Modifying for literals

pyasm/pyasm

@@ -33,10 +33,20 @@ import string
 import getopt
 import traceback
 
+
+
+
 ######
 # Globals
 ######
 
+# list of Reloc blocks
+RelocList = []
+
+# the next literal block number
+# used to generate literal labels like '$L1'
+Literal = 0
+
 # the input assembler filename
 AsmFile = None
 
@@ -105,6 +115,132 @@ ListFileExtension = '.lst'
 # number of bytes in the 'zero' leader
 ZeroLeaderSize = 16
 
+#################################
+
+######
+# The Reloc class
+######
+
+class Reloc(object):
+    """A Reloc object holding attributes of a block of code:
+        .base   the base address of the block (may be None)
+        .code   a list of generated code words
+        .reloc  a list of relocation tuples:
+                  (offset, label)
+                  where offset is the offset from the block base to the word
+                               to be relocated
+                  and   label  is the symtab label of the value to relocate with
+        .label  is the symbolic label for the block, only used for literals
+    """
+
+    def __init__(self, base=None, label=None):
+        """Initialize a Reloc object
+
+        base   is the base address of the block (None for literals)
+        label  is the symbolic label for the literal start adress
+        """
+
+        self.base = base
+        self.label = label
+        self.code = []
+        self.reloc = []
+
+    def add_code(self, code):
+        """Add a code word to the internal code list.
+
+        code  the binary value to add
+        """
+
+        self.code.append(code)
+
+    def add_reloc(self, reloc):
+        """Add a relocation tuple to the Reloc object.
+
+        reloc  the relocation tuple to add:
+                 (offset, label)
+        """
+
+        self.reloc.append(reloc)
+
+######
+# Utility functions
+######
+
+def next_lit_label():
+    """Generate the next literal label string.
+
+    Returns a string like '$L25'.  This label is not legally
+    usable by the user.
+    """
+
+    global Literal
+
+    Literal += 1
+    return '$L%d' % Literal
+
+def assemble_source_block(base, code, start_line, indent=0, label=None):
+    """Assemble a source block.
+
+    base        base address of the block, may be None
+    code        list of source lines to assemble
+    start_line  line number of first code line
+    indent      column number to be considered leftmost column
+                only non-zero for a literal block
+    label       label for a literal block if defined
+
+    Appends zero or more Reloc blocks to the global Reloc list.  Zero blocks
+    because the source block may contain only comments.
+    """
+
+    # if indent is non-zero, strip leading space
+    # raise an error if leading chars not all whitespace
+    if indent > 0:
+        new_code = []
+        for (loffset, line) in enumerate(code):
+            leader = line[:indent]
+            rest = line[indent:]
+            if leader.strip():
+                fatal('%s\n%d: None-blank indent in literal'
+                      % (line, start_line+loffset))
+            new_code.append(rest)
+        code = new_code
+
+    # assemble the source, line by line
+    code_reloc = Reloc(base, label)
+    for line in code:
+        result = assemble_line(line)
+        if result is None:
+            # no code generated
+            continue
+        (word, reloc) = result
+        code_reloc.add_code(word)
+        if reloc:
+            code_reloc.add_reloc(reloc)
+
+    # maybe add the Reloc block to the global list
+    if code_reloc.code:
+        global RelocList
+        RelocList.append(code_reloc)
+
+def assemble_line(line, block_base, offset):
+    """Assemble one line of code.
+
+    line        the line of code to assemble
+    block_base  base address of the block this line is in (may be None)
+    offset      offset of this word from the block_base address
+
+    Returns a tuple of generated word and reloc value (if any):
+        (word, label)
+
+    Note, the relocation value may be:
+        . string - the undefined label
+        . integer - the offset from the base address for operand expressions
+                    like '.+4'
+    """
+
+    pass
+
+
 ######
 # dict mapping opcode to generated word, address opts, address mask & indirect allowed
 ######