Got a lot of regex test cases working

src/parsers/bits.c

@@ -97,6 +97,8 @@ static bool h_svm_action_bits(HArena *arena, HSVMContext *ctx, void* env) {
   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.
+  uint64_t msb = (env_->signedp ? 1LL:0) << (top->bytes.len * 8 - 1);
+  res = (res ^ msb) - msb;
   top->uint = res; // possibly cast to signed through union
   top->token_type = (env_->signedp ? TT_SINT : TT_UINT);
   return true;
@@ -105,7 +107,7 @@ static bool h_svm_action_bits(HArena *arena, HSVMContext *ctx, void* env) {
 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
+  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);
   }

src/parsers/ch.c

@@ -1,3 +1,4 @@
+#include <assert.h>
 #include "parser_internal.h"
 
 static HParseResult* parse_ch(void* env, HParseState *state) {
@@ -20,11 +21,26 @@ static HCFChoice* desugar_ch(HAllocator *mm__, void *env) {
   return ret;
 }
 
+static bool h_svm_action_ch(HArena *arena, HSVMContext *ctx, void* env) {
+  // BUG: relies un undefined behaviour: int64_t is a signed uint64_t; not necessarily true on 32-bit
+  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 = TT_UINT;
+  return true;
+}
+
 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_PUSH, 0);
+  h_rvm_insert_insn(prog, RVM_MATCH, c | c << 8);
   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_ch, env));
   return true;
 }
 

src/parsers/charset.c

@@ -1,3 +1,4 @@
+#include <assert.h>
 #include <string.h>
 #include "../internal.h"
 #include "parser_internal.h"
@@ -22,23 +23,42 @@ static HCFChoice* desugar_charset(HAllocator *mm__, void *env) {
   return ret;
 }
 
+static bool h_svm_action_ch(HArena *arena, HSVMContext *ctx, void* env) {
+  // BUG: relies un undefined behaviour: int64_t is a signed uint64_t; not necessarily true on 32-bit
+  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 = TT_UINT;
+  return true;
+}
+
 // FUTURE: this is horribly inefficient
 static bool cs_ctrvm(HRVMProg *prog, void *env) {
   HCharset cs = (HCharset)env;
+  h_rvm_insert_insn(prog, RVM_PUSH, 0);
+
   uint16_t start = h_rvm_get_ip(prog);
   for (size_t i=0; i<256; ++i) {
+    // TODO: merge ranges.
     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_MATCH, i | i << 8);
       h_rvm_insert_insn(prog, RVM_GOTO, 0);
       h_rvm_patch_arg(prog, insn, h_rvm_get_ip(prog));
     }
   }
+  h_rvm_insert_insn(prog, RVM_MATCH, 0x00FF);
   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);
   }
+
+  h_rvm_insert_insn(prog, RVM_CAPTURE, 0);
+  h_rvm_insert_insn(prog, RVM_ACTION, h_rvm_create_action(prog, h_svm_action_ch, env));
   return true;
 }
 

src/parsers/token.c

@@ -69,7 +69,7 @@ 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_MATCH, t->str[i] | t->str[i] << 8);
     h_rvm_insert_insn(prog, RVM_STEP, 0);
   }
   h_rvm_insert_insn(prog, RVM_CAPTURE, 0);

src/parsers/whitespace.c

@@ -73,7 +73,7 @@ static bool ws_ctrvm(HRVMProg *prog, void *env) {
     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);
+  return h_compile_regex(prog, p);
 }
 
 static const HParserVtable whitespace_vt = {