Commit | Line | Data |
---|---|---|
964561d6 FT |
1 | from struct import pack, unpack, calcsize |
2 | ||
3 | class fmterror(Exception): | |
4 | pass | |
5 | ||
6 | class eomerror(fmterror): | |
7 | pass | |
8 | ||
9 | def mutf8dec(bs): | |
10 | ret = "" | |
11 | i = 0 | |
12 | while i < len(bs): | |
13 | b = bs[i] | |
14 | i += 1 | |
15 | if b & 0x80 == 0: | |
16 | ret += chr(b) | |
17 | else: | |
18 | c = 0 | |
19 | while (c < 7) and (b & (1 << (6 - c))): | |
20 | c += 1 | |
21 | if c == 0 or c == 7: raise fmterror("invalid utf8 start-byte") | |
22 | iacc = acc = b & ((1 << (6 - c)) - 1) | |
23 | ic = c | |
24 | while c > 0: | |
25 | if i >= len(bs): raise fmterror("unterminated utf8 compound") | |
26 | b = bs[i] | |
27 | i += 1 | |
28 | if b & 0xc0 != 0x80: raise fmterror("invalid utf8 continuation byte") | |
eaf86c37 | 29 | acc = (acc << 6) | b & 0x3f |
964561d6 FT |
30 | c -= 1 |
31 | if iacc == 0 and ic != 2 and acc != 0: raise fmterror("invalid utf8 compound") | |
32 | ret += chr(acc) | |
33 | return ret | |
34 | ||
35 | def mutf8enc(cs): | |
36 | ret = bytearray() | |
37 | for c in cs: | |
38 | c = ord(c) | |
39 | if c == 0: | |
40 | ret.extend(b"\xc0\x80") | |
41 | elif 1 <= c < 128: | |
42 | ret.append(c) | |
43 | elif 128 <= c < 2048: | |
44 | ret.append(0xc0 | ((c & 0x7c0) >> 6)) | |
45 | ret.append(0x80 | (c & 0x03f)) | |
46 | elif 2048 <= c < 65536: | |
47 | ret.append(0xe0 | ((c & 0xf000) >> 12)) | |
48 | ret.append(0x80 | ((c & 0x0fc0) >> 6)) | |
49 | ret.append(0x80 | (c & 0x003f)) | |
50 | else: | |
51 | raise fmterror("non-BMP unicode not supported by Java") | |
52 | return bytes(ret) | |
53 | ||
54 | class decoder(object): | |
55 | def destruct(self, fmt): | |
56 | return unpack(fmt, self.splice(calcsize(fmt))) | |
57 | ||
58 | def skip(self, ln): | |
59 | self.splice(ln) | |
60 | ||
61 | def int8(self): | |
62 | return self.destruct(">b")[0] | |
63 | def uint8(self): | |
64 | return self.destruct(">B")[0] | |
65 | def int16(self): | |
66 | return self.destruct(">h")[0] | |
67 | def uint16(self): | |
68 | return self.destruct(">H")[0] | |
69 | def int32(self): | |
70 | return self.destruct(">i")[0] | |
71 | def uint32(self): | |
72 | return self.destruct(">I")[0] | |
73 | def int64(self): | |
74 | return self.destruct(">q")[0] | |
75 | def uint64(self): | |
76 | return self.destruct(">Q")[0] | |
77 | def float32(self): | |
78 | return self.destruct(">f")[0] | |
79 | def float64(self): | |
80 | return self.destruct(">d")[0] | |
81 | ||
82 | class decstream(decoder): | |
83 | def __init__(self, bk): | |
84 | self.bk = bk | |
85 | self.buf = bytearray() | |
86 | ||
87 | def eom(self): | |
88 | if len(self.buf) > 0: | |
89 | return False | |
90 | ret = self.bk.read(1024) | |
91 | if ret == b"": | |
92 | return True | |
93 | self.buf.extend(ret) | |
94 | return False | |
95 | ||
7db3019f FT |
96 | def tell(self): |
97 | return self.bk.tell() - len(self.buf) | |
98 | ||
964561d6 FT |
99 | def splice(self, ln=-1): |
100 | buf = self.buf | |
101 | if ln < 0: | |
102 | while True: | |
103 | ret = self.bk.read() | |
104 | if ret == b"": | |
105 | self.buf = bytearray() | |
106 | return bytes(buf) | |
107 | buf.extend(ret) | |
108 | else: | |
109 | while len(buf) < ln: | |
110 | rl = max(ln - len(buf), 1024) | |
111 | ret = self.bk.read(rl) | |
112 | if ret == b"": | |
113 | raise eomerror("unexpected end-of-file") | |
114 | buf.extend(ret) | |
115 | self.buf = buf[ln:] | |
116 | return bytes(buf[:ln]) | |
117 | ||
118 | def skip(self, ln): | |
119 | if ln < len(self.buf): | |
120 | self.buf = self.buf[ln:] | |
121 | else: | |
122 | ln -= len(self.buf) | |
123 | self.buf = bytearray() | |
124 | if hasattr(self.bk, "seek"): | |
125 | self.bk.seek(ln - 1, 1) | |
126 | if len(self.bk.read(1)) != 1: | |
127 | raise eomerror("unexpected end-of-file") | |
128 | else: | |
129 | while ln > 0: | |
130 | r = self.bk.read(ln) | |
131 | if r == b"": | |
132 | raise eomerror("unexpected end-of-file") | |
133 | ln -= len(r) | |
134 | ||
135 | def str(self): | |
136 | buf = self.buf | |
137 | p = 0 | |
138 | while True: | |
139 | p2 = buf.find(b'\0', p) | |
140 | if p2 > 0: | |
141 | self.buf = buf[p2 + 1:] | |
142 | return str(buf[:p2], "utf-8") | |
143 | ret = self.bk.read(1024) | |
144 | if ret == b"": | |
145 | if len(buf) == 0: | |
146 | raise eomerror("unexpected end-of-file") | |
147 | raise fmterror("no string terminator found") | |
148 | p = len(buf) | |
149 | buf.extend(ret) | |
150 | ||
151 | def close(self): | |
152 | self.bk.close() | |
153 | ||
154 | def __enter__(self): | |
155 | return self | |
156 | ||
157 | def __exit__(self, *excinfo): | |
158 | self.close() | |
159 | return False | |
160 | ||
161 | class decbuf(decoder): | |
162 | def __init__(self, data): | |
163 | self.data = data | |
164 | self.offset = 0 | |
165 | ||
166 | def __len__(self): | |
167 | return len(self.data) - self.offset | |
168 | ||
169 | def eom(self): | |
170 | return self.offset >= len(self.data) | |
171 | ||
172 | def splice(self, ln=-1): | |
173 | if ln < 0: | |
174 | ret = self.data[self.offset:] | |
175 | self.offset = len(self.data) | |
176 | return ret | |
177 | else: | |
178 | if self.offset + ln > len(self.data): | |
179 | raise eomerror("out of data to decode") | |
180 | ret = self.data[self.offset:self.offset + ln] | |
181 | self.offset += ln | |
182 | return ret | |
183 | ||
184 | def str(self): | |
185 | p = self.data.find(b'\0', self.offset) | |
186 | if p < 0: | |
187 | if self.offset == len(self.data): | |
188 | raise eomerror("out of data to decode") | |
189 | raise fmterror("no string terminator found") | |
190 | ret = str(self.data[self.offset:p], "utf-8") | |
191 | self.offset = p + 1 | |
192 | return str(ret) | |
193 | ||
194 | class encoder(object): | |
195 | def enstruct(self, fmt, *args): | |
196 | self.extend(pack(fmt, *args)) | |
197 | return self | |
198 | ||
199 | def int8(self, val): | |
200 | self.enstruct(">b", val) | |
201 | return self | |
202 | def uint8(self, val): | |
203 | self.enstruct(">B", val) | |
204 | return self | |
205 | def int16(self, val): | |
206 | self.enstruct(">h", val) | |
207 | return self | |
208 | def uint16(self, val): | |
209 | self.enstruct(">H", val) | |
210 | return self | |
211 | def int32(self, val): | |
212 | self.enstruct(">i", val) | |
213 | return self | |
214 | def uint32(self, val): | |
215 | self.enstruct(">I", val) | |
216 | return self | |
217 | def int64(self, val): | |
218 | self.enstruct(">q", val) | |
219 | return self | |
220 | def uint64(self, val): | |
221 | self.enstruct(">Q", val) | |
222 | return self | |
223 | def float32(self, val): | |
224 | self.enstruct(">f", val) | |
225 | return self | |
226 | def float64(self, val): | |
227 | self.enstruct(">d", val) | |
228 | return self | |
229 | ||
230 | def str(self, val): | |
231 | if val.find('\0') >= 0: | |
232 | raise ValueError("encoded strings must not contain NULs") | |
233 | self.extend(val.encode("utf-8")) | |
234 | self.extend(b"\0") | |
235 | return self | |
236 | ||
237 | def ttol(self, val, term=False): | |
238 | for obj in val: | |
239 | if isinstance(obj, int): | |
240 | if 0 <= obj < 256: | |
241 | self.uint8(T_UINT8) | |
242 | self.uint8(obj) | |
243 | elif 0 <= obj < 65536: | |
244 | self.uint8(T_UINT16) | |
245 | self.uint16(obj) | |
246 | else: | |
247 | self.uint8(T_INT) | |
248 | self.int32(obj) | |
249 | elif isinstance(obj, str): | |
250 | self.uint8(T_STR) | |
251 | self.str(obj) | |
252 | elif isinstance(obj, utils.coord): | |
253 | self.uint8(T_COORD) | |
254 | self.coord(obj) | |
255 | elif isinstance(obj, utils.color): | |
256 | self.uint8(T_COLOR) | |
257 | self.color(obj) | |
258 | elif isinstance(obj, list): | |
259 | self.uint8(T_TTOL) | |
260 | self.ttol(obj, True) | |
261 | elif isinstance(obj, float): | |
262 | self.uint8(T_FLOAT32) | |
263 | self.float32(obj) | |
264 | elif obj is None: | |
265 | self.uint8(T_NIL) | |
266 | elif isinstance(obj, collections.ByteString): | |
267 | self.uint8(T_BYTES) | |
268 | if len(obj) < 128: | |
269 | self.uint8(len(obj)) | |
270 | else: | |
271 | self.uint8(0x80).int32(len(obj)) | |
272 | self.extend(obj) | |
273 | else: | |
274 | raise ValueError("unexpected type in tto-list: %s" % type(obj)) | |
275 | if term: | |
276 | self.uint8(T_END) | |
277 | return self | |
278 | ||
279 | class encstream(encoder): | |
280 | def __init__(self, bk): | |
281 | self.bk = bk | |
282 | ||
283 | def extend(self, data): | |
284 | self.bk.write(data) | |
285 | return self | |
286 | ||
287 | def close(self): | |
288 | self.bk.close() | |
289 | ||
290 | def __enter__(self): | |
291 | return self | |
292 | ||
293 | def __exit__(self, *excinfo): | |
294 | self.close() | |
295 | return False | |
296 | ||
297 | class encbuf(encoder, bytearray): | |
298 | def extend(self, data): | |
299 | bytearray.extend(self, data) | |
300 | return self |