;
+}
+
 static const HParserVtable choice_vt = {
   .parse = parse_choice,
   .isValidRegular = choice_isValidRegular,
   .isValidCF = choice_isValidCF,
   .desugar = desugar_choice,
+  .compile_to_rvm = choice_ctrvm,
 };
 
-const HParser* h_choice(const HParser* p, ...) {
+HParser* h_choice(const HParser* p, ...) {
   va_list ap;
   va_start(ap, p);
-  const HParser* ret = h_choice__mv(&system_allocator, p,  ap);
+  HParser* ret = h_choice__mv(&system_allocator, p,  ap);
   va_end(ap);
   return ret;
 }
 
-const HParser* h_choice__m(HAllocator* mm__, const HParser* p, ...) {
+HParser* h_choice__m(HAllocator* mm__, const HParser* p, ...) {
   va_list ap;
   va_start(ap, p);
-  const HParser* ret = h_choice__mv(mm__, p,  ap);
+  HParser* ret = h_choice__mv(mm__, p,  ap);
   va_end(ap);
   return ret;
 }
 
-const HParser* h_choice__v(const HParser* p, va_list ap) {
+HParser* h_choice__v(const HParser* p, va_list ap) {
   return h_choice__mv(&system_allocator, p, ap);
 }
 
-const HParser* h_choice__mv(HAllocator* mm__, const HParser* p, va_list ap_) {
+HParser* h_choice__mv(HAllocator* mm__, const HParser* p, va_list ap_) {
   va_list ap;
   size_t len = 0;
   HSequence *s = h_new(HSequence, 1);

src/parsers/difference.c

@@ -42,14 +42,15 @@ static HCFChoice* desugar_difference(HAllocator *mm__, void *env) {
 static HParserVtable difference_vt = {
   .parse = parse_difference,
   .isValidRegular = h_false,
-  .isValidCF = h_false,
+  .isValidCF = h_false, // XXX should this be true if both p1 and p2 are CF?
   .desugar = desugar_difference,
+  .compile_to_rvm = h_not_regular,
 };
 
-const HParser* h_difference(const HParser* p1, const HParser* p2) {
+HParser* h_difference(const HParser* p1, const HParser* p2) {
   return h_difference__m(&system_allocator, p1, p2);
 }
-const HParser* h_difference__m(HAllocator* mm__, const HParser* p1, const HParser* p2) { 
+HParser* h_difference__m(HAllocator* mm__, const HParser* p1, const HParser* p2) { 
   HTwoParsers *env = h_new(HTwoParsers, 1);
   env->p1 = p1;
   env->p2 = p2;

src/parsers/end.c

@@ -17,17 +17,23 @@ static HCFChoice* desugar_end(HAllocator *mm__, void *env) {
   return &ret;
 }
 
+static bool end_ctrvm(HRVMProg *prog, void *env) {
+  h_rvm_insert_insn(prog, RVM_EOF, 0);
+  return true;
+}
+
 static const HParserVtable end_vt = {
   .parse = parse_end,
   .isValidRegular = h_true,
   .isValidCF = h_true,
   .desugar = desugar_end,
+  .compile_to_rvm = end_ctrvm,
 };
 
-const HParser* h_end_p() {
+HParser* h_end_p() {
   return h_end_p__m(&system_allocator);
 }
 
-const HParser* h_end_p__m(HAllocator* mm__) { 
+HParser* h_end_p__m(HAllocator* mm__) { 
   return h_new_parser(mm__, &end_vt, NULL);
 }

src/parsers/epsilon.c

@@ -8,21 +8,23 @@ static HParseResult* parse_epsilon(void* env, HParseState* state) {
   return res;
 }
 
+static bool epsilon_ctrvm(HRVMProg *prog, void* env) {
+  return true;
+}
+
 static const HParserVtable epsilon_vt = {
   .parse = parse_epsilon,
   .isValidRegular = h_true,
   .isValidCF = h_true,
   .desugar = desugar_epsilon,
+  .compile_to_rvm = epsilon_ctrvm,
 };
 
-static HParser epsilon_p = {
+HParser* h_epsilon_p() {
-  .vtable = &epsilon_vt,
+  return h_epsilon_p__m(&system_allocator);
-  .env = NULL
-};
-
-const HParser* h_epsilon_p() {
-  return &epsilon_p;
 }
-const HParser* h_epsilon_p__m(HAllocator* mm__) {
+HParser* h_epsilon_p__m(HAllocator* mm__) {
-  return &epsilon_p;
+  HParser *epsilon_p = h_new(HParser, 1);
+  epsilon_p->vtable = &epsilon_vt;
+  return epsilon_p;
 }

src/parsers/ignore.c

@@ -1,3 +1,4 @@
+#include <assert.h>
 #include "parser_internal.h"
 
 static HParseResult* parse_ignore(void* env, HParseState* state) {
@@ -25,16 +26,30 @@ static HCFChoice* desugar_ignore(HAllocator *mm__, void *env) {
   return (h_desugar(mm__, p));
 }
 
+static bool h_svm_action_pop(HArena *arena, HSVMContext *ctx, void* arg) {
+  assert(ctx->stack_count > 0);
+  ctx->stack_count--;
+  return true;
+}
+
+static bool ignore_ctrvm(HRVMProg *prog, void *env) {
+  HParser *p = (HParser*)env;
+  h_compile_regex(prog, p->env);
+  h_rvm_insert_insn(prog, RVM_ACTION, h_rvm_create_action(prog, h_svm_action_pop, NULL));
+  return true;
+}
+
 static const HParserVtable ignore_vt = {
   .parse = parse_ignore,
   .isValidRegular = ignore_isValidRegular,
   .isValidCF = ignore_isValidCF,
   .desugar = desugar_ignore,
+  .compile_to_rvm = ignore_ctrvm,
 };
 
-const HParser* h_ignore(const HParser* p) {
+HParser* h_ignore(const HParser* p) {
   return h_ignore__m(&system_allocator, p);
 }
-const HParser* h_ignore__m(HAllocator* mm__, const HParser* p) {
+HParser* h_ignore__m(HAllocator* mm__, const HParser* p) {
   return h_new_parser(mm__, &ignore_vt, (void *)p);
 }

src/parsers/ignoreseq.c

@@ -1,3 +1,4 @@
+#include <assert.h>
 #include "parser_internal.h"
 
 
@@ -5,7 +6,7 @@
 // general case: parse sequence, pick one result
 //
 
-typedef struct {
+typedef struct HIgnoreSeq_ {
   const HParser **parsers;
   size_t len;         // how many parsers in 'ps'
   size_t which;         // whose result to return
@@ -61,11 +62,40 @@ static bool is_isValidCF(void *env) {
   return true;
 }
 
+static bool h_svm_action_ignoreseq(HArena *arena, HSVMContext *ctx, void* env) {
+  HIgnoreSeq *seq = (HIgnoreSeq*)env;
+  HParsedToken* save;
+  // We can assume that each subitem generated at most one item on the
+  // stack.
+  assert(seq->len >= 1);
+  for (int i = seq->len - 1; i>=0; i--) {
+    if (i == (int)seq->which && ctx->stack[ctx->stack_count]->token_type != TT_MARK) 
+      save = ctx->stack[ctx->stack_count-1];
+    // skip over everything up to and including the mark.
+    while (ctx->stack[--ctx->stack_count]->token_type != TT_MARK)
+      ;
+  }
+  ctx->stack[ctx->stack_count++] = save;
+  return true;
+}
+
+static bool is_ctrvm(HRVMProg *prog, void* env) {
+  HIgnoreSeq *seq = (HIgnoreSeq*)env;
+  for (size_t i=0; i<seq->len; ++i) {
+    h_rvm_insert_insn(prog, RVM_PUSH, 0);
+    if (!h_compile_regex(prog, seq->parsers[i]->env))
+      return false;
+  }
+  h_rvm_insert_insn(prog, RVM_ACTION, h_rvm_create_action(prog, h_svm_action_ignoreseq, env));
+  return true;
+}
+
 static const HParserVtable ignoreseq_vt = {
   .parse = parse_ignoreseq,
   .isValidRegular = is_isValidRegular,
   .isValidCF = is_isValidCF,
   .desugar = desugar_ignoreseq,
+  .compile_to_rvm = is_ctrvm,
 };
 
 
@@ -73,7 +103,7 @@ static const HParserVtable ignoreseq_vt = {
 // API frontends
 //
 
-static const HParser* h_leftright__m(HAllocator* mm__, const HParser* p, const HParser* q, size_t which) {
+static HParser* h_leftright__m(HAllocator* mm__, const HParser* p, const HParser* q, size_t which) {
   HIgnoreSeq *seq = h_new(HIgnoreSeq, 1);
   seq->parsers = h_new(const HParser*, 2);
   seq->parsers[0] = p;
@@ -84,25 +114,25 @@ static const HParser* h_leftright__m(HAllocator* mm__, const HParser* p, const H
   return h_new_parser(mm__, &ignoreseq_vt, seq);
 }
 
-const HParser* h_left(const HParser* p, const HParser* q) {
+HParser* h_left(const HParser* p, const HParser* q) {
   return h_leftright__m(&system_allocator, p, q, 0);
 }
-const HParser* h_left__m(HAllocator* mm__, const HParser* p, const HParser* q) {
+HParser* h_left__m(HAllocator* mm__, const HParser* p, const HParser* q) {
   return h_leftright__m(mm__, p, q, 0);
 }
 
-const HParser* h_right(const HParser* p, const HParser* q) {
+HParser* h_right(const HParser* p, const HParser* q) {
   return h_leftright__m(&system_allocator, p, q, 1);
 }
-const HParser* h_right__m(HAllocator* mm__, const HParser* p, const HParser* q) {
+HParser* h_right__m(HAllocator* mm__, const HParser* p, const HParser* q) {
   return h_leftright__m(mm__, p, q, 1);
 }
 
 
-const HParser* h_middle(const HParser* p, const HParser* x, const HParser* q) {
+HParser* h_middle(const HParser* p, const HParser* x, const HParser* q) {
   return h_middle__m(&system_allocator, p, x, q);
 }
-const HParser* h_middle__m(HAllocator* mm__, const HParser* p, const HParser* x, const HParser* q) {
+HParser* h_middle__m(HAllocator* mm__, const HParser* p, const HParser* x, const HParser* q) {
   HIgnoreSeq *seq = h_new(HIgnoreSeq, 1);
   seq->parsers = h_new(const HParser*, 3);
   seq->parsers[0] = p;

src/parsers/indirect.c

@@ -21,6 +21,7 @@ static const HParserVtable indirect_vt = {
   .isValidRegular = h_false,
   .isValidCF = indirect_isValidCF,
   .desugar = desugar_indirect,
+  .compile_to_rvm = h_not_regular,
 };
 
 void h_bind_indirect(HParser* indirect, const HParser* inner) {

src/parsers/int_range.c

@@ -121,17 +121,39 @@ static HCFChoice* desugar_int_range(HAllocator *mm__, void *env) {
   return gen_int_range(mm__, r->lower, r->upper, bytes);
 }
 
+bool h_svm_action_validate_int_range(HArena *arena, HSVMContext *ctx, void* env) {
+  HRange *r_env = (HRange*)env;
+  HParsedToken *head = ctx->stack[ctx->stack_count-1];
+  switch (head-> token_type) {
+  case TT_SINT: 
+    return head->sint >= r_env->lower && head->sint <= r_env->upper;
+  case TT_UINT: 
+    return head->uint >= (uint64_t)r_env->lower && head->uint <= (uint64_t)r_env->upper;
+  default:
+    return false;
+  }
+}
+
+static bool ir_ctrvm(HRVMProg *prog, void *env) {
+  HRange *r_env = (HRange*)env;
+  
+  h_compile_regex(prog, r_env->p);
+  h_rvm_insert_insn(prog, RVM_ACTION, h_rvm_create_action(prog, h_svm_action_validate_int_range, env));
+  return false;
+}
+
 static const HParserVtable int_range_vt = {
   .parse = parse_int_range,
   .isValidRegular = h_true,
   .isValidCF = h_true,
   .desugar = desugar_int_range,
+  .compile_to_rvm = ir_ctrvm,
 };
 
-const HParser* h_int_range(const HParser *p, const int64_t lower, const int64_t upper) {
+HParser* h_int_range(const HParser *p, const int64_t lower, const int64_t upper) {
   return h_int_range__m(&system_allocator, p, lower, upper);
 }
-const HParser* h_int_range__m(HAllocator* mm__, const HParser *p, const int64_t lower, const int64_t upper) {
+HParser* h_int_range__m(HAllocator* mm__, const HParser *p, const int64_t lower, const int64_t upper) {
   // p must be an integer parser, which means it's using parse_bits
   // TODO: re-add this check
   //assert_message(p->vtable == &bits_vt, "int_range requires an integer parser"); 

src/parsers/many.c

@@ -33,7 +33,7 @@ static HParseResult *parse_many(void* env, HParseState *state) {
   HParsedToken *res = a_new(HParsedToken, 1);
   res->token_type = TT_SEQUENCE;
   res->seq = seq;
-  return make_result(state, res);
+  return make_result(state->arena, res);
  err0:
   if (count >= env_->count) {
     state->input_stream = bak;
@@ -112,17 +112,40 @@ static HCFChoice* desugar_many(HAllocator *mm__, void *env) {
   return ma;
 }
 
+static bool many_ctrvm(HRVMProg *prog, void *env) {
+  HRepeat *repeat = (HRepeat*)env;
+  // FIXME: Implement clear_to_mark
+  uint16_t clear_to_mark = h_rvm_create_action(prog, h_svm_action_clear_to_mark, NULL);
+  h_rvm_insert_insn(prog, RVM_PUSH, 0);
+  // TODO: implement min and max properly. Right now, it's always min==0, max==inf
+  uint16_t insn = h_rvm_insert_insn(prog, RVM_FORK, 0);
+  if (!h_compile_regex(prog, repeat->p))
+    return false;
+  if (repeat->sep != NULL) {
+    h_rvm_insert_insn(prog, RVM_PUSH, 0);
+    if (!h_compile_regex(prog, repeat->sep))
+      return false;
+    h_rvm_insert_insn(prog, RVM_ACTION, clear_to_mark);
+  }
+  h_rvm_insert_insn(prog, RVM_GOTO, insn);
+  h_rvm_patch_arg(prog, insn, h_rvm_get_ip(prog));
+
+  h_rvm_insert_insn(prog, RVM_ACTION, h_rvm_create_action(prog, h_svm_action_make_sequence, NULL));
+  return true;
+}
+
 static const HParserVtable many_vt = {
   .parse = parse_many,
   .isValidRegular = many_isValidRegular,
   .isValidCF = many_isValidCF,
   .desugar = desugar_many,
+  .compile_to_rvm = many_ctrvm,
 };
 
-const HParser* h_many(const HParser* p) {
+HParser* h_many(const HParser* p) {
   return h_many__m(&system_allocator, p);
 }
-const HParser* h_many__m(HAllocator* mm__, const HParser* p) {
+HParser* h_many__m(HAllocator* mm__, const HParser* p) {
   HRepeat *env = h_new(HRepeat, 1);
   env->p = p;
   env->sep = h_epsilon_p__m(mm__);
@@ -131,10 +154,10 @@ const HParser* h_many__m(HAllocator* mm__, const HParser* p) {
   return h_new_parser(mm__, &many_vt, env);
 }
 
-const HParser* h_many1(const HParser* p) {
+HParser* h_many1(const HParser* p) {
   return h_many1__m(&system_allocator, p);
 }
-const HParser* h_many1__m(HAllocator* mm__, const HParser* p) {
+HParser* h_many1__m(HAllocator* mm__, const HParser* p) {
   HRepeat *env = h_new(HRepeat, 1);
   env->p = p;
   env->sep = h_epsilon_p__m(mm__);
@@ -143,10 +166,10 @@ const HParser* h_many1__m(HAllocator* mm__, const HParser* p) {
   return h_new_parser(mm__, &many_vt, env);
 }
 
-const HParser* h_repeat_n(const HParser* p, const size_t n) {
+HParser* h_repeat_n(const HParser* p, const size_t n) {
   return h_repeat_n__m(&system_allocator, p, n);
 }
-const HParser* h_repeat_n__m(HAllocator* mm__, const HParser* p, const size_t n) {
+HParser* h_repeat_n__m(HAllocator* mm__, const HParser* p, const size_t n) {
   HRepeat *env = h_new(HRepeat, 1);
   env->p = p;
   env->sep = h_epsilon_p__m(mm__);
@@ -155,10 +178,10 @@ const HParser* h_repeat_n__m(HAllocator* mm__, const HParser* p, const size_t n)
   return h_new_parser(mm__, &many_vt, env);
 }
 
-const HParser* h_sepBy(const HParser* p, const HParser* sep) {
+HParser* h_sepBy(const HParser* p, const HParser* sep) {
   return h_sepBy__m(&system_allocator, p, sep);
 }
-const HParser* h_sepBy__m(HAllocator* mm__, const HParser* p, const HParser* sep) {
+HParser* h_sepBy__m(HAllocator* mm__, const HParser* p, const HParser* sep) {
   HRepeat *env = h_new(HRepeat, 1);
   env->p = p;
   env->sep = sep;
@@ -167,10 +190,10 @@ const HParser* h_sepBy__m(HAllocator* mm__, const HParser* p, const HParser* sep
   return h_new_parser(mm__, &many_vt, env);
 }
 
-const HParser* h_sepBy1(const HParser* p, const HParser* sep) {
+HParser* h_sepBy1(const HParser* p, const HParser* sep) {
   return h_sepBy1__m(&system_allocator, p, sep);
 }
-const HParser* h_sepBy1__m(HAllocator* mm__, const HParser* p, const HParser* sep) {
+HParser* h_sepBy1__m(HAllocator* mm__, const HParser* p, const HParser* sep) {
   HRepeat *env = h_new(HRepeat, 1);
   env->p = p;
   env->sep = sep;
@@ -213,10 +236,10 @@ static const HParserVtable length_value_vt = {
   .desugar = desugar_length_value,
 };
 
-const HParser* h_length_value(const HParser* length, const HParser* value) {
+HParser* h_length_value(const HParser* length, const HParser* value) {
   return h_length_value__m(&system_allocator, length, value);
 }
-const HParser* h_length_value__m(HAllocator* mm__, const HParser* length, const HParser* value) {
+HParser* h_length_value__m(HAllocator* mm__, const HParser* length, const HParser* value) {
   HLenVal *env = h_new(HLenVal, 1);
   env->length = length;
   env->value = value;

src/parsers/not.c

@@ -6,7 +6,7 @@ static HParseResult* parse_not(void* env, HParseState* state) {
     return NULL;
   else {
     state->input_stream = bak;
-    return make_result(state, NULL);
+    return make_result(state->arena, NULL);
   }
 }
 
@@ -20,11 +20,12 @@ static const HParserVtable not_vt = {
   .isValidRegular = h_false,  /* see and.c for why */
   .isValidCF = h_false,       /* also see and.c for why */
   .desugar = desugar_not,
+  .compile_to_rvm = h_not_regular, // Is actually regular, but the generation step is currently unable to handle it. TODO: fix this.
 };
 
-const HParser* h_not(const HParser* p) {
+HParser* h_not(const HParser* p) {
   return h_not__m(&system_allocator, p);
 }
-const HParser* h_not__m(HAllocator* mm__, const HParser* p) {
+HParser* h_not__m(HAllocator* mm__, const HParser* p) {
   return h_new_parser(mm__, &not_vt, (void *)p);
 }

src/parsers/nothing.c

@@ -1,6 +1,5 @@
 #include "parser_internal.h"
 
-
 static HParseResult* parse_nothing() {
   // not a mistake, this parser always fails
   return NULL;
@@ -15,16 +14,23 @@ static HCFChoice *desugar_nothing(HAllocator *mm__, void *env) {
   return ret;
 }
 
+static bool nothing_ctrvm(HRVMProg *prog, void* env) {
+  h_rvm_insert_insn(prog, RVM_MATCH, 0x0000);
+  h_rvm_insert_insn(prog, RVM_MATCH, 0xFFFF);
+  return true;
+}
+
 static const HParserVtable nothing_vt = {
   .parse = parse_nothing,
   .isValidRegular = h_true,
   .isValidCF = h_true,
   .desugar = desugar_nothing,
+  .compile_to_rvm = nothing_ctrvm,
 };
 
-const HParser* h_nothing_p() {
+HParser* h_nothing_p() {
   return h_nothing_p__m(&system_allocator);
 }
-const HParser* h_nothing_p__m(HAllocator* mm__) { 
+HParser* h_nothing_p__m(HAllocator* mm__) { 
   return h_new_parser(mm__, &nothing_vt, NULL);
 }

src/parsers/optional.c

@@ -1,3 +1,4 @@
+#include <assert.h>
 #include "parser_internal.h"
 
 static HParseResult* parse_optional(void* env, HParseState* state) {
@@ -8,7 +9,7 @@ static HParseResult* parse_optional(void* env, HParseState* state) {
   state->input_stream = bak;
   HParsedToken *ast = a_new(HParsedToken, 1);
   ast->token_type = TT_NONE;
-  return make_result(state, ast);
+  return make_result(state->arena, ast);
 }
 
 static bool opt_isValidRegular(void *env) {
@@ -26,17 +27,40 @@ static HCFChoice* desugar_optional(HAllocator *mm__, void *env) {
   return h_desugar(mm__, p);
 }
 
+static bool h_svm_action_optional(HArena *arena, HSVMContext *ctx, void *env) {
+  if (ctx->stack[ctx->stack_count-1]->token_type == TT_MARK) {
+    ctx->stack[ctx->stack_count-1]->token_type = TT_NONE;
+  } else {
+    ctx->stack_count--;
+    assert(ctx->stack[ctx->stack_count-1]->token_type == TT_MARK);
+    ctx->stack[ctx->stack_count-1] = ctx->stack[ctx->stack_count];
+  }
+  return true;
+}
+
+static bool opt_ctrvm(HRVMProg *prog, void* env) {
+  h_rvm_insert_insn(prog, RVM_PUSH, 0);
+  uint16_t insn = h_rvm_insert_insn(prog, RVM_FORK, 0);
+  HParser *p = (HParser*) env;
+  if (!h_compile_regex(prog, p->env))
+    return false;
+  h_rvm_patch_arg(prog, insn, h_rvm_get_ip(prog));
+  h_rvm_insert_insn(prog, RVM_ACTION, h_rvm_create_action(prog, h_svm_action_optional, NULL));
+  return true;
+}
+
 static const HParserVtable optional_vt = {
   .parse = parse_optional,
   .isValidRegular = opt_isValidRegular,
   .isValidCF = opt_isValidCF,
   .desugar = desugar_optional,
+  .compile_to_rvm = opt_ctrvm,
 };
 
-const HParser* h_optional(const HParser* p) {
+HParser* h_optional(const HParser* p) {
   return h_optional__m(&system_allocator, p);
 }
-const HParser* h_optional__m(HAllocator* mm__, const HParser* p) {
+HParser* h_optional__m(HAllocator* mm__, const HParser* p) {
   // TODO: re-add this
   //assert_message(p->vtable != &ignore_vt, "Thou shalt ignore an option, rather than the other way 'round.");
   return h_new_parser(mm__, &optional_vt, (void *)p);

src/parsers/parser_internal.h

@@ -2,15 +2,16 @@
 #define HAMMER_PARSE_INTERNAL__H
 #include "../hammer.h"
 #include "../internal.h"
+#include "../backends/regex.h"
 
 #define a_new_(arena, typ, count) ((typ*)h_arena_malloc((arena), sizeof(typ)*(count)))
 #define a_new(typ, count) a_new_(state->arena, typ, count)
 // we can create a_new0 if necessary. It would allocate some memory and immediately zero it out.
 
-static inline HParseResult* make_result(HParseState *state, HParsedToken *tok) {
+static inline HParseResult* make_result(HArena *arena, HParsedToken *tok) {
-  HParseResult *ret = a_new(HParseResult, 1);
+  HParseResult *ret = h_arena_malloc(arena, sizeof(HParseResult));
   ret->ast = tok;
-  ret->arena = state->arena;
+  ret->arena = arena;
   return ret;
 }
 
@@ -23,9 +24,6 @@ static inline size_t token_length(HParseResult *pr) {
   }
 }
 
-static inline bool h_true(void *env) { return true; }
-static inline bool h_false(void *env) { return false; }
-
 /* Epsilon rules happen during desugaring. This handles them. */
 static inline HCFChoice* desugar_epsilon(HAllocator *mm__, void *env) {
   static HCFChoice *res_seq_l[] = {NULL};

src/parsers/sequence.c

@@ -21,7 +21,7 @@ static HParseResult* parse_sequence(void *env, HParseState *state) {
   }
   HParsedToken *tok = a_new(HParsedToken, 1);
   tok->token_type = TT_SEQUENCE; tok->seq = seq;
-  return make_result(state, tok);
+  return make_result(state->arena, tok);
 }
 
 static bool sequence_isValidRegular(void *env) {
@@ -59,34 +59,46 @@ static HCFChoice* desugar_sequence(HAllocator *mm__, void *env) {
   return ret;
 }
 
+static bool sequence_ctrvm(HRVMProg *prog, void *env) {
+  HSequence *s = (HSequence*)env;
+  h_rvm_insert_insn(prog, RVM_PUSH, 0);
+  for (size_t i=0; i<s->len; ++i) {
+    if (!s->p_array[i]->vtable->compile_to_rvm(prog, s->p_array[i]->env))
+      return false;
+  }
+  h_rvm_insert_insn(prog, RVM_ACTION, h_rvm_create_action(prog, h_svm_action_make_sequence, NULL));
+  return true;
+}
+
 static const HParserVtable sequence_vt = {
   .parse = parse_sequence,
   .isValidRegular = sequence_isValidRegular,
   .isValidCF = sequence_isValidCF,
   .desugar = desugar_sequence,
+  .compile_to_rvm = sequence_ctrvm,
 };
 
-const HParser* h_sequence(const HParser* p, ...) {
+HParser* h_sequence(const HParser* p, ...) {
   va_list ap;
   va_start(ap, p);
-  const HParser* ret = h_sequence__mv(&system_allocator, p,  ap);
+  HParser* ret = h_sequence__mv(&system_allocator, p,  ap);
   va_end(ap);
   return ret;
 }
 
-const HParser* h_sequence__m(HAllocator* mm__, const HParser* p, ...) {
+HParser* h_sequence__m(HAllocator* mm__, const HParser* p, ...) {
   va_list ap;
   va_start(ap, p);
-  const HParser* ret = h_sequence__mv(mm__, p,  ap);
+  HParser* ret = h_sequence__mv(mm__, p,  ap);
   va_end(ap);
   return ret;
 }
 
-const HParser* h_sequence__v(const HParser* p, va_list ap) {
+HParser* h_sequence__v(const HParser* p, va_list ap) {
   return h_sequence__mv(&system_allocator, p, ap);
 }
 
-const HParser* h_sequence__mv(HAllocator* mm__, const HParser *p, va_list ap_) {
+HParser* h_sequence__mv(HAllocator* mm__, const HParser *p, va_list ap_) {
   va_list ap;
   size_t len = 0;
   const HParser *arg;

src/parsers/token.c

@@ -15,9 +15,10 @@ static HParseResult* parse_token(void *env, HParseState *state) {
   }
   HParsedToken *tok = a_new(HParsedToken, 1);
   tok->token_type = TT_BYTES; tok->bytes.token = t->str; tok->bytes.len = t->len;
-  return make_result(state, tok);
+  return make_result(state->arena, tok);
 }
 
+
 static HCFChoice* desugar_token(HAllocator *mm__, void *env) {
   HToken *tok = (HToken*)env;
   HCFSequence *seq = h_new(HCFSequence, 1);
@@ -37,17 +38,29 @@ static HCFChoice* desugar_token(HAllocator *mm__, void *env) {
   return ret;
 }
 
+static bool token_ctrvm(HRVMProg *prog, void *env) {
+  HToken *t = (HToken*)env;
+  h_rvm_insert_insn(prog, RVM_PUSH, 0);
+  for (int i=0; i<t->len; ++i) {
+    h_rvm_insert_insn(prog, RVM_MATCH, t->str[i] & t->str[i] << 8);
+    h_rvm_insert_insn(prog, RVM_STEP, 0);
+  }
+  h_rvm_insert_insn(prog, RVM_CAPTURE, 0);
+  return true;
+}
+
 const HParserVtable token_vt = {
   .parse = parse_token,
   .isValidRegular = h_true,
   .isValidCF = h_true,
   .desugar = desugar_token,
+  .compile_to_rvm = token_ctrvm,
 };
 
-const HParser* h_token(const uint8_t *str, const size_t len) {
+HParser* h_token(const uint8_t *str, const size_t len) {
   return h_token__m(&system_allocator, str, len);
 }
-const HParser* h_token__m(HAllocator* mm__, const uint8_t *str, const size_t len) { 
+HParser* h_token__m(HAllocator* mm__, const uint8_t *str, const size_t len) { 
   HToken *t = h_new(HToken, 1);
   t->str = (uint8_t*)str, t->len = len;
   return h_new_parser(mm__, &token_vt, t);

src/parsers/unimplemented.c

@@ -22,6 +22,7 @@ static const HParserVtable unimplemented_vt = {
   .isValidRegular = h_false,
   .isValidCF = h_false,
   .desugar = desugar_unimplemented,
+  .compile_to_rvm = h_not_regular,
 };
 
 static HParser unimplemented = {

src/parsers/whitespace.c

@@ -49,16 +49,32 @@ static bool ws_isValidCF(void *env) {
   return p->vtable->isValidCF(p->env);
 }
 
+static bool ws_ctrvm(HRVMProg *prog, void *env) {
+  HParser *p = (HParser*)env;
+  uint16_t start = h_rvm_get_ip(prog);
+  uint16_t next;
+  const char SPACE_CHRS[6] = {' ', '\f', '\n', '\r', '\t', '\v'};
+
+  for (int i = 0; i < 6; i++) {
+    next = h_rvm_insert_insn(prog, RVM_FORK, 0);
+    h_rvm_insert_insn(prog, RVM_MATCH, (SPACE_CHRS[i] << 8) | (SPACE_CHRS[i]));
+    h_rvm_insert_insn(prog, RVM_GOTO, start);
+    h_rvm_patch_arg(prog, next, h_rvm_get_ip(prog));
+  }
+  return h_compile_regex(prog, p->env);
+}
+
 static const HParserVtable whitespace_vt = {
   .parse = parse_whitespace,
   .isValidRegular = ws_isValidRegular,
   .isValidCF = ws_isValidCF,
   .desugar = desugar_whitespace,
+  .compile_to_rvm = ws_ctrvm,
 };
 
-const HParser* h_whitespace(const HParser* p) {
+HParser* h_whitespace(const HParser* p) {
   return h_whitespace__m(&system_allocator, p);
 }
-const HParser* h_whitespace__m(HAllocator* mm__, const HParser* p) {
+HParser* h_whitespace__m(HAllocator* mm__, const HParser* p) {
   return h_new_parser(mm__, &whitespace_vt, (void *)p);
 }

src/parsers/xor.c

@@ -39,14 +39,15 @@ static HCFChoice* desugar_xor(HAllocator *mm__, void *env) {
 static const HParserVtable xor_vt = {
   .parse = parse_xor,
   .isValidRegular = h_false,
-  .isValidCF = h_false,
+  .isValidCF = h_false, // XXX should this be true if both p1 and p2 are CF?
   .desugar = desugar_xor,
+  .compile_to_rvm = h_not_regular,
 };
 
-const HParser* h_xor(const HParser* p1, const HParser* p2) {
+HParser* h_xor(const HParser* p1, const HParser* p2) {
   return h_xor__m(&system_allocator, p1, p2);
 }
-const HParser* h_xor__m(HAllocator* mm__, const HParser* p1, const HParser* p2) { 
+HParser* h_xor__m(HAllocator* mm__, const HParser* p1, const HParser* p2) { 
   HTwoParsers *env = h_new(HTwoParsers, 1);
   env->p1 = p1;
   env->p2 = p2;

src/system_allocator.c

@@ -1,16 +1,27 @@
+#include <string.h>
 #include <stdlib.h>
 #include "internal.h"
 
 static void* system_alloc(HAllocator *allocator, size_t size) {
-  return malloc(size);
+  
+  void* ptr = calloc(size + sizeof(size_t), 1);
+  *(size_t*)ptr = size;
+  return ptr + sizeof(size_t);
 }
 
 static void* system_realloc(HAllocator *allocator, void* ptr, size_t size) {
-  return realloc(ptr, size);
+  if (ptr == NULL)
+    return system_alloc(allocator, size);
+  ptr = realloc(ptr - sizeof(size_t), size + sizeof(size_t));
+  size_t old_size = *(size_t*)ptr;
+  *(size_t*)ptr = size;
+  if (size > old_size)
+    memset(ptr+sizeof(size_t)+old_size, 0, size - old_size);
+  return ptr + sizeof(size_t);
 }
 
 static void system_free(HAllocator *allocator, void* ptr) {
-  free(ptr);
+  free(ptr - sizeof(size_t));
 }
 
 HAllocator system_allocator = {

src/t_benchmark.c

@@ -11,7 +11,7 @@ HParserTestcase testcases[] = {
 };
 
 static void test_benchmark_1() {
-  const HParser *parser = h_sepBy1(h_choice(h_ch('1'), h_ch('2'), h_ch('3'), NULL), h_ch(',')); 
+  HParser *parser = h_sepBy1(h_choice(h_ch('1'), h_ch('2'), h_ch('3'), NULL), h_ch(',')); 
 
   HBenchmarkResults *res = h_benchmark(parser, testcases);
   h_benchmark_report(stderr, res);

src/test_suite.h

@@ -52,6 +52,28 @@
     }							\
   } while(0)
 
+#define g_check_regular(lang) do {			\
+    if (!lang->isValidRegular(lang->env)) {		\
+      g_test_message("Language is not regular");	\
+      g_test_fail();					\
+    }							\
+  } while(0)
+
+#define g_check_contextfree(lang) do {			\
+    if (!lang->isValidCF(lang->env)) {			\
+      g_test_message("Language is not context-free");	\
+      g_test_fail();					\
+    }							\
+  } while(0)
+
+#define g_check_compilable(lang, backend, params) do {	\
+    if (!h_compile(lang, backend, params)) {		\
+      g_test_message("Language is not %s(%s)", #backend, params);	\
+      g_test_fail();					\
+    }							\
+  } while(0)
+
+  
 // TODO: replace uses of this with g_check_parse_failed
 #define g_check_failed(res) do {					\
     const HParseResult *result = (res);					\
@@ -77,7 +99,7 @@
     } else {								\
       char* cres = h_write_result_unamb(res->ast);			\
       g_check_string(cres, ==, result);					\
-      g_free(cres);							\
+      system_allocator.free(&system_allocator, cres);			\
       HArenaStats stats;						\
       h_allocator_stats(res->arena, &stats);				\
       g_test_message("Parse used %zd bytes, wasted %zd bytes. "		\
@@ -149,4 +171,5 @@
 #define g_check_cmpdouble(n1, op, n2) g_check_inttype("%g", double, n1, op, n2)
 
 
+
 #endif // #ifndef HAMMER_TEST_SUITE__H