factor out a struct HLREngine

src/backends/lalr.c

@@ -61,6 +61,16 @@ typedef struct HLREnhGrammar_ {
   HArena *arena;
 } HLREnhGrammar;
 
+typedef struct HLREngine_ {
+  const HLRTable *table;
+  HSlist *left;     // left stack; reductions happen here
+  HSlist *right;    // right stack; input appears here
+  size_t state;
+  bool running;
+  HArena *arena;    // will hold the results
+  HArena *tarena;   // tmp, deleted after parse
+} HLREngine;
+
 
 // XXX move to internal.h or something
 // XXX replace other hashtable iterations with this
@@ -733,24 +743,33 @@ h_lr_lookup(const HLRTable *table, size_t state, const HCFChoice *symbol)
   }
 }
 
-HParseResult *h_lr_parse(HAllocator* mm__, const HParser* parser, HInputStream* stream)
+HLREngine *h_lrengine_new(HArena *arena, HArena *tarena, const HLRTable *table)
 {
-  HLRTable *table = parser->backend_data;
+  HLREngine *engine = h_arena_malloc(tarena, sizeof(HLREngine));
-  if(!table)
-    return NULL;
 
-  HArena *arena  = h_new_arena(mm__, 0);    // will hold the results
+  engine->table = table;
-  HArena *tarena = h_new_arena(mm__, 0);    // tmp, deleted after parse
+  engine->left = h_slist_new(tarena);
-  HSlist *left = h_slist_new(tarena);   // left stack; reductions happen here
+  engine->right = h_slist_new(tarena);
-  HSlist *right = h_slist_new(tarena);  // right stack; input appears here
+  engine->state = 0;
+  engine->running = 1;
+  engine->arena = arena;
+  engine->tarena = tarena;
+
+  return engine;
+}
+
+void h_lrengine_step(HLREngine *engine, HInputStream *stream)
+{
+  // short-hand names
+  HSlist *left = engine->left;
+  HSlist *right = engine->right;
+  HArena *arena = engine->arena;
+  HArena *tarena = engine->tarena;
 
   // stack layout:
   // on the left stack, we put pairs:  (saved state, semantic value)
   // on the right stack, we put pairs: (symbol, semantic value)
 
-  // run while the recognizer finds handles in the input
-  size_t state = 0;
-  while(1) {
   // make sure there is input on the right stack
   if(h_slist_empty(right)) {
     // XXX use statically-allocated terminal symbols
@@ -778,15 +797,18 @@ HParseResult *h_lr_parse(HAllocator* mm__, const HParser* parser, HInputStream*
   HCFChoice *symbol = right->head->elem;
 
   // table lookup
-    const HLRAction *action = h_lr_lookup(table, state, symbol);
+  const HLRAction *action = h_lr_lookup(engine->table, engine->state, symbol);
-    if(action == NULL)
+  if(action == NULL) {
-      break;    // no handle recognizable in input, terminate parsing
+    // no handle recognizable in input, terminate
+    engine->running = false;
+    return;
+  }
 
   if(action->type == HLR_SHIFT) {
-      h_slist_push(left, (void *)(uintptr_t)state);
+    h_slist_push(left, (void *)(uintptr_t)engine->state);
     h_slist_pop(right);                       // symbol (discard)
     h_slist_push(left, h_slist_pop(right));   // semantic value
-      state = action->nextstate;
+    engine->state = action->nextstate;
   } else {
     assert(action->type == HLR_REDUCE);
     size_t len = action->production.length;
@@ -801,7 +823,7 @@ HParseResult *h_lr_parse(HAllocator* mm__, const HParser* parser, HInputStream*
     HParsedToken *v = NULL;
     for(size_t i=0; i<len; i++) {
       v = h_slist_pop(left);
-        state = (uintptr_t)h_slist_pop(left);
+      engine->state = (uintptr_t)h_slist_pop(left);
 
       // collect values in result sequence
       value->seq->elements[len-1-i] = v;
@@ -820,8 +842,11 @@ HParseResult *h_lr_parse(HAllocator* mm__, const HParser* parser, HInputStream*
       value = (HParsedToken *)symbol->reshape(make_result(arena, value));
 
     // call validation and semantic action, if present
-      if(symbol->pred && !symbol->pred(make_result(tarena, value)))
+    if(symbol->pred && !symbol->pred(make_result(tarena, value))) {
-        break;  // validation failed -> no parse
+      // validation failed -> no parse; terminate
+      engine->running = false;
+      return;
+    }
     if(symbol->action)
       value = (HParsedToken *)symbol->action(make_result(arena, value));
 
@@ -831,20 +856,36 @@ HParseResult *h_lr_parse(HAllocator* mm__, const HParser* parser, HInputStream*
   }
 }
 
-
+HParseResult *h_lrengine_result(HLREngine *engine)
-
+{
   // parsing was successful iff the start symbol is on top of the right stack
-  HParseResult *result = NULL;
+  if(h_slist_pop(engine->right) == engine->table->start) {
-  if(h_slist_pop(right) == table->start) {
     // next on the right stack is the start symbol's semantic value
-    assert(!h_slist_empty(right));
+    assert(!h_slist_empty(engine->right));
-    HParsedToken *tok = h_slist_pop(right);
+    HParsedToken *tok = h_slist_pop(engine->right);
-    result = make_result(arena, tok);
+    return make_result(engine->arena, tok);
   } else {
-    h_delete_arena(arena);
+    return NULL;
-    result = NULL;
+  }
 }
 
+HParseResult *h_lr_parse(HAllocator* mm__, const HParser* parser, HInputStream* stream)
+{
+  HLRTable *table = parser->backend_data;
+  if(!table)
+    return NULL;
+
+  HArena *arena  = h_new_arena(mm__, 0);    // will hold the results
+  HArena *tarena = h_new_arena(mm__, 0);    // tmp, deleted after parse
+  HLREngine *engine = h_lrengine_new(arena, tarena, table);
+
+  // run while the recognizer finds handles in the input
+  while(engine->running)
+    h_lrengine_step(engine, stream);
+
+  HParseResult *result = h_lrengine_result(engine);
+  if(!result)
+    h_delete_arena(arena);
   h_delete_arena(tarena);
   return result;
 }