Lichen

Annotated encoders.py

1000:f646afbcfe1c
1039:16990afa2754
989:23edd8f55d40
5 months ago Paul Boddie file changeset files shortlog Eliminated redundant temporary result and attribute reference storage. value-replacement
paul@0 1 
#!/usr/bin/env python
paul@0 2 
 
 paul@0      3 
"""
 
 paul@0      4 
Encoder functions, producing representations of program objects.
 
 paul@0      5 
 
 paul@833      6 
Copyright (C) 2016, 2017, 2018 Paul Boddie <paul@boddie.org.uk>
 
 paul@0      7 
 
 paul@0      8 
This program is free software; you can redistribute it and/or modify it under
 
 paul@0      9 
the terms of the GNU General Public License as published by the Free Software
 
 paul@0     10 
Foundation; either version 3 of the License, or (at your option) any later
 
 paul@0     11 
version.
 
 paul@0     12 
 
 paul@0     13 
This program is distributed in the hope that it will be useful, but WITHOUT
 
 paul@0     14 
ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
 
 paul@0     15 
FOR A PARTICULAR PURPOSE.  See the GNU General Public License for more
 
 paul@0     16 
details.
 
 paul@0     17 
 
 paul@0     18 
You should have received a copy of the GNU General Public License along with
 
 paul@0     19 
this program.  If not, see <http://www.gnu.org/licenses/>.
 
 paul@0     20 
"""
 
 paul@0     21 
 
 paul@498     22 
from common import first, InstructionSequence
 
 paul@56     23 
 
 paul@609     24 
 
 paul@609     25 
 
 paul@609     26 
# Value digest computation.
 
 paul@609     27 
 
 paul@609     28 
from base64 import b64encode
 
 paul@609     29 
from hashlib import sha1
 
 paul@609     30 
 
 paul@609     31 
def digest(values):
 
 paul@609     32 
    m = sha1()
 
 paul@609     33 
    for value in values:
 
 paul@609     34 
        m.update(str(value))
 
 paul@609     35 
    return b64encode(m.digest()).replace("+", "__").replace("/", "_").rstrip("=")
 
 paul@609     36 
 
 paul@609     37 
 
 paul@609     38 
 
 paul@0     39 
# Output encoding and decoding for the summary files.
 
 paul@0     40 
 
 paul@0     41 
def encode_attrnames(attrnames):
 
 paul@0     42 
 
 paul@0     43 
    "Encode the 'attrnames' representing usage."
 
 paul@0     44 
 
 paul@0     45 
    return ", ".join(attrnames) or "{}"
 
 paul@0     46 
 
 paul@0     47 
def encode_constrained(constrained):
 
 paul@0     48 
 
 paul@0     49 
    "Encode the 'constrained' status for program summaries."
 
 paul@0     50 
 
 paul@0     51 
    return constrained and "constrained" or "deduced"
 
 paul@0     52 
 
 paul@0     53 
def encode_usage(usage):
 
 paul@0     54 
 
 paul@0     55 
    "Encode attribute details from 'usage'."
 
 paul@0     56 
 
 paul@0     57 
    all_attrnames = []
 
 paul@0     58 
    for t in usage:
 
 paul@107     59 
        attrname, invocation, assignment = t
 
 paul@107     60 
        all_attrnames.append("%s%s" % (attrname, invocation and "!" or assignment and "=" or ""))
 
 paul@0     61 
    return ", ".join(all_attrnames) or "{}"
 
 paul@0     62 
 
 paul@88     63 
def decode_usage(s):
 
 paul@88     64 
 
 paul@88     65 
    "Decode attribute details from 's'."
 
 paul@88     66 
 
 paul@88     67 
    all_attrnames = set()
 
 paul@88     68 
    for attrname_str in s.split(", "):
 
 paul@107     69 
        all_attrnames.add((attrname_str.rstrip("!="), attrname_str.endswith("!"), attrname_str.endswith("=")))
 
 paul@88     70 
 
 paul@88     71 
    all_attrnames = list(all_attrnames)
 
 paul@88     72 
    all_attrnames.sort()
 
 paul@88     73 
    return tuple(all_attrnames)
 
 paul@88     74 
 
 paul@0     75 
def encode_access_location(t):
 
 paul@0     76 
 
 paul@0     77 
    "Encode the access location 't'."
 
 paul@0     78 
 
 paul@791     79 
    return "%s:%s:%s:%d" % (t.path, t.name or "{}", t.attrnames or "{}", t.access_number)
 
 paul@0     80 
 
 paul@841     81 
def decode_access_location(s):
 
 paul@841     82 
 
 paul@841     83 
    "Decode the access location 's'."
 
 paul@841     84 
 
 paul@841     85 
    path, name, attrnames, access_number = s.split(":")
 
 paul@841     86 
    return path, name, attrnames, access_number
 
 paul@841     87 
 
 paul@789     88 
def encode_alias_location(t, invocation=False):
 
 paul@789     89 
 
 paul@789     90 
    "Encode the alias location 't'."
 
 paul@789     91 
 
 paul@791     92 
    return "%s:%s:%s%s%s%s" % (t.path, t.name or "{}", t.attrnames or "{}",
 
 paul@791     93 
        t.version is not None and ":=%d" % t.version or "",
 
 paul@791     94 
        t.access_number is not None and ":#%d" % t.access_number or "",
 
 paul@791     95 
        invocation and "!" or "")
 
 paul@789     96 
 
 paul@833     97 
def decode_alias_location(s):
 
 paul@833     98 
 
 paul@833     99 
    "Decode the alias location 's'."
 
 paul@833    100 
 
 paul@833    101 
    path, name, rest = s.split(":", 2)
 
 paul@833    102 
    attrnames = version = access_number = None
 
 paul@833    103 
    invocation = rest.endswith("!")
 
 paul@833    104 
 
 paul@833    105 
    t = rest.rstrip("!").split(":#")
 
 paul@833    106 
    if len(t) > 1:
 
 paul@833    107 
        rest = t[0]; access_number = int(t[1])
 
 paul@833    108 
 
 paul@833    109 
    t = rest.split(":=")
 
 paul@833    110 
    if len(t) > 1:
 
 paul@833    111 
        attrnames = t[0]; version = int(t[1])
 
 paul@833    112 
    else:
 
 paul@833    113 
        attrnames = rest
 
 paul@833    114 
 
 paul@833    115 
    return path, name, attrnames, version, access_number, invocation
 
 paul@833    116 
 
 paul@0    117 
def encode_location(t):
 
 paul@0    118 
 
 paul@0    119 
    "Encode the general location 't' in a concise form."
 
 paul@0    120 
 
 paul@791    121 
    if t.name is not None and t.version is not None:
 
 paul@791    122 
        return "%s:%s:%d" % (t.path, t.name, t.version)
 
 paul@791    123 
    elif t.name is not None:
 
 paul@791    124 
        return "%s:%s" % (t.path, t.name)
 
 paul@0    125 
    else:
 
 paul@791    126 
        return "%s::%s" % (t.path, t.attrnames)
 
 paul@0    127 
 
 paul@0    128 
def encode_modifiers(modifiers):
 
 paul@0    129 
 
 paul@553    130 
    "Encode assignment and invocation details from 'modifiers'."
 
 paul@0    131 
 
 paul@0    132 
    all_modifiers = []
 
 paul@0    133 
    for t in modifiers:
 
 paul@0    134 
        all_modifiers.append(encode_modifier_term(t))
 
 paul@0    135 
    return "".join(all_modifiers)
 
 paul@0    136 
 
 paul@0    137 
def encode_modifier_term(t):
 
 paul@0    138 
 
 paul@553    139 
    "Encode modifier 't' representing an assignment or an invocation."
 
 paul@0    140 
 
 paul@117    141 
    assignment, invocation = t
 
 paul@553    142 
    if assignment:
 
 paul@553    143 
        return "="
 
 paul@553    144 
    elif invocation is not None:
 
 paul@557    145 
        arguments, keywords = invocation
 
 paul@557    146 
        return "(%d;%s)" % (arguments, ",".join(keywords))
 
 paul@553    147 
    else:
 
 paul@553    148 
        return "_"
 
 paul@0    149 
 
 paul@553    150 
def decode_modifiers(s):
 
 paul@553    151 
 
 paul@553    152 
    "Decode 's' containing modifiers."
 
 paul@553    153 
 
 paul@553    154 
    i = 0
 
 paul@553    155 
    end = len(s)
 
 paul@0    156 
 
 paul@553    157 
    modifiers = []
 
 paul@0    158 
 
 paul@553    159 
    while i < end:
 
 paul@553    160 
        if s[i] == "=":
 
 paul@553    161 
            modifiers.append((True, None))
 
 paul@553    162 
            i += 1
 
 paul@553    163 
        elif s[i] == "(":
 
 paul@557    164 
            j = s.index(";", i)
 
 paul@557    165 
            arguments = int(s[i+1:j])
 
 paul@557    166 
            i = j
 
 paul@553    167 
            j = s.index(")", i)
 
 paul@557    168 
            keywords = s[i+1:j]
 
 paul@557    169 
            keywords = keywords and keywords.split(",") or []
 
 paul@557    170 
            modifiers.append((False, (arguments, keywords)))
 
 paul@553    171 
            i = j + 1
 
 paul@553    172 
        else:
 
 paul@553    173 
            modifiers.append((False, None))
 
 paul@553    174 
            i += 1
 
 paul@553    175 
 
 paul@553    176 
    return modifiers
 
 paul@0    177 
 
 paul@56    178 
 
 paul@56    179 
 
 paul@56    180 
# Test generation functions.
 
 paul@56    181 
 
 paul@56    182 
def get_kinds(all_types):
 
 paul@56    183 
 
 paul@56    184 
    """ 
 
 paul@56    185 
    Return object kind details for 'all_types', being a collection of
 
 paul@56    186 
    references for program types.
 
 paul@56    187 
    """
 
 paul@56    188 
 
 paul@56    189 
    return map(lambda ref: ref.get_kind(), all_types)
 
 paul@56    190 
 
 paul@237    191 
def test_label_for_kind(kind):
 
 paul@56    192 
 
 paul@237    193 
    "Return the label used for 'kind' in test details."
 
 paul@56    194 
 
 paul@237    195 
    return kind == "<instance>" and "instance" or "type"
 
 paul@56    196 
 
 paul@237    197 
def test_label_for_type(ref):
 
 paul@56    198 
 
 paul@237    199 
    "Return the label used for 'ref' in test details."
 
 paul@56    200 
 
 paul@237    201 
    return test_label_for_kind(ref.get_kind())
 
 paul@56    202 
 
 paul@56    203 
 
 paul@56    204 
 
 paul@94    205 
# Instruction representation encoding.
 
 paul@94    206 
 
 paul@94    207 
def encode_instruction(instruction):
 
 paul@94    208 
 
 paul@94    209 
    """
 
 paul@94    210 
    Encode the 'instruction' - a sequence starting with an operation and
 
 paul@94    211 
    followed by arguments, each of which may be an instruction sequence or a
 
 paul@94    212 
    plain value - to produce a function call string representation.
 
 paul@94    213 
    """
 
 paul@94    214 
 
 paul@94    215 
    op = instruction[0]
 
 paul@94    216 
    args = instruction[1:]
 
 paul@94    217 
 
 paul@94    218 
    if args:
 
 paul@94    219 
        a = []
 
 paul@113    220 
        for arg in args:
 
 paul@113    221 
            if isinstance(arg, tuple):
 
 paul@113    222 
                a.append(encode_instruction(arg))
 
 paul@94    223 
            else:
 
 paul@113    224 
                a.append(arg or "{}")
 
 paul@94    225 
        argstr = "(%s)" % ", ".join(a)
 
 paul@94    226 
        return "%s%s" % (op, argstr)
 
 paul@94    227 
    else:
 
 paul@94    228 
        return op
 
 paul@94    229 
 
 paul@94    230 
 
 paul@94    231 
 
 paul@0    232 
# Output program encoding.
 
 paul@0    233 
 
 paul@153    234 
attribute_loading_ops = (
 
 paul@153    235 
    "__load_via_class", "__load_via_object", "__get_class_and_load",
 
 paul@113    236 
    "__check_and_load_via_class", "__check_and_load_via_object", "__check_and_load_via_any",
 
 paul@153    237 
    )
 
 paul@153    238 
 
 paul@989    239 
attribute_ref_lookup_ops = (
 
 paul@989    240 
    "__get_object_attr_ref", "__get_class_attr_ref",
 
 paul@989    241 
    "__check_and_get_object_attr_ref",
 
 paul@113    242 
    )
 
 paul@113    243 
 
 paul@113    244 
typename_ops = (
 
 paul@144    245 
    "__test_common_instance", "__test_common_object", "__test_common_type",
 
 paul@113    246 
    )
 
 paul@113    247 
 
 paul@385    248 
type_ops = (
 
 paul@385    249 
    "__test_specific_instance", "__test_specific_object", "__test_specific_type",
 
 paul@385    250 
    )
 
 paul@385    251 
 
 paul@141    252 
static_ops = (
 
 paul@595    253 
    "__load_static_ignore", "__load_static_replace", "__load_static_test", "<test_context_static>",
 
 paul@141    254 
    )
 
 paul@141    255 
 
 paul@858    256 
accessor_values = (
 
 paul@858    257 
    "<accessor>",
 
 paul@858    258 
    )
 
 paul@858    259 
 
 paul@858    260 
accessor_ops = (
 
 paul@858    261 
    "<accessor>", "<set_accessor>",
 
 paul@858    262 
    )
 
 paul@858    263 
 
 paul@989    264 
attribute_ref_values = (
 
 paul@989    265 
    "<attr_ref>",
 
 paul@989    266 
    )
 
 paul@989    267 
 
 paul@989    268 
attribute_ref_ops = (
 
 paul@989    269 
    "<attr_ref>", "<set_attr_ref>",
 
 paul@989    270 
    )
 
 paul@989    271 
 
 paul@591    272 
context_values = (
 
 paul@591    273 
    "<context>",
 
 paul@591    274 
    )
 
 paul@591    275 
 
 paul@591    276 
context_ops = (
 
 paul@601    277 
    "<context>", "<set_context>", "<test_context_revert>", "<test_context_static>",
 
 paul@591    278 
    )
 
 paul@591    279 
 
 paul@602    280 
context_op_functions = (
 
 paul@602    281 
    "<test_context_revert>", "<test_context_static>",
 
 paul@602    282 
    )
 
 paul@602    283 
 
 paul@989    284 
reference_acting_ops = attribute_ref_lookup_ops + attribute_loading_ops + type_ops + typename_ops
 
 paul@989    285 
attribute_producing_ops = attribute_loading_ops
 
 paul@153    286 
 
 paul@757    287 
attribute_producing_variables = (
 
 paul@757    288 
    "<accessor>", "<context>", "<name>", "<private_context>", "<target_accessor>"
 
 paul@757    289 
    )
 
 paul@757    290 
 
 paul@989    291 
def encode_access_instruction(instruction, subs, accessor_index, context_index,
 
 paul@989    292 
    attribute_ref_index):
 
 paul@113    293 
 
 paul@113    294 
    """
 
 paul@113    295 
    Encode the 'instruction' - a sequence starting with an operation and
 
 paul@113    296 
    followed by arguments, each of which may be an instruction sequence or a
 
 paul@113    297 
    plain value - to produce a function call string representation.
 
 paul@113    298 
 
 paul@113    299 
    The 'subs' parameter defines a mapping of substitutions for special values
 
 paul@113    300 
    used in instructions.
 
 paul@482    301 
 
 paul@858    302 
    The 'accessor_index' parameter defines the position in local accessor
 
 paul@858    303 
    storage for the referenced accessor or affected by an accessor operation.
 
 paul@858    304 
 
 paul@591    305 
    The 'context_index' parameter defines the position in local context storage
 
 paul@591    306 
    for the referenced context or affected by a context operation.
 
 paul@591    307 
 
 paul@989    308 
    The 'attribute_ref_index' parameter defines the position in local attribute
 
 paul@989    309 
    reference storage for a referenced attribute.
 
 paul@989    310 
 
 paul@482    311 
    Return both the encoded instruction and a collection of substituted names.
 
 paul@113    312 
    """
 
 paul@113    313 
 
 paul@113    314 
    op = instruction[0]
 
 paul@113    315 
    args = instruction[1:]
 
 paul@482    316 
    substituted = set()
 
 paul@113    317 
 
 paul@591    318 
    # Encode the arguments.
 
 paul@113    319 
 
 paul@591    320 
    a = []
 
 paul@591    321 
    if args:
 
 paul@153    322 
        converting_op = op
 
 paul@113    323 
        for arg in args:
 
 paul@989    324 
            s, _substituted = encode_access_instruction_arg(arg, subs,
 
 paul@989    325 
                converting_op, accessor_index, context_index, attribute_ref_index)
 
 paul@482    326 
            substituted.update(_substituted)
 
 paul@482    327 
            a.append(s)
 
 paul@153    328 
            converting_op = None
 
 paul@113    329 
 
 paul@591    330 
    # Modify certain arguments.
 
 paul@113    331 
 
 paul@624    332 
    # Convert type name arguments.
 
 paul@113    333 
 
 paul@624    334 
    if op in typename_ops:
 
 paul@624    335 
        a[1] = encode_path(encode_type_attribute(args[1]))
 
 paul@113    336 
 
 paul@591    337 
    # Obtain addresses of type arguments.
 
 paul@591    338 
 
 paul@591    339 
    elif op in type_ops:
 
 paul@591    340 
        a[1] = "&%s" % a[1]
 
 paul@113    341 
 
 paul@591    342 
    # Obtain addresses of static objects.
 
 paul@591    343 
 
 paul@591    344 
    elif op in static_ops:
 
 paul@591    345 
        a[-1] = "&%s" % a[-1]
 
 paul@385    346 
 
 paul@858    347 
    # Add accessor storage information to certain operations.
 
 paul@858    348 
 
 paul@858    349 
    if op in accessor_ops:
 
 paul@858    350 
        a.insert(0, accessor_index)
 
 paul@858    351 
 
 paul@989    352 
    # Add attribute reference storage information to certain operations.
 
 paul@989    353 
 
 paul@989    354 
    if op in attribute_ref_ops:
 
 paul@989    355 
        a.insert(0, attribute_ref_index)
 
 paul@989    356 
 
 paul@591    357 
    # Add context storage information to certain operations.
 
 paul@385    358 
 
 paul@595    359 
    if op in context_ops:
 
 paul@591    360 
        a.insert(0, context_index)
 
 paul@141    361 
 
 paul@602    362 
    # Add the local context array to certain operations.
 
 paul@602    363 
 
 paul@602    364 
    if op in context_op_functions:
 
 paul@602    365 
        a.append("__tmp_contexts")
 
 paul@602    366 
 
 paul@591    367 
    # Define any argument string.
 
 paul@141    368 
 
 paul@591    369 
    if a:
 
 paul@491    370 
        argstr = "(%s)" % ", ".join(map(str, a))
 
 paul@591    371 
    else:
 
 paul@591    372 
        argstr = ""
 
 paul@113    373 
 
 paul@113    374 
    # Substitute the first element of the instruction, which may not be an
 
 paul@113    375 
    # operation at all.
 
 paul@113    376 
 
 paul@144    377 
    if subs.has_key(op):
 
 paul@482    378 
        substituted.add(op)
 
 paul@498    379 
 
 paul@498    380 
        # Break accessor initialisation into initialisation and value-yielding
 
 paul@498    381 
        # parts:
 
 paul@498    382 
 
 paul@498    383 
        if op == "<set_accessor>" and isinstance(a[0], InstructionSequence):
 
 paul@498    384 
            ops = []
 
 paul@498    385 
            ops += a[0].get_init_instructions()
 
 paul@498    386 
            ops.append("%s(%s)" % (subs[op], a[0].get_value_instruction()))
 
 paul@498    387 
            return ", ".join(map(str, ops)), substituted
 
 paul@498    388 
 
 paul@144    389 
        op = subs[op]
 
 paul@498    390 
 
 paul@144    391 
    elif not args:
 
 paul@144    392 
        op = "&%s" % encode_path(op)
 
 paul@144    393 
 
 paul@482    394 
    return "%s%s" % (op, argstr), substituted
 
 paul@113    395 
 
 paul@989    396 
def encode_access_instruction_arg(arg, subs, op, accessor_index, context_index, attribute_ref_index):
 
 paul@113    397 
 
 paul@482    398 
    """
 
 paul@591    399 
    Encode 'arg' using 'subs' to define substitutions, 'op' to indicate the
 
 paul@858    400 
    operation to which the argument belongs, and with 'accessor_index' and
 
 paul@858    401 
    'context_index' indicating any affected accessor and context storage.
 
 paul@591    402 
 
 paul@591    403 
    Return a tuple containing the encoded form of 'arg' along with a collection
 
 paul@591    404 
    of any substituted values.
 
 paul@482    405 
    """
 
 paul@113    406 
 
 paul@113    407 
    if isinstance(arg, tuple):
 
 paul@989    408 
        encoded, substituted = encode_access_instruction(arg, subs,
 
 paul@989    409 
            accessor_index, context_index, attribute_ref_index)
 
 paul@757    410 
        return attribute_to_reference(op, arg[0], encoded, substituted)
 
 paul@113    411 
 
 paul@113    412 
    # Special values only need replacing, not encoding.
 
 paul@113    413 
 
 paul@113    414 
    elif subs.has_key(arg):
 
 paul@591    415 
 
 paul@591    416 
        # Handle values modified by storage details.
 
 paul@591    417 
 
 paul@858    418 
        if arg in accessor_values or arg in context_values:
 
 paul@757    419 
            encoded = "%s(%s)" % (subs.get(arg), context_index)
 
 paul@989    420 
        elif arg in attribute_ref_values:
 
 paul@989    421 
            encoded = "%s(%s)" % (subs.get(arg), attribute_ref_index)
 
 paul@591    422 
        else:
 
 paul@757    423 
            encoded = subs.get(arg)
 
 paul@757    424 
 
 paul@757    425 
        substituted = set([arg])
 
 paul@757    426 
        return attribute_to_reference(op, arg, encoded, substituted)
 
 paul@113    427 
 
 paul@258    428 
    # Convert static references to the appropriate type.
 
 paul@258    429 
 
 paul@757    430 
    elif op and op in reference_acting_ops and \
 
 paul@757    431 
         arg not in attribute_producing_variables:
 
 paul@757    432 
 
 paul@482    433 
        return "&%s" % encode_path(arg), set()
 
 paul@258    434 
 
 paul@113    435 
    # Other values may need encoding.
 
 paul@113    436 
 
 paul@113    437 
    else:
 
 paul@482    438 
        return encode_path(arg), set()
 
 paul@113    439 
 
 paul@757    440 
def attribute_to_reference(op, arg, encoded, substituted):
 
 paul@757    441 
 
 paul@757    442 
    # Convert attribute results to references where required.
 
 paul@757    443 
 
 paul@757    444 
    if op and op in reference_acting_ops and (
 
 paul@757    445 
       arg in attribute_producing_ops or
 
 paul@757    446 
       arg in attribute_producing_variables):
 
 paul@757    447 
 
 paul@757    448 
        return "__VALUE(%s)" % encoded, substituted
 
 paul@757    449 
    else:
 
 paul@757    450 
        return encoded, substituted
 
 paul@757    451 
 
 paul@0    452 
def encode_function_pointer(path):
 
 paul@0    453 
 
 paul@0    454 
    "Encode 'path' as a reference to an output program function."
 
 paul@0    455 
 
 paul@0    456 
    return "__fn_%s" % encode_path(path)
 
 paul@0    457 
 
 paul@0    458 
def encode_instantiator_pointer(path):
 
 paul@0    459 
 
 paul@0    460 
    "Encode 'path' as a reference to an output program instantiator."
 
 paul@0    461 
 
 paul@0    462 
    return "__new_%s" % encode_path(path)
 
 paul@0    463 
 
 paul@491    464 
def encode_instructions(instructions):
 
 paul@491    465 
 
 paul@491    466 
    "Encode 'instructions' as a sequence."
 
 paul@491    467 
 
 paul@491    468 
    if len(instructions) == 1:
 
 paul@491    469 
        return instructions[0]
 
 paul@491    470 
    else:
 
 paul@491    471 
        return "(\n%s\n)" % ",\n".join(instructions)
 
 paul@491    472 
 
 paul@136    473 
def encode_literal_constant(n):
 
 paul@136    474 
 
 paul@136    475 
    "Encode a name for the literal constant with the number 'n'."
 
 paul@136    476 
 
 paul@609    477 
    return "__const%s" % n
 
 paul@136    478 
 
 paul@378    479 
def encode_literal_constant_size(value):
 
 paul@378    480 
 
 paul@378    481 
    "Encode a size for the literal constant with the given 'value'."
 
 paul@378    482 
 
 paul@378    483 
    if isinstance(value, basestring):
 
 paul@378    484 
        return len(value)
 
 paul@378    485 
    else:
 
 paul@378    486 
        return 0
 
 paul@378    487 
 
 paul@136    488 
def encode_literal_constant_member(value):
 
 paul@136    489 
 
 paul@136    490 
    "Encode the member name for the 'value' in the final program."
 
 paul@136    491 
 
 paul@136    492 
    return "%svalue" % value.__class__.__name__
 
 paul@136    493 
 
 paul@136    494 
def encode_literal_constant_value(value):
 
 paul@136    495 
 
 paul@136    496 
    "Encode the given 'value' in the final program."
 
 paul@136    497 
 
 paul@136    498 
    if isinstance(value, (int, float)):
 
 paul@136    499 
        return str(value)
 
 paul@136    500 
    else:
 
 paul@451    501 
        l = []
 
 paul@451    502 
 
 paul@451    503 
        # Encode characters including non-ASCII ones.
 
 paul@451    504 
 
 paul@451    505 
        for c in str(value):
 
 paul@451    506 
            if c == '"': l.append('\\"')
 
 paul@451    507 
            elif c == '\n': l.append('\\n')
 
 paul@451    508 
            elif c == '\t': l.append('\\t')
 
 paul@451    509 
            elif c == '\r': l.append('\\r')
 
 paul@512    510 
            elif c == '\\': l.append('\\\\')
 
 paul@451    511 
            elif 0x20 <= ord(c) < 0x80: l.append(c)
 
 paul@451    512 
            else: l.append("\\x%02x" % ord(c))
 
 paul@451    513 
 
 paul@451    514 
        return '"%s"' % "".join(l)
 
 paul@136    515 
 
 paul@283    516 
def encode_literal_data_initialiser(style):
 
 paul@283    517 
 
 paul@283    518 
    """
 
 paul@283    519 
    Encode a reference to a function populating the data for a literal having
 
 paul@283    520 
    the given 'style' ("mapping" or "sequence").
 
 paul@283    521 
    """
 
 paul@283    522 
 
 paul@283    523 
    return "__newdata_%s" % style
 
 paul@283    524 
 
 paul@159    525 
def encode_literal_instantiator(path):
 
 paul@159    526 
 
 paul@159    527 
    """
 
 paul@159    528 
    Encode a reference to an instantiator for a literal having the given 'path'.
 
 paul@159    529 
    """
 
 paul@159    530 
 
 paul@159    531 
    return "__newliteral_%s" % encode_path(path)
 
 paul@159    532 
 
 paul@136    533 
def encode_literal_reference(n):
 
 paul@136    534 
 
 paul@136    535 
    "Encode a reference to a literal constant with the number 'n'."
 
 paul@136    536 
 
 paul@609    537 
    return "__constvalue%s" % n
 
 paul@136    538 
 
 paul@850    539 
def encode_trailing_area(path):
 
 paul@850    540 
 
 paul@850    541 
    """
 
 paul@850    542 
    Encode any reference to trailing data members for instances of the type
 
 paul@850    543 
    given by 'path'.
 
 paul@850    544 
    """
 
 paul@850    545 
 
 paul@850    546 
    return "__TRAILING_%s" % encode_path(path)
 
 paul@850    547 
 
 paul@512    548 
 
 paul@512    549 
 
 paul@340    550 
# Track all encoded paths, detecting and avoiding conflicts.
 
 paul@340    551 
 
 paul@340    552 
all_encoded_paths = {}
 
 paul@340    553 
 
 paul@0    554 
def encode_path(path):
 
 paul@0    555 
 
 paul@0    556 
    "Encode 'path' as an output program object, translating special symbols."
 
 paul@0    557 
 
 paul@0    558 
    if path in reserved_words:
 
 paul@0    559 
        return "__%s" % path
 
 paul@0    560 
    else:
 
 paul@340    561 
        part_encoded = path.replace("#", "__").replace("$", "__")
 
 paul@349    562 
 
 paul@349    563 
        if "." not in path:
 
 paul@349    564 
            return part_encoded
 
 paul@349    565 
 
 paul@340    566 
        encoded = part_encoded.replace(".", "_")
 
 paul@340    567 
 
 paul@340    568 
        # Test for a conflict with the encoding of a different path, re-encoding
 
 paul@340    569 
        # if necessary.
 
 paul@340    570 
 
 paul@340    571 
        previous = all_encoded_paths.get(encoded)
 
 paul@340    572 
        replacement = "_"
 
 paul@340    573 
 
 paul@340    574 
        while previous:
 
 paul@340    575 
            if path == previous:
 
 paul@340    576 
                return encoded
 
 paul@340    577 
            replacement += "_"
 
 paul@340    578 
            encoded = part_encoded.replace(".", replacement)
 
 paul@340    579 
            previous = all_encoded_paths.get(encoded)
 
 paul@340    580 
 
 paul@340    581 
        # Store any new or re-encoded path.
 
 paul@340    582 
 
 paul@340    583 
        all_encoded_paths[encoded] = path
 
 paul@340    584 
        return encoded
 
 paul@0    585 
 
 paul@623    586 
def encode_code(name):
 
 paul@623    587 
 
 paul@623    588 
    "Encode 'name' as an attribute code indicator."
 
 paul@623    589 
 
 paul@623    590 
    return "__ATTRCODE(%s)" % encode_path(name)
 
 paul@623    591 
 
 paul@623    592 
def encode_pcode(name):
 
 paul@623    593 
 
 paul@623    594 
    "Encode 'name' as an parameter code indicator."
 
 paul@623    595 
 
 paul@623    596 
    return "__PARAMCODE(%s)" % encode_path(name)
 
 paul@623    597 
 
 paul@623    598 
def encode_pos(name):
 
 paul@623    599 
 
 paul@623    600 
    "Encode 'name' as an attribute position indicator."
 
 paul@623    601 
 
 paul@623    602 
    return "__ATTRPOS(%s)" % encode_path(name)
 
 paul@623    603 
 
 paul@623    604 
def encode_ppos(name):
 
 paul@623    605 
 
 paul@623    606 
    "Encode 'name' as an parameter position indicator."
 
 paul@623    607 
 
 paul@623    608 
    return "__PARAMPOS(%s)" % encode_path(name)
 
 paul@623    609 
 
 paul@136    610 
def encode_predefined_reference(path):
 
 paul@136    611 
 
 paul@136    612 
    "Encode a reference to a predefined constant value for 'path'."
 
 paul@136    613 
 
 paul@136    614 
    return "__predefined_%s" % encode_path(path)
 
 paul@136    615 
 
 paul@150    616 
def encode_size(kind, path=None):
 
 paul@150    617 
 
 paul@150    618 
    """
 
 paul@150    619 
    Encode a structure size reference for the given 'kind' of structure, with
 
 paul@150    620 
    'path' indicating a specific structure name.
 
 paul@150    621 
    """
 
 paul@150    622 
 
 paul@150    623 
    return "__%ssize%s" % (structure_size_prefixes.get(kind, kind), path and "_%s" % encode_path(path) or "")
 
 paul@150    624 
 
 paul@0    625 
def encode_symbol(symbol_type, path=None):
 
 paul@0    626 
 
 paul@0    627 
    "Encode a symbol with the given 'symbol_type' and optional 'path'."
 
 paul@0    628 
 
 paul@0    629 
    return "__%s%s" % (symbol_type, path and "_%s" % encode_path(path) or "")
 
 paul@0    630 
 
 paul@150    631 
def encode_tablename(kind, path):
 
 paul@150    632 
 
 paul@150    633 
    """
 
 paul@150    634 
    Encode a table reference for the given 'kind' of table structure, indicating
 
 paul@150    635 
    a 'path' for the specific object concerned.
 
 paul@150    636 
    """
 
 paul@150    637 
 
 paul@150    638 
    return "__%sTable_%s" % (table_name_prefixes[kind], encode_path(path))
 
 paul@150    639 
 
 paul@131    640 
def encode_type_attribute(path):
 
 paul@131    641 
 
 paul@131    642 
    "Encode the special type attribute for 'path'."
 
 paul@131    643 
 
 paul@131    644 
    return "#%s" % path
 
 paul@131    645 
 
 paul@318    646 
def decode_type_attribute(s):
 
 paul@318    647 
 
 paul@318    648 
    "Decode the special type attribute 's'."
 
 paul@318    649 
 
 paul@318    650 
    return s[1:]
 
 paul@318    651 
 
 paul@318    652 
def is_type_attribute(s):
 
 paul@318    653 
 
 paul@318    654 
    "Return whether 's' is a type attribute name."
 
 paul@318    655 
 
 paul@318    656 
    return s.startswith("#")
 
 paul@318    657 
 
 paul@56    658 
 
 paul@56    659 
 
 paul@150    660 
# A mapping from kinds to structure size reference prefixes.
 
 paul@150    661 
 
 paul@150    662 
structure_size_prefixes = {
 
 paul@150    663 
    "<class>" : "c",
 
 paul@150    664 
    "<module>" : "m",
 
 paul@150    665 
    "<instance>" : "i"
 
 paul@150    666 
    }
 
 paul@150    667 
 
 paul@150    668 
# A mapping from kinds to table name prefixes.
 
 paul@150    669 
 
 paul@150    670 
table_name_prefixes = {
 
 paul@150    671 
    "<class>" : "Class",
 
 paul@150    672 
    "<function>" : "Function",
 
 paul@150    673 
    "<module>" : "Module",
 
 paul@150    674 
    "<instance>" : "Instance"
 
 paul@150    675 
    }
 
 paul@150    676 
 
 paul@150    677 
 
 paul@150    678 
 
 paul@0    679 
# Output language reserved words.
 
 paul@0    680 
 
 paul@0    681 
reserved_words = [
 
 paul@0    682 
    "break", "char", "const", "continue",
 
 paul@0    683 
    "default", "double", "else",
 
 paul@0    684 
    "float", "for",
 
 paul@0    685 
    "if", "int", "long",
 
 paul@0    686 
    "NULL",
 
 paul@0    687 
    "return", "struct",
 
 paul@0    688 
    "typedef",
 
 paul@0    689 
    "void", "while",
 
 paul@0    690 
    ]
 
 paul@0    691 
 
 paul@0    692 
# vim: tabstop=4 expandtab shiftwidth=4