Lichen

Annotated encoders.py

964:e98699004465 989:23edd8f55d40
1037:2ef21d25c6e5
860:21f92740b225
5 months ago Paul Boddie file changeset files shortlog Merged changes from the default branch. trailing-data
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@153    236 
    )
 
 paul@153    237 
 
 paul@153    238 
attribute_ops = attribute_loading_ops + (
 
 paul@815    239 
    "__store_via_class", "__store_via_object",
 
 paul@113    240 
    )
 
 paul@113    241 
 
 paul@153    242 
checked_loading_ops = (
 
 paul@113    243 
    "__check_and_load_via_class", "__check_and_load_via_object", "__check_and_load_via_any",
 
 paul@153    244 
    )
 
 paul@153    245 
 
 paul@153    246 
checked_ops = checked_loading_ops + (
 
 paul@113    247 
    "__check_and_store_via_class", "__check_and_store_via_object", "__check_and_store_via_any",
 
 paul@113    248 
    )
 
 paul@113    249 
 
 paul@113    250 
typename_ops = (
 
 paul@144    251 
    "__test_common_instance", "__test_common_object", "__test_common_type",
 
 paul@113    252 
    )
 
 paul@113    253 
 
 paul@385    254 
type_ops = (
 
 paul@385    255 
    "__test_specific_instance", "__test_specific_object", "__test_specific_type",
 
 paul@385    256 
    )
 
 paul@385    257 
 
 paul@141    258 
static_ops = (
 
 paul@595    259 
    "__load_static_ignore", "__load_static_replace", "__load_static_test", "<test_context_static>",
 
 paul@141    260 
    )
 
 paul@141    261 
 
 paul@858    262 
accessor_values = (
 
 paul@858    263 
    "<accessor>",
 
 paul@858    264 
    )
 
 paul@858    265 
 
 paul@858    266 
accessor_ops = (
 
 paul@858    267 
    "<accessor>", "<set_accessor>",
 
 paul@858    268 
    )
 
 paul@858    269 
 
 paul@591    270 
context_values = (
 
 paul@591    271 
    "<context>",
 
 paul@591    272 
    )
 
 paul@591    273 
 
 paul@591    274 
context_ops = (
 
 paul@601    275 
    "<context>", "<set_context>", "<test_context_revert>", "<test_context_static>",
 
 paul@591    276 
    )
 
 paul@591    277 
 
 paul@602    278 
context_op_functions = (
 
 paul@602    279 
    "<test_context_revert>", "<test_context_static>",
 
 paul@602    280 
    )
 
 paul@602    281 
 
 paul@757    282 
reference_acting_ops = attribute_ops + checked_ops + type_ops + typename_ops
 
 paul@153    283 
attribute_producing_ops = attribute_loading_ops + checked_loading_ops
 
 paul@153    284 
 
 paul@757    285 
attribute_producing_variables = (
 
 paul@757    286 
    "<accessor>", "<context>", "<name>", "<private_context>", "<target_accessor>"
 
 paul@757    287 
    )
 
 paul@757    288 
 
 paul@858    289 
def encode_access_instruction(instruction, subs, accessor_index, context_index):
 
 paul@113    290 
 
 paul@113    291 
    """
 
 paul@113    292 
    Encode the 'instruction' - a sequence starting with an operation and
 
 paul@113    293 
    followed by arguments, each of which may be an instruction sequence or a
 
 paul@113    294 
    plain value - to produce a function call string representation.
 
 paul@113    295 
 
 paul@113    296 
    The 'subs' parameter defines a mapping of substitutions for special values
 
 paul@113    297 
    used in instructions.
 
 paul@482    298 
 
 paul@858    299 
    The 'accessor_index' parameter defines the position in local accessor
 
 paul@858    300 
    storage for the referenced accessor or affected by an accessor operation.
 
 paul@858    301 
 
 paul@591    302 
    The 'context_index' parameter defines the position in local context storage
 
 paul@591    303 
    for the referenced context or affected by a context operation.
 
 paul@591    304 
 
 paul@482    305 
    Return both the encoded instruction and a collection of substituted names.
 
 paul@113    306 
    """
 
 paul@113    307 
 
 paul@113    308 
    op = instruction[0]
 
 paul@113    309 
    args = instruction[1:]
 
 paul@482    310 
    substituted = set()
 
 paul@113    311 
 
 paul@591    312 
    # Encode the arguments.
 
 paul@113    313 
 
 paul@591    314 
    a = []
 
 paul@591    315 
    if args:
 
 paul@153    316 
        converting_op = op
 
 paul@113    317 
        for arg in args:
 
 paul@858    318 
            s, _substituted = encode_access_instruction_arg(arg, subs, converting_op, accessor_index, context_index)
 
 paul@482    319 
            substituted.update(_substituted)
 
 paul@482    320 
            a.append(s)
 
 paul@153    321 
            converting_op = None
 
 paul@113    322 
 
 paul@591    323 
    # Modify certain arguments.
 
 paul@113    324 
 
 paul@624    325 
    # Convert type name arguments.
 
 paul@113    326 
 
 paul@624    327 
    if op in typename_ops:
 
 paul@624    328 
        a[1] = encode_path(encode_type_attribute(args[1]))
 
 paul@113    329 
 
 paul@591    330 
    # Obtain addresses of type arguments.
 
 paul@591    331 
 
 paul@591    332 
    elif op in type_ops:
 
 paul@591    333 
        a[1] = "&%s" % a[1]
 
 paul@113    334 
 
 paul@591    335 
    # Obtain addresses of static objects.
 
 paul@591    336 
 
 paul@591    337 
    elif op in static_ops:
 
 paul@591    338 
        a[-1] = "&%s" % a[-1]
 
 paul@385    339 
 
 paul@858    340 
    # Add accessor storage information to certain operations.
 
 paul@858    341 
 
 paul@858    342 
    if op in accessor_ops:
 
 paul@858    343 
        a.insert(0, accessor_index)
 
 paul@858    344 
 
 paul@591    345 
    # Add context storage information to certain operations.
 
 paul@385    346 
 
 paul@595    347 
    if op in context_ops:
 
 paul@591    348 
        a.insert(0, context_index)
 
 paul@141    349 
 
 paul@602    350 
    # Add the local context array to certain operations.
 
 paul@602    351 
 
 paul@602    352 
    if op in context_op_functions:
 
 paul@602    353 
        a.append("__tmp_contexts")
 
 paul@602    354 
 
 paul@591    355 
    # Define any argument string.
 
 paul@141    356 
 
 paul@591    357 
    if a:
 
 paul@491    358 
        argstr = "(%s)" % ", ".join(map(str, a))
 
 paul@591    359 
    else:
 
 paul@591    360 
        argstr = ""
 
 paul@113    361 
 
 paul@113    362 
    # Substitute the first element of the instruction, which may not be an
 
 paul@113    363 
    # operation at all.
 
 paul@113    364 
 
 paul@144    365 
    if subs.has_key(op):
 
 paul@482    366 
        substituted.add(op)
 
 paul@498    367 
 
 paul@498    368 
        # Break accessor initialisation into initialisation and value-yielding
 
 paul@498    369 
        # parts:
 
 paul@498    370 
 
 paul@498    371 
        if op == "<set_accessor>" and isinstance(a[0], InstructionSequence):
 
 paul@498    372 
            ops = []
 
 paul@498    373 
            ops += a[0].get_init_instructions()
 
 paul@498    374 
            ops.append("%s(%s)" % (subs[op], a[0].get_value_instruction()))
 
 paul@498    375 
            return ", ".join(map(str, ops)), substituted
 
 paul@498    376 
 
 paul@144    377 
        op = subs[op]
 
 paul@498    378 
 
 paul@144    379 
    elif not args:
 
 paul@144    380 
        op = "&%s" % encode_path(op)
 
 paul@144    381 
 
 paul@482    382 
    return "%s%s" % (op, argstr), substituted
 
 paul@113    383 
 
 paul@858    384 
def encode_access_instruction_arg(arg, subs, op, accessor_index, context_index):
 
 paul@113    385 
 
 paul@482    386 
    """
 
 paul@591    387 
    Encode 'arg' using 'subs' to define substitutions, 'op' to indicate the
 
 paul@858    388 
    operation to which the argument belongs, and with 'accessor_index' and
 
 paul@858    389 
    'context_index' indicating any affected accessor and context storage.
 
 paul@591    390 
 
 paul@591    391 
    Return a tuple containing the encoded form of 'arg' along with a collection
 
 paul@591    392 
    of any substituted values.
 
 paul@482    393 
    """
 
 paul@113    394 
 
 paul@113    395 
    if isinstance(arg, tuple):
 
 paul@858    396 
        encoded, substituted = encode_access_instruction(arg, subs, accessor_index, context_index)
 
 paul@757    397 
        return attribute_to_reference(op, arg[0], encoded, substituted)
 
 paul@113    398 
 
 paul@113    399 
    # Special values only need replacing, not encoding.
 
 paul@113    400 
 
 paul@113    401 
    elif subs.has_key(arg):
 
 paul@591    402 
 
 paul@591    403 
        # Handle values modified by storage details.
 
 paul@591    404 
 
 paul@858    405 
        if arg in accessor_values or arg in context_values:
 
 paul@757    406 
            encoded = "%s(%s)" % (subs.get(arg), context_index)
 
 paul@591    407 
        else:
 
 paul@757    408 
            encoded = subs.get(arg)
 
 paul@757    409 
 
 paul@757    410 
        substituted = set([arg])
 
 paul@757    411 
        return attribute_to_reference(op, arg, encoded, substituted)
 
 paul@113    412 
 
 paul@258    413 
    # Convert static references to the appropriate type.
 
 paul@258    414 
 
 paul@757    415 
    elif op and op in reference_acting_ops and \
 
 paul@757    416 
         arg not in attribute_producing_variables:
 
 paul@757    417 
 
 paul@482    418 
        return "&%s" % encode_path(arg), set()
 
 paul@258    419 
 
 paul@113    420 
    # Other values may need encoding.
 
 paul@113    421 
 
 paul@113    422 
    else:
 
 paul@482    423 
        return encode_path(arg), set()
 
 paul@113    424 
 
 paul@757    425 
def attribute_to_reference(op, arg, encoded, substituted):
 
 paul@757    426 
 
 paul@757    427 
    # Convert attribute results to references where required.
 
 paul@757    428 
 
 paul@757    429 
    if op and op in reference_acting_ops and (
 
 paul@757    430 
       arg in attribute_producing_ops or
 
 paul@757    431 
       arg in attribute_producing_variables):
 
 paul@757    432 
 
 paul@757    433 
        return "__VALUE(%s)" % encoded, substituted
 
 paul@757    434 
    else:
 
 paul@757    435 
        return encoded, substituted
 
 paul@757    436 
 
 paul@0    437 
def encode_function_pointer(path):
 
 paul@0    438 
 
 paul@0    439 
    "Encode 'path' as a reference to an output program function."
 
 paul@0    440 
 
 paul@0    441 
    return "__fn_%s" % encode_path(path)
 
 paul@0    442 
 
 paul@0    443 
def encode_instantiator_pointer(path):
 
 paul@0    444 
 
 paul@0    445 
    "Encode 'path' as a reference to an output program instantiator."
 
 paul@0    446 
 
 paul@0    447 
    return "__new_%s" % encode_path(path)
 
 paul@0    448 
 
 paul@491    449 
def encode_instructions(instructions):
 
 paul@491    450 
 
 paul@491    451 
    "Encode 'instructions' as a sequence."
 
 paul@491    452 
 
 paul@491    453 
    if len(instructions) == 1:
 
 paul@491    454 
        return instructions[0]
 
 paul@491    455 
    else:
 
 paul@491    456 
        return "(\n%s\n)" % ",\n".join(instructions)
 
 paul@491    457 
 
 paul@136    458 
def encode_literal_constant(n):
 
 paul@136    459 
 
 paul@136    460 
    "Encode a name for the literal constant with the number 'n'."
 
 paul@136    461 
 
 paul@609    462 
    return "__const%s" % n
 
 paul@136    463 
 
 paul@378    464 
def encode_literal_constant_size(value):
 
 paul@378    465 
 
 paul@378    466 
    "Encode a size for the literal constant with the given 'value'."
 
 paul@378    467 
 
 paul@378    468 
    if isinstance(value, basestring):
 
 paul@378    469 
        return len(value)
 
 paul@378    470 
    else:
 
 paul@378    471 
        return 0
 
 paul@378    472 
 
 paul@136    473 
def encode_literal_constant_member(value):
 
 paul@136    474 
 
 paul@136    475 
    "Encode the member name for the 'value' in the final program."
 
 paul@136    476 
 
 paul@136    477 
    return "%svalue" % value.__class__.__name__
 
 paul@136    478 
 
 paul@136    479 
def encode_literal_constant_value(value):
 
 paul@136    480 
 
 paul@136    481 
    "Encode the given 'value' in the final program."
 
 paul@136    482 
 
 paul@136    483 
    if isinstance(value, (int, float)):
 
 paul@136    484 
        return str(value)
 
 paul@136    485 
    else:
 
 paul@451    486 
        l = []
 
 paul@451    487 
 
 paul@451    488 
        # Encode characters including non-ASCII ones.
 
 paul@451    489 
 
 paul@451    490 
        for c in str(value):
 
 paul@451    491 
            if c == '"': l.append('\\"')
 
 paul@451    492 
            elif c == '\n': l.append('\\n')
 
 paul@451    493 
            elif c == '\t': l.append('\\t')
 
 paul@451    494 
            elif c == '\r': l.append('\\r')
 
 paul@512    495 
            elif c == '\\': l.append('\\\\')
 
 paul@451    496 
            elif 0x20 <= ord(c) < 0x80: l.append(c)
 
 paul@451    497 
            else: l.append("\\x%02x" % ord(c))
 
 paul@451    498 
 
 paul@451    499 
        return '"%s"' % "".join(l)
 
 paul@136    500 
 
 paul@283    501 
def encode_literal_data_initialiser(style):
 
 paul@283    502 
 
 paul@283    503 
    """
 
 paul@283    504 
    Encode a reference to a function populating the data for a literal having
 
 paul@283    505 
    the given 'style' ("mapping" or "sequence").
 
 paul@283    506 
    """
 
 paul@283    507 
 
 paul@283    508 
    return "__newdata_%s" % style
 
 paul@283    509 
 
 paul@159    510 
def encode_literal_instantiator(path):
 
 paul@159    511 
 
 paul@159    512 
    """
 
 paul@159    513 
    Encode a reference to an instantiator for a literal having the given 'path'.
 
 paul@159    514 
    """
 
 paul@159    515 
 
 paul@159    516 
    return "__newliteral_%s" % encode_path(path)
 
 paul@159    517 
 
 paul@136    518 
def encode_literal_reference(n):
 
 paul@136    519 
 
 paul@136    520 
    "Encode a reference to a literal constant with the number 'n'."
 
 paul@136    521 
 
 paul@609    522 
    return "__constvalue%s" % n
 
 paul@136    523 
 
 paul@850    524 
def encode_trailing_area(path):
 
 paul@850    525 
 
 paul@850    526 
    """
 
 paul@850    527 
    Encode any reference to trailing data members for instances of the type
 
 paul@850    528 
    given by 'path'.
 
 paul@850    529 
    """
 
 paul@850    530 
 
 paul@850    531 
    return "__TRAILING_%s" % encode_path(path)
 
 paul@850    532 
 
 paul@512    533 
 
 paul@512    534 
 
 paul@340    535 
# Track all encoded paths, detecting and avoiding conflicts.
 
 paul@340    536 
 
 paul@340    537 
all_encoded_paths = {}
 
 paul@340    538 
 
 paul@0    539 
def encode_path(path):
 
 paul@0    540 
 
 paul@0    541 
    "Encode 'path' as an output program object, translating special symbols."
 
 paul@0    542 
 
 paul@0    543 
    if path in reserved_words:
 
 paul@0    544 
        return "__%s" % path
 
 paul@0    545 
    else:
 
 paul@340    546 
        part_encoded = path.replace("#", "__").replace("$", "__")
 
 paul@349    547 
 
 paul@349    548 
        if "." not in path:
 
 paul@349    549 
            return part_encoded
 
 paul@349    550 
 
 paul@340    551 
        encoded = part_encoded.replace(".", "_")
 
 paul@340    552 
 
 paul@340    553 
        # Test for a conflict with the encoding of a different path, re-encoding
 
 paul@340    554 
        # if necessary.
 
 paul@340    555 
 
 paul@340    556 
        previous = all_encoded_paths.get(encoded)
 
 paul@340    557 
        replacement = "_"
 
 paul@340    558 
 
 paul@340    559 
        while previous:
 
 paul@340    560 
            if path == previous:
 
 paul@340    561 
                return encoded
 
 paul@340    562 
            replacement += "_"
 
 paul@340    563 
            encoded = part_encoded.replace(".", replacement)
 
 paul@340    564 
            previous = all_encoded_paths.get(encoded)
 
 paul@340    565 
 
 paul@340    566 
        # Store any new or re-encoded path.
 
 paul@340    567 
 
 paul@340    568 
        all_encoded_paths[encoded] = path
 
 paul@340    569 
        return encoded
 
 paul@0    570 
 
 paul@623    571 
def encode_code(name):
 
 paul@623    572 
 
 paul@623    573 
    "Encode 'name' as an attribute code indicator."
 
 paul@623    574 
 
 paul@623    575 
    return "__ATTRCODE(%s)" % encode_path(name)
 
 paul@623    576 
 
 paul@623    577 
def encode_pcode(name):
 
 paul@623    578 
 
 paul@623    579 
    "Encode 'name' as an parameter code indicator."
 
 paul@623    580 
 
 paul@623    581 
    return "__PARAMCODE(%s)" % encode_path(name)
 
 paul@623    582 
 
 paul@623    583 
def encode_pos(name):
 
 paul@623    584 
 
 paul@623    585 
    "Encode 'name' as an attribute position indicator."
 
 paul@623    586 
 
 paul@623    587 
    return "__ATTRPOS(%s)" % encode_path(name)
 
 paul@623    588 
 
 paul@623    589 
def encode_ppos(name):
 
 paul@623    590 
 
 paul@623    591 
    "Encode 'name' as an parameter position indicator."
 
 paul@623    592 
 
 paul@623    593 
    return "__PARAMPOS(%s)" % encode_path(name)
 
 paul@623    594 
 
 paul@136    595 
def encode_predefined_reference(path):
 
 paul@136    596 
 
 paul@136    597 
    "Encode a reference to a predefined constant value for 'path'."
 
 paul@136    598 
 
 paul@136    599 
    return "__predefined_%s" % encode_path(path)
 
 paul@136    600 
 
 paul@150    601 
def encode_size(kind, path=None):
 
 paul@150    602 
 
 paul@150    603 
    """
 
 paul@150    604 
    Encode a structure size reference for the given 'kind' of structure, with
 
 paul@150    605 
    'path' indicating a specific structure name.
 
 paul@150    606 
    """
 
 paul@150    607 
 
 paul@150    608 
    return "__%ssize%s" % (structure_size_prefixes.get(kind, kind), path and "_%s" % encode_path(path) or "")
 
 paul@150    609 
 
 paul@0    610 
def encode_symbol(symbol_type, path=None):
 
 paul@0    611 
 
 paul@0    612 
    "Encode a symbol with the given 'symbol_type' and optional 'path'."
 
 paul@0    613 
 
 paul@0    614 
    return "__%s%s" % (symbol_type, path and "_%s" % encode_path(path) or "")
 
 paul@0    615 
 
 paul@150    616 
def encode_tablename(kind, path):
 
 paul@150    617 
 
 paul@150    618 
    """
 
 paul@150    619 
    Encode a table reference for the given 'kind' of table structure, indicating
 
 paul@150    620 
    a 'path' for the specific object concerned.
 
 paul@150    621 
    """
 
 paul@150    622 
 
 paul@150    623 
    return "__%sTable_%s" % (table_name_prefixes[kind], encode_path(path))
 
 paul@150    624 
 
 paul@131    625 
def encode_type_attribute(path):
 
 paul@131    626 
 
 paul@131    627 
    "Encode the special type attribute for 'path'."
 
 paul@131    628 
 
 paul@131    629 
    return "#%s" % path
 
 paul@131    630 
 
 paul@318    631 
def decode_type_attribute(s):
 
 paul@318    632 
 
 paul@318    633 
    "Decode the special type attribute 's'."
 
 paul@318    634 
 
 paul@318    635 
    return s[1:]
 
 paul@318    636 
 
 paul@318    637 
def is_type_attribute(s):
 
 paul@318    638 
 
 paul@318    639 
    "Return whether 's' is a type attribute name."
 
 paul@318    640 
 
 paul@318    641 
    return s.startswith("#")
 
 paul@318    642 
 
 paul@56    643 
 
 paul@56    644 
 
 paul@150    645 
# A mapping from kinds to structure size reference prefixes.
 
 paul@150    646 
 
 paul@150    647 
structure_size_prefixes = {
 
 paul@150    648 
    "<class>" : "c",
 
 paul@150    649 
    "<module>" : "m",
 
 paul@150    650 
    "<instance>" : "i"
 
 paul@150    651 
    }
 
 paul@150    652 
 
 paul@150    653 
# A mapping from kinds to table name prefixes.
 
 paul@150    654 
 
 paul@150    655 
table_name_prefixes = {
 
 paul@150    656 
    "<class>" : "Class",
 
 paul@150    657 
    "<function>" : "Function",
 
 paul@150    658 
    "<module>" : "Module",
 
 paul@150    659 
    "<instance>" : "Instance"
 
 paul@150    660 
    }
 
 paul@150    661 
 
 paul@150    662 
 
 paul@150    663 
 
 paul@0    664 
# Output language reserved words.
 
 paul@0    665 
 
 paul@0    666 
reserved_words = [
 
 paul@0    667 
    "break", "char", "const", "continue",
 
 paul@0    668 
    "default", "double", "else",
 
 paul@0    669 
    "float", "for",
 
 paul@0    670 
    "if", "int", "long",
 
 paul@0    671 
    "NULL",
 
 paul@0    672 
    "return", "struct",
 
 paul@0    673 
    "typedef",
 
 paul@0    674 
    "void", "while",
 
 paul@0    675 
    ]
 
 paul@0    676 
 
 paul@0    677 
# vim: tabstop=4 expandtab shiftwidth=4