]> git.stg.codes - stg.git/blob - stglibs/smux.lib/xer_decoder.c
convertor -> sgconv in docs.
[stg.git] / stglibs / smux.lib / xer_decoder.c
1 /*
2  * Copyright (c) 2004, 2005 Lev Walkin <vlm@lionet.info>. All rights reserved.
3  * Redistribution and modifications are permitted subject to BSD license.
4  */
5 #include <asn_application.h>
6 #include <asn_internal.h>
7 #include <xer_support.h>                /* XER/XML parsing support */
8
9
10 /*
11  * Decode the XER encoding of a given type.
12  */
13 asn_dec_rval_t
14 xer_decode(asn_codec_ctx_t *opt_codec_ctx, asn_TYPE_descriptor_t *td,
15                 void **struct_ptr, const void *buffer, size_t size) {
16         asn_codec_ctx_t s_codec_ctx;
17
18         /*
19          * Stack checker requires that the codec context
20          * must be allocated on the stack.
21          */
22         if(opt_codec_ctx) {
23                 if(opt_codec_ctx->max_stack_size) {
24                         s_codec_ctx = *opt_codec_ctx;
25                         opt_codec_ctx = &s_codec_ctx;
26                 }
27         } else {
28                 /* If context is not given, be security-conscious anyway */
29                 memset(&s_codec_ctx, 0, sizeof(s_codec_ctx));
30                 s_codec_ctx.max_stack_size = _ASN_DEFAULT_STACK_MAX;
31                 opt_codec_ctx = &s_codec_ctx;
32         }
33
34         /*
35          * Invoke type-specific decoder.
36          */
37         return td->xer_decoder(opt_codec_ctx, td, struct_ptr, 0, buffer, size);
38 }
39
40
41
42 struct xer__cb_arg {
43         pxml_chunk_type_e       chunk_type;
44         size_t                  chunk_size;
45         const void              *chunk_buf;
46         int callback_not_invoked;
47 };
48
49 static int
50 xer__token_cb(pxml_chunk_type_e type, const void *_chunk_data, size_t _chunk_size, void *key) {
51         struct xer__cb_arg *arg = (struct xer__cb_arg *)key;
52         arg->chunk_type = type;
53         arg->chunk_size = _chunk_size;
54         arg->chunk_buf = _chunk_data;
55         arg->callback_not_invoked = 0;
56         return -1;      /* Terminate the XML parsing */
57 }
58
59 /*
60  * Fetch the next token from the XER/XML stream.
61  */
62 ssize_t
63 xer_next_token(int *stateContext, const void *buffer, size_t size, pxer_chunk_type_e *ch_type) {
64         struct xer__cb_arg arg;
65         int new_stateContext = *stateContext;
66         ssize_t ret;
67
68         arg.callback_not_invoked = 1;
69         ret = pxml_parse(&new_stateContext, buffer, size, xer__token_cb, &arg);
70         if(ret < 0) return -1;
71         if(arg.callback_not_invoked) {
72                 assert(ret == 0);       /* No data was consumed */
73                 return 0;               /* Try again with more data */
74         } else {
75                 assert(arg.chunk_size);
76                 assert(arg.chunk_buf == buffer);
77         }
78
79         /*
80          * Translate the XML chunk types into more convenient ones.
81          */
82         switch(arg.chunk_type) {
83         case PXML_TEXT:
84                 *ch_type = PXER_TEXT;
85                 break;
86         case PXML_TAG: return 0;        /* Want more */
87         case PXML_TAG_END:
88                 *ch_type = PXER_TAG;
89                 break;
90         case PXML_COMMENT:
91         case PXML_COMMENT_END:
92                 *ch_type = PXER_COMMENT;
93                 break;
94         }
95
96         *stateContext = new_stateContext;
97         return arg.chunk_size;
98 }
99
100 #define CSLASH  0x2f    /* '/' */
101 #define LANGLE  0x3c    /* '<' */
102 #define RANGLE  0x3e    /* '>' */
103
104 xer_check_tag_e
105 xer_check_tag(const void *buf_ptr, int size, const char *need_tag) {
106         const char *buf = (const char *)buf_ptr;
107         const char *end;
108         xer_check_tag_e ct = XCT_OPENING;
109
110         if(size < 2 || buf[0] != LANGLE || buf[size-1] != RANGLE) {
111                 if(size >= 2)
112                 ASN_DEBUG("Broken XML tag: \"%c...%c\"", buf[0], buf[size - 1]);
113                 return XCT_BROKEN;
114         }
115
116         /*
117          * Determine the tag class.
118          */
119         if(buf[1] == CSLASH) {
120                 buf += 2;       /* advance past "</" */
121                 size -= 3;      /* strip "</" and ">" */
122                 ct = XCT_CLOSING;
123                 if(size > 0 && buf[size-1] == CSLASH)
124                         return XCT_BROKEN;      /* </abc/> */
125         } else {
126                 buf++;          /* advance past "<" */
127                 size -= 2;      /* strip "<" and ">" */
128                 if(size > 0 && buf[size-1] == CSLASH) {
129                         ct = XCT_BOTH;
130                         size--; /* One more, for "/" */
131                 }
132         }
133
134         /* Sometimes we don't care about the tag */
135         if(!need_tag || !*need_tag)
136                 return (xer_check_tag_e)(XCT__UNK__MASK | ct);
137
138         /*
139          * Determine the tag name.
140          */
141         for(end = buf + size; buf < end; buf++, need_tag++) {
142                 int b = *buf, n = *need_tag;
143                 if(b != n) {
144                         if(n == 0) {
145                                 switch(b) {
146                                 case 0x09: case 0x0a: case 0x0c: case 0x0d:
147                                 case 0x20:
148                                         /* "<abc def/>": whitespace is normal */
149                                         return ct;
150                                 }
151                         }
152                         return (xer_check_tag_e)(XCT__UNK__MASK | ct);
153                 }
154                 if(b == 0)
155                         return XCT_BROKEN;      /* Embedded 0 in buf?! */
156         }
157         if(*need_tag)
158                 return (xer_check_tag_e)(XCT__UNK__MASK | ct);
159
160         return ct;
161 }
162
163
164 #undef  ADVANCE
165 #define ADVANCE(num_bytes)      do {                            \
166                 size_t num = (num_bytes);                       \
167                 buf_ptr = ((const char *)buf_ptr) + num;        \
168                 size -= num;                                    \
169                 consumed_myself += num;                         \
170         } while(0)
171
172 #undef  RETURN
173 #define RETURN(_code)   do {                                    \
174                 rval.code = _code;                              \
175                 rval.consumed = consumed_myself;                \
176                 if(rval.code != RC_OK)                          \
177                         ASN_DEBUG("Failed with %d", rval.code); \
178                 return rval;                                    \
179         } while(0)
180
181 #define XER_GOT_BODY(chunk_buf, chunk_size, size)       do {    \
182                 ssize_t converted_size = body_receiver          \
183                         (struct_key, chunk_buf, chunk_size,     \
184                                 (size_t)chunk_size < size);     \
185                 if(converted_size == -1) RETURN(RC_FAIL);       \
186                 if(converted_size == 0                          \
187                         && size == (size_t)chunk_size)          \
188                         RETURN(RC_WMORE);                       \
189                 chunk_size = converted_size;                    \
190         } while(0)
191 #define XER_GOT_EMPTY() do {                                    \
192         if(body_receiver(struct_key, 0, 0, size > 0) == -1)     \
193                         RETURN(RC_FAIL);                        \
194         } while(0)
195
196 /*
197  * Generalized function for decoding the primitive values.
198  */
199 asn_dec_rval_t
200 xer_decode_general(asn_codec_ctx_t *opt_codec_ctx,
201         asn_struct_ctx_t *ctx,  /* Type decoder context */
202         void *struct_key,
203         const char *xml_tag,    /* Expected XML tag */
204         const void *buf_ptr, size_t size,
205         int (*opt_unexpected_tag_decoder)
206                 (void *struct_key, const void *chunk_buf, size_t chunk_size),
207         ssize_t (*body_receiver)
208                 (void *struct_key, const void *chunk_buf, size_t chunk_size,
209                         int have_more)
210         ) {
211
212         asn_dec_rval_t rval;
213         ssize_t consumed_myself = 0;
214
215         (void)opt_codec_ctx;
216
217         /*
218          * Phases of XER/XML processing:
219          * Phase 0: Check that the opening tag matches our expectations.
220          * Phase 1: Processing body and reacting on closing tag.
221          */
222         if(ctx->phase > 1) RETURN(RC_FAIL);
223         for(;;) {
224                 pxer_chunk_type_e ch_type;      /* XER chunk type */
225                 ssize_t ch_size;                /* Chunk size */
226                 xer_check_tag_e tcv;            /* Tag check value */
227
228                 /*
229                  * Get the next part of the XML stream.
230                  */
231                 ch_size = xer_next_token(&ctx->context, buf_ptr, size,
232                         &ch_type);
233                 switch(ch_size) {
234                 case -1: RETURN(RC_FAIL);
235                 case 0:
236                         RETURN(RC_WMORE);
237                 default:
238                         switch(ch_type) {
239                         case PXER_COMMENT:              /* Got XML comment */
240                                 ADVANCE(ch_size);       /* Skip silently */
241                                 continue;
242                         case PXER_TEXT:
243                                 if(ctx->phase == 0) {
244                                         /*
245                                          * We have to ignore whitespace here,
246                                          * but in order to be forward compatible
247                                          * with EXTENDED-XER (EMBED-VALUES, #25)
248                                          * any text is just ignored here.
249                                          */
250                                 } else {
251                                         XER_GOT_BODY(buf_ptr, ch_size, size);
252                                 }
253                                 ADVANCE(ch_size);
254                                 continue;
255                         case PXER_TAG:
256                                 break;  /* Check the rest down there */
257                         }
258                 }
259
260                 assert(ch_type == PXER_TAG && size);
261
262                 tcv = xer_check_tag(buf_ptr, ch_size, xml_tag);
263                 /*
264                  * Phase 0:
265                  *      Expecting the opening tag
266                  *      for the type being processed.
267                  * Phase 1:
268                  *      Waiting for the closing XML tag.
269                  */
270                 switch(tcv) {
271                 case XCT_BOTH:
272                         if(ctx->phase) break;
273                         /* Finished decoding of an empty element */
274                         XER_GOT_EMPTY();
275                         ADVANCE(ch_size);
276                         ctx->phase = 2; /* Phase out */
277                         RETURN(RC_OK);
278                 case XCT_OPENING:
279                         if(ctx->phase) break;
280                         ADVANCE(ch_size);
281                         ctx->phase = 1; /* Processing body phase */
282                         continue;
283                 case XCT_CLOSING:
284                         if(!ctx->phase) break;
285                         ADVANCE(ch_size);
286                         ctx->phase = 2; /* Phase out */
287                         RETURN(RC_OK);
288                 case XCT_UNKNOWN_BO:
289                         /*
290                          * Certain tags in the body may be expected.
291                          */
292                         if(opt_unexpected_tag_decoder
293                         && opt_unexpected_tag_decoder(struct_key,
294                                         buf_ptr, ch_size) >= 0) {
295                                 /* Tag's processed fine */
296                                 ADVANCE(ch_size);
297                                 if(!ctx->phase) {
298                                         /* We are not expecting
299                                          * the closing tag anymore. */
300                                         ctx->phase = 2; /* Phase out */
301                                         RETURN(RC_OK);
302                                 }
303                                 continue;
304                         }
305                         /* Fall through */
306                 default:
307                         break;          /* Unexpected tag */
308                 }
309
310                 ASN_DEBUG("Unexpected XML tag (expected \"%s\")", xml_tag);
311                 break;  /* Dark and mysterious things have just happened */
312         }
313
314         RETURN(RC_FAIL);
315 }
316
317
318 int
319 xer_is_whitespace(const void *chunk_buf, size_t chunk_size) {
320         const char *p = (const char *)chunk_buf;
321         const char *pend = p + chunk_size;
322
323         for(; p < pend; p++) {
324                 switch(*p) {
325                 /* X.693, #8.1.4
326                  * HORISONTAL TAB (9)
327                  * LINE FEED (10) 
328                  * CARRIAGE RETURN (13) 
329                  * SPACE (32)
330                  */
331                 case 0x09: case 0x0a: case 0x0d: case 0x20:
332                         break;
333                 default:
334                         return 0;
335                 }
336         }
337         return 1;       /* All whitespace */
338 }
339
340 /*
341  * This is a vastly simplified, non-validating XML tree skipper.
342  */
343 int
344 xer_skip_unknown(xer_check_tag_e tcv, ber_tlv_len_t *depth) {
345         assert(*depth > 0);
346         switch(tcv) {
347         case XCT_BOTH:
348         case XCT_UNKNOWN_BO:
349                 /* These negate each other. */
350                 return 0;
351         case XCT_OPENING:
352         case XCT_UNKNOWN_OP:
353                 ++(*depth);
354                 return 0;
355         case XCT_CLOSING:
356         case XCT_UNKNOWN_CL:
357                 if(--(*depth) == 0)
358                         return (tcv == XCT_CLOSING) ? 2 : 1;
359                 return 0;
360         default:
361                 return -1;
362         }
363 }