2 $Id: marcomp.c,v 1.2 2003-02-28 14:34:37 adam Exp $
4 marcomp.c - compiler of MARC statements.
12 #include <yaz/yaz-util.h>
16 static mc_token mc_gettoken(mc_context *c);
17 static void mc_ungettoken(mc_context *c);
18 static int mc_getval(mc_context *c);
19 static int mc_getdata(mc_context *c, char *s, int sz);
20 static void mc_getinterval(mc_context *c, int *start, int *end);
22 static mc_subfield *mc_mk_subfield(mc_subfield *parent);
23 static mc_field *mc_mk_field(void);
25 static struct mc_errmsg
32 {EMCF, "not complete field"},
33 {EMCSF, "not complete subfield"},
34 {EMCSFGROUP, "not closed GROUP"},
35 {EMCSFVAR, "not closed VARIANT"},
36 {EMCSFINLINE, "not closed IN-LINE"},
37 {EMCEND, "not correct errno"}
39 mc_errcode mc_errno(mc_context *c)
43 const char *mc_error(mc_errcode no)
45 if (no >= EMCOK && no<EMCEND)
46 return mc_errmsg[no].msg;
48 return mc_errmsg[EMCEND].msg;
50 mc_context *mc_mk_context(const char *s)
56 p = (mc_context*) xmalloc(sizeof(*p));
61 memset(p, 0, sizeof(*p));
70 void mc_destroy_context(mc_context *c)
74 mc_token mc_gettoken(mc_context *c)
76 if (c->offset >= c->len)
79 switch (*(c->data+c->offset))
81 case '{': c->crrtok = LVARIANT; break;
82 case '}': c->crrtok = RVARIANT; break;
83 case '(': c->crrtok = LGROUP; break;
84 case ')': c->crrtok = RGROUP; break;
85 case '<': c->crrtok = LINLINE; break;
86 case '>': c->crrtok = RINLINE; break;
87 case '$': c->crrtok = SUBFIELD; break;
88 case '[': c->crrtok = LINTERVAL; break;
89 case ']': c->crrtok = RINTERVAL; break;
91 if (isspace(*(c->data+c->offset)) || *(c->data+c->offset) == '\n')
98 c->crrval = *(c->data+c->offset);
102 fprintf(stderr, "gettoken(): offset: %d", c->offset);
103 if (c->crrtok == REGULAR)
104 fprintf(stderr, "<%c>", c->crrval);
105 fprintf(stderr, "\n");
110 void mc_ungettoken(mc_context *c)
115 int mc_getval(mc_context *c)
119 int mc_getdata(mc_context *c, char *s, int sz)
125 if (mc_gettoken(c)!=REGULAR)
136 void mc_getinterval(mc_context *c, int *start, int *end)
139 int start_pos, end_pos;
141 start_pos = end_pos = -1;
143 if (mc_gettoken(c) == LINTERVAL)
149 mc_token tok = mc_gettoken(c);
151 if (tok == RINTERVAL || tok == NOP)
154 buf[i] = mc_getval(c);
158 i = sscanf(buf, "%d-%d", &start_pos, &end_pos);
170 mc_field *mc_mk_field(void)
172 mc_field *p = (mc_field *)xmalloc(sizeof(*p));
176 memset(p, 0, sizeof(*p));
177 p->name = (char *)xmalloc(SZ_FNAME+1);
179 p->ind1 = (char *)xmalloc(SZ_IND+1);
181 p->ind2 = (char *)xmalloc(SZ_IND+1);
183 p->interval.start = p->interval.end = -1;
187 void mc_destroy_field(mc_field *p)
191 if (p->name) xfree(p->name);
192 if (p->ind1) xfree(p->ind1);
193 if (p->ind2) xfree(p->ind2);
194 if (p->list) mc_destroy_subfields_recursive(p->list);
197 mc_field *mc_getfield(mc_context *c)
205 c->errcode = EMCNOMEM;
209 if (mc_getdata(c, pf->name, SZ_FNAME) == SZ_FNAME)
211 mc_token nexttok = mc_gettoken(c);
215 if (nexttok == LINTERVAL)
217 mc_getinterval(c, &pf->interval.start, &pf->interval.end);
219 fprintf(stderr, "ineterval (%d)-(%d)\n", pf->interval.start,
224 if ((mc_getdata(c, pf->ind1, SZ_IND) == SZ_IND) &&
225 (mc_getdata(c, pf->ind2, SZ_IND) == SZ_IND))
227 pf->list = mc_getsubfields(c, 0);
233 mc_destroy_field(pf);
239 mc_subfield *mc_mk_subfield(mc_subfield *parent)
241 mc_subfield *p = (mc_subfield*)xmalloc(sizeof(*p));
245 memset(p, 0, sizeof(*p));
247 p->name = (char *)xmalloc(SZ_SFNAME+1);
249 p->prefix = (char *)xmalloc(SZ_PREFIX+1);
251 p->suffix = (char *)xmalloc(SZ_SUFFIX+1);
254 p->interval.start = p->interval.end = -1;
258 void mc_destroy_subfield(mc_subfield *p)
263 if (p->which == MC_SFGROUP || p->which == MC_SFVARIANT)
266 mc_destroy_subfields_recursive(p->u.child);
268 else if (p->which == MC_SF)
271 mc_destroy_field(p->u.in_line);
273 if (p->name) xfree(p->name);
274 if (p->prefix) xfree(p->prefix);
275 if (p->suffix) xfree(p->suffix);
276 if (p->parent) p->parent->next = p->next;
279 void mc_destroy_subfields_recursive(mc_subfield *p)
284 mc_destroy_subfields_recursive(p->next);
286 if (p->which == MC_SFGROUP || p->which == MC_SFVARIANT)
289 mc_destroy_subfields_recursive(p->u.child);
291 else if (p->which == MC_SF)
294 mc_destroy_field(p->u.in_line);
297 if (p->name) xfree(p->name);
298 if (p->prefix) xfree(p->prefix);
299 if (p->suffix) xfree(p->suffix);
300 if (p->parent) p->parent->next = 0;
303 mc_subfield *mc_getsubfields(mc_context *c, mc_subfield *parent)
306 mc_token tok = mc_gettoken(c);
313 if (!(psf = mc_mk_subfield(parent)))
315 c->errcode = EMCNOMEM;
319 psf->which = MC_SFGROUP;
320 psf->u.child = mc_getsubfields(c, psf);
322 if (mc_gettoken(c) == RGROUP)
323 psf->next = mc_getsubfields(c, psf);
326 c->errcode = EMCSFGROUP;
327 mc_destroy_subfield(psf);
331 else if (tok == LVARIANT)
333 if (!(psf = mc_mk_subfield(parent)))
335 c->errcode = EMCNOMEM;
339 psf->which = MC_SFVARIANT;
340 psf->u.child = mc_getsubfields(c, psf);
342 if (mc_gettoken(c) == RVARIANT)
343 psf->next = mc_getsubfields(c, psf);
346 c->errcode = EMCSFVAR;
347 mc_destroy_subfield(psf);
351 else if (tok == RGROUP || tok == RVARIANT || tok == RINLINE)
356 else if (tok == REGULAR)
358 if (!(psf = mc_mk_subfield(parent)))
360 c->errcode = EMCNOMEM;
366 if((mc_getdata(c, psf->prefix, SZ_PREFIX) == SZ_PREFIX) &&
367 (mc_gettoken(c) == SUBFIELD) &&
368 (mc_getdata(c, psf->name, SZ_SFNAME) == SZ_SFNAME))
370 mc_token tok = mc_gettoken(c);
374 if (tok == LINTERVAL)
376 mc_getinterval(c, &psf->interval.start, &psf->interval.end);
378 else if (tok == LINLINE)
381 psf->u.in_line = mc_getfield(c);
382 if (mc_gettoken(c) != RINLINE)
384 c->errcode = EMCSFINLINE;
385 mc_destroy_subfield(psf);
390 if (mc_getdata(c, psf->suffix, SZ_SUFFIX) == SZ_SUFFIX)
393 psf->next = mc_getsubfields(c, psf);
398 mc_destroy_subfield(psf);