fix labels and variables in data directives

lips/Collector.lua

@@ -35,6 +35,10 @@ function Collector:push_data(datum, size)
 
     -- TODO: consider not scrunching data statements, just their tokens
 
+    if type(datum) == 'number' then
+        datum = self:token(datum)
+    end
+
     local last_statement = self.statements[#self.statements]
     local s
     if last_statement and last_statement.type == '!DATA' then
@@ -44,16 +48,25 @@ function Collector:push_data(datum, size)
         insert(self.statements, s)
     end
 
-    if type(datum) == 'string' and size == 'WORD' then
-        -- labels will be assembled to words
-        insert(s, Token('LABEL', datum))
-        return
-    end
-
     if size ~= 'BYTE' and size ~= 'HALFWORD' and size ~= 'WORD' then
         error('Internal Error: unknown data size argument')
     end
 
+    if datum.tt == 'LABELSYM' then
+        if size == 'WORD' then
+            -- labels will be assembled to words
+            insert(s, datum)
+            return
+        else
+            self:error('labels are too large to be used in this directive')
+        end
+    elseif datum.tt == 'VARSYM' then
+        insert(s, datum:set('size', size))
+        return
+    elseif datum.tt ~= 'NUM' then
+        self:error('unsupported data type', datum.tt)
+    end
+
     local sizes = size..'S'
 
     local last_token = s[#s]
@@ -65,7 +78,7 @@ function Collector:push_data(datum, size)
         insert(s, t)
         s:validate()
     end
-    insert(t.tok, datum)
+    insert(t.tok, datum.tok)
 end
 
 function Collector:variable()
@@ -104,10 +117,10 @@ function Collector:directive()
             end
         end
     elseif name == 'BYTE' or name == 'HALFWORD' or name == 'WORD' then
-        self:push_data(self:const().tok, name)
+        self:push_data(self:const(), name)
         while not self:is_EOL() do
             self:optional_comma()
-            self:push_data(self:const().tok, name)
+            self:push_data(self:const(), name)
         end
     elseif name == 'HEX' then
         if self.tt ~= 'OPEN' then
@@ -119,7 +132,7 @@ function Collector:directive()
             if self.tt == 'EOL' then
                 self:advance()
             else
-                self:push_data(self:const().tok, 'BYTE')
+                self:push_data(self:const(), 'BYTE')
             end
         end
         self:advance()

lips/Dumper.lua

@@ -365,13 +365,17 @@ function Dumper:load(statements)
             insert(new_statements, new)
         elseif s.type == '!DATA' then
             for i, t in ipairs(s) do
-                if t.tt == 'LABEL' then
+                if t.tt == 'LABELSYM' then
                     local label = self.labels[t.tok]
                     if label == nil then
                         self:error('undefined label', t.tok)
                     end
                     t.tt = 'WORDS'
                     t.tok = {label}
+                elseif t.tt == 'NUM' then
+                    t.tt = t.size..'S'
+                    t.tok = {t.tok}
+                    t.size = nil
                 end
             end
             self.pos = self.pos + (s.length or util.measure_data(s))

lips/util.lua

@@ -39,12 +39,27 @@ else -- 5.2, 5.3
     end
 end
 
+local data_sizes = {
+    BYTE = 1,
+    HALFWORD = 2,
+    WORD = 4,
+}
+
 local function measure_data(s)
     assert(s and s.type == '!DATA', 'Internal Error: expected !DATA statement')
     local n = 0
     for i, t in ipairs(s) do
-        if t.tt == 'LABEL' then
+        if t.tt == 'LABELSYM' then
             n = n + 4
+        elseif t.tt == 'NUM' then
+            if t.size == nil then
+                error('Internal Error: unspecified data size in NUM')
+            end
+            local size = data_sizes[t.size]
+            if size == nil then
+                error('Internal Error: unknown data size in NUM, got '..tostring(t.size))
+            end
+            n = n + size
         elseif t.tt == 'WORDS' then
             n = n + #t.tok * 4
         elseif t.tt == 'HALFWORDS' then
@@ -52,7 +67,7 @@ local function measure_data(s)
         elseif t.tt == 'BYTES' then
             n = n + #t.tok * 1
         else
-            error('Internal Error: unknown data type in !DATA')
+            error('Internal Error: unknown data type in !DATA, got '..tostring(t.tt))
         end
     end
     return n