rino 0.1.0
Sign up to get free protection for your applications and to get access to all the features.
- data/README +44 -0
- data/Rakefile +123 -0
- data/ext/extconf.rb +26 -0
- data/ext/ruby_inchi_main.so +0 -0
- data/ext/src/aux2atom.h +2786 -0
- data/ext/src/comdef.h +148 -0
- data/ext/src/e_0dstereo.c +3014 -0
- data/ext/src/e_0dstereo.h +31 -0
- data/ext/src/e_comdef.h +57 -0
- data/ext/src/e_ctl_data.h +147 -0
- data/ext/src/e_ichi_io.c +498 -0
- data/ext/src/e_ichi_io.h +40 -0
- data/ext/src/e_ichi_parms.c +37 -0
- data/ext/src/e_ichi_parms.h +41 -0
- data/ext/src/e_ichicomp.h +50 -0
- data/ext/src/e_ichierr.h +40 -0
- data/ext/src/e_ichimain.c +593 -0
- data/ext/src/e_ichisize.h +43 -0
- data/ext/src/e_inchi_atom.c +75 -0
- data/ext/src/e_inchi_atom.h +33 -0
- data/ext/src/e_inpdef.h +41 -0
- data/ext/src/e_mode.h +706 -0
- data/ext/src/e_mol2atom.c +649 -0
- data/ext/src/e_readinch.c +58 -0
- data/ext/src/e_readmol.c +54 -0
- data/ext/src/e_readmol.h +180 -0
- data/ext/src/e_readstru.c +251 -0
- data/ext/src/e_readstru.h +33 -0
- data/ext/src/e_util.c +284 -0
- data/ext/src/e_util.h +61 -0
- data/ext/src/extr_ct.h +251 -0
- data/ext/src/ichi.h +206 -0
- data/ext/src/ichi_bns.c +7999 -0
- data/ext/src/ichi_bns.h +231 -0
- data/ext/src/ichican2.c +5000 -0
- data/ext/src/ichicano.c +2195 -0
- data/ext/src/ichicano.h +49 -0
- data/ext/src/ichicans.c +1625 -0
- data/ext/src/ichicant.h +379 -0
- data/ext/src/ichicomn.h +260 -0
- data/ext/src/ichicomp.h +50 -0
- data/ext/src/ichidrp.h +119 -0
- data/ext/src/ichierr.h +124 -0
- data/ext/src/ichiisot.c +101 -0
- data/ext/src/ichilnct.c +286 -0
- data/ext/src/ichimain.h +132 -0
- data/ext/src/ichimak2.c +1189 -0
- data/ext/src/ichimake.c +3812 -0
- data/ext/src/ichimake.h +205 -0
- data/ext/src/ichimap1.c +851 -0
- data/ext/src/ichimap2.c +2856 -0
- data/ext/src/ichimap4.c +1609 -0
- data/ext/src/ichinorm.c +741 -0
- data/ext/src/ichinorm.h +67 -0
- data/ext/src/ichiparm.c +45 -0
- data/ext/src/ichiparm.h +1441 -0
- data/ext/src/ichiprt1.c +3612 -0
- data/ext/src/ichiprt2.c +1511 -0
- data/ext/src/ichiprt3.c +3011 -0
- data/ext/src/ichiqueu.c +1003 -0
- data/ext/src/ichiring.c +326 -0
- data/ext/src/ichiring.h +49 -0
- data/ext/src/ichisize.h +35 -0
- data/ext/src/ichisort.c +539 -0
- data/ext/src/ichister.c +3538 -0
- data/ext/src/ichister.h +35 -0
- data/ext/src/ichitaut.c +3843 -0
- data/ext/src/ichitaut.h +387 -0
- data/ext/src/ichitime.h +74 -0
- data/ext/src/inchi_api.h +670 -0
- data/ext/src/inchi_dll.c +1480 -0
- data/ext/src/inchi_dll.h +34 -0
- data/ext/src/inchi_dll_main.c +23 -0
- data/ext/src/inchi_dll_main.h +31 -0
- data/ext/src/inpdef.h +328 -0
- data/ext/src/lreadmol.h +1246 -0
- data/ext/src/mode.h +706 -0
- data/ext/src/ruby_inchi_main.c +558 -0
- data/ext/src/runichi.c +4179 -0
- data/ext/src/strutil.c +3861 -0
- data/ext/src/strutil.h +182 -0
- data/ext/src/util.c +1130 -0
- data/ext/src/util.h +85 -0
- data/lib/clean_tempfile.rb +220 -0
- data/lib/rino.rb +111 -0
- data/test/test.rb +386 -0
- metadata +130 -0
data/ext/src/aux2atom.h
ADDED
@@ -0,0 +1,2786 @@
|
|
1
|
+
/*
|
2
|
+
* International Union of Pure and Applied Chemistry (IUPAC)
|
3
|
+
* International Chemical Identifier (InChI)
|
4
|
+
* Version 1
|
5
|
+
* Software version 1.00
|
6
|
+
* April 13, 2005
|
7
|
+
* Developed at NIST
|
8
|
+
*/
|
9
|
+
|
10
|
+
/*
|
11
|
+
The code in this #include file reads InChI AuxInfo
|
12
|
+
*/
|
13
|
+
|
14
|
+
/****************************************************************************/
|
15
|
+
#define MIN_BOND_LENGTH (1.0e-6)
|
16
|
+
#define INCHI_LINE_LEN 512 /*1024*/ /*256*/
|
17
|
+
#define INCHI_LINE_ADD 384 /*128*/ /*64*/
|
18
|
+
/* Note: (INCHI_LINE_LEN - INCHI_LINE_ADD) > (length of the longest item: szCoord) = 33 */
|
19
|
+
/*****************************************************************************/
|
20
|
+
|
21
|
+
#if( defined(INCHI_LIBRARY) || defined(INCHI_MAIN) )
|
22
|
+
|
23
|
+
#define AB_MAX_WELL_DEFINED_PARITY inchi_max(INCHI_PARITY_ODD, INCHI_PARITY_EVEN) /* 1, 2 => well defined parities, uncluding 'unknown' */
|
24
|
+
#define AB_MIN_WELL_DEFINED_PARITY inchi_min(INCHI_PARITY_ODD, INCHI_PARITY_EVEN) /* min(INCHI_PARITY_ODD, INCHI_PARITY_EVEN) */
|
25
|
+
#define ATOM_PARITY_WELL_DEF(X) (AB_MIN_WELL_DEFINED_PARITY <= (X) && (X) <= AB_MAX_WELL_DEFINED_PARITY)
|
26
|
+
|
27
|
+
#define inchi_NUMH2(AT,CUR_AT) ((AT[CUR_AT].num_iso_H[0]>0?AT[CUR_AT].num_iso_H[0]:0) +AT[CUR_AT].num_iso_H[1]+AT[CUR_AT].num_iso_H[2]+AT[CUR_AT].num_iso_H[3])
|
28
|
+
|
29
|
+
#define SB_PARITY_FLAG 0x38 /* disconnected structure has undef. parity */
|
30
|
+
#define SB_PARITY_SHFT 3
|
31
|
+
#define SB_PARITY_MASK 0x07
|
32
|
+
#define SB_PARITY_1(X) (X & SB_PARITY_MASK) /* refers to connected structure */
|
33
|
+
#define SB_PARITY_2(X) (((X) >> SB_PARITY_SHFT) & SB_PARITY_MASK) /* refers to connected structure */
|
34
|
+
|
35
|
+
|
36
|
+
int str_fgetc( INCHI_FILE *f );
|
37
|
+
char *str_fgets( char *szLine, int len, INCHI_FILE *f );
|
38
|
+
int my_fgets( char *szLine, int len, INCHI_FILE *f, int *bTooLongLine );
|
39
|
+
|
40
|
+
#endif
|
41
|
+
|
42
|
+
|
43
|
+
#ifdef INCHI_LIBRARY
|
44
|
+
|
45
|
+
void FreeInchi_Atom( inchi_Atom **at );
|
46
|
+
inchi_Atom *CreateInchi_Atom( int num_atoms );
|
47
|
+
void FreeInchi_Input( inchi_Input *inp_at_data );
|
48
|
+
S_SHORT *is_in_the_slist( S_SHORT *pathAtom, S_SHORT nNextAtom, int nPathLen );
|
49
|
+
int is_element_a_metal( char szEl[] );
|
50
|
+
|
51
|
+
char *str_fgetsTab( char *szLine, int len, INCHI_FILE *f );
|
52
|
+
int my_fgetsTab( char *szLine, int len, INCHI_FILE *f, int *bTooLongLine );
|
53
|
+
int my_fgetsTab1( char *szLine, int len, INCHI_FILE *f, int *bTooLongLine );
|
54
|
+
|
55
|
+
#endif
|
56
|
+
|
57
|
+
#ifndef INCHI_MAIN
|
58
|
+
|
59
|
+
void FreeInchi_Stereo0D( inchi_Stereo0D **stereo0D );
|
60
|
+
inchi_Stereo0D *CreateInchi_Stereo0D( int num_stereo0D );
|
61
|
+
int Extract0DParities( inp_ATOM *at, int nNumAtoms, inchi_Stereo0D *stereo0D,
|
62
|
+
int num_stereo0D, char *pStrErr, int *err );
|
63
|
+
|
64
|
+
#endif
|
65
|
+
|
66
|
+
|
67
|
+
#if( defined(INCHI_LIBRARY) || defined(INCHI_MAIN) )
|
68
|
+
/*******************************************************************/
|
69
|
+
int str_fgetc( INCHI_FILE *f )
|
70
|
+
{
|
71
|
+
#ifdef INCHI_LIBRARY
|
72
|
+
if ( f->nPtr < f->nUsedLength ) {
|
73
|
+
return (int)f->pStr[f->nPtr++];
|
74
|
+
}
|
75
|
+
return EOF;
|
76
|
+
#else
|
77
|
+
return fgetc( f );
|
78
|
+
#endif
|
79
|
+
}
|
80
|
+
/*******************************************************************/
|
81
|
+
char *str_fgets( char *szLine, int len, INCHI_FILE *f )
|
82
|
+
{
|
83
|
+
int length=0, c;
|
84
|
+
len --;
|
85
|
+
while ( length < len && EOF != (c = str_fgetc( f )) ) {
|
86
|
+
szLine[length++] = (char)c;
|
87
|
+
if ( c == '\n' )
|
88
|
+
break;
|
89
|
+
}
|
90
|
+
if ( !length && EOF == c ) {
|
91
|
+
return NULL;
|
92
|
+
}
|
93
|
+
szLine[length] = '\0';
|
94
|
+
return szLine;
|
95
|
+
}
|
96
|
+
/*******************************************************************/
|
97
|
+
int my_fgets( char *szLine, int len, INCHI_FILE *f, int *bTooLongLine )
|
98
|
+
{
|
99
|
+
int length;
|
100
|
+
char *p;
|
101
|
+
do {
|
102
|
+
p = str_fgets( szLine, len-1, f );
|
103
|
+
if ( !p ) {
|
104
|
+
*bTooLongLine = 0;
|
105
|
+
return -1; /* end of file or cannot read */
|
106
|
+
}
|
107
|
+
szLine[len-1] = '\0';
|
108
|
+
/*
|
109
|
+
*bTooLongLine = !strchr( szLine, '\n' );
|
110
|
+
*/
|
111
|
+
p = strchr( szLine, '\n' );
|
112
|
+
*bTooLongLine = ( !p && ((int)strlen(szLine)) == len-2 );
|
113
|
+
LtrimRtrim( szLine, &length );
|
114
|
+
} while ( !length );
|
115
|
+
return length;
|
116
|
+
}
|
117
|
+
|
118
|
+
#endif
|
119
|
+
|
120
|
+
|
121
|
+
#ifdef INCHI_LIBRARY
|
122
|
+
/******************************************************************************************************/
|
123
|
+
void FreeInchi_Atom( inchi_Atom **at )
|
124
|
+
{
|
125
|
+
if ( at && *at ) {
|
126
|
+
inchi_free( *at );
|
127
|
+
*at = NULL;
|
128
|
+
}
|
129
|
+
}
|
130
|
+
/******************************************************************************************************/
|
131
|
+
inchi_Atom *CreateInchi_Atom( int num_atoms )
|
132
|
+
{
|
133
|
+
inchi_Atom *p = (inchi_Atom* ) inchi_calloc(num_atoms, sizeof(inchi_Atom) );
|
134
|
+
return p;
|
135
|
+
}
|
136
|
+
/******************************************************************************************************/
|
137
|
+
void FreeInchi_Input( inchi_Input *inp_at_data )
|
138
|
+
{
|
139
|
+
FreeInchi_Atom( &inp_at_data->atom );
|
140
|
+
FreeInchi_Stereo0D( &inp_at_data->stereo0D );
|
141
|
+
memset( inp_at_data, 0, sizeof(*inp_at_data) );
|
142
|
+
}
|
143
|
+
/*************************************************************************/
|
144
|
+
S_SHORT *is_in_the_slist( S_SHORT *pathAtom, S_SHORT nNextAtom, int nPathLen )
|
145
|
+
{
|
146
|
+
for ( ; nPathLen && *pathAtom != nNextAtom; nPathLen--, pathAtom++ )
|
147
|
+
;
|
148
|
+
return nPathLen? pathAtom : NULL;
|
149
|
+
}
|
150
|
+
/************************************************/
|
151
|
+
int is_element_a_metal( char szEl[] )
|
152
|
+
{
|
153
|
+
static char szMetals[] = "K;V;Y;W;U;"
|
154
|
+
"Li;Be;Na;Mg;Al;Ca;Sc;Ti;Cr;Mn;Fe;Co;Ni;Cu;Zn;Ga;Rb;Sr;Zr;"
|
155
|
+
"Nb;Mo;Tc;Ru;Rh;Pd;Ag;Cd;In;Sn;Sb;Cs;Ba;La;Ce;Pr;Nd;Pm;Sm;"
|
156
|
+
"Eu;Gd;Tb;Dy;Ho;Er;Tm;Yb;Lu;Hf;Ta;Re;Os;Ir;Pt;Au;Hg;Tl;Pb;"
|
157
|
+
"Bi;Po;Fr;Ra;Ac;Th;Pa;Np;Pu;Am;Cm;Bk;Cf;Es;Fm;Md;No;Lr;Rf;";
|
158
|
+
int len = strlen(szEl);
|
159
|
+
char *p;
|
160
|
+
|
161
|
+
if ( 0 < len && len <= 2 &&
|
162
|
+
isalpha( UCINT szEl[0] ) && isupper( szEl[0] ) &&
|
163
|
+
(p = strstr(szMetals, szEl) ) && p[len] == ';' ) {
|
164
|
+
|
165
|
+
return 1; /*return AtType_Metal;*/
|
166
|
+
}
|
167
|
+
return 0;
|
168
|
+
}
|
169
|
+
/*******************************************************************/
|
170
|
+
/* read up to len or tab or LF; if empty read next until finds non-empty line */
|
171
|
+
/* remove leading and trailing white spaces; keep zero termination */
|
172
|
+
/*******************************************************************/
|
173
|
+
char *str_fgetsTab( char *szLine, int len, INCHI_FILE *f )
|
174
|
+
{
|
175
|
+
int length=0, c;
|
176
|
+
len --;
|
177
|
+
while ( length < len && EOF != (c = str_fgetc( f )) ) {
|
178
|
+
if ( c == '\t' )
|
179
|
+
c = '\n';
|
180
|
+
szLine[length++] = (char)c;
|
181
|
+
if ( c == '\n' )
|
182
|
+
break;
|
183
|
+
}
|
184
|
+
if ( !length && EOF == c ) {
|
185
|
+
return NULL;
|
186
|
+
}
|
187
|
+
szLine[length] = '\0';
|
188
|
+
return szLine;
|
189
|
+
}
|
190
|
+
/*******************************************************************/
|
191
|
+
/* read up to len or tab or LF; if empty read next until finds non-empty line */
|
192
|
+
/* remove leading and trailing white spaces; keep zero termination */
|
193
|
+
/*******************************************************************/
|
194
|
+
int my_fgetsTab( char *szLine, int len, INCHI_FILE *f, int *bTooLongLine )
|
195
|
+
{
|
196
|
+
int length;
|
197
|
+
char *p;
|
198
|
+
do {
|
199
|
+
p = str_fgetsTab( szLine, len-1, f );
|
200
|
+
if ( !p ) {
|
201
|
+
*bTooLongLine = 0;
|
202
|
+
return -1; /* end of file or cannot read */
|
203
|
+
}
|
204
|
+
szLine[len-1] = '\0';
|
205
|
+
/*
|
206
|
+
*bTooLongLine = !strchr( szLine, '\n' );
|
207
|
+
*/
|
208
|
+
p = strchr( szLine, '\n' );
|
209
|
+
*bTooLongLine = ( !p && ((int)strlen(szLine)) == len-2 );
|
210
|
+
LtrimRtrim( szLine, &length );
|
211
|
+
} while ( !length );
|
212
|
+
return length;
|
213
|
+
}
|
214
|
+
/*******************************************************************/
|
215
|
+
int my_fgetsTab1( char *szLine, int len, INCHI_FILE *f, int *bTooLongLine )
|
216
|
+
{
|
217
|
+
int length;
|
218
|
+
char *p;
|
219
|
+
/*do {*/
|
220
|
+
p = str_fgetsTab( szLine, len-1, f );
|
221
|
+
if ( !p ) {
|
222
|
+
*bTooLongLine = 0;
|
223
|
+
return -1; /* end of file or cannot read */
|
224
|
+
}
|
225
|
+
szLine[len-1] = '\0';
|
226
|
+
/*
|
227
|
+
*bTooLongLine = !strchr( szLine, '\n' );
|
228
|
+
*/
|
229
|
+
p = strchr( szLine, '\n' );
|
230
|
+
*bTooLongLine = ( !p && ((int)strlen(szLine)) == len-2 );
|
231
|
+
LtrimRtrim( szLine, &length );
|
232
|
+
/*} while ( !length );*/
|
233
|
+
return length;
|
234
|
+
}
|
235
|
+
|
236
|
+
#endif
|
237
|
+
|
238
|
+
|
239
|
+
#ifndef INCHI_MAIN
|
240
|
+
/******************************************************************************************************/
|
241
|
+
inchi_Stereo0D *CreateInchi_Stereo0D( int num_stereo0D )
|
242
|
+
{
|
243
|
+
return (inchi_Stereo0D* ) inchi_calloc(num_stereo0D, sizeof(inchi_Stereo0D) );
|
244
|
+
}
|
245
|
+
/******************************************************************************************************/
|
246
|
+
void FreeInchi_Stereo0D( inchi_Stereo0D **stereo0D )
|
247
|
+
{
|
248
|
+
if ( stereo0D && *stereo0D ) {
|
249
|
+
inchi_free( *stereo0D );
|
250
|
+
*stereo0D = NULL;
|
251
|
+
}
|
252
|
+
}
|
253
|
+
#endif
|
254
|
+
|
255
|
+
|
256
|
+
#if( defined(INCHI_LIBRARY) || defined(INCHI_MAIN) )
|
257
|
+
|
258
|
+
#ifdef INCHI_LIBRARY
|
259
|
+
#define INChITo_Atom ll_INChIToInchi_Atom
|
260
|
+
#else
|
261
|
+
#define INChITo_Atom ee_INChIToIbChI_Atom
|
262
|
+
#define FindToken e_FindToken
|
263
|
+
#define LoadLine e_LoadLine
|
264
|
+
#endif
|
265
|
+
|
266
|
+
#define AT_NUM_BONDS(AT) (AT).num_bonds
|
267
|
+
#define ATOM_NUMBER AT_NUM
|
268
|
+
#define IN_NEIGH_LIST is_in_the_slist
|
269
|
+
#define INPUT_FILE INCHI_FILE
|
270
|
+
#define Create_Atom CreateInchi_Atom
|
271
|
+
#define AT_BONDS_VAL(AT,I) AT[I].num_iso_H[0]
|
272
|
+
#define ISOLATED_ATOM (-15)
|
273
|
+
#define NUM_ISO_Hk(AT,I,K) AT[I].num_iso_H[K+1]
|
274
|
+
#define IS_METAL_ATOM(AT,I) is_element_a_metal( AT[I].elname )
|
275
|
+
|
276
|
+
#else
|
277
|
+
|
278
|
+
#define inchi_Atom inp_ATOM
|
279
|
+
#define AT_NUM_BONDS(AT) (AT).valence
|
280
|
+
#define ATOM_NUMBER AT_NUMB
|
281
|
+
#define IN_NEIGH_LIST is_in_the_list
|
282
|
+
#define inchi_NUMH2(AT,N) NUMH(AT,N)
|
283
|
+
#define INChITo_Atom cc_INChIToInpAtom
|
284
|
+
#define INPUT_FILE FILE
|
285
|
+
#define Create_Atom CreateInpAtom
|
286
|
+
#define AT_BONDS_VAL(AT,I) AT[I].chem_bonds_valence
|
287
|
+
#define ISOLATED_ATOM 15
|
288
|
+
#define NUM_ISO_Hk(AT,I,K) AT[I].num_iso_H[K]
|
289
|
+
#define IS_METAL_ATOM(AT,I) is_el_a_metal( AT[I].el_number )
|
290
|
+
|
291
|
+
#endif
|
292
|
+
|
293
|
+
/*****************************************************************************/
|
294
|
+
/* local prototypes */
|
295
|
+
char *FindToken( INPUT_FILE *inp_molfile, int *bTooLongLine, const char *sToken, int lToken,
|
296
|
+
char *szLine, int nLenLine, char *p, int *res );
|
297
|
+
char *LoadLine( INPUT_FILE *inp_molfile, int *bTooLongLine, int *bItemIsOver, char **s,
|
298
|
+
char *szLine, int nLenLine, int nMinLen2Load, char *p, int *res );
|
299
|
+
|
300
|
+
|
301
|
+
int INChITo_Atom(INPUT_FILE *inp_molfile, MOL_COORD **szCoord,
|
302
|
+
inchi_Stereo0D **stereo0D, int *num_stereo0D,
|
303
|
+
int bDoNotAddH, INPUT_TYPE nInputType, inchi_Atom **at,
|
304
|
+
int max_num_at,
|
305
|
+
int *num_dimensions, int *num_bonds, char *pSdfLabel, char *pSdfValue,
|
306
|
+
long *Id, INCHI_MODE *pInpAtomFlags, int *err, char *pStrErr );
|
307
|
+
|
308
|
+
|
309
|
+
#if( defined(INCHI_LIBRARY) || defined(INCHI_MAIN) )
|
310
|
+
/*****************************************************************************/
|
311
|
+
int INChIToInchi_Atom ( INCHI_FILE *inp_molfile, inchi_Stereo0D **stereo0D, int *num_stereo0D,
|
312
|
+
int bDoNotAddH, INPUT_TYPE nInputType, inchi_Atom **at,
|
313
|
+
int max_num_at,
|
314
|
+
int *num_dimensions, int *num_bonds, char *pSdfLabel, char *pSdfValue,
|
315
|
+
long *Id, INCHI_MODE *pInpAtomFlags, int *err, char *pStrErr );
|
316
|
+
|
317
|
+
int INChIToInchi_Atom ( INCHI_FILE *inp_molfile, inchi_Stereo0D **stereo0D, int *num_stereo0D,
|
318
|
+
int bDoNotAddH, INPUT_TYPE nInputType, inchi_Atom **at,
|
319
|
+
int max_num_at,
|
320
|
+
int *num_dimensions, int *num_bonds, char *pSdfLabel, char *pSdfValue,
|
321
|
+
long *Id, INCHI_MODE *pInpAtomFlags, int *err, char *pStrErr )
|
322
|
+
{
|
323
|
+
return INChITo_Atom ( inp_molfile, NULL, stereo0D, num_stereo0D,
|
324
|
+
bDoNotAddH, nInputType, at, max_num_at,
|
325
|
+
num_dimensions, num_bonds, pSdfLabel, pSdfValue,
|
326
|
+
Id, pInpAtomFlags, err, pStrErr );
|
327
|
+
}
|
328
|
+
|
329
|
+
#else
|
330
|
+
|
331
|
+
/*****************************************************************************/
|
332
|
+
int INChIToInpAtom ( FILE *inp_molfile, MOL_COORD **szCoord,
|
333
|
+
int bDoNotAddH, INPUT_TYPE nInputType, inp_ATOM **at,
|
334
|
+
int max_num_at,
|
335
|
+
int *num_dimensions, int *num_bonds, char *pSdfLabel, char *pSdfValue,
|
336
|
+
long *Id, INCHI_MODE *pInpAtomFlags, int *err, char *pStrErr );
|
337
|
+
|
338
|
+
int INChIToInpAtom ( FILE *inp_molfile, MOL_COORD **szCoord,
|
339
|
+
int bDoNotAddH, INPUT_TYPE nInputType, inp_ATOM **at,
|
340
|
+
int max_num_at,
|
341
|
+
int *num_dimensions, int *num_bonds, char *pSdfLabel, char *pSdfValue,
|
342
|
+
long *Id, INCHI_MODE *pInpAtomFlags, int *err, char *pStrErr )
|
343
|
+
{
|
344
|
+
return INChITo_Atom ( inp_molfile, szCoord, NULL, NULL,
|
345
|
+
bDoNotAddH, nInputType, at, max_num_at,
|
346
|
+
num_dimensions, num_bonds, pSdfLabel, pSdfValue,
|
347
|
+
Id, pInpAtomFlags, err, pStrErr );
|
348
|
+
}
|
349
|
+
|
350
|
+
#endif
|
351
|
+
|
352
|
+
/*****************************************************************************/
|
353
|
+
char *FindToken( INPUT_FILE *inp_molfile, int *bTooLongLine, const char *sToken, int lToken,
|
354
|
+
char *szLine, int nLenLine, char *p, int *res )
|
355
|
+
{
|
356
|
+
char *q;
|
357
|
+
int res2;
|
358
|
+
|
359
|
+
while ( !(q = strstr( p, sToken ) ) ) {
|
360
|
+
if ( (q = strrchr( p, '/' )) && (q + lToken > szLine + *res) ) {
|
361
|
+
*res -= q - szLine; /* res = the length of the szLine to be left in */
|
362
|
+
memmove( szLine, q, *res + 1);
|
363
|
+
} else {
|
364
|
+
*res = 0;
|
365
|
+
}
|
366
|
+
if ( !*bTooLongLine ||
|
367
|
+
0 > (res2 = my_fgetsTab1( szLine + *res, nLenLine - *res - 1,
|
368
|
+
inp_molfile, bTooLongLine ) ) ) {
|
369
|
+
/* the line is over or end of file */
|
370
|
+
return NULL;
|
371
|
+
} else {
|
372
|
+
*res += res2;
|
373
|
+
p = szLine;
|
374
|
+
}
|
375
|
+
}
|
376
|
+
|
377
|
+
return q + lToken;
|
378
|
+
}
|
379
|
+
/*****************************************************************************/
|
380
|
+
char *LoadLine( INPUT_FILE *inp_molfile, int *bTooLongLine, int *bItemIsOver, char **s,
|
381
|
+
char *szLine, int nLenLine, int nMinLen2Load, char *p, int *res )
|
382
|
+
{
|
383
|
+
int pos = p - szLine, res2;
|
384
|
+
if ( !*bItemIsOver && nLenLine - (*res - pos) > nMinLen2Load ) {
|
385
|
+
/* load the next portion if possible */
|
386
|
+
if ( pos ) {
|
387
|
+
*res -= pos;
|
388
|
+
memmove( szLine, p, *res+1 );
|
389
|
+
p = szLine;
|
390
|
+
if ( *s ) {
|
391
|
+
*s -= pos;
|
392
|
+
}
|
393
|
+
pos = 0;
|
394
|
+
}
|
395
|
+
res2 = my_fgetsTab1( szLine + *res, nLenLine - *res - 1, inp_molfile, bTooLongLine );
|
396
|
+
if ( res2 > 0 ) {
|
397
|
+
*bItemIsOver = ( (*s = strchr( p + *res, '/') ) || !*bTooLongLine );
|
398
|
+
*res += res2;
|
399
|
+
} else {
|
400
|
+
*bItemIsOver = 1;
|
401
|
+
}
|
402
|
+
}
|
403
|
+
return p;
|
404
|
+
}
|
405
|
+
/*****************************************************************************/
|
406
|
+
int INChITo_Atom(INPUT_FILE *inp_molfile, MOL_COORD **szCoord,
|
407
|
+
inchi_Stereo0D **stereo0D, int *num_stereo0D,
|
408
|
+
int bDoNotAddH, INPUT_TYPE nInputType, inchi_Atom **at,
|
409
|
+
int max_num_at,
|
410
|
+
int *num_dimensions, int *num_bonds, char *pSdfLabel, char *pSdfValue,
|
411
|
+
long *Id, INCHI_MODE *pInpAtomFlags, int *err, char *pStrErr )
|
412
|
+
{
|
413
|
+
int num_atoms = 0, bFindNext = 0, len, bHeaderRead, bItemIsOver, bErrorMsg, bRestoreInfo;
|
414
|
+
int bFatal = 0, num_struct = 0;
|
415
|
+
int i, k, k2, res, bond_type, bond_stereo1, bond_stereo2, bond_char, neigh, bond_parity, bond_parityNM;
|
416
|
+
int bTooLongLine, res2, bTooLongLine2, pos, hlen, hk;
|
417
|
+
long longID;
|
418
|
+
char szLine[INCHI_LINE_LEN], szNextLine[INCHI_LINE_ADD], *p, *q, *s, parity;
|
419
|
+
int b2D=0, b3D=0, b23D, nNumBonds = 0, bNonZeroXYZ, bNonMetal;
|
420
|
+
int len_stereo0D = 0, max_len_stereo0D = 0;
|
421
|
+
inchi_Stereo0D *atom_stereo0D = NULL;
|
422
|
+
inchi_Atom *atom = NULL;
|
423
|
+
MOL_COORD *pszCoord = NULL;
|
424
|
+
INCHI_MODE InpAtomFlags = 0; /* 0 or FLAG_INP_AT_NONCHIRAL or FLAG_INP_AT_CHIRAL */
|
425
|
+
static char szIsoH[] = "hdt";
|
426
|
+
/* plain tags */
|
427
|
+
static char sStructHdrPln[] = "Structure:";
|
428
|
+
static char sStructHdrPlnNoLblVal[] = " is missing";
|
429
|
+
static char sStructHdrPlnAuxStart[64] =""; /*"$1.1Beta/";*/
|
430
|
+
static int lenStructHdrPlnAuxStart = 0;
|
431
|
+
static char sStructHdrPlnRevAt[] = "/rA:";
|
432
|
+
static char sStructHdrPlnRevBn[] = "/rB:";
|
433
|
+
static char sStructHdrPlnRevXYZ[] = "/rC:";
|
434
|
+
char *sToken;
|
435
|
+
int lToken;
|
436
|
+
if ( !lenStructHdrPlnAuxStart ) {
|
437
|
+
lenStructHdrPlnAuxStart = sprintf( sStructHdrPlnAuxStart, "AuxInfo=" );
|
438
|
+
}
|
439
|
+
|
440
|
+
if ( at ) {
|
441
|
+
if ( *at && max_num_at ) {
|
442
|
+
memset( *at, 0, max_num_at * sizeof(**at) );
|
443
|
+
}
|
444
|
+
#if( defined(INCHI_LIBRARY) || defined(INCHI_MAIN) )
|
445
|
+
if ( stereo0D && num_stereo0D ) {
|
446
|
+
if ( *stereo0D && *num_stereo0D ) {
|
447
|
+
max_len_stereo0D = *num_stereo0D;
|
448
|
+
memset( *stereo0D, 0, max_len_stereo0D * sizeof( **stereo0D ));
|
449
|
+
} else {
|
450
|
+
max_len_stereo0D = 0;
|
451
|
+
}
|
452
|
+
}
|
453
|
+
#else
|
454
|
+
if ( szCoord && *szCoord ) {
|
455
|
+
inchi_free( *szCoord );
|
456
|
+
*szCoord = NULL;
|
457
|
+
}
|
458
|
+
#endif
|
459
|
+
} else {
|
460
|
+
bFindNext = 1;
|
461
|
+
}
|
462
|
+
bHeaderRead = bErrorMsg = bRestoreInfo = 0;
|
463
|
+
*num_dimensions = *num_bonds = 0;
|
464
|
+
|
465
|
+
/*************************************************************/
|
466
|
+
/* extract reversibility info from plain text INChI format */
|
467
|
+
/*************************************************************/
|
468
|
+
if ( nInputType == INPUT_INCHI_PLAIN ) {
|
469
|
+
bHeaderRead = hk = 0;
|
470
|
+
while ( 0 < (res = my_fgetsTab( szLine, sizeof(szLine)-1, inp_molfile, &bTooLongLine ) ) ) {
|
471
|
+
|
472
|
+
/********************* find and interpret structure header ************/
|
473
|
+
if ( !bTooLongLine &&
|
474
|
+
(hlen=sizeof(sStructHdrPln)-1, !memcmp(szLine, sStructHdrPln, hlen)) ) {
|
475
|
+
p = szLine + hlen;
|
476
|
+
longID = 0;
|
477
|
+
num_atoms = 0;
|
478
|
+
/* structure number */
|
479
|
+
longID = strtol( p, &q, 10 );
|
480
|
+
if ( q && q[0] == '.' && q[1] == ' ' ) {
|
481
|
+
p = q+2;
|
482
|
+
}
|
483
|
+
p = p + strspn( p, " \n\r" );
|
484
|
+
|
485
|
+
if ( pSdfLabel ) {
|
486
|
+
pSdfLabel[0] = '\0';
|
487
|
+
}
|
488
|
+
if ( pSdfValue ) {
|
489
|
+
pSdfValue[0] = '\0';
|
490
|
+
}
|
491
|
+
|
492
|
+
if ( *p ) {
|
493
|
+
/* has label name */
|
494
|
+
/*p ++;*/
|
495
|
+
if ( q = strchr( p, '=' ) ) {
|
496
|
+
/* '=' separates label name from the value */
|
497
|
+
len = inchi_min( q-p+1, MAX_SDF_HEADER-1);
|
498
|
+
if ( pSdfLabel ) {
|
499
|
+
mystrncpy( pSdfLabel, p, len );
|
500
|
+
LtrimRtrim( pSdfLabel, &len );
|
501
|
+
}
|
502
|
+
p = q+1;
|
503
|
+
q = p + (int)strlen( p );
|
504
|
+
if ( q-p > 0 ) {
|
505
|
+
len = inchi_min( q-p+1, MAX_SDF_VALUE-1);
|
506
|
+
if ( pSdfValue ) {
|
507
|
+
mystrncpy( pSdfValue, p, len );
|
508
|
+
}
|
509
|
+
p = q;
|
510
|
+
}
|
511
|
+
|
512
|
+
} else
|
513
|
+
if ( q = strstr( p, sStructHdrPlnNoLblVal ) ) {
|
514
|
+
len = inchi_min( q-p+1, MAX_SDF_HEADER-1);
|
515
|
+
if ( pSdfLabel ) {
|
516
|
+
mystrncpy( pSdfLabel, p, len );
|
517
|
+
}
|
518
|
+
p = q+1;
|
519
|
+
}
|
520
|
+
}
|
521
|
+
if ( Id )
|
522
|
+
*Id = longID;
|
523
|
+
|
524
|
+
bHeaderRead = 1;
|
525
|
+
bErrorMsg = bRestoreInfo = 0;
|
526
|
+
} else
|
527
|
+
if ( !memcmp( szLine, sStructHdrPlnAuxStart, lenStructHdrPlnAuxStart) ) {
|
528
|
+
/* found the header of the AuxInfo, read AuxInfo head of the line */
|
529
|
+
if ( !bHeaderRead ) {
|
530
|
+
longID = 0;
|
531
|
+
if ( Id )
|
532
|
+
*Id = longID;
|
533
|
+
if ( pSdfLabel ) {
|
534
|
+
pSdfLabel[0] = '\0';
|
535
|
+
}
|
536
|
+
if ( pSdfValue ) {
|
537
|
+
pSdfValue[0] = '\0';
|
538
|
+
}
|
539
|
+
}
|
540
|
+
bHeaderRead = 0;
|
541
|
+
/* check for empty "AuxInfo=ver//" */
|
542
|
+
p = strchr( szLine + lenStructHdrPlnAuxStart, '/' );
|
543
|
+
if ( p && p[1] == '/' && (!p[2] || '\n' == p[2]) ) {
|
544
|
+
goto bypass_end_of_INChI_plain;
|
545
|
+
}
|
546
|
+
/***************** search for atoms block (plain) **********************/
|
547
|
+
p = szLine;
|
548
|
+
sToken = sStructHdrPlnRevAt;
|
549
|
+
lToken = sizeof(sStructHdrPlnRevAt)-1;
|
550
|
+
/* search for sToken in the line; load next segments of the line if sToken has not found */
|
551
|
+
p = FindToken( inp_molfile, &bTooLongLine, sToken, lToken,
|
552
|
+
szLine, sizeof(szLine), p, &res );
|
553
|
+
if ( !p ) {
|
554
|
+
*err = INCHI_INP_ERROR_ERR;
|
555
|
+
num_atoms = INCHI_INP_ERROR_RET;
|
556
|
+
MOLFILE_ERR_SET (*err, 0, "Missing atom data");
|
557
|
+
goto bypass_end_of_INChI_plain;
|
558
|
+
} else {
|
559
|
+
/* atoms block started */
|
560
|
+
i = 0;
|
561
|
+
res2 = bTooLongLine2 = -1;
|
562
|
+
bItemIsOver = (s = strchr( p, '/') ) || !bTooLongLine;
|
563
|
+
while ( 1 ) {
|
564
|
+
p = LoadLine( inp_molfile, &bTooLongLine, &bItemIsOver, &s,
|
565
|
+
szLine, sizeof(szLine), INCHI_LINE_ADD, p, &res );
|
566
|
+
if ( !i ) {
|
567
|
+
/* allocate atom */
|
568
|
+
num_atoms = strtol( p, &q, 10 );
|
569
|
+
if ( !num_atoms || !q || !*q ) {
|
570
|
+
num_atoms = 0; /* no atom data */
|
571
|
+
goto bypass_end_of_INChI_plain;
|
572
|
+
}
|
573
|
+
p = q;
|
574
|
+
/* Molfile chirality flag */
|
575
|
+
switch( *p ) {
|
576
|
+
case 'c':
|
577
|
+
InpAtomFlags |= FLAG_INP_AT_CHIRAL;
|
578
|
+
p ++;
|
579
|
+
break;
|
580
|
+
case 'n':
|
581
|
+
InpAtomFlags |= FLAG_INP_AT_NONCHIRAL;
|
582
|
+
p ++;
|
583
|
+
break;
|
584
|
+
}
|
585
|
+
if ( at && *at ) {
|
586
|
+
if ( num_atoms > max_num_at ) {
|
587
|
+
inchi_free( *at );
|
588
|
+
*at = NULL;
|
589
|
+
} else {
|
590
|
+
memset( *at, 0, max_num_at * sizeof( **at ) );
|
591
|
+
atom = *at;
|
592
|
+
}
|
593
|
+
}
|
594
|
+
if ( !at || !*at ) {
|
595
|
+
atom = Create_Atom( num_atoms+1 );
|
596
|
+
if ( !atom ) {
|
597
|
+
num_atoms = INCHI_INP_FATAL_RET; /* was -1; error */
|
598
|
+
*err = INCHI_INP_FATAL_ERR;
|
599
|
+
MOLFILE_ERR_SET (*err, 0, "Out of RAM");
|
600
|
+
goto bypass_end_of_INChI_plain;
|
601
|
+
}
|
602
|
+
}
|
603
|
+
if ( stereo0D && *stereo0D ) {
|
604
|
+
if ( num_atoms > max_len_stereo0D ) {
|
605
|
+
FreeInchi_Stereo0D( stereo0D );
|
606
|
+
} else {
|
607
|
+
memset( *stereo0D, 0, max_len_stereo0D * sizeof( **stereo0D ) );
|
608
|
+
atom_stereo0D = *stereo0D;
|
609
|
+
}
|
610
|
+
}
|
611
|
+
if ( !stereo0D || !*stereo0D ) {
|
612
|
+
max_len_stereo0D = num_atoms+1;
|
613
|
+
atom_stereo0D = CreateInchi_Stereo0D( max_len_stereo0D );
|
614
|
+
if ( !atom_stereo0D ) {
|
615
|
+
num_atoms = INCHI_INP_FATAL_RET; /* fatal error: cannot allocate */
|
616
|
+
*err = INCHI_INP_FATAL_ERR;
|
617
|
+
MOLFILE_ERR_SET (*err, 0, "Out of RAM");
|
618
|
+
goto bypass_end_of_INChI_plain;
|
619
|
+
}
|
620
|
+
}
|
621
|
+
}
|
622
|
+
/* element, first char */
|
623
|
+
if ( !isalpha( UCINT *p ) || !isupper( UCINT *p ) || i >= num_atoms ) {
|
624
|
+
break; /* end of atoms block */
|
625
|
+
}
|
626
|
+
atom[i].elname[0] = *p ++;
|
627
|
+
/* element, second char */
|
628
|
+
if ( isalpha( UCINT *p ) && islower( UCINT *p ) ) {
|
629
|
+
atom[i].elname[1] = *p ++;
|
630
|
+
}
|
631
|
+
#if( defined(INCHI_LIBRARY) || defined(INCHI_MAIN) )
|
632
|
+
#else
|
633
|
+
atom[i].el_number = get_periodic_table_number( atom[i].elname );
|
634
|
+
#endif
|
635
|
+
/* bonds' valence */
|
636
|
+
if ( isdigit( UCINT *p ) ) {
|
637
|
+
AT_BONDS_VAL(atom,i) = (char)strtol( p, &q, 10 );
|
638
|
+
if ( !AT_BONDS_VAL(atom,i) )
|
639
|
+
AT_BONDS_VAL(atom,i) = ISOLATED_ATOM; /* same convention as in MOLfile, found zero bonds valence */
|
640
|
+
p = q;
|
641
|
+
}
|
642
|
+
/* charge */
|
643
|
+
atom[i].charge = (*p == '+')? 1 : (*p == '-')? -1 : 0;
|
644
|
+
if ( atom[i].charge ) {
|
645
|
+
p ++;
|
646
|
+
if ( isdigit( UCINT *p ) ) {
|
647
|
+
atom[i].charge *= (S_CHAR)(strtol( p, &q, 10 ) & CHAR_MASK);
|
648
|
+
p = q;
|
649
|
+
}
|
650
|
+
}
|
651
|
+
/* radical */
|
652
|
+
if ( *p == '.' ) {
|
653
|
+
p ++;
|
654
|
+
if ( isdigit( UCINT *p ) ) {
|
655
|
+
atom[i].radical = (S_CHAR)strtol( p, &q, 10 );
|
656
|
+
p = q;
|
657
|
+
}
|
658
|
+
}
|
659
|
+
/* isotopic mass */
|
660
|
+
if ( *p == 'i' ) {
|
661
|
+
p ++;
|
662
|
+
if ( isdigit( UCINT *p ) ) {
|
663
|
+
int mw = strtol( p, &q, 10 );
|
664
|
+
p = q;
|
665
|
+
#if( defined(INCHI_LIBRARY) || defined(INCHI_MAIN) )
|
666
|
+
atom[i].isotopic_mass = mw;
|
667
|
+
#else
|
668
|
+
mw -= get_atw_from_elnum( atom[i].el_number );
|
669
|
+
if ( mw >= 0 )
|
670
|
+
mw ++;
|
671
|
+
atom[i].iso_atw_diff = mw;
|
672
|
+
#endif
|
673
|
+
}
|
674
|
+
}
|
675
|
+
/* parity */
|
676
|
+
switch( *p ) {
|
677
|
+
case 'o':
|
678
|
+
parity = INCHI_PARITY_ODD;
|
679
|
+
p ++;
|
680
|
+
break;
|
681
|
+
case 'e':
|
682
|
+
parity = INCHI_PARITY_EVEN;
|
683
|
+
p ++;
|
684
|
+
break;
|
685
|
+
case 'u':
|
686
|
+
parity = INCHI_PARITY_UNKNOWN;
|
687
|
+
p ++;
|
688
|
+
break;
|
689
|
+
case '?':
|
690
|
+
parity = INCHI_PARITY_UNDEFINED;
|
691
|
+
p ++;
|
692
|
+
break;
|
693
|
+
default:
|
694
|
+
parity = 0;
|
695
|
+
break;
|
696
|
+
}
|
697
|
+
if ( parity ) {
|
698
|
+
atom_stereo0D[len_stereo0D].central_atom = i;
|
699
|
+
atom_stereo0D[len_stereo0D].parity = parity;
|
700
|
+
atom_stereo0D[len_stereo0D].type = INCHI_StereoType_Tetrahedral;
|
701
|
+
len_stereo0D ++;
|
702
|
+
}
|
703
|
+
/* isotopic h, d, t */
|
704
|
+
for ( k = 0; k < NUM_H_ISOTOPES; k ++ ) {
|
705
|
+
if ( *p == szIsoH[k] ) {
|
706
|
+
NUM_ISO_Hk(atom,i,k) = 1;
|
707
|
+
p ++;
|
708
|
+
if ( isdigit( UCINT *p ) ) {
|
709
|
+
NUM_ISO_Hk(atom,i,k) = (char)strtol( p, &q, 10 );
|
710
|
+
p = q;
|
711
|
+
}
|
712
|
+
}
|
713
|
+
}
|
714
|
+
i ++;
|
715
|
+
}
|
716
|
+
if ( !bItemIsOver || i != num_atoms || s && p != s ) {
|
717
|
+
num_atoms = INCHI_INP_ERROR_RET; /* error */
|
718
|
+
*err = INCHI_INP_ERROR_ERR;
|
719
|
+
MOLFILE_ERR_SET (*err, 0, "Wrong number of atoms");
|
720
|
+
goto bypass_end_of_INChI_plain;
|
721
|
+
}
|
722
|
+
}
|
723
|
+
/***************** search for bonds block (plain) and read it *****************/
|
724
|
+
/*p = szLine;*/
|
725
|
+
sToken = sStructHdrPlnRevBn;
|
726
|
+
lToken = sizeof(sStructHdrPlnRevBn)-1;
|
727
|
+
/* search for sToken in the line; load next segments of the line if sToken has not found */
|
728
|
+
p = FindToken( inp_molfile, &bTooLongLine, sToken, lToken,
|
729
|
+
szLine, sizeof(szLine), p, &res );
|
730
|
+
if ( !p ) {
|
731
|
+
num_atoms = INCHI_INP_ERROR_RET; /* error */
|
732
|
+
*err = INCHI_INP_ERROR_ERR;
|
733
|
+
MOLFILE_ERR_SET (*err, 0, "Missing bonds data");
|
734
|
+
goto bypass_end_of_INChI_plain;
|
735
|
+
} else {
|
736
|
+
/* bonds block started */
|
737
|
+
i = 1;
|
738
|
+
res2 = bTooLongLine2 = -1;
|
739
|
+
bItemIsOver = (s = strchr( p, '/') ) || !bTooLongLine;
|
740
|
+
if ( 1 == num_atoms ) {
|
741
|
+
/* needed because the next '/' may be still out of szLine */
|
742
|
+
p = LoadLine( inp_molfile, &bTooLongLine, &bItemIsOver, &s,
|
743
|
+
szLine, sizeof(szLine), INCHI_LINE_ADD, p, &res );
|
744
|
+
}
|
745
|
+
while ( i < num_atoms ) {
|
746
|
+
p = LoadLine( inp_molfile, &bTooLongLine, &bItemIsOver, &s,
|
747
|
+
szLine, sizeof(szLine), INCHI_LINE_ADD, p, &res );
|
748
|
+
if ( i >= num_atoms || s && p >= s ) {
|
749
|
+
break; /* end of bonds (plain) */
|
750
|
+
}
|
751
|
+
/* bond, first char */
|
752
|
+
if ( *p == ';' ) {
|
753
|
+
p ++;
|
754
|
+
i ++;
|
755
|
+
continue;
|
756
|
+
}
|
757
|
+
if ( !isalpha( UCINT *p ) ) {
|
758
|
+
num_atoms = INCHI_INP_ERROR_RET; /* error */
|
759
|
+
*err = INCHI_INP_ERROR_ERR;
|
760
|
+
MOLFILE_ERR_SET (*err, 0, "Wrong bonds data");
|
761
|
+
goto bypass_end_of_INChI_plain;
|
762
|
+
}
|
763
|
+
bond_char = *p ++;
|
764
|
+
/* bond parity */
|
765
|
+
switch( *p ) {
|
766
|
+
case '-':
|
767
|
+
bond_parity = INCHI_PARITY_ODD;
|
768
|
+
p ++;
|
769
|
+
break;
|
770
|
+
case '+':
|
771
|
+
bond_parity = INCHI_PARITY_EVEN;
|
772
|
+
p ++;
|
773
|
+
break;
|
774
|
+
case 'u':
|
775
|
+
bond_parity = INCHI_PARITY_UNKNOWN;
|
776
|
+
p ++;
|
777
|
+
break;
|
778
|
+
case '?':
|
779
|
+
bond_parity = INCHI_PARITY_UNDEFINED;
|
780
|
+
p ++;
|
781
|
+
break;
|
782
|
+
default:
|
783
|
+
bond_parity = 0;
|
784
|
+
break;
|
785
|
+
}
|
786
|
+
if ( bond_parity ) {
|
787
|
+
switch( *p ) {
|
788
|
+
case '-':
|
789
|
+
bond_parityNM = INCHI_PARITY_ODD;
|
790
|
+
p ++;
|
791
|
+
break;
|
792
|
+
case '+':
|
793
|
+
bond_parityNM = INCHI_PARITY_EVEN;
|
794
|
+
p ++;
|
795
|
+
break;
|
796
|
+
case 'u':
|
797
|
+
bond_parityNM = INCHI_PARITY_UNKNOWN;
|
798
|
+
p ++;
|
799
|
+
break;
|
800
|
+
case '?':
|
801
|
+
bond_parityNM = INCHI_PARITY_UNDEFINED;
|
802
|
+
p ++;
|
803
|
+
break;
|
804
|
+
default:
|
805
|
+
bond_parityNM = 0;
|
806
|
+
break;
|
807
|
+
}
|
808
|
+
} else {
|
809
|
+
bond_parityNM = 0;
|
810
|
+
}
|
811
|
+
|
812
|
+
/* neighbor of the current atom */
|
813
|
+
if ( !isdigit( UCINT *p ) ) {
|
814
|
+
num_atoms = INCHI_INP_ERROR_RET; /* error */
|
815
|
+
*err = INCHI_INP_ERROR_ERR;
|
816
|
+
MOLFILE_ERR_SET (*err, 0, "Wrong bonds data");
|
817
|
+
goto bypass_end_of_INChI_plain;
|
818
|
+
}
|
819
|
+
neigh = (int)strtol( p, &q, 10 )-1;
|
820
|
+
|
821
|
+
if ( i >= num_atoms || neigh >= num_atoms ) {
|
822
|
+
num_atoms = INCHI_INP_ERROR_RET; /* error */
|
823
|
+
*err = INCHI_INP_ERROR_ERR;
|
824
|
+
MOLFILE_ERR_SET (*err, 0, "Bond to nonexistent atom");
|
825
|
+
goto bypass_end_of_INChI_plain;
|
826
|
+
}
|
827
|
+
p = q;
|
828
|
+
bond_stereo1 = bond_stereo2 = 0;
|
829
|
+
|
830
|
+
/* bond type & 2D stereo */
|
831
|
+
switch( bond_char ) {
|
832
|
+
case 'v':
|
833
|
+
bond_type = INCHI_BOND_TYPE_SINGLE;
|
834
|
+
bond_stereo1 = INCHI_BOND_STEREO_SINGLE_1EITHER;
|
835
|
+
bond_stereo2 = INCHI_BOND_STEREO_SINGLE_2EITHER;
|
836
|
+
break;
|
837
|
+
case 'V':
|
838
|
+
bond_type = INCHI_BOND_TYPE_SINGLE;
|
839
|
+
bond_stereo1 = INCHI_BOND_STEREO_SINGLE_2EITHER;
|
840
|
+
bond_stereo2 = INCHI_BOND_STEREO_SINGLE_1EITHER;
|
841
|
+
break;
|
842
|
+
case 'w':
|
843
|
+
bond_type = INCHI_BOND_TYPE_DOUBLE;
|
844
|
+
bond_stereo1 =
|
845
|
+
bond_stereo2 = INCHI_BOND_STEREO_DOUBLE_EITHER;
|
846
|
+
break;
|
847
|
+
case 's':
|
848
|
+
bond_type = INCHI_BOND_TYPE_SINGLE;
|
849
|
+
break;
|
850
|
+
case 'd':
|
851
|
+
bond_type = INCHI_BOND_TYPE_DOUBLE;
|
852
|
+
break;
|
853
|
+
case 't':
|
854
|
+
bond_type = INCHI_BOND_TYPE_TRIPLE;
|
855
|
+
break;
|
856
|
+
case 'a':
|
857
|
+
bond_type = INCHI_BOND_TYPE_ALTERN;
|
858
|
+
break;
|
859
|
+
case 'p':
|
860
|
+
bond_type = INCHI_BOND_TYPE_SINGLE;
|
861
|
+
bond_stereo1 = INCHI_BOND_STEREO_SINGLE_1UP;
|
862
|
+
bond_stereo2 = INCHI_BOND_STEREO_SINGLE_2UP;
|
863
|
+
break;
|
864
|
+
case 'P':
|
865
|
+
bond_type = INCHI_BOND_TYPE_SINGLE;
|
866
|
+
bond_stereo1 = INCHI_BOND_STEREO_SINGLE_2UP;
|
867
|
+
bond_stereo2 = INCHI_BOND_STEREO_SINGLE_1UP;
|
868
|
+
break;
|
869
|
+
case 'n':
|
870
|
+
bond_type = INCHI_BOND_TYPE_SINGLE;
|
871
|
+
bond_stereo1 = INCHI_BOND_STEREO_SINGLE_1DOWN;
|
872
|
+
bond_stereo2 = INCHI_BOND_STEREO_SINGLE_2DOWN;
|
873
|
+
break;
|
874
|
+
case 'N':
|
875
|
+
bond_type = INCHI_BOND_TYPE_SINGLE;
|
876
|
+
bond_stereo1 = INCHI_BOND_STEREO_SINGLE_2DOWN;
|
877
|
+
bond_stereo2 = INCHI_BOND_STEREO_SINGLE_1DOWN;
|
878
|
+
break;
|
879
|
+
default:
|
880
|
+
num_atoms = INCHI_INP_ERROR_RET; /* error */
|
881
|
+
*err = INCHI_INP_ERROR_ERR;
|
882
|
+
MOLFILE_ERR_SET (*err, 0, "Wrong bond type");
|
883
|
+
goto bypass_end_of_INChI_plain;
|
884
|
+
}
|
885
|
+
k = AT_NUM_BONDS(atom[i]) ++;
|
886
|
+
atom[i].bond_type[k] = bond_type;
|
887
|
+
atom[i].bond_stereo[k] = bond_stereo1;
|
888
|
+
atom[i].neighbor[k] = (ATOM_NUMBER)neigh;
|
889
|
+
k2 = AT_NUM_BONDS(atom[neigh]) ++;
|
890
|
+
atom[neigh].bond_type[k2] = bond_type;
|
891
|
+
atom[neigh].bond_stereo[k2] = bond_stereo2;
|
892
|
+
atom[neigh].neighbor[k2] = (ATOM_NUMBER)i;
|
893
|
+
bond_parity |= (bond_parityNM << SB_PARITY_SHFT);
|
894
|
+
|
895
|
+
if ( bond_parity ) {
|
896
|
+
if ( max_len_stereo0D <= len_stereo0D ) {
|
897
|
+
/* realloc atom_Stereo0D */
|
898
|
+
inchi_Stereo0D *new_atom_stereo0D = CreateInchi_Stereo0D( max_len_stereo0D+num_atoms );
|
899
|
+
if ( !new_atom_stereo0D ) {
|
900
|
+
num_atoms = INCHI_INP_FATAL_RET; /* fatal error: cannot allocate */
|
901
|
+
*err = INCHI_INP_FATAL_ERR;
|
902
|
+
MOLFILE_ERR_SET (*err, 0, "Out of RAM");
|
903
|
+
goto bypass_end_of_INChI_plain;
|
904
|
+
}
|
905
|
+
memcpy( new_atom_stereo0D, atom_stereo0D, len_stereo0D * sizeof(*atom_stereo0D) );
|
906
|
+
FreeInchi_Stereo0D( &atom_stereo0D );
|
907
|
+
atom_stereo0D = new_atom_stereo0D;
|
908
|
+
max_len_stereo0D += num_atoms;
|
909
|
+
}
|
910
|
+
/* (a) i may be allene endpoint and neigh = allene middle point or
|
911
|
+
(b) i may be allene middle point and neigh = allene endpoint
|
912
|
+
!!!!! CURRENTLY ONLY (b) IS ALLOWED !!!!!
|
913
|
+
*/
|
914
|
+
atom_stereo0D[len_stereo0D].neighbor[1] = neigh; /* neigh < i */
|
915
|
+
atom_stereo0D[len_stereo0D].neighbor[2] = i;
|
916
|
+
atom_stereo0D[len_stereo0D].parity = bond_parity;
|
917
|
+
atom_stereo0D[len_stereo0D].type = INCHI_StereoType_DoubleBond; /* incl allenes & cumulenes */
|
918
|
+
len_stereo0D ++;
|
919
|
+
}
|
920
|
+
}
|
921
|
+
if ( !bItemIsOver || i != num_atoms || s && p != s ) {
|
922
|
+
num_atoms = INCHI_INP_ERROR_RET; /* error */
|
923
|
+
*err = INCHI_INP_ERROR_ERR;
|
924
|
+
MOLFILE_ERR_SET (*err, 0, "Wrong number of bonds");
|
925
|
+
goto bypass_end_of_INChI_plain;
|
926
|
+
}
|
927
|
+
}
|
928
|
+
/***************** search for coordinates block (plain) **********************/
|
929
|
+
/*p = szLine;*/
|
930
|
+
sToken = sStructHdrPlnRevXYZ;
|
931
|
+
lToken = sizeof(sStructHdrPlnRevXYZ)-1;
|
932
|
+
/* search for sToken in the line; load next segments of the line if sToken has not found */
|
933
|
+
p = FindToken( inp_molfile, &bTooLongLine, sToken, lToken,
|
934
|
+
szLine, sizeof(szLine), p, &res );
|
935
|
+
if ( !p ) {
|
936
|
+
num_atoms = INCHI_INP_ERROR_RET; /* error */
|
937
|
+
*err = INCHI_INP_ERROR_ERR;
|
938
|
+
MOLFILE_ERR_SET (*err, 0, "Missing atom coordinates data");
|
939
|
+
goto bypass_end_of_INChI_plain;
|
940
|
+
} else {
|
941
|
+
/* coordinates block started */
|
942
|
+
if ( pszCoord = (MOL_COORD*)inchi_malloc(inchi_max(num_atoms,1) * sizeof(MOL_COORD)) ) {
|
943
|
+
memset( pszCoord, ' ', inchi_max(num_atoms,1) * sizeof(MOL_COORD));
|
944
|
+
} else {
|
945
|
+
num_atoms = INCHI_INP_FATAL_RET; /* allocation error */
|
946
|
+
*err = INCHI_INP_FATAL_ERR;
|
947
|
+
MOLFILE_ERR_SET (*err, 0, "Out of RAM");
|
948
|
+
goto bypass_end_of_INChI_plain;
|
949
|
+
}
|
950
|
+
i = 0;
|
951
|
+
res2 = bTooLongLine2 = -1;
|
952
|
+
bItemIsOver = (s = strchr( p, '/') ) || !bTooLongLine;
|
953
|
+
while ( i < num_atoms ) {
|
954
|
+
p = LoadLine( inp_molfile, &bTooLongLine, &bItemIsOver, &s,
|
955
|
+
szLine, sizeof(szLine), INCHI_LINE_ADD, p, &res );
|
956
|
+
if ( i >= num_atoms || s && p >= s ) {
|
957
|
+
break; /* end of bonds (plain) */
|
958
|
+
}
|
959
|
+
|
960
|
+
/* coord, first char */
|
961
|
+
if ( *p == ';' ) {
|
962
|
+
for ( k = 0; k < NUM_COORD; k ++ ) {
|
963
|
+
pszCoord[i][LEN_COORD*k + 4] = '0';
|
964
|
+
}
|
965
|
+
p ++;
|
966
|
+
i ++;
|
967
|
+
continue;
|
968
|
+
}
|
969
|
+
for ( k = 0; k < 3; k ++ ) {
|
970
|
+
double xyz;
|
971
|
+
bNonZeroXYZ = 0;
|
972
|
+
if ( *p == ';' ) {
|
973
|
+
pszCoord[i][LEN_COORD*k + 4] = '0';
|
974
|
+
xyz = 0.0;
|
975
|
+
} else
|
976
|
+
if ( *p == ',' ) {
|
977
|
+
/* empty */
|
978
|
+
pszCoord[i][LEN_COORD*k + 4] = '0';
|
979
|
+
xyz = 0.0;
|
980
|
+
p ++;
|
981
|
+
} else {
|
982
|
+
xyz = strtod( p, &q );
|
983
|
+
bNonZeroXYZ = fabs(xyz) > MIN_BOND_LENGTH;
|
984
|
+
if ( q != NULL ) {
|
985
|
+
memcpy( pszCoord[i]+LEN_COORD*k, p, q-p );
|
986
|
+
if ( *q == ',' )
|
987
|
+
q ++;
|
988
|
+
p = q;
|
989
|
+
} else {
|
990
|
+
pszCoord[i][LEN_COORD*k + 4] = '0';
|
991
|
+
}
|
992
|
+
}
|
993
|
+
switch( k ) {
|
994
|
+
case 0:
|
995
|
+
atom[i].x = xyz;
|
996
|
+
b2D |= bNonZeroXYZ;
|
997
|
+
break;
|
998
|
+
case 1:
|
999
|
+
atom[i].y = xyz;
|
1000
|
+
b2D |= bNonZeroXYZ;
|
1001
|
+
break;
|
1002
|
+
case 2:
|
1003
|
+
b3D |= bNonZeroXYZ;
|
1004
|
+
atom[i].z = xyz;
|
1005
|
+
break;
|
1006
|
+
}
|
1007
|
+
}
|
1008
|
+
if ( *p == ';' ) {
|
1009
|
+
p ++; /* end of this triple of coordinates */
|
1010
|
+
i ++;
|
1011
|
+
} else {
|
1012
|
+
num_atoms = INCHI_INP_ERROR_RET; /* error in input data: atoms, bonds & coord must be present together */
|
1013
|
+
*err = INCHI_INP_ERROR_ERR;
|
1014
|
+
MOLFILE_ERR_SET (*err, 0, "Wrong atom coordinates data");
|
1015
|
+
goto bypass_end_of_INChI_plain;
|
1016
|
+
}
|
1017
|
+
}
|
1018
|
+
if ( !bItemIsOver || s && p != s || i != num_atoms ) {
|
1019
|
+
num_atoms = INCHI_INP_ERROR_RET; /* error */
|
1020
|
+
*err = INCHI_INP_ERROR_ERR;
|
1021
|
+
MOLFILE_ERR_SET (*err, 0, "Wrong number of coordinates");
|
1022
|
+
goto bypass_end_of_INChI_plain;
|
1023
|
+
}
|
1024
|
+
} /* end of coordinates */
|
1025
|
+
/* set special valences and implicit H (xml) */
|
1026
|
+
b23D = b2D | b3D;
|
1027
|
+
b2D = b3D = 0;
|
1028
|
+
if ( at ) {
|
1029
|
+
if ( !*at ) {
|
1030
|
+
int a1, a2, n1, n2, valence;
|
1031
|
+
int chem_bonds_valence;
|
1032
|
+
int nX=0, nY=0, nZ=0, nXYZ;
|
1033
|
+
*at = atom;
|
1034
|
+
/* special valences */
|
1035
|
+
for ( bNonMetal = 0; bNonMetal < 1; bNonMetal ++ ) {
|
1036
|
+
for ( a1 = 0; a1 < num_atoms; a1 ++ ) {
|
1037
|
+
int num_bond_type[MAX_INPUT_BOND_TYPE - MIN_INPUT_BOND_TYPE + 1];
|
1038
|
+
#if( defined(INCHI_LIBRARY) || defined(INCHI_MAIN) )
|
1039
|
+
#else
|
1040
|
+
int bHasMetalNeighbor=0;
|
1041
|
+
#endif
|
1042
|
+
memset( num_bond_type, 0, sizeof(num_bond_type) );
|
1043
|
+
|
1044
|
+
valence = AT_BONDS_VAL(atom, a1); /* save atom valence if available */
|
1045
|
+
AT_BONDS_VAL(atom, a1) = 0;
|
1046
|
+
#if( defined(INCHI_LIBRARY) || defined(INCHI_MAIN) )
|
1047
|
+
#else
|
1048
|
+
atom[a1].orig_at_number = a1+1;
|
1049
|
+
#endif
|
1050
|
+
nX = nY = nZ = 0;
|
1051
|
+
for ( n1 = 0; n1 < AT_NUM_BONDS(atom[a1]); n1 ++ ) {
|
1052
|
+
bond_type = atom[a1].bond_type[n1] - MIN_INPUT_BOND_TYPE;
|
1053
|
+
if ( bond_type < 0 || bond_type > MAX_INPUT_BOND_TYPE - MIN_INPUT_BOND_TYPE ) {
|
1054
|
+
bond_type = 0;
|
1055
|
+
MOLFILE_ERR_SET (*err, 0, "Unknown bond type in InChI aux assigned as a single bond");
|
1056
|
+
}
|
1057
|
+
|
1058
|
+
num_bond_type[ bond_type ] ++;
|
1059
|
+
nNumBonds ++;
|
1060
|
+
if ( b23D ) {
|
1061
|
+
neigh = atom[a1].neighbor[n1];
|
1062
|
+
nX |= (fabs(atom[a1].x - atom[neigh].x) > MIN_BOND_LENGTH);
|
1063
|
+
nY |= (fabs(atom[a1].y - atom[neigh].y) > MIN_BOND_LENGTH);
|
1064
|
+
nZ |= (fabs(atom[a1].z - atom[neigh].z) > MIN_BOND_LENGTH);
|
1065
|
+
}
|
1066
|
+
}
|
1067
|
+
chem_bonds_valence = 0;
|
1068
|
+
for ( n1 = 0; MIN_INPUT_BOND_TYPE + n1 <= 3 && MIN_INPUT_BOND_TYPE + n1 <= MAX_INPUT_BOND_TYPE; n1 ++ ) {
|
1069
|
+
chem_bonds_valence += (MIN_INPUT_BOND_TYPE + n1) * num_bond_type[n1];
|
1070
|
+
}
|
1071
|
+
if ( MIN_INPUT_BOND_TYPE <= INCHI_BOND_TYPE_ALTERN && INCHI_BOND_TYPE_ALTERN <= MAX_INPUT_BOND_TYPE &&
|
1072
|
+
( n2 = num_bond_type[INCHI_BOND_TYPE_ALTERN-MIN_INPUT_BOND_TYPE] ) ) {
|
1073
|
+
/* accept input aromatic bonds for now */
|
1074
|
+
switch ( n2 ) {
|
1075
|
+
case 2:
|
1076
|
+
chem_bonds_valence += 3; /* =A- */
|
1077
|
+
break;
|
1078
|
+
case 3:
|
1079
|
+
chem_bonds_valence += 4; /* =A< */
|
1080
|
+
break;
|
1081
|
+
default:
|
1082
|
+
/* if 1 or >= 4 aromatic bonds then replace such bonds with single bonds */
|
1083
|
+
for ( n1 = 0; n1 < AT_NUM_BONDS(atom[a1]); n1 ++ ) {
|
1084
|
+
if ( atom[a1].bond_type[n1] == INCHI_BOND_TYPE_ALTERN ) {
|
1085
|
+
ATOM_NUMBER *p1;
|
1086
|
+
a2 = atom[a1].neighbor[n1];
|
1087
|
+
p1 = IN_NEIGH_LIST( atom[a2].neighbor, (ATOM_NUMBER)a1, AT_NUM_BONDS(atom[a2]) );
|
1088
|
+
if ( p1 ) {
|
1089
|
+
atom[a1].bond_type[n1] =
|
1090
|
+
atom[a2].bond_type[p1-atom[a2].neighbor] = INCHI_BOND_TYPE_SINGLE;
|
1091
|
+
} else {
|
1092
|
+
*err = -2; /* Program error */
|
1093
|
+
MOLFILE_ERR_SET (*err, 0, "Program error interpreting InChI aux");
|
1094
|
+
num_atoms = 0;
|
1095
|
+
goto bypass_end_of_INChI_plain; /* no structure */
|
1096
|
+
}
|
1097
|
+
}
|
1098
|
+
}
|
1099
|
+
chem_bonds_valence += n2;
|
1100
|
+
*err |= 32; /* Unrecognized aromatic bond(s) replaced with single */
|
1101
|
+
MOLFILE_ERR_SET (*err, 0, "Atom has more than 3 aromatic bonds");
|
1102
|
+
break;
|
1103
|
+
}
|
1104
|
+
}
|
1105
|
+
#if( defined(INCHI_LIBRARY) || defined(INCHI_MAIN) )
|
1106
|
+
/*************************************************************************************
|
1107
|
+
*
|
1108
|
+
* Set number of hydrogen atoms
|
1109
|
+
*/
|
1110
|
+
{
|
1111
|
+
int num_iso_H;
|
1112
|
+
num_iso_H = atom[a1].num_iso_H[1] + atom[a1].num_iso_H[2] + atom[a1].num_iso_H[3];
|
1113
|
+
if ( valence == ISOLATED_ATOM ) {
|
1114
|
+
atom[a1].num_iso_H[0] = 0;
|
1115
|
+
} else
|
1116
|
+
if ( valence && valence >= chem_bonds_valence ) {
|
1117
|
+
atom[a1].num_iso_H[0] = valence - chem_bonds_valence;
|
1118
|
+
} else
|
1119
|
+
if ( valence || bDoNotAddH ) {
|
1120
|
+
atom[a1].num_iso_H[0] = 0;
|
1121
|
+
} else
|
1122
|
+
if ( !bDoNotAddH ) {
|
1123
|
+
atom[a1].num_iso_H[0] = -1; /* auto add H */
|
1124
|
+
}
|
1125
|
+
}
|
1126
|
+
#else
|
1127
|
+
atom[a1].chem_bonds_valence = chem_bonds_valence;
|
1128
|
+
atom[a1].num_H = get_num_H( atom[a1].elname, atom[a1].num_H, atom[a1].num_iso_H, atom[a1].charge, atom[a1].radical,
|
1129
|
+
atom[a1].chem_bonds_valence,
|
1130
|
+
valence,
|
1131
|
+
0, bDoNotAddH, bHasMetalNeighbor );
|
1132
|
+
#endif
|
1133
|
+
}
|
1134
|
+
}
|
1135
|
+
nNumBonds /= 2;
|
1136
|
+
if ( b23D && nNumBonds ) {
|
1137
|
+
nXYZ = nX+nY+nZ;
|
1138
|
+
b2D = (nXYZ > 0);
|
1139
|
+
b3D = (nXYZ == 3);
|
1140
|
+
*num_dimensions = b3D? 3 : b2D? 2 : 0;
|
1141
|
+
*num_bonds = nNumBonds;
|
1142
|
+
}
|
1143
|
+
/*======= 0D parities =================================*/
|
1144
|
+
#if( defined(INCHI_LIBRARY) || defined(INCHI_MAIN) )
|
1145
|
+
if ( len_stereo0D > 0 && atom_stereo0D && stereo0D ) {
|
1146
|
+
*stereo0D = atom_stereo0D;
|
1147
|
+
*num_stereo0D = len_stereo0D;
|
1148
|
+
} else {
|
1149
|
+
FreeInchi_Stereo0D( &atom_stereo0D );
|
1150
|
+
*num_stereo0D = len_stereo0D = 0;
|
1151
|
+
}
|
1152
|
+
#endif
|
1153
|
+
for ( i = 0; i < len_stereo0D; i ++ ) {
|
1154
|
+
ATOM_NUMBER *p1, *p2;
|
1155
|
+
int sb_ord_from_a1 = -1, sb_ord_from_a2 = -1, bEnd1 = 0, bEnd2 = 0;
|
1156
|
+
switch( atom_stereo0D[i].type ) {
|
1157
|
+
|
1158
|
+
case INCHI_StereoType_Tetrahedral:
|
1159
|
+
a1 = atom_stereo0D[i].central_atom;
|
1160
|
+
if ( atom_stereo0D[i].parity && (AT_NUM_BONDS(atom[a1]) == 3 || AT_NUM_BONDS(atom[a1]) == 4) ) {
|
1161
|
+
int ii, kk = 0;
|
1162
|
+
if ( AT_NUM_BONDS(atom[a1]) == 3 ) {
|
1163
|
+
atom_stereo0D[i].neighbor[kk++] = a1;
|
1164
|
+
}
|
1165
|
+
for ( ii = 0; ii < AT_NUM_BONDS(atom[a1]); ii ++ ) {
|
1166
|
+
atom_stereo0D[i].neighbor[kk++] = atom[a1].neighbor[ii];
|
1167
|
+
}
|
1168
|
+
}
|
1169
|
+
|
1170
|
+
break;
|
1171
|
+
|
1172
|
+
case INCHI_StereoType_DoubleBond:
|
1173
|
+
#define MAX_CHAIN_LEN 20
|
1174
|
+
a1 = atom_stereo0D[i].neighbor[1];
|
1175
|
+
a2 = atom_stereo0D[i].neighbor[2];
|
1176
|
+
p1 = IN_NEIGH_LIST( atom[a1].neighbor, (ATOM_NUMBER)a2, AT_NUM_BONDS(atom[a1]) );
|
1177
|
+
p2 = IN_NEIGH_LIST( atom[a2].neighbor, (ATOM_NUMBER)a1, AT_NUM_BONDS(atom[a2]) );
|
1178
|
+
if ( !p1 || !p2 ) {
|
1179
|
+
atom_stereo0D[i].type = INCHI_StereoType_None;
|
1180
|
+
atom_stereo0D[i].central_atom = NO_ATOM;
|
1181
|
+
atom_stereo0D[i].neighbor[0] =
|
1182
|
+
atom_stereo0D[i].neighbor[4] = -1;
|
1183
|
+
*err |= 64; /* Error in cumulene stereo */
|
1184
|
+
MOLFILE_ERR_SET (*err, 0, "0D stereobond not recognized");
|
1185
|
+
break;
|
1186
|
+
}
|
1187
|
+
/* streobond, allene, or cumulene */
|
1188
|
+
|
1189
|
+
sb_ord_from_a1 = p1 - atom[a1].neighbor;
|
1190
|
+
sb_ord_from_a2 = p2 - atom[a2].neighbor;
|
1191
|
+
|
1192
|
+
if ( AT_NUM_BONDS(atom[a1]) == 2 &&
|
1193
|
+
atom[a1].bond_type[0] + atom[a1].bond_type[1] == 2*INCHI_BOND_TYPE_DOUBLE &&
|
1194
|
+
0 == inchi_NUMH2(atom, a1) &&
|
1195
|
+
(AT_NUM_BONDS(atom[a2]) != 2 ||
|
1196
|
+
atom[a2].bond_type[0] + atom[a2].bond_type[1] != 2*INCHI_BOND_TYPE_DOUBLE ) ) {
|
1197
|
+
bEnd2 = 1; /* a2 is the end-atom, a1 is middle atom */
|
1198
|
+
}
|
1199
|
+
if ( AT_NUM_BONDS(atom[a2]) == 2 &&
|
1200
|
+
atom[a2].bond_type[0] + atom[a2].bond_type[1] == 2*INCHI_BOND_TYPE_DOUBLE &&
|
1201
|
+
0 == inchi_NUMH2(atom, a2) &&
|
1202
|
+
(AT_NUM_BONDS(atom[a1]) != 2 ||
|
1203
|
+
atom[a1].bond_type[0] + atom[a1].bond_type[1] != 2*INCHI_BOND_TYPE_DOUBLE ) ) {
|
1204
|
+
bEnd1 = 1; /* a1 is the end-atom, a2 is middle atom */
|
1205
|
+
}
|
1206
|
+
|
1207
|
+
if ( bEnd2 + bEnd1 == 1 ) {
|
1208
|
+
/* allene or cumulene */
|
1209
|
+
ATOM_NUMBER chain[MAX_CHAIN_LEN+1], prev, cur, next;
|
1210
|
+
if ( bEnd2 && !bEnd1 ) {
|
1211
|
+
cur = a1;
|
1212
|
+
a1 = a2;
|
1213
|
+
a2 = cur;
|
1214
|
+
sb_ord_from_a1 = sb_ord_from_a2;
|
1215
|
+
}
|
1216
|
+
sb_ord_from_a2 = -1;
|
1217
|
+
cur = a1;
|
1218
|
+
next = a2;
|
1219
|
+
len = 0;
|
1220
|
+
chain[len++] = cur;
|
1221
|
+
chain[len++] = next;
|
1222
|
+
while ( len < MAX_CHAIN_LEN ) { /* arbitrary very high upper limit to prevent infinite loop */
|
1223
|
+
prev = cur;
|
1224
|
+
cur = next;
|
1225
|
+
/* follow double bond path && avoid going back */
|
1226
|
+
if ( AT_NUM_BONDS(atom[cur]) == 2 &&
|
1227
|
+
atom[cur].bond_type[0]+atom[cur].bond_type[1] == 2*INCHI_BOND_TYPE_DOUBLE &&
|
1228
|
+
0 == inchi_NUMH2(atom, cur) ) {
|
1229
|
+
next = atom[cur].neighbor[atom[cur].neighbor[0] == prev];
|
1230
|
+
chain[len++] = next;
|
1231
|
+
} else {
|
1232
|
+
break;
|
1233
|
+
}
|
1234
|
+
}
|
1235
|
+
if ( len > 2 &&
|
1236
|
+
(p2 = IN_NEIGH_LIST( atom[cur].neighbor, (ATOM_NUMBER)prev, AT_NUM_BONDS(atom[cur]))) ) {
|
1237
|
+
sb_ord_from_a2 = p2 - atom[cur].neighbor;
|
1238
|
+
a2 = cur;
|
1239
|
+
/* by design we need to pick up the first non-stereo-bond-neighbor as "sn"-atom */
|
1240
|
+
atom_stereo0D[i].neighbor[0] = atom[a1].neighbor[sb_ord_from_a1 == 0];
|
1241
|
+
atom_stereo0D[i].neighbor[1] = a1;
|
1242
|
+
atom_stereo0D[i].neighbor[2] = a2;
|
1243
|
+
atom_stereo0D[i].neighbor[3] = atom[a2].neighbor[sb_ord_from_a2 == 0];
|
1244
|
+
if ( len % 2 ) {
|
1245
|
+
atom_stereo0D[i].central_atom = chain[len/2];
|
1246
|
+
atom_stereo0D[i].type = INCHI_StereoType_Allene;
|
1247
|
+
} else {
|
1248
|
+
atom_stereo0D[i].central_atom = NO_ATOM;
|
1249
|
+
}
|
1250
|
+
} else {
|
1251
|
+
/* error */
|
1252
|
+
atom_stereo0D[i].type = INCHI_StereoType_None;
|
1253
|
+
atom_stereo0D[i].central_atom = NO_ATOM;
|
1254
|
+
atom_stereo0D[i].neighbor[0] =
|
1255
|
+
atom_stereo0D[i].neighbor[4] = -1;
|
1256
|
+
*err |= 64; /* Error in cumulene stereo */
|
1257
|
+
MOLFILE_ERR_SET (*err, 0, "Cumulene stereo not recognized (0D)");
|
1258
|
+
|
1259
|
+
}
|
1260
|
+
#undef MAX_CHAIN_LEN
|
1261
|
+
} else {
|
1262
|
+
/****** a normal possibly stereogenic bond -- not an allene or cumulene *******/
|
1263
|
+
/* by design we need to pick up the first non-stereo-bond-neighbor as "sn"-atom */
|
1264
|
+
sb_ord_from_a1 = p1 - atom[a1].neighbor;
|
1265
|
+
sb_ord_from_a2 = p2 - atom[a2].neighbor;
|
1266
|
+
atom_stereo0D[i].neighbor[0] = atom[a1].neighbor[p1 == atom[a1].neighbor];
|
1267
|
+
atom_stereo0D[i].neighbor[3] = atom[a2].neighbor[p2 == atom[a2].neighbor];
|
1268
|
+
atom_stereo0D[i].central_atom = NO_ATOM;
|
1269
|
+
}
|
1270
|
+
if ( atom_stereo0D[i].type != INCHI_StereoType_None &&
|
1271
|
+
sb_ord_from_a1 >= 0 && sb_ord_from_a2 >= 0 &&
|
1272
|
+
ATOM_PARITY_WELL_DEF( SB_PARITY_2(atom_stereo0D[i].parity) ) ) {
|
1273
|
+
/* Detected well-defined disconnected stereo
|
1274
|
+
* locate first non-metal neighbors */
|
1275
|
+
int a, n, j, /* k,*/ sb_ord, cur_neigh, min_neigh;
|
1276
|
+
for ( k = 0; k < 2; k ++ ) {
|
1277
|
+
a = k? atom_stereo0D[i].neighbor[2] : atom_stereo0D[i].neighbor[1];
|
1278
|
+
sb_ord = k? sb_ord_from_a2 : sb_ord_from_a1;
|
1279
|
+
min_neigh = num_atoms;
|
1280
|
+
for ( n = j = 0; j < AT_NUM_BONDS(atom[a]); j ++ ) {
|
1281
|
+
cur_neigh = atom[a].neighbor[j];
|
1282
|
+
if ( j != sb_ord && !IS_METAL_ATOM(atom, cur_neigh) ) {
|
1283
|
+
min_neigh = inchi_min( cur_neigh, min_neigh );
|
1284
|
+
}
|
1285
|
+
}
|
1286
|
+
if ( min_neigh < num_atoms ) {
|
1287
|
+
atom_stereo0D[i].neighbor[k?3:0] = min_neigh;
|
1288
|
+
} else {
|
1289
|
+
MOLFILE_ERR_SET (*err, 0, "Cannot find non-metal stereobond neighor (0D)");
|
1290
|
+
}
|
1291
|
+
}
|
1292
|
+
}
|
1293
|
+
|
1294
|
+
break;
|
1295
|
+
}
|
1296
|
+
}
|
1297
|
+
/* end of 0D parities extraction */
|
1298
|
+
/*exit_cycle:;*/
|
1299
|
+
}
|
1300
|
+
#if( defined(INCHI_LIBRARY) || defined(INCHI_MAIN) )
|
1301
|
+
#else
|
1302
|
+
/* transfer atom_stereo0D[] to atom[] */
|
1303
|
+
if ( len_stereo0D ) {
|
1304
|
+
Extract0DParities( atom, num_atoms, atom_stereo0D, len_stereo0D, pStrErr, err );
|
1305
|
+
}
|
1306
|
+
#endif
|
1307
|
+
if ( pInpAtomFlags ) {
|
1308
|
+
/* save chirality flag */
|
1309
|
+
*pInpAtomFlags |= InpAtomFlags;
|
1310
|
+
}
|
1311
|
+
} else
|
1312
|
+
if ( atom ) {
|
1313
|
+
inchi_free( atom );
|
1314
|
+
atom = NULL;
|
1315
|
+
}
|
1316
|
+
#if( defined(INCHI_LIBRARY) || defined(INCHI_MAIN) )
|
1317
|
+
#else
|
1318
|
+
if ( szCoord ) {
|
1319
|
+
*szCoord = pszCoord;
|
1320
|
+
pszCoord = NULL;
|
1321
|
+
} else
|
1322
|
+
#endif
|
1323
|
+
if ( pszCoord ) {
|
1324
|
+
inchi_free( pszCoord );
|
1325
|
+
pszCoord = NULL;
|
1326
|
+
}
|
1327
|
+
goto bypass_end_of_INChI_plain;
|
1328
|
+
/*return num_atoms;*/
|
1329
|
+
}
|
1330
|
+
}
|
1331
|
+
if ( atom_stereo0D ) {
|
1332
|
+
FreeInchi_Stereo0D( &atom_stereo0D );
|
1333
|
+
}
|
1334
|
+
/* end of struct. reading cycle */
|
1335
|
+
if ( res <= 0 ) {
|
1336
|
+
if ( *err == INCHI_INP_ERROR_ERR ) {
|
1337
|
+
return num_atoms;
|
1338
|
+
}
|
1339
|
+
*err = INCHI_INP_EOF_ERR;
|
1340
|
+
return INCHI_INP_EOF_RET; /* no more data */
|
1341
|
+
}
|
1342
|
+
bypass_end_of_INChI_plain:
|
1343
|
+
/* cleanup */
|
1344
|
+
if ( num_atoms == INCHI_INP_ERROR_RET && atom_stereo0D ) {
|
1345
|
+
if ( stereo0D && *stereo0D == atom_stereo0D ) {
|
1346
|
+
*stereo0D = NULL;
|
1347
|
+
*num_stereo0D = 0;
|
1348
|
+
}
|
1349
|
+
FreeInchi_Stereo0D( &atom_stereo0D );
|
1350
|
+
}
|
1351
|
+
while ( bTooLongLine &&
|
1352
|
+
0 < my_fgetsTab1( szLine, sizeof(szLine)-1, inp_molfile, &bTooLongLine ) ) {
|
1353
|
+
;
|
1354
|
+
}
|
1355
|
+
#if( defined(INCHI_LIBRARY) || defined(INCHI_MAIN) )
|
1356
|
+
/* cleanup */
|
1357
|
+
if ( !*at ) {
|
1358
|
+
if ( atom ) {
|
1359
|
+
inchi_free( atom );
|
1360
|
+
atom = NULL;
|
1361
|
+
}
|
1362
|
+
if ( pszCoord ) {
|
1363
|
+
inchi_free( pszCoord );
|
1364
|
+
pszCoord = NULL;
|
1365
|
+
}
|
1366
|
+
}
|
1367
|
+
#endif
|
1368
|
+
|
1369
|
+
return num_atoms;
|
1370
|
+
}
|
1371
|
+
|
1372
|
+
/***********************************************************/
|
1373
|
+
/* extract reversibility info from xml text INChI format */
|
1374
|
+
/* */
|
1375
|
+
/* OBSOLETE CODE because InChI output in XML */
|
1376
|
+
/* does not exist anymore. Unsupported. */
|
1377
|
+
/* */
|
1378
|
+
/***********************************************************/
|
1379
|
+
if ( nInputType == INPUT_INCHI_XML ) {
|
1380
|
+
/* xml tags */
|
1381
|
+
static char sStructHdrXml[] = "<structure";
|
1382
|
+
static char sStructHdrXmlEnd[] = "</structure";
|
1383
|
+
static char sStructHdrXmlNumber[] = "number=\"";
|
1384
|
+
static char sStructHdrXmlIdName[] = "id.name=\"";
|
1385
|
+
static char sStructHdrXmlIdValue[] = "id.value=\"";
|
1386
|
+
static char sStructMsgXmlErr[] = "<message type=\"error (no InChI)\" value=\"";
|
1387
|
+
static char sStructMsgXmlErrFatal[] = "<message type=\"fatal (aborted)\" value=\"";
|
1388
|
+
static char sStructRevXmlRevHdr[] = "<reversibility>";
|
1389
|
+
static char sStructRevXmlRevAt[] = "<atoms>";
|
1390
|
+
static char sStructRevXmlRevAtEnd[] = "</atoms>";
|
1391
|
+
static char sStructRevXmlRevBn[] = "<bonds>";
|
1392
|
+
static char sStructRevXmlRevBnEnd[] = "</bonds>";
|
1393
|
+
static char sStructRevXmlRevXYZ[] = "<xyz>";
|
1394
|
+
static char sStructRevXmlRevXYZEnd[]= "</xyz>";
|
1395
|
+
static char sStructAuxXml[] = "<identifier.auxiliary-info";
|
1396
|
+
static char sStructAuxXmlEnd[] = "</identifier.auxiliary-info";
|
1397
|
+
int bInTheAuxInfo = 0;
|
1398
|
+
|
1399
|
+
while ( 0 < (res = my_fgets( szLine, sizeof(szLine)-1, inp_molfile, &bTooLongLine ) ) ) {
|
1400
|
+
|
1401
|
+
/********************* find and interpret structure header ************/
|
1402
|
+
if ( !memcmp(szLine, sStructHdrXml, sizeof(sStructHdrXml)-1) ) {
|
1403
|
+
num_struct = 1;
|
1404
|
+
p = szLine + sizeof(sStructHdrXml)-1;
|
1405
|
+
longID = 0;
|
1406
|
+
num_atoms = 0;
|
1407
|
+
/* structure number */
|
1408
|
+
if ( q = strstr( p, sStructHdrXmlNumber ) ) {
|
1409
|
+
p = q + sizeof(sStructHdrXmlNumber)-1;
|
1410
|
+
longID = strtol( p, &q, 10);
|
1411
|
+
if ( q && *q == '\"' )
|
1412
|
+
p = q+1;
|
1413
|
+
}
|
1414
|
+
if ( pSdfLabel ) {
|
1415
|
+
pSdfLabel[0] = '\0';
|
1416
|
+
}
|
1417
|
+
if ( pSdfValue ) {
|
1418
|
+
pSdfValue[0] = '\0';
|
1419
|
+
}
|
1420
|
+
/* pSdfLabel */
|
1421
|
+
if ( q = strstr( p, sStructHdrXmlIdName ) ) {
|
1422
|
+
p = q + sizeof(sStructHdrXmlIdName)-1;
|
1423
|
+
q = strchr( p, '\"' );
|
1424
|
+
if ( q ) {
|
1425
|
+
len = inchi_min( q-p+1, MAX_SDF_HEADER-1);
|
1426
|
+
if ( pSdfLabel ) {
|
1427
|
+
mystrncpy( pSdfLabel, p, len );
|
1428
|
+
}
|
1429
|
+
p = q+1;
|
1430
|
+
}
|
1431
|
+
}
|
1432
|
+
/* pSdfValue */
|
1433
|
+
if ( q = strstr( p, sStructHdrXmlIdValue ) ) {
|
1434
|
+
p = q + sizeof(sStructHdrXmlIdValue)-1;
|
1435
|
+
q = strchr( p, '\"' );
|
1436
|
+
if ( q ) {
|
1437
|
+
len = inchi_min( q-p+1, MAX_SDF_VALUE-1);
|
1438
|
+
if ( pSdfValue ) {
|
1439
|
+
mystrncpy( pSdfValue, p, len );
|
1440
|
+
}
|
1441
|
+
p = q+1;
|
1442
|
+
}
|
1443
|
+
}
|
1444
|
+
if ( Id )
|
1445
|
+
*Id = longID;
|
1446
|
+
bHeaderRead = 1;
|
1447
|
+
bErrorMsg = bRestoreInfo = 0;
|
1448
|
+
} else
|
1449
|
+
if ( bHeaderRead && (bFatal=0, len=sizeof(sStructMsgXmlErr)-1, !memcmp(szLine, sStructMsgXmlErr, len)) ||
|
1450
|
+
bHeaderRead && (len=sizeof(sStructMsgXmlErrFatal)-1, !memcmp(szLine, sStructMsgXmlErrFatal, len))&&(bFatal=1)) {
|
1451
|
+
p = szLine+len;
|
1452
|
+
q = strchr( p, '\"' );
|
1453
|
+
if ( q && !bFindNext ) {
|
1454
|
+
int c;
|
1455
|
+
bErrorMsg = 1;
|
1456
|
+
pStrErr[0] = '\0';
|
1457
|
+
c = *q;
|
1458
|
+
*q = '\0';
|
1459
|
+
MOLFILE_ERR_SET (*err, 0, p);
|
1460
|
+
*q = c;
|
1461
|
+
}
|
1462
|
+
*err = bFatal? INCHI_INP_FATAL_ERR : INCHI_INP_ERROR_ERR;
|
1463
|
+
num_atoms = bFatal? INCHI_INP_FATAL_RET : INCHI_INP_ERROR_RET;
|
1464
|
+
goto bypass_end_of_INChI;
|
1465
|
+
} else
|
1466
|
+
if ( bHeaderRead && !memcmp(szLine, sStructAuxXml, sizeof(sStructAuxXml)-1) ) {
|
1467
|
+
bInTheAuxInfo = 1;
|
1468
|
+
} else
|
1469
|
+
if ( bHeaderRead && !memcmp(szLine, sStructAuxXmlEnd, sizeof(sStructAuxXmlEnd)-1) ) {
|
1470
|
+
*err = INCHI_INP_ERROR_ERR;
|
1471
|
+
num_atoms = INCHI_INP_ERROR_RET;
|
1472
|
+
MOLFILE_ERR_SET (*err, 0, "Missing reversibility info" );
|
1473
|
+
goto bypass_end_of_INChI; /* reversibility info not found */
|
1474
|
+
} else
|
1475
|
+
if ( bHeaderRead && bInTheAuxInfo && !memcmp(szLine, sStructRevXmlRevHdr, sizeof(sStructRevXmlRevHdr)-1) ) {
|
1476
|
+
/*********************** atoms xml ***************************/
|
1477
|
+
num_struct = 1;
|
1478
|
+
res = my_fgets( szLine, sizeof(szLine)-1, inp_molfile, &bTooLongLine );
|
1479
|
+
if ( res <= 0 ) {
|
1480
|
+
num_atoms = INCHI_INP_EOF_RET; /* no data, probably end of file */
|
1481
|
+
*err = INCHI_INP_EOF_ERR;
|
1482
|
+
goto bypass_end_of_INChI;
|
1483
|
+
}
|
1484
|
+
if ( memcmp(szLine, sStructRevXmlRevAt, sizeof(sStructRevXmlRevAt)-1) ) {
|
1485
|
+
bHeaderRead = 0; /* invalid reversibility info; look for another header */
|
1486
|
+
continue;
|
1487
|
+
}
|
1488
|
+
/* read (the head of) the atoms line */
|
1489
|
+
res = my_fgets( szLine, sizeof(szLine)-1, inp_molfile, &bTooLongLine );
|
1490
|
+
if ( res <= 0 ) {
|
1491
|
+
num_atoms = INCHI_INP_EOF_RET; /* no data */
|
1492
|
+
*err = INCHI_INP_EOF_ERR;
|
1493
|
+
goto bypass_end_of_INChI;
|
1494
|
+
}
|
1495
|
+
p = szLine;
|
1496
|
+
num_atoms = strtol( p, &q, 10 );
|
1497
|
+
if ( !num_atoms || !q || !*q ) {
|
1498
|
+
num_atoms = INCHI_INP_EOF_RET; /* no atom data */
|
1499
|
+
*err = INCHI_INP_EOF_ERR;
|
1500
|
+
goto bypass_end_of_INChI;
|
1501
|
+
}
|
1502
|
+
p = q;
|
1503
|
+
/* Molfile chirality flag */
|
1504
|
+
switch( *p ) {
|
1505
|
+
case 'c':
|
1506
|
+
InpAtomFlags |= FLAG_INP_AT_CHIRAL;
|
1507
|
+
p ++;
|
1508
|
+
break;
|
1509
|
+
case 'n':
|
1510
|
+
InpAtomFlags |= FLAG_INP_AT_NONCHIRAL;
|
1511
|
+
p ++;
|
1512
|
+
break;
|
1513
|
+
}
|
1514
|
+
if ( at && *at ) {
|
1515
|
+
if ( num_atoms > max_num_at ) {
|
1516
|
+
inchi_free( *at );
|
1517
|
+
*at = NULL;
|
1518
|
+
} else {
|
1519
|
+
memset( *at, 0, max_num_at * sizeof( **at ) );
|
1520
|
+
atom = *at;
|
1521
|
+
}
|
1522
|
+
}
|
1523
|
+
if ( !at || !*at ) {
|
1524
|
+
atom = Create_Atom( num_atoms+1 );
|
1525
|
+
if ( !atom ) {
|
1526
|
+
num_atoms = INCHI_INP_FATAL_RET; /* fatal error: cannot allocate */
|
1527
|
+
*err = INCHI_INP_FATAL_ERR;
|
1528
|
+
MOLFILE_ERR_SET (*err, 0, "Out of RAM");
|
1529
|
+
goto bypass_end_of_INChI;
|
1530
|
+
}
|
1531
|
+
}
|
1532
|
+
if ( stereo0D && *stereo0D ) {
|
1533
|
+
if ( num_atoms > max_len_stereo0D ) {
|
1534
|
+
FreeInchi_Stereo0D( stereo0D );
|
1535
|
+
} else {
|
1536
|
+
memset( *stereo0D, 0, max_len_stereo0D * sizeof( **stereo0D ) );
|
1537
|
+
atom_stereo0D = *stereo0D;
|
1538
|
+
}
|
1539
|
+
}
|
1540
|
+
if ( !stereo0D || !*stereo0D ) {
|
1541
|
+
max_len_stereo0D = num_atoms+1;
|
1542
|
+
atom_stereo0D = CreateInchi_Stereo0D( max_len_stereo0D );
|
1543
|
+
if ( !atom_stereo0D ) {
|
1544
|
+
num_atoms = INCHI_INP_FATAL_RET; /* fatal error: cannot allocate */
|
1545
|
+
*err = INCHI_INP_FATAL_ERR;
|
1546
|
+
MOLFILE_ERR_SET (*err, 0, "Out of RAM");
|
1547
|
+
goto bypass_end_of_INChI;
|
1548
|
+
}
|
1549
|
+
}
|
1550
|
+
|
1551
|
+
i = 0;
|
1552
|
+
bItemIsOver = 0;
|
1553
|
+
res2 = bTooLongLine2 = -1;
|
1554
|
+
|
1555
|
+
/* read all atoms xml */
|
1556
|
+
while ( i < num_atoms ) {
|
1557
|
+
pos = p - szLine;
|
1558
|
+
if ( !bItemIsOver && (int)sizeof(szLine)-res + pos > (int)sizeof(szNextLine) ) {
|
1559
|
+
/* load next line if possible */
|
1560
|
+
res2 = my_fgets( szNextLine, sizeof(szNextLine)-1, inp_molfile, &bTooLongLine2 );
|
1561
|
+
if ( res2 > 0 && memcmp(szNextLine, sStructRevXmlRevAtEnd, sizeof(sStructRevXmlRevAtEnd)-1) ) {
|
1562
|
+
if ( pos ) {
|
1563
|
+
res -= pos; /* number of chars left to process in szLine */
|
1564
|
+
memmove( szLine, p, res*sizeof(szLine[0]) ); /* move them to the start of the line */
|
1565
|
+
}
|
1566
|
+
memcpy( szLine+res, szNextLine, (res2+1)*sizeof(szNextLine[0]) );
|
1567
|
+
res += res2;
|
1568
|
+
szLine[res] = '\0';
|
1569
|
+
bTooLongLine = bTooLongLine2;
|
1570
|
+
p = szLine;
|
1571
|
+
} else {
|
1572
|
+
bItemIsOver = 1;
|
1573
|
+
}
|
1574
|
+
}
|
1575
|
+
/* element, first char */
|
1576
|
+
if ( !isalpha( UCINT *p ) || !isupper( UCINT *p ) || i >= num_atoms ) {
|
1577
|
+
bHeaderRead = 0; /* wrong atom data */
|
1578
|
+
num_atoms = INCHI_INP_ERROR_RET; /* was 0, error */
|
1579
|
+
*err = INCHI_INP_ERROR_ERR; /* 40 */
|
1580
|
+
MOLFILE_ERR_SET (*err, 0, "Wrong atoms data");
|
1581
|
+
goto bypass_end_of_INChI;
|
1582
|
+
}
|
1583
|
+
atom[i].elname[0] = *p ++;
|
1584
|
+
/* element, second char */
|
1585
|
+
if ( isalpha( UCINT *p ) && islower( UCINT *p ) ) {
|
1586
|
+
atom[i].elname[1] = *p ++;
|
1587
|
+
}
|
1588
|
+
#if( defined(INCHI_LIBRARY) || defined(INCHI_MAIN) )
|
1589
|
+
#else
|
1590
|
+
atom[i].el_number = get_periodic_table_number( atom[i].elname );
|
1591
|
+
#endif
|
1592
|
+
/* bonds' valence */
|
1593
|
+
if ( isdigit( UCINT *p ) ) {
|
1594
|
+
AT_BONDS_VAL(atom,i) = (char)strtol( p, &q, 10 );
|
1595
|
+
if ( !AT_BONDS_VAL(atom,i) )
|
1596
|
+
AT_BONDS_VAL(atom,i) = ISOLATED_ATOM; /* same convention as in MOLfile, found zero bonds valence */
|
1597
|
+
p = q;
|
1598
|
+
}
|
1599
|
+
/* charge */
|
1600
|
+
atom[i].charge = (*p == '+')? 1 : (*p == '-')? -1 : 0;
|
1601
|
+
if ( atom[i].charge ) {
|
1602
|
+
p ++;
|
1603
|
+
if ( isdigit( UCINT *p ) ) {
|
1604
|
+
atom[i].charge *= (S_CHAR)(strtol( p, &q, 10 ) & CHAR_MASK);
|
1605
|
+
p = q;
|
1606
|
+
}
|
1607
|
+
}
|
1608
|
+
/* radical */
|
1609
|
+
if ( *p == '.' ) {
|
1610
|
+
p ++;
|
1611
|
+
if ( isdigit( UCINT *p ) ) {
|
1612
|
+
atom[i].radical = (S_CHAR)strtol( p, &q, 10 );
|
1613
|
+
p = q;
|
1614
|
+
}
|
1615
|
+
}
|
1616
|
+
/* isotopic mass */
|
1617
|
+
if ( *p == 'i' ) {
|
1618
|
+
p ++;
|
1619
|
+
if ( isdigit( UCINT *p ) ) {
|
1620
|
+
int mw = strtol( p, &q, 10 );
|
1621
|
+
p = q;
|
1622
|
+
#if( defined(INCHI_LIBRARY) || defined(INCHI_MAIN) )
|
1623
|
+
atom[i].isotopic_mass = mw;
|
1624
|
+
#else
|
1625
|
+
mw -= get_atw_from_elnum( atom[i].el_number );
|
1626
|
+
if ( mw >= 0 )
|
1627
|
+
mw ++;
|
1628
|
+
atom[i].iso_atw_diff = mw;
|
1629
|
+
#endif
|
1630
|
+
}
|
1631
|
+
}
|
1632
|
+
/* parity */
|
1633
|
+
switch( *p ) {
|
1634
|
+
case 'o':
|
1635
|
+
parity = INCHI_PARITY_ODD;
|
1636
|
+
p ++;
|
1637
|
+
break;
|
1638
|
+
case 'e':
|
1639
|
+
parity = INCHI_PARITY_EVEN;
|
1640
|
+
p ++;
|
1641
|
+
break;
|
1642
|
+
case 'u':
|
1643
|
+
parity = INCHI_PARITY_UNKNOWN;
|
1644
|
+
p ++;
|
1645
|
+
break;
|
1646
|
+
case '?':
|
1647
|
+
parity = INCHI_PARITY_UNDEFINED;
|
1648
|
+
p ++;
|
1649
|
+
break;
|
1650
|
+
default:
|
1651
|
+
parity = 0;
|
1652
|
+
break;
|
1653
|
+
}
|
1654
|
+
if ( parity ) {
|
1655
|
+
atom_stereo0D[len_stereo0D].central_atom = i;
|
1656
|
+
atom_stereo0D[len_stereo0D].parity = parity;
|
1657
|
+
atom_stereo0D[len_stereo0D].type = INCHI_StereoType_Tetrahedral;
|
1658
|
+
len_stereo0D ++;
|
1659
|
+
}
|
1660
|
+
/* isotopic h, d, t */
|
1661
|
+
for ( k = 0; k < NUM_H_ISOTOPES; k ++ ) {
|
1662
|
+
if ( *p == szIsoH[k] ) {
|
1663
|
+
NUM_ISO_Hk(atom,i,k) = 1;
|
1664
|
+
p ++;
|
1665
|
+
if ( isdigit( UCINT *p ) ) {
|
1666
|
+
NUM_ISO_Hk(atom,i,k) = (char)strtol( p, &q, 10 );
|
1667
|
+
p = q;
|
1668
|
+
}
|
1669
|
+
}
|
1670
|
+
}
|
1671
|
+
i ++;
|
1672
|
+
}
|
1673
|
+
if ( !bItemIsOver || p - szLine != res || i != num_atoms ) {
|
1674
|
+
num_atoms = INCHI_INP_ERROR_RET; /* error */
|
1675
|
+
*err = INCHI_INP_ERROR_ERR;
|
1676
|
+
MOLFILE_ERR_SET (*err, 0, "Wrong number of atoms");
|
1677
|
+
goto bypass_end_of_INChI;
|
1678
|
+
}
|
1679
|
+
/********************** bonds xml ****************************/
|
1680
|
+
res = my_fgets( szLine, sizeof(szLine)-1, inp_molfile, &bTooLongLine );
|
1681
|
+
if ( res <= 0 ) {
|
1682
|
+
num_atoms = 0; /* no data */
|
1683
|
+
goto bypass_end_of_INChI;
|
1684
|
+
}
|
1685
|
+
if ( memcmp(szLine, sStructRevXmlRevBn, sizeof(sStructRevXmlRevBn)-1) ) {
|
1686
|
+
bHeaderRead = 0; /* invalid reversibility info; look for another header */
|
1687
|
+
continue;
|
1688
|
+
}
|
1689
|
+
/* read (the head of) the xml bonds line */
|
1690
|
+
res = my_fgets( szLine, sizeof(szLine)-1, inp_molfile, &bTooLongLine );
|
1691
|
+
if ( res <= 0 ) {
|
1692
|
+
num_atoms = INCHI_INP_ERROR_RET; /* was 0; error: no data -- eof? */
|
1693
|
+
*err = INCHI_INP_ERROR_ERR;
|
1694
|
+
goto bypass_end_of_INChI;
|
1695
|
+
}
|
1696
|
+
i = 1;
|
1697
|
+
bItemIsOver = 0;
|
1698
|
+
res2 = bTooLongLine2 = -1;
|
1699
|
+
p = szLine;
|
1700
|
+
if ( !memcmp(szLine, sStructRevXmlRevBnEnd, sizeof(sStructRevXmlRevBnEnd)-1) ) {
|
1701
|
+
/* empty bonds section */
|
1702
|
+
res = 0;
|
1703
|
+
bItemIsOver = 1;
|
1704
|
+
}
|
1705
|
+
/* read all bonds (xml), starting from atom 1 (not 0) */
|
1706
|
+
while ( i < num_atoms ) {
|
1707
|
+
pos = p - szLine;
|
1708
|
+
if ( !bItemIsOver && (int)sizeof(szLine)-res + pos > (int)sizeof(szNextLine) ) {
|
1709
|
+
/* load next line if possible */
|
1710
|
+
res2 = my_fgets( szNextLine, sizeof(szNextLine)-1, inp_molfile, &bTooLongLine2 );
|
1711
|
+
if ( res2 > 0 && memcmp(szNextLine, sStructRevXmlRevBnEnd, sizeof(sStructRevXmlRevBnEnd)-1) ) {
|
1712
|
+
if ( pos ) {
|
1713
|
+
res -= pos; /* number of chars left to process in szLine */
|
1714
|
+
memmove( szLine, p, res*sizeof(szLine[0]) ); /* move them to the start of the line */
|
1715
|
+
}
|
1716
|
+
memcpy( szLine+res, szNextLine, (res2+1)*sizeof(szNextLine[0]) );
|
1717
|
+
res += res2;
|
1718
|
+
szLine[res] = '\0';
|
1719
|
+
bTooLongLine = bTooLongLine2;
|
1720
|
+
p = szLine;
|
1721
|
+
} else {
|
1722
|
+
bItemIsOver = 1;
|
1723
|
+
}
|
1724
|
+
}
|
1725
|
+
if ( i >= num_atoms ) {
|
1726
|
+
break;
|
1727
|
+
}
|
1728
|
+
/* bond, first char */
|
1729
|
+
if ( *p == ';' ) {
|
1730
|
+
p ++;
|
1731
|
+
i ++;
|
1732
|
+
continue;
|
1733
|
+
}
|
1734
|
+
if ( !isalpha( UCINT *p ) ) {
|
1735
|
+
num_atoms = INCHI_INP_ERROR_RET; /* error in input data */
|
1736
|
+
*err = INCHI_INP_ERROR_ERR;
|
1737
|
+
MOLFILE_ERR_SET (*err, 0, "Wrong bonds data");
|
1738
|
+
goto bypass_end_of_INChI;
|
1739
|
+
}
|
1740
|
+
bond_char = *p ++;
|
1741
|
+
/* bond parity */
|
1742
|
+
switch( *p ) {
|
1743
|
+
case '-':
|
1744
|
+
bond_parity = INCHI_PARITY_ODD;
|
1745
|
+
p ++;
|
1746
|
+
break;
|
1747
|
+
case '+':
|
1748
|
+
bond_parity = INCHI_PARITY_EVEN;
|
1749
|
+
p ++;
|
1750
|
+
break;
|
1751
|
+
case 'u':
|
1752
|
+
bond_parity = INCHI_PARITY_UNKNOWN;
|
1753
|
+
p ++;
|
1754
|
+
break;
|
1755
|
+
case '?':
|
1756
|
+
bond_parity = INCHI_PARITY_UNDEFINED;
|
1757
|
+
p ++;
|
1758
|
+
break;
|
1759
|
+
default:
|
1760
|
+
bond_parity = 0;
|
1761
|
+
break;
|
1762
|
+
}
|
1763
|
+
if ( bond_parity ) {
|
1764
|
+
switch( *p ) {
|
1765
|
+
case '-':
|
1766
|
+
bond_parityNM = INCHI_PARITY_ODD;
|
1767
|
+
p ++;
|
1768
|
+
break;
|
1769
|
+
case '+':
|
1770
|
+
bond_parityNM = INCHI_PARITY_EVEN;
|
1771
|
+
p ++;
|
1772
|
+
break;
|
1773
|
+
case 'u':
|
1774
|
+
bond_parityNM = INCHI_PARITY_UNKNOWN;
|
1775
|
+
p ++;
|
1776
|
+
break;
|
1777
|
+
case '?':
|
1778
|
+
bond_parityNM = INCHI_PARITY_UNDEFINED;
|
1779
|
+
p ++;
|
1780
|
+
break;
|
1781
|
+
default:
|
1782
|
+
bond_parityNM = 0;
|
1783
|
+
break;
|
1784
|
+
}
|
1785
|
+
} else {
|
1786
|
+
bond_parityNM = 0;
|
1787
|
+
}
|
1788
|
+
|
1789
|
+
/* neighbor of the current atom */
|
1790
|
+
if ( !isdigit( UCINT *p ) ) {
|
1791
|
+
num_atoms = INCHI_INP_ERROR_RET; /* error in input data */
|
1792
|
+
*err = INCHI_INP_ERROR_ERR;
|
1793
|
+
MOLFILE_ERR_SET (*err, 0, "Wrong bonds data");
|
1794
|
+
goto bypass_end_of_INChI;
|
1795
|
+
}
|
1796
|
+
neigh = (int)strtol( p, &q, 10 )-1;
|
1797
|
+
|
1798
|
+
if ( i >= num_atoms || neigh >= num_atoms ) {
|
1799
|
+
num_atoms = INCHI_INP_ERROR_RET; /* error in input data */
|
1800
|
+
*err = INCHI_INP_ERROR_ERR;
|
1801
|
+
MOLFILE_ERR_SET (*err, 0, "Bond to nonexistent atom");
|
1802
|
+
goto bypass_end_of_INChI;
|
1803
|
+
}
|
1804
|
+
p = q;
|
1805
|
+
bond_stereo1 = bond_stereo2 = 0;
|
1806
|
+
|
1807
|
+
/* bond type & 2D stereo */
|
1808
|
+
switch( bond_char ) {
|
1809
|
+
case 'v':
|
1810
|
+
bond_type = INCHI_BOND_TYPE_SINGLE;
|
1811
|
+
bond_stereo1 = INCHI_BOND_STEREO_SINGLE_1EITHER;
|
1812
|
+
bond_stereo2 = INCHI_BOND_STEREO_SINGLE_2EITHER;
|
1813
|
+
break;
|
1814
|
+
case 'V':
|
1815
|
+
bond_type = INCHI_BOND_TYPE_SINGLE;
|
1816
|
+
bond_stereo1 = INCHI_BOND_STEREO_SINGLE_2EITHER;
|
1817
|
+
bond_stereo2 = INCHI_BOND_STEREO_SINGLE_1EITHER;
|
1818
|
+
break;
|
1819
|
+
case 'w':
|
1820
|
+
bond_type = INCHI_BOND_TYPE_DOUBLE;
|
1821
|
+
bond_stereo1 =
|
1822
|
+
bond_stereo2 = INCHI_BOND_STEREO_DOUBLE_EITHER;
|
1823
|
+
break;
|
1824
|
+
case 's':
|
1825
|
+
bond_type = INCHI_BOND_TYPE_SINGLE;
|
1826
|
+
break;
|
1827
|
+
case 'd':
|
1828
|
+
bond_type = INCHI_BOND_TYPE_DOUBLE;
|
1829
|
+
break;
|
1830
|
+
case 't':
|
1831
|
+
bond_type = INCHI_BOND_TYPE_TRIPLE;
|
1832
|
+
break;
|
1833
|
+
case 'a':
|
1834
|
+
bond_type = INCHI_BOND_TYPE_ALTERN;
|
1835
|
+
break;
|
1836
|
+
case 'p':
|
1837
|
+
bond_type = INCHI_BOND_TYPE_SINGLE;
|
1838
|
+
bond_stereo1 = INCHI_BOND_STEREO_SINGLE_1UP;
|
1839
|
+
bond_stereo2 = INCHI_BOND_STEREO_SINGLE_2UP;
|
1840
|
+
break;
|
1841
|
+
case 'P':
|
1842
|
+
bond_type = INCHI_BOND_TYPE_SINGLE;
|
1843
|
+
bond_stereo1 = INCHI_BOND_STEREO_SINGLE_2UP;
|
1844
|
+
bond_stereo2 = INCHI_BOND_STEREO_SINGLE_1UP;
|
1845
|
+
break;
|
1846
|
+
case 'n':
|
1847
|
+
bond_type = INCHI_BOND_TYPE_SINGLE;
|
1848
|
+
bond_stereo1 = INCHI_BOND_STEREO_SINGLE_1DOWN;
|
1849
|
+
bond_stereo2 = INCHI_BOND_STEREO_SINGLE_2DOWN;
|
1850
|
+
break;
|
1851
|
+
case 'N':
|
1852
|
+
bond_type = INCHI_BOND_TYPE_SINGLE;
|
1853
|
+
bond_stereo1 = INCHI_BOND_STEREO_SINGLE_2DOWN;
|
1854
|
+
bond_stereo2 = INCHI_BOND_STEREO_SINGLE_1DOWN;
|
1855
|
+
break;
|
1856
|
+
default:
|
1857
|
+
num_atoms = INCHI_INP_ERROR_RET; /* error */
|
1858
|
+
*err = INCHI_INP_ERROR_ERR;
|
1859
|
+
MOLFILE_ERR_SET (*err, 0, "Wrong bond type");
|
1860
|
+
goto bypass_end_of_INChI;
|
1861
|
+
}
|
1862
|
+
k = AT_NUM_BONDS(atom[i]) ++;
|
1863
|
+
atom[i].bond_type[k] = bond_type;
|
1864
|
+
atom[i].bond_stereo[k] = bond_stereo1;
|
1865
|
+
atom[i].neighbor[k] = (ATOM_NUMBER)neigh;
|
1866
|
+
k2 = AT_NUM_BONDS(atom[neigh]) ++;
|
1867
|
+
atom[neigh].bond_type[k2] = bond_type;
|
1868
|
+
atom[neigh].bond_stereo[k2] = bond_stereo2;
|
1869
|
+
atom[neigh].neighbor[k2] = (ATOM_NUMBER)i;
|
1870
|
+
bond_parity |= (bond_parityNM << SB_PARITY_SHFT);
|
1871
|
+
|
1872
|
+
if ( bond_parity ) {
|
1873
|
+
if ( max_len_stereo0D <= len_stereo0D ) {
|
1874
|
+
/* realloc atom_Stereo0D */
|
1875
|
+
inchi_Stereo0D *new_atom_stereo0D = CreateInchi_Stereo0D( max_len_stereo0D+num_atoms );
|
1876
|
+
if ( !new_atom_stereo0D ) {
|
1877
|
+
num_atoms = INCHI_INP_FATAL_RET; /* fatal error: cannot allocate */
|
1878
|
+
*err = INCHI_INP_FATAL_ERR;
|
1879
|
+
MOLFILE_ERR_SET (*err, 0, "Out of RAM");
|
1880
|
+
goto bypass_end_of_INChI;
|
1881
|
+
}
|
1882
|
+
memcpy( new_atom_stereo0D, atom_stereo0D, len_stereo0D * sizeof(*atom_stereo0D) );
|
1883
|
+
FreeInchi_Stereo0D( &atom_stereo0D );
|
1884
|
+
atom_stereo0D = new_atom_stereo0D;
|
1885
|
+
max_len_stereo0D += num_atoms;
|
1886
|
+
}
|
1887
|
+
/* (a) i may be allene endpoint and neigh = allene middle point or
|
1888
|
+
(b) i may be allene middle point and neigh = allene endpoint
|
1889
|
+
!!!!! CURRENTLY ONLY (b) IS ALLOWED !!!!!
|
1890
|
+
*/
|
1891
|
+
atom_stereo0D[len_stereo0D].neighbor[1] = neigh; /* neigh < i */
|
1892
|
+
atom_stereo0D[len_stereo0D].neighbor[2] = i;
|
1893
|
+
atom_stereo0D[len_stereo0D].parity = bond_parity;
|
1894
|
+
atom_stereo0D[len_stereo0D].type = INCHI_StereoType_DoubleBond; /* incl allenes & cumulenes */
|
1895
|
+
len_stereo0D ++;
|
1896
|
+
}
|
1897
|
+
}
|
1898
|
+
if ( !bItemIsOver || p - szLine != res || i != num_atoms ) {
|
1899
|
+
num_atoms = INCHI_INP_ERROR_RET; /* error in input data */
|
1900
|
+
*err = INCHI_INP_ERROR_ERR;
|
1901
|
+
MOLFILE_ERR_SET (*err, 0, "Wrong number of bonds");
|
1902
|
+
goto bypass_end_of_INChI;
|
1903
|
+
}
|
1904
|
+
/********************** coordinates xml ****************************/
|
1905
|
+
if ( pszCoord = (MOL_COORD*)inchi_malloc(inchi_max(num_atoms,1) * sizeof(MOL_COORD)) ) {
|
1906
|
+
memset( pszCoord, ' ', inchi_max(num_atoms,1) * sizeof(MOL_COORD));
|
1907
|
+
res = my_fgets( szLine, sizeof(szLine)-1, inp_molfile, &bTooLongLine );
|
1908
|
+
if ( res <= 0 ||
|
1909
|
+
/* compare the header */
|
1910
|
+
memcmp(szLine, sStructRevXmlRevXYZ, sizeof(sStructRevXmlRevXYZ)-1) ||
|
1911
|
+
/* read (the head of) the coordinates (xml) line */
|
1912
|
+
0 >= (res = my_fgets( szLine, sizeof(szLine)-1, inp_molfile, &bTooLongLine ))) {
|
1913
|
+
num_atoms = INCHI_INP_ERROR_RET; /* error in input data: atoms, bonds & coord must be present together */
|
1914
|
+
*err = INCHI_INP_ERROR_ERR;
|
1915
|
+
MOLFILE_ERR_SET (*err, 0, "Missing atom coordinates data");
|
1916
|
+
goto bypass_end_of_INChI;
|
1917
|
+
}
|
1918
|
+
i = 0;
|
1919
|
+
bItemIsOver = 0;
|
1920
|
+
res2 = bTooLongLine2 = -1;
|
1921
|
+
p = szLine;
|
1922
|
+
if ( !memcmp(szLine, sStructRevXmlRevXYZEnd, sizeof(sStructRevXmlRevXYZEnd)-1) ) {
|
1923
|
+
/* empty bonds section */
|
1924
|
+
res = 0;
|
1925
|
+
bItemIsOver = 1;
|
1926
|
+
}
|
1927
|
+
/* read all coordinates (xml), starting from atom 1 (not 0) */
|
1928
|
+
while ( i < num_atoms ) {
|
1929
|
+
pos = p - szLine;
|
1930
|
+
if ( !bItemIsOver && (int)sizeof(szLine)-res + pos > (int)sizeof(szNextLine) ) {
|
1931
|
+
/* load next line if possible */
|
1932
|
+
res2 = my_fgets( szNextLine, sizeof(szNextLine)-1, inp_molfile, &bTooLongLine2 );
|
1933
|
+
if ( res2 > 0 && memcmp(szNextLine, sStructRevXmlRevXYZEnd, sizeof(sStructRevXmlRevXYZEnd)-1) ) {
|
1934
|
+
if ( pos ) {
|
1935
|
+
res -= pos; /* number of chars left to process in szLine */
|
1936
|
+
memmove( szLine, p, res*sizeof(szLine[0]) ); /* move them to the start of the line */
|
1937
|
+
}
|
1938
|
+
memcpy( szLine+res, szNextLine, (res2+1)*sizeof(szNextLine[0]) );
|
1939
|
+
res += res2;
|
1940
|
+
szLine[res] = '\0';
|
1941
|
+
bTooLongLine = bTooLongLine2;
|
1942
|
+
p = szLine;
|
1943
|
+
} else {
|
1944
|
+
bItemIsOver = 1;
|
1945
|
+
}
|
1946
|
+
}
|
1947
|
+
/* coord, first char */
|
1948
|
+
if ( *p == ';' ) {
|
1949
|
+
for ( k = 0; k < NUM_COORD; k ++ ) {
|
1950
|
+
pszCoord[i][LEN_COORD*k + 4] = '0';
|
1951
|
+
}
|
1952
|
+
p ++;
|
1953
|
+
i ++;
|
1954
|
+
continue;
|
1955
|
+
}
|
1956
|
+
for ( k = 0; k < 3; k ++ ) {
|
1957
|
+
double xyz;
|
1958
|
+
bNonZeroXYZ = 0;
|
1959
|
+
if ( *p == ';' ) {
|
1960
|
+
pszCoord[i][LEN_COORD*k + 4] = '0';
|
1961
|
+
xyz = 0.0;
|
1962
|
+
} else
|
1963
|
+
if ( *p == ',' ) {
|
1964
|
+
/* empty */
|
1965
|
+
pszCoord[i][LEN_COORD*k + 4] = '0';
|
1966
|
+
xyz = 0.0;
|
1967
|
+
p ++;
|
1968
|
+
} else {
|
1969
|
+
xyz = strtod( p, &q );
|
1970
|
+
bNonZeroXYZ = fabs(xyz) > MIN_BOND_LENGTH;
|
1971
|
+
if ( q != NULL ) {
|
1972
|
+
memcpy( pszCoord[i]+LEN_COORD*k, p, q-p );
|
1973
|
+
if ( *q == ',' )
|
1974
|
+
q ++;
|
1975
|
+
p = q;
|
1976
|
+
} else {
|
1977
|
+
pszCoord[i][LEN_COORD*k + 4] = '0';
|
1978
|
+
}
|
1979
|
+
}
|
1980
|
+
switch( k ) {
|
1981
|
+
case 0:
|
1982
|
+
atom[i].x = xyz;
|
1983
|
+
b2D |= bNonZeroXYZ;
|
1984
|
+
break;
|
1985
|
+
case 1:
|
1986
|
+
atom[i].y = xyz;
|
1987
|
+
b2D |= bNonZeroXYZ;
|
1988
|
+
break;
|
1989
|
+
case 2:
|
1990
|
+
b3D |= bNonZeroXYZ;
|
1991
|
+
atom[i].z = xyz;
|
1992
|
+
break;
|
1993
|
+
}
|
1994
|
+
}
|
1995
|
+
if ( *p == ';' ) {
|
1996
|
+
p ++; /* end of this triple of coordinates */
|
1997
|
+
i ++;
|
1998
|
+
} else {
|
1999
|
+
num_atoms = INCHI_INP_ERROR_RET; /* error in input data: atoms, bonds & coord must be present together */
|
2000
|
+
*err = INCHI_INP_ERROR_ERR;
|
2001
|
+
MOLFILE_ERR_SET (*err, 0, "Wrong atom coordinates data");
|
2002
|
+
goto bypass_end_of_INChI;
|
2003
|
+
}
|
2004
|
+
}
|
2005
|
+
if ( !bItemIsOver || p - szLine != res || i != num_atoms ) {
|
2006
|
+
num_atoms = INCHI_INP_ERROR_RET; /* error in input data: atoms, bonds & coord must be present together */
|
2007
|
+
*err = INCHI_INP_ERROR_ERR;
|
2008
|
+
MOLFILE_ERR_SET (*err, 0, "Wrong number of coordinates");
|
2009
|
+
goto bypass_end_of_INChI;
|
2010
|
+
}
|
2011
|
+
} else { /* allocation failed */
|
2012
|
+
num_atoms = INCHI_INP_FATAL_RET;
|
2013
|
+
*err = INCHI_INP_FATAL_ERR;
|
2014
|
+
MOLFILE_ERR_SET (*err, 0, "Out of RAM");
|
2015
|
+
goto bypass_end_of_INChI;
|
2016
|
+
}
|
2017
|
+
/* set special valences and implicit H (xml) */
|
2018
|
+
b23D = b2D | b3D;
|
2019
|
+
b2D = b3D = 0;
|
2020
|
+
if ( at ) {
|
2021
|
+
if ( !*at ) {
|
2022
|
+
int a1, a2, n1, n2, valence;
|
2023
|
+
int chem_bonds_valence;
|
2024
|
+
int nX=0, nY=0, nZ=0, nXYZ;
|
2025
|
+
*at = atom;
|
2026
|
+
/* special valences */
|
2027
|
+
for ( bNonMetal = 0; bNonMetal < 1 /*2*/; bNonMetal ++ ) {
|
2028
|
+
for ( a1 = 0; a1 < num_atoms; a1 ++ ) {
|
2029
|
+
int num_bond_type[MAX_INPUT_BOND_TYPE - MIN_INPUT_BOND_TYPE + 1];
|
2030
|
+
#if( defined(INCHI_LIBRARY) || defined(INCHI_MAIN) )
|
2031
|
+
#else
|
2032
|
+
int bHasMetalNeighbor=0;
|
2033
|
+
#endif
|
2034
|
+
memset( num_bond_type, 0, sizeof(num_bond_type) );
|
2035
|
+
|
2036
|
+
valence = AT_BONDS_VAL(atom, a1); /* save atom valence if available */
|
2037
|
+
AT_BONDS_VAL(atom, a1) = 0;
|
2038
|
+
#if( defined(INCHI_LIBRARY) || defined(INCHI_MAIN) )
|
2039
|
+
#else
|
2040
|
+
atom[a1].orig_at_number = a1+1;
|
2041
|
+
#endif
|
2042
|
+
nX = nY = nZ = 0;
|
2043
|
+
for ( n1 = 0; n1 < AT_NUM_BONDS(atom[a1]); n1 ++ ) {
|
2044
|
+
bond_type = atom[a1].bond_type[n1] - MIN_INPUT_BOND_TYPE;
|
2045
|
+
if ( bond_type < 0 || bond_type > MAX_INPUT_BOND_TYPE - MIN_INPUT_BOND_TYPE ) {
|
2046
|
+
bond_type = 0; /* cannot happen */
|
2047
|
+
MOLFILE_ERR_SET (*err, 0, "Unknown bond type in InChI aux assigned as a single bond");
|
2048
|
+
}
|
2049
|
+
|
2050
|
+
num_bond_type[ bond_type ] ++;
|
2051
|
+
nNumBonds ++;
|
2052
|
+
if ( b23D ) {
|
2053
|
+
neigh = atom[a1].neighbor[n1];
|
2054
|
+
nX |= (fabs(atom[a1].x - atom[neigh].x) > MIN_BOND_LENGTH);
|
2055
|
+
nY |= (fabs(atom[a1].y - atom[neigh].y) > MIN_BOND_LENGTH);
|
2056
|
+
nZ |= (fabs(atom[a1].z - atom[neigh].z) > MIN_BOND_LENGTH);
|
2057
|
+
}
|
2058
|
+
}
|
2059
|
+
chem_bonds_valence = 0;
|
2060
|
+
for ( n1 = 0; MIN_INPUT_BOND_TYPE + n1 <= 3 && MIN_INPUT_BOND_TYPE + n1 <= MAX_INPUT_BOND_TYPE; n1 ++ ) {
|
2061
|
+
chem_bonds_valence += (MIN_INPUT_BOND_TYPE + n1) * num_bond_type[n1];
|
2062
|
+
}
|
2063
|
+
if ( MIN_INPUT_BOND_TYPE <= INCHI_BOND_TYPE_ALTERN && INCHI_BOND_TYPE_ALTERN <= MAX_INPUT_BOND_TYPE &&
|
2064
|
+
( n2 = num_bond_type[INCHI_BOND_TYPE_ALTERN-MIN_INPUT_BOND_TYPE] ) ) {
|
2065
|
+
/* accept input aromatic bonds for now */
|
2066
|
+
switch ( n2 ) {
|
2067
|
+
case 2:
|
2068
|
+
chem_bonds_valence += 3; /* =A- */
|
2069
|
+
break;
|
2070
|
+
case 3:
|
2071
|
+
chem_bonds_valence += 4; /* =A< */
|
2072
|
+
break;
|
2073
|
+
default:
|
2074
|
+
/* if 1 or >= 4 aromatic bonds then replace such bonds with single bonds */
|
2075
|
+
for ( n1 = 0; n1 < AT_NUM_BONDS(atom[a1]); n1 ++ ) {
|
2076
|
+
if ( atom[a1].bond_type[n1] == INCHI_BOND_TYPE_ALTERN ) {
|
2077
|
+
ATOM_NUMBER *p1;
|
2078
|
+
a2 = atom[a1].neighbor[n1];
|
2079
|
+
p1 = IN_NEIGH_LIST( atom[a2].neighbor, (ATOM_NUMBER)a1, AT_NUM_BONDS(atom[a2]) );
|
2080
|
+
if ( p1 ) {
|
2081
|
+
atom[a1].bond_type[n1] =
|
2082
|
+
atom[a2].bond_type[p1-atom[a2].neighbor] = INCHI_BOND_TYPE_SINGLE;
|
2083
|
+
} else {
|
2084
|
+
*err = -2; /* Program error */
|
2085
|
+
MOLFILE_ERR_SET (*err, 0, "Program error interpreting InChI aux");
|
2086
|
+
num_atoms = 0;
|
2087
|
+
goto bypass_end_of_INChI; /* no structure */
|
2088
|
+
}
|
2089
|
+
}
|
2090
|
+
}
|
2091
|
+
chem_bonds_valence += n2;
|
2092
|
+
*err |= 32; /* Unrecognized aromatic bond(s) replaced with single */
|
2093
|
+
MOLFILE_ERR_SET (*err, 0, "Atom has more than 3 aromatic bonds");
|
2094
|
+
break;
|
2095
|
+
}
|
2096
|
+
}
|
2097
|
+
|
2098
|
+
#if( defined(INCHI_LIBRARY) || defined(INCHI_MAIN) )
|
2099
|
+
/*************************************************************************************
|
2100
|
+
*
|
2101
|
+
* Set number of hydrogen atoms
|
2102
|
+
*/
|
2103
|
+
{
|
2104
|
+
int num_iso_H;
|
2105
|
+
num_iso_H = atom[a1].num_iso_H[1] + atom[a1].num_iso_H[2] + atom[a1].num_iso_H[3];
|
2106
|
+
if ( valence == ISOLATED_ATOM ) {
|
2107
|
+
atom[a1].num_iso_H[0] = 0;
|
2108
|
+
} else
|
2109
|
+
if ( valence && valence >= chem_bonds_valence ) {
|
2110
|
+
atom[a1].num_iso_H[0] = valence - chem_bonds_valence;
|
2111
|
+
} else
|
2112
|
+
if ( valence || bDoNotAddH ) {
|
2113
|
+
atom[a1].num_iso_H[0] = 0;
|
2114
|
+
} else
|
2115
|
+
if ( !bDoNotAddH ) {
|
2116
|
+
atom[a1].num_iso_H[0] = -1; /* auto add H */
|
2117
|
+
}
|
2118
|
+
}
|
2119
|
+
#else
|
2120
|
+
atom[a1].chem_bonds_valence = chem_bonds_valence;
|
2121
|
+
atom[a1].num_H = get_num_H( atom[a1].elname, atom[a1].num_H, atom[a1].num_iso_H, atom[a1].charge, atom[a1].radical,
|
2122
|
+
atom[a1].chem_bonds_valence,
|
2123
|
+
valence,
|
2124
|
+
0, bDoNotAddH, bHasMetalNeighbor );
|
2125
|
+
#endif
|
2126
|
+
}
|
2127
|
+
}
|
2128
|
+
nNumBonds /= 2;
|
2129
|
+
if ( b23D && nNumBonds ) {
|
2130
|
+
nXYZ = nX+nY+nZ;
|
2131
|
+
b2D = (nXYZ > 0);
|
2132
|
+
b3D = (nXYZ == 3);
|
2133
|
+
*num_dimensions = b3D? 3 : b2D? 2 : 0;
|
2134
|
+
*num_bonds = nNumBonds;
|
2135
|
+
}
|
2136
|
+
/*======= 0D parities =================================*/
|
2137
|
+
#if( defined(INCHI_LIBRARY) || defined(INCHI_MAIN) )
|
2138
|
+
if ( len_stereo0D > 0 && atom_stereo0D && stereo0D ) {
|
2139
|
+
*stereo0D = atom_stereo0D;
|
2140
|
+
*num_stereo0D = len_stereo0D;
|
2141
|
+
} else {
|
2142
|
+
FreeInchi_Stereo0D( &atom_stereo0D );
|
2143
|
+
*num_stereo0D = len_stereo0D = 0;
|
2144
|
+
}
|
2145
|
+
#endif
|
2146
|
+
for ( i = 0; i < len_stereo0D; i ++ ) {
|
2147
|
+
ATOM_NUMBER *p1, *p2;
|
2148
|
+
int sb_ord_from_a1 = -1, sb_ord_from_a2 = -1, bEnd1 = 0, bEnd2 = 0;
|
2149
|
+
switch( atom_stereo0D[i].type ) {
|
2150
|
+
|
2151
|
+
case INCHI_StereoType_Tetrahedral:
|
2152
|
+
a1 = atom_stereo0D[i].central_atom;
|
2153
|
+
if ( atom_stereo0D[i].parity && (AT_NUM_BONDS(atom[a1]) == 3 || AT_NUM_BONDS(atom[a1]) == 4) ) {
|
2154
|
+
int ii, kk = 0;
|
2155
|
+
if ( AT_NUM_BONDS(atom[a1]) == 3 ) {
|
2156
|
+
atom_stereo0D[i].neighbor[kk++] = a1;
|
2157
|
+
}
|
2158
|
+
for ( ii = 0; ii < AT_NUM_BONDS(atom[a1]); ii ++ ) {
|
2159
|
+
atom_stereo0D[i].neighbor[kk++] = atom[a1].neighbor[ii];
|
2160
|
+
}
|
2161
|
+
}
|
2162
|
+
|
2163
|
+
break;
|
2164
|
+
|
2165
|
+
case INCHI_StereoType_DoubleBond:
|
2166
|
+
#define MAX_CHAIN_LEN 20
|
2167
|
+
a1 = atom_stereo0D[i].neighbor[1];
|
2168
|
+
a2 = atom_stereo0D[i].neighbor[2];
|
2169
|
+
p1 = IN_NEIGH_LIST( atom[a1].neighbor, (ATOM_NUMBER)a2, AT_NUM_BONDS(atom[a1]) );
|
2170
|
+
p2 = IN_NEIGH_LIST( atom[a2].neighbor, (ATOM_NUMBER)a1, AT_NUM_BONDS(atom[a2]) );
|
2171
|
+
if ( !p1 || !p2 ) {
|
2172
|
+
atom_stereo0D[i].type = INCHI_StereoType_None;
|
2173
|
+
atom_stereo0D[i].central_atom = NO_ATOM;
|
2174
|
+
atom_stereo0D[i].neighbor[0] =
|
2175
|
+
atom_stereo0D[i].neighbor[4] = -1;
|
2176
|
+
*err |= 64; /* Error in cumulene stereo */
|
2177
|
+
MOLFILE_ERR_SET (*err, 0, "0D stereobond not recognized");
|
2178
|
+
break;
|
2179
|
+
}
|
2180
|
+
/* streobond, allene, or cumulene */
|
2181
|
+
|
2182
|
+
sb_ord_from_a1 = p1 - atom[a1].neighbor;
|
2183
|
+
sb_ord_from_a2 = p2 - atom[a2].neighbor;
|
2184
|
+
|
2185
|
+
if ( AT_NUM_BONDS(atom[a1]) == 2 &&
|
2186
|
+
atom[a1].bond_type[0] + atom[a1].bond_type[1] == 2*INCHI_BOND_TYPE_DOUBLE &&
|
2187
|
+
0 == inchi_NUMH2(atom, a1) &&
|
2188
|
+
(AT_NUM_BONDS(atom[a2]) != 2 ||
|
2189
|
+
atom[a2].bond_type[0] + atom[a2].bond_type[1] != 2*INCHI_BOND_TYPE_DOUBLE ) ) {
|
2190
|
+
bEnd2 = 1; /* a2 is the end-atom, a1 is middle atom */
|
2191
|
+
}
|
2192
|
+
if ( AT_NUM_BONDS(atom[a2]) == 2 &&
|
2193
|
+
atom[a2].bond_type[0] + atom[a2].bond_type[1] == 2*INCHI_BOND_TYPE_DOUBLE &&
|
2194
|
+
0 == inchi_NUMH2(atom, a2) &&
|
2195
|
+
(AT_NUM_BONDS(atom[a1]) != 2 ||
|
2196
|
+
atom[a1].bond_type[0] + atom[a1].bond_type[1] != 2*INCHI_BOND_TYPE_DOUBLE ) ) {
|
2197
|
+
bEnd1 = 1; /* a1 is the end-atom, a2 is middle atom */
|
2198
|
+
}
|
2199
|
+
|
2200
|
+
if ( bEnd2 + bEnd1 == 1 ) {
|
2201
|
+
/* allene or cumulene */
|
2202
|
+
ATOM_NUMBER chain[MAX_CHAIN_LEN+1], prev, cur, next;
|
2203
|
+
if ( bEnd2 && !bEnd1 ) {
|
2204
|
+
cur = a1;
|
2205
|
+
a1 = a2;
|
2206
|
+
a2 = cur;
|
2207
|
+
sb_ord_from_a1 = sb_ord_from_a2;
|
2208
|
+
}
|
2209
|
+
sb_ord_from_a2 = -1;
|
2210
|
+
cur = a1;
|
2211
|
+
next = a2;
|
2212
|
+
len = 0;
|
2213
|
+
chain[len++] = cur;
|
2214
|
+
chain[len++] = next;
|
2215
|
+
while ( len < MAX_CHAIN_LEN ) { /* arbitrary very high upper limit to prevent infinite loop */
|
2216
|
+
prev = cur;
|
2217
|
+
cur = next;
|
2218
|
+
/* follow double bond path && avoid going back */
|
2219
|
+
if ( AT_NUM_BONDS(atom[cur]) == 2 &&
|
2220
|
+
atom[cur].bond_type[0]+atom[cur].bond_type[1] == 2*INCHI_BOND_TYPE_DOUBLE &&
|
2221
|
+
0 == inchi_NUMH2(atom, cur) ) {
|
2222
|
+
next = atom[cur].neighbor[atom[cur].neighbor[0] == prev];
|
2223
|
+
chain[len++] = next;
|
2224
|
+
} else {
|
2225
|
+
break;
|
2226
|
+
}
|
2227
|
+
}
|
2228
|
+
if ( len > 2 &&
|
2229
|
+
(p2 = IN_NEIGH_LIST( atom[cur].neighbor, (ATOM_NUMBER)prev, AT_NUM_BONDS(atom[cur]))) ) {
|
2230
|
+
sb_ord_from_a2 = p2 - atom[cur].neighbor;
|
2231
|
+
a2 = cur;
|
2232
|
+
/* by design we need to pick up the first non-stereo-bond-neighbor as "sn"-atom */
|
2233
|
+
atom_stereo0D[i].neighbor[0] = atom[a1].neighbor[sb_ord_from_a1 == 0];
|
2234
|
+
atom_stereo0D[i].neighbor[1] = a1;
|
2235
|
+
atom_stereo0D[i].neighbor[2] = a2;
|
2236
|
+
atom_stereo0D[i].neighbor[3] = atom[a2].neighbor[sb_ord_from_a2 == 0];
|
2237
|
+
if ( len % 2 ) {
|
2238
|
+
atom_stereo0D[i].central_atom = chain[len/2];
|
2239
|
+
atom_stereo0D[i].type = INCHI_StereoType_Allene;
|
2240
|
+
} else {
|
2241
|
+
atom_stereo0D[i].central_atom = NO_ATOM;
|
2242
|
+
}
|
2243
|
+
} else {
|
2244
|
+
/* error */
|
2245
|
+
atom_stereo0D[i].type = INCHI_StereoType_None;
|
2246
|
+
atom_stereo0D[i].central_atom = NO_ATOM;
|
2247
|
+
atom_stereo0D[i].neighbor[0] =
|
2248
|
+
atom_stereo0D[i].neighbor[4] = -1;
|
2249
|
+
*err |= 64; /* Error in cumulene stereo */
|
2250
|
+
MOLFILE_ERR_SET (*err, 0, "Cumulene stereo not recognized (0D)");
|
2251
|
+
|
2252
|
+
}
|
2253
|
+
#undef MAX_CHAIN_LEN
|
2254
|
+
} else {
|
2255
|
+
/****** a normal possibly stereogenic bond -- not an allene or cumulene *******/
|
2256
|
+
/* by design we need to pick up the first non-stereo-bond-neighbor as "sn"-atom */
|
2257
|
+
sb_ord_from_a1 = p1 - atom[a1].neighbor;
|
2258
|
+
sb_ord_from_a2 = p2 - atom[a2].neighbor;
|
2259
|
+
atom_stereo0D[i].neighbor[0] = atom[a1].neighbor[p1 == atom[a1].neighbor];
|
2260
|
+
atom_stereo0D[i].neighbor[3] = atom[a2].neighbor[p2 == atom[a2].neighbor];
|
2261
|
+
atom_stereo0D[i].central_atom = NO_ATOM;
|
2262
|
+
}
|
2263
|
+
if ( atom_stereo0D[i].type != INCHI_StereoType_None &&
|
2264
|
+
sb_ord_from_a1 >= 0 && sb_ord_from_a2 >= 0 &&
|
2265
|
+
ATOM_PARITY_WELL_DEF( SB_PARITY_2(atom_stereo0D[i].parity) ) ) {
|
2266
|
+
/* Detected well-defined disconnected stereo
|
2267
|
+
* locate first non-metal neighbors */
|
2268
|
+
int a, n, j, /* k,*/ sb_ord, cur_neigh, min_neigh;
|
2269
|
+
for ( k = 0; k < 2; k ++ ) {
|
2270
|
+
a = k? atom_stereo0D[i].neighbor[2] : atom_stereo0D[i].neighbor[1];
|
2271
|
+
sb_ord = k? sb_ord_from_a2 : sb_ord_from_a1;
|
2272
|
+
min_neigh = num_atoms;
|
2273
|
+
for ( n = j = 0; j < AT_NUM_BONDS(atom[a]); j ++ ) {
|
2274
|
+
cur_neigh = atom[a].neighbor[j];
|
2275
|
+
if ( j != sb_ord && !IS_METAL_ATOM(atom, cur_neigh) ) {
|
2276
|
+
min_neigh = inchi_min( cur_neigh, min_neigh );
|
2277
|
+
}
|
2278
|
+
}
|
2279
|
+
if ( min_neigh < num_atoms ) {
|
2280
|
+
atom_stereo0D[i].neighbor[k?3:0] = min_neigh;
|
2281
|
+
} else {
|
2282
|
+
MOLFILE_ERR_SET (*err, 0, "Cannot find non-metal stereobond neighor (0D)");
|
2283
|
+
}
|
2284
|
+
}
|
2285
|
+
}
|
2286
|
+
|
2287
|
+
break;
|
2288
|
+
}
|
2289
|
+
}
|
2290
|
+
/* end of 0D parities extraction */
|
2291
|
+
/*exit_cycle:;*/
|
2292
|
+
}
|
2293
|
+
#if( defined(INCHI_LIBRARY) || defined(INCHI_MAIN) )
|
2294
|
+
#else
|
2295
|
+
/* transfer atom_stereo0D[] to atom[] */
|
2296
|
+
if ( len_stereo0D ) {
|
2297
|
+
Extract0DParities( atom, num_atoms, atom_stereo0D, len_stereo0D, pStrErr, err );
|
2298
|
+
}
|
2299
|
+
#endif
|
2300
|
+
if ( pInpAtomFlags ) {
|
2301
|
+
/* save chirality flag */
|
2302
|
+
*pInpAtomFlags |= InpAtomFlags;
|
2303
|
+
}
|
2304
|
+
} else
|
2305
|
+
if ( atom ) {
|
2306
|
+
inchi_free( atom );
|
2307
|
+
atom = NULL;
|
2308
|
+
}
|
2309
|
+
#if( defined(INCHI_LIBRARY) || defined(INCHI_MAIN) )
|
2310
|
+
#else
|
2311
|
+
if ( szCoord ) {
|
2312
|
+
*szCoord = pszCoord;
|
2313
|
+
pszCoord = NULL;
|
2314
|
+
} else
|
2315
|
+
#endif
|
2316
|
+
if ( pszCoord ) {
|
2317
|
+
inchi_free( pszCoord );
|
2318
|
+
}
|
2319
|
+
goto bypass_end_of_INChI;
|
2320
|
+
/*return num_atoms;*/
|
2321
|
+
}
|
2322
|
+
}
|
2323
|
+
if ( atom_stereo0D ) {
|
2324
|
+
FreeInchi_Stereo0D( &atom_stereo0D );
|
2325
|
+
}
|
2326
|
+
/* end of struct. reading cycle, code never used? */
|
2327
|
+
if ( res <= 0 ) {
|
2328
|
+
if ( *err == INCHI_INP_ERROR_ERR ) {
|
2329
|
+
return num_atoms;
|
2330
|
+
}
|
2331
|
+
*err = INCHI_INP_EOF_ERR;
|
2332
|
+
return INCHI_INP_EOF_RET; /* no more data */
|
2333
|
+
}
|
2334
|
+
bypass_end_of_INChI:
|
2335
|
+
/* cleanup */
|
2336
|
+
if ( num_atoms == INCHI_INP_ERROR_RET && atom_stereo0D ) {
|
2337
|
+
if ( stereo0D && *stereo0D == atom_stereo0D ) {
|
2338
|
+
*stereo0D = NULL;
|
2339
|
+
*num_stereo0D = 0;
|
2340
|
+
}
|
2341
|
+
FreeInchi_Stereo0D( &atom_stereo0D );
|
2342
|
+
}
|
2343
|
+
if ( !memcmp(szLine, sStructHdrXmlEnd, sizeof(sStructHdrXmlEnd)-1) )
|
2344
|
+
num_struct --;
|
2345
|
+
if ( !memcmp(szLine, sStructHdrXml, sizeof(sStructHdrXml)-1) )
|
2346
|
+
num_struct ++;
|
2347
|
+
|
2348
|
+
while ( num_struct > 0 && 0 < my_fgets( szLine, sizeof(szLine)-1, inp_molfile, &bTooLongLine ) ) {
|
2349
|
+
if ( !memcmp(szLine, sStructHdrXmlEnd, sizeof(sStructHdrXmlEnd)-1) )
|
2350
|
+
num_struct --;
|
2351
|
+
else
|
2352
|
+
if ( !memcmp(szLine, sStructHdrXml, sizeof(sStructHdrXml)-1) )
|
2353
|
+
num_struct ++;
|
2354
|
+
}
|
2355
|
+
return num_atoms;
|
2356
|
+
|
2357
|
+
}
|
2358
|
+
|
2359
|
+
return num_atoms;
|
2360
|
+
|
2361
|
+
#undef AT_NUM_BONDS
|
2362
|
+
#undef ATOM_NUMBER
|
2363
|
+
#undef IN_NEIGH_LIST
|
2364
|
+
#undef inchi_NUMH2
|
2365
|
+
|
2366
|
+
#if( defined(INCHI_LIBRARY) || defined(INCHI_MAIN) )
|
2367
|
+
#else
|
2368
|
+
#undef inchi_Atom
|
2369
|
+
#endif
|
2370
|
+
|
2371
|
+
#undef AT_NUM_BONDS
|
2372
|
+
#undef ATOM_NUMBER
|
2373
|
+
#undef IN_NEIGH_LIST
|
2374
|
+
#undef inchi_NUMH2
|
2375
|
+
#undef INChITo_Atom
|
2376
|
+
#undef MoreParms
|
2377
|
+
#undef INPUT_FILE
|
2378
|
+
#undef Create_Atom
|
2379
|
+
#undef AT_BONDS_VAL
|
2380
|
+
#undef ISOLATED_ATOM
|
2381
|
+
#undef NUM_ISO_Hk
|
2382
|
+
#undef IS_METAL_ATOM
|
2383
|
+
|
2384
|
+
|
2385
|
+
}
|
2386
|
+
#ifdef INCHI_MAIN
|
2387
|
+
|
2388
|
+
/**********************************************************************************/
|
2389
|
+
int INChIToInchi_Input( INCHI_FILE *inp_molfile, inchi_Input *orig_at_data, int bMergeAllInputStructures,
|
2390
|
+
int bDoNotAddH, INPUT_TYPE nInputType,
|
2391
|
+
char *pSdfLabel, char *pSdfValue, long *lSdfId, INCHI_MODE *pInpAtomFlags,
|
2392
|
+
int *err, char *pStrErr )
|
2393
|
+
{
|
2394
|
+
/* inp_ATOM *at = NULL; */
|
2395
|
+
int num_dimensions_new;
|
2396
|
+
int num_inp_bonds_new;
|
2397
|
+
int num_inp_atoms_new;
|
2398
|
+
int num_inp_0D_new;
|
2399
|
+
inchi_Atom *at_new = NULL;
|
2400
|
+
inchi_Atom *at_old = NULL;
|
2401
|
+
inchi_Stereo0D *stereo0D_new = NULL;
|
2402
|
+
inchi_Stereo0D *stereo0D_old = NULL;
|
2403
|
+
int nNumAtoms = 0, nNumStereo0D = 0;
|
2404
|
+
MOL_COORD *szCoordNew = NULL;
|
2405
|
+
MOL_COORD *szCoordOld = NULL;
|
2406
|
+
int i, j;
|
2407
|
+
|
2408
|
+
if ( pStrErr ) {
|
2409
|
+
pStrErr[0] = '\0';
|
2410
|
+
}
|
2411
|
+
|
2412
|
+
/*FreeOrigAtData( orig_at_data );*/
|
2413
|
+
if ( lSdfId )
|
2414
|
+
*lSdfId = 0;
|
2415
|
+
do {
|
2416
|
+
|
2417
|
+
at_old = orig_at_data? orig_at_data->atom : NULL; /* save pointer to the previous allocation */
|
2418
|
+
stereo0D_old = orig_at_data? orig_at_data->stereo0D : NULL;
|
2419
|
+
szCoordOld = NULL;
|
2420
|
+
num_inp_atoms_new =
|
2421
|
+
INChIToInchi_Atom( inp_molfile, orig_at_data? &stereo0D_new:NULL, &num_inp_0D_new,
|
2422
|
+
bDoNotAddH, nInputType, orig_at_data? &at_new:NULL, MAX_ATOMS,
|
2423
|
+
&num_dimensions_new, &num_inp_bonds_new,
|
2424
|
+
pSdfLabel, pSdfValue, lSdfId, pInpAtomFlags, err, pStrErr );
|
2425
|
+
if ( num_inp_atoms_new <= 0 && !*err ) {
|
2426
|
+
MOLFILE_ERR_SET (*err, 0, "Empty structure");
|
2427
|
+
*err = 98;
|
2428
|
+
} else
|
2429
|
+
if ( orig_at_data && !num_inp_atoms_new && 10 < *err && *err < 20 && orig_at_data->num_atoms > 0 && bMergeAllInputStructures ) {
|
2430
|
+
*err = 0; /* end of file */
|
2431
|
+
break;
|
2432
|
+
} else
|
2433
|
+
if ( num_inp_atoms_new > 0 && orig_at_data ) {
|
2434
|
+
/* merge pOrigDataTmp + orig_at_data => pOrigDataTmp; */
|
2435
|
+
nNumAtoms = num_inp_atoms_new + orig_at_data->num_atoms;
|
2436
|
+
nNumStereo0D = num_inp_0D_new + orig_at_data->num_stereo0D;
|
2437
|
+
if ( nNumAtoms >= MAX_ATOMS ) {
|
2438
|
+
MOLFILE_ERR_SET (*err, 0, "Too many atoms");
|
2439
|
+
*err = 70;
|
2440
|
+
orig_at_data->num_atoms = -1;
|
2441
|
+
} else
|
2442
|
+
if ( !at_old ) {
|
2443
|
+
/* the first structure */
|
2444
|
+
orig_at_data->atom = at_new; at_new = NULL;
|
2445
|
+
orig_at_data->num_atoms = num_inp_atoms_new; num_inp_atoms_new = 0;
|
2446
|
+
orig_at_data->stereo0D = stereo0D_new; stereo0D_new = NULL;
|
2447
|
+
orig_at_data->num_stereo0D = num_inp_0D_new; num_inp_0D_new = 0;
|
2448
|
+
} else
|
2449
|
+
if ( orig_at_data->atom = CreateInchi_Atom( nNumAtoms ) ) {
|
2450
|
+
/* switch at_new <--> orig_at_data->at; */
|
2451
|
+
if ( orig_at_data->num_atoms ) {
|
2452
|
+
memcpy( orig_at_data->atom, at_old, orig_at_data->num_atoms * sizeof(orig_at_data->atom[0]) );
|
2453
|
+
/* adjust numbering in the newly read structure */
|
2454
|
+
for ( i = 0; i < num_inp_atoms_new; i ++ ) {
|
2455
|
+
for ( j = 0; j < at_new[i].num_bonds; j ++ ) {
|
2456
|
+
at_new[i].neighbor[j] += orig_at_data->num_atoms;
|
2457
|
+
}
|
2458
|
+
}
|
2459
|
+
}
|
2460
|
+
FreeInchi_Atom( &at_old );
|
2461
|
+
/* copy newly read structure */
|
2462
|
+
memcpy( orig_at_data->atom + orig_at_data->num_atoms,
|
2463
|
+
at_new,
|
2464
|
+
num_inp_atoms_new * sizeof(orig_at_data->atom[0]) );
|
2465
|
+
/* cpy newly read 0D stereo */
|
2466
|
+
if ( num_inp_0D_new > 0 && stereo0D_new ) {
|
2467
|
+
if ( orig_at_data->stereo0D = CreateInchi_Stereo0D( nNumStereo0D ) ) {
|
2468
|
+
memcpy( orig_at_data->stereo0D, stereo0D_old, orig_at_data->num_stereo0D * sizeof(orig_at_data->stereo0D[0]) );
|
2469
|
+
/* adjust numbering in the newly read structure */
|
2470
|
+
for ( i = 0; i < num_inp_0D_new; i ++ ) {
|
2471
|
+
if ( stereo0D_new[i].central_atom >= 0 ) {
|
2472
|
+
stereo0D_new[i].central_atom += orig_at_data->num_atoms;
|
2473
|
+
}
|
2474
|
+
for ( j = 0; j < 4; j ++ ) {
|
2475
|
+
stereo0D_new[i].neighbor[j] += orig_at_data->num_atoms;
|
2476
|
+
}
|
2477
|
+
}
|
2478
|
+
FreeInchi_Stereo0D( &stereo0D_old );
|
2479
|
+
memcpy( orig_at_data->stereo0D+orig_at_data->num_stereo0D,
|
2480
|
+
stereo0D_new,
|
2481
|
+
num_inp_0D_new * sizeof(orig_at_data->stereo0D[0]) );
|
2482
|
+
} else {
|
2483
|
+
num_inp_0D_new = 0;
|
2484
|
+
MOLFILE_ERR_SET (*err, 0, "Out of RAM");
|
2485
|
+
*err = -1;
|
2486
|
+
}
|
2487
|
+
} else {
|
2488
|
+
num_inp_0D_new = 0;
|
2489
|
+
}
|
2490
|
+
/* update lengths */
|
2491
|
+
orig_at_data->num_atoms += num_inp_atoms_new;
|
2492
|
+
orig_at_data->num_stereo0D += num_inp_0D_new;
|
2493
|
+
} else {
|
2494
|
+
MOLFILE_ERR_SET (*err, 0, "Out of RAM");
|
2495
|
+
*err = -1;
|
2496
|
+
}
|
2497
|
+
} else
|
2498
|
+
if ( num_inp_atoms_new > 0 ) {
|
2499
|
+
nNumAtoms += num_inp_atoms_new;
|
2500
|
+
}
|
2501
|
+
FreeInchi_Atom( &at_new );
|
2502
|
+
num_inp_atoms_new = 0;
|
2503
|
+
FreeInchi_Stereo0D( &stereo0D_new );
|
2504
|
+
num_inp_0D_new = 0;
|
2505
|
+
|
2506
|
+
} while ( !*err && bMergeAllInputStructures );
|
2507
|
+
/*
|
2508
|
+
if ( !*err ) {
|
2509
|
+
orig_at_data->num_components =
|
2510
|
+
MarkDisconnectedComponents( orig_at_data );
|
2511
|
+
if ( orig_at_data->num_components == 0 ) {
|
2512
|
+
MOLFILE_ERR_SET (*err, 0, "No components found");
|
2513
|
+
*err = 99;
|
2514
|
+
}
|
2515
|
+
if ( orig_at_data->num_components < 0 ) {
|
2516
|
+
MOLFILE_ERR_SET (*err, 0, "Too many components");
|
2517
|
+
*err = 99;
|
2518
|
+
}
|
2519
|
+
}
|
2520
|
+
*/
|
2521
|
+
if ( szCoordNew ) {
|
2522
|
+
inchi_free( szCoordNew );
|
2523
|
+
}
|
2524
|
+
if ( at_new ) {
|
2525
|
+
inchi_free( at_new );
|
2526
|
+
}
|
2527
|
+
/*
|
2528
|
+
if ( !*err ) {
|
2529
|
+
if ( ReconcileAllCmlBondParities( orig_at_data->atom, orig_at_data->num_atoms ) ) {
|
2530
|
+
MOLFILE_ERR_SET (*err, 0, "Cannot reconcile stereobond parities");
|
2531
|
+
if (!orig_at_data->num_atoms) {
|
2532
|
+
*err = 1;
|
2533
|
+
}
|
2534
|
+
}
|
2535
|
+
}
|
2536
|
+
*/
|
2537
|
+
if ( *err ) {
|
2538
|
+
FreeInchi_Input( orig_at_data );
|
2539
|
+
}
|
2540
|
+
if ( *err && !(10 < *err && *err < 20) && pStrErr && !pStrErr[0] ) {
|
2541
|
+
MOLFILE_ERR_SET (*err, 0, "Unknown error"); /* <BRKPT> */
|
2542
|
+
}
|
2543
|
+
return orig_at_data? orig_at_data->num_atoms : nNumAtoms;
|
2544
|
+
}
|
2545
|
+
|
2546
|
+
#endif
|
2547
|
+
|
2548
|
+
#ifndef INCHI_MAIN
|
2549
|
+
#undef AB_MAX_WELL_DEFINED_PARITY
|
2550
|
+
#undef AB_MIN_WELL_DEFINED_PARITY
|
2551
|
+
#include "extr_ct.h"
|
2552
|
+
/****************************************************************************************/
|
2553
|
+
int Extract0DParities( inp_ATOM *at, int nNumAtoms, inchi_Stereo0D *stereo0D,
|
2554
|
+
int num_stereo0D, char *pStrErr, int *err )
|
2555
|
+
{
|
2556
|
+
if ( stereo0D && num_stereo0D > 0 ) {
|
2557
|
+
int i0D, a2, k, k_prev, type, j, j1, j2, len, parity, parityNM;
|
2558
|
+
int sb_ord_from_i1, sb_ord_from_i2, sn_ord_from_i1, sn_ord_from_i2;
|
2559
|
+
AT_NUMB i1n, i2n, i1, i2;
|
2560
|
+
for ( i0D = 0; i0D < num_stereo0D; i0D ++ ) {
|
2561
|
+
parity = (stereo0D[i0D].parity & SB_PARITY_MASK);
|
2562
|
+
parityNM = (stereo0D[i0D].parity & SB_PARITY_FLAG) >> SB_PARITY_SHFT;
|
2563
|
+
if ( parity == INCHI_PARITY_NONE ||
|
2564
|
+
parity != INCHI_PARITY_ODD && parity != INCHI_PARITY_EVEN &&
|
2565
|
+
parity != INCHI_PARITY_UNKNOWN && parity != INCHI_PARITY_UNDEFINED ) {
|
2566
|
+
char szTemp[16];
|
2567
|
+
sprintf( szTemp, "#%d", i0D+1 );
|
2568
|
+
MOLFILE_ERR_SET (*err, 0, "Wrong 0D stereo descriptor(s):");
|
2569
|
+
MOLFILE_ERR_SET (*err, 0, szTemp);
|
2570
|
+
continue; /* warning */
|
2571
|
+
}
|
2572
|
+
type = stereo0D[i0D].type;
|
2573
|
+
a2 = stereo0D[i0D].central_atom; /* central atom or -1 */
|
2574
|
+
j = -1;
|
2575
|
+
len = 0;
|
2576
|
+
sb_ord_from_i1 = sb_ord_from_i2 = sn_ord_from_i1 = sn_ord_from_i2 = -1;
|
2577
|
+
i1n = i2n = i1 = i2 = MAX_ATOMS+1;
|
2578
|
+
|
2579
|
+
if ( (type == INCHI_StereoType_Tetrahedral ||
|
2580
|
+
type == INCHI_StereoType_Allene ) &&
|
2581
|
+
0 <= a2 && a2 < nNumAtoms ||
|
2582
|
+
type == INCHI_StereoType_DoubleBond &&
|
2583
|
+
a2 == NO_ATOM) {
|
2584
|
+
/* test the quadruplet */
|
2585
|
+
for ( j = 0, k_prev = -1; j < 4; j ++, k_prev = k ) {
|
2586
|
+
k = stereo0D[i0D].neighbor[j];
|
2587
|
+
if ( k < 0 || k >= nNumAtoms || k_prev == k )
|
2588
|
+
break;
|
2589
|
+
/* tetrahedral atom connectivity test */
|
2590
|
+
if ( type == INCHI_StereoType_Tetrahedral &&
|
2591
|
+
k != a2 &&
|
2592
|
+
!is_in_the_list( at[a2].neighbor, (AT_NUMB)k, at[a2].valence) ) {
|
2593
|
+
break;
|
2594
|
+
}
|
2595
|
+
/* Double bond, Cumulene and allene are tested in the next if() */
|
2596
|
+
}
|
2597
|
+
}
|
2598
|
+
/* find in the adjacency lists the double bond neighbor that leads to the opposite atom */
|
2599
|
+
if ( j == 4 && (type == INCHI_StereoType_Allene ||
|
2600
|
+
type == INCHI_StereoType_DoubleBond) ) {
|
2601
|
+
AT_NUMB *p1 = NULL, *p2 = NULL, *q1 = NULL, *q2 = NULL;
|
2602
|
+
i1n = (AT_NUMB)stereo0D[i0D].neighbor[0];
|
2603
|
+
i1 = (AT_NUMB)stereo0D[i0D].neighbor[1];
|
2604
|
+
i2 = (AT_NUMB)stereo0D[i0D].neighbor[2];
|
2605
|
+
i2n = (AT_NUMB)stereo0D[i0D].neighbor[3];
|
2606
|
+
/* find q1 and q2 */
|
2607
|
+
if ( !(q1 = is_in_the_list( at[i1].neighbor, i1n, at[i1].valence)) ||
|
2608
|
+
!(q2 = is_in_the_list( at[i2].neighbor, i2n, at[i2].valence)) ) {
|
2609
|
+
j = -2; /* error flag */
|
2610
|
+
} else
|
2611
|
+
/* allene or cumulene; follow double bonds from i1 to i2 */
|
2612
|
+
if ( !(p1 = is_in_the_list( at[i1].neighbor, i2, at[i1].valence)) ) {
|
2613
|
+
/* at[i1] and at[i2] are not connected: can be only allene or cumulene */
|
2614
|
+
AT_NUMB prev, cur, next;
|
2615
|
+
int num_dbond, i, next_ord, half_len;
|
2616
|
+
|
2617
|
+
cur = next = i1;
|
2618
|
+
len = half_len = 0;
|
2619
|
+
while ( len < 20 ) { /* arbitrary very high upper limit to prevent infinite loop */
|
2620
|
+
prev = cur;
|
2621
|
+
cur = next;
|
2622
|
+
for ( i = 0, num_dbond = 0; i < at[cur].valence; i ++ ) {
|
2623
|
+
/* follow double bond path && avoid going back */
|
2624
|
+
if ( at[cur].bond_type[i] == BOND_TYPE_DOUBLE &&
|
2625
|
+
prev != at[cur].neighbor[i] ) {
|
2626
|
+
next = at[cur].neighbor[i];
|
2627
|
+
next_ord = i;
|
2628
|
+
num_dbond ++;
|
2629
|
+
}
|
2630
|
+
}
|
2631
|
+
if ( num_dbond == 1 && next != i1 ) {
|
2632
|
+
len ++;
|
2633
|
+
if ( len == 1 ) {
|
2634
|
+
sb_ord_from_i1 = next_ord;
|
2635
|
+
}
|
2636
|
+
if ( type == INCHI_StereoType_Allene && next == (AT_NUMB)a2 ) {
|
2637
|
+
half_len = len;
|
2638
|
+
}
|
2639
|
+
} else {
|
2640
|
+
break;
|
2641
|
+
}
|
2642
|
+
}
|
2643
|
+
if ( cur == i2 && prev != cur && 0 == num_dbond && len > 1 &&
|
2644
|
+
(p2 = is_in_the_list( at[i2].neighbor, prev, at[i2].valence)) &&
|
2645
|
+
(type != INCHI_StereoType_Allene || len == 2*half_len )) {
|
2646
|
+
sb_ord_from_i2 = p2 - at[i2].neighbor;
|
2647
|
+
sn_ord_from_i1 = q1 - at[i1].neighbor;
|
2648
|
+
sn_ord_from_i2 = q2 - at[i2].neighbor;
|
2649
|
+
} else {
|
2650
|
+
j = -5; /* error flag */
|
2651
|
+
}
|
2652
|
+
} else
|
2653
|
+
/* allene must have been already processed, otherwise error */
|
2654
|
+
if ( type == INCHI_StereoType_Allene ) {
|
2655
|
+
/* error: atoms #1 and #2 of allene are connected */
|
2656
|
+
j = -3; /* error flag */
|
2657
|
+
} else
|
2658
|
+
/* double bond only; the bond type is not checked because at the end
|
2659
|
+
of the normalization it may happen to be alternating */
|
2660
|
+
if ( type == INCHI_StereoType_DoubleBond &&
|
2661
|
+
(p2 = is_in_the_list( at[i2].neighbor, i1, at[i2].valence) ) ) {
|
2662
|
+
sb_ord_from_i1 = p1 - at[i1].neighbor;
|
2663
|
+
sb_ord_from_i2 = p2 - at[i2].neighbor;
|
2664
|
+
sn_ord_from_i1 = q1 - at[i1].neighbor;
|
2665
|
+
sn_ord_from_i2 = q2 - at[i2].neighbor;
|
2666
|
+
} else {
|
2667
|
+
j = -4; /* error flag */
|
2668
|
+
}
|
2669
|
+
}
|
2670
|
+
if ( j != 4 ) {
|
2671
|
+
char szTemp[16];
|
2672
|
+
sprintf( szTemp, "#%d", i0D+1 );
|
2673
|
+
MOLFILE_ERR_SET (*err, 0, "Wrong 0D stereo descriptor(s):");
|
2674
|
+
MOLFILE_ERR_SET (*err, 0, szTemp);
|
2675
|
+
continue; /* error */
|
2676
|
+
}
|
2677
|
+
|
2678
|
+
switch ( type ) {
|
2679
|
+
case INCHI_StereoType_None:
|
2680
|
+
continue;
|
2681
|
+
case INCHI_StereoType_DoubleBond:
|
2682
|
+
case INCHI_StereoType_Allene:
|
2683
|
+
for ( j1 = 0; j1 < MAX_NUM_STEREO_BONDS && at[i1].sb_parity[j1]; j1 ++ )
|
2684
|
+
;
|
2685
|
+
for ( j2 = 0; j2 < MAX_NUM_STEREO_BONDS && at[i2].sb_parity[j2]; j2 ++ )
|
2686
|
+
;
|
2687
|
+
if ( j1 < MAX_NUM_STEREO_BONDS && j2 < MAX_NUM_STEREO_BONDS &&
|
2688
|
+
sb_ord_from_i1 >= 0 && sb_ord_from_i2 >= 0 &&
|
2689
|
+
sn_ord_from_i1 >= 0 && sn_ord_from_i2 >= 0) {
|
2690
|
+
|
2691
|
+
switch( parity ) {
|
2692
|
+
case INCHI_PARITY_ODD:
|
2693
|
+
at[i1].sb_parity[j1] = AB_PARITY_ODD;
|
2694
|
+
at[i2].sb_parity[j2] = AB_PARITY_EVEN;
|
2695
|
+
break;
|
2696
|
+
case INCHI_PARITY_EVEN:
|
2697
|
+
at[i1].sb_parity[j1] = AB_PARITY_ODD;
|
2698
|
+
at[i2].sb_parity[j2] = AB_PARITY_ODD;
|
2699
|
+
break;
|
2700
|
+
case INCHI_PARITY_UNKNOWN:
|
2701
|
+
at[i1].sb_parity[j1] = AB_PARITY_UNKN;
|
2702
|
+
at[i2].sb_parity[j2] = AB_PARITY_UNKN;
|
2703
|
+
break;
|
2704
|
+
case INCHI_PARITY_UNDEFINED:
|
2705
|
+
at[i1].sb_parity[j1] = AB_PARITY_UNDF;
|
2706
|
+
at[i2].sb_parity[j2] = AB_PARITY_UNDF;
|
2707
|
+
break;
|
2708
|
+
default:
|
2709
|
+
at[i1].sb_parity[j1] = AB_PARITY_NONE;
|
2710
|
+
at[i2].sb_parity[j2] = AB_PARITY_NONE;
|
2711
|
+
}
|
2712
|
+
|
2713
|
+
switch( parityNM ) {
|
2714
|
+
case INCHI_PARITY_ODD:
|
2715
|
+
at[i1].sb_parity[j1] |= AB_PARITY_ODD << SB_PARITY_SHFT;
|
2716
|
+
at[i2].sb_parity[j2] |= AB_PARITY_EVEN << SB_PARITY_SHFT;
|
2717
|
+
break;
|
2718
|
+
case INCHI_PARITY_EVEN:
|
2719
|
+
at[i1].sb_parity[j1] |= AB_PARITY_ODD << SB_PARITY_SHFT;
|
2720
|
+
at[i2].sb_parity[j2] |= AB_PARITY_ODD << SB_PARITY_SHFT;
|
2721
|
+
break;
|
2722
|
+
case INCHI_PARITY_UNKNOWN:
|
2723
|
+
at[i1].sb_parity[j1] |= AB_PARITY_UNKN << SB_PARITY_SHFT;
|
2724
|
+
at[i2].sb_parity[j2] |= AB_PARITY_UNKN << SB_PARITY_SHFT;
|
2725
|
+
break;
|
2726
|
+
case INCHI_PARITY_UNDEFINED:
|
2727
|
+
at[i1].sb_parity[j1] |= AB_PARITY_UNDF << SB_PARITY_SHFT;
|
2728
|
+
at[i2].sb_parity[j2] |= AB_PARITY_UNDF << SB_PARITY_SHFT;
|
2729
|
+
break;
|
2730
|
+
default:
|
2731
|
+
break;
|
2732
|
+
}
|
2733
|
+
|
2734
|
+
at[i1].sb_ord[j1] = sb_ord_from_i1;
|
2735
|
+
at[i1].sn_ord[j1] = sn_ord_from_i1;
|
2736
|
+
at[i1].sn_orig_at_num[j1] = at[i1n].orig_at_number;
|
2737
|
+
|
2738
|
+
at[i2].sb_ord[j2] = sb_ord_from_i2;
|
2739
|
+
at[i2].sn_ord[j2] = sn_ord_from_i2;
|
2740
|
+
at[i2].sn_orig_at_num[j2] = at[i2n].orig_at_number;
|
2741
|
+
}
|
2742
|
+
break;
|
2743
|
+
case INCHI_StereoType_Tetrahedral:
|
2744
|
+
switch( parity ) {
|
2745
|
+
case INCHI_PARITY_ODD:
|
2746
|
+
at[a2].p_parity = AB_PARITY_ODD;
|
2747
|
+
break;
|
2748
|
+
case INCHI_PARITY_EVEN:
|
2749
|
+
at[a2].p_parity = AB_PARITY_EVEN;
|
2750
|
+
break;
|
2751
|
+
case INCHI_PARITY_UNKNOWN:
|
2752
|
+
at[a2].p_parity = AB_PARITY_UNKN;
|
2753
|
+
break;
|
2754
|
+
case INCHI_PARITY_UNDEFINED:
|
2755
|
+
at[a2].p_parity = AB_PARITY_UNDF;
|
2756
|
+
break;
|
2757
|
+
default:
|
2758
|
+
continue;
|
2759
|
+
}
|
2760
|
+
for ( j = 0; j < 4; j ++ ) {
|
2761
|
+
k = stereo0D[i0D].neighbor[j];
|
2762
|
+
at[a2].p_orig_at_num[j] = at[k].orig_at_number;
|
2763
|
+
}
|
2764
|
+
break;
|
2765
|
+
default:
|
2766
|
+
break;
|
2767
|
+
}
|
2768
|
+
}
|
2769
|
+
|
2770
|
+
#ifdef INCHI_LIBRARY
|
2771
|
+
|
2772
|
+
if ( k = ReconcileAllCmlBondParities( at, nNumAtoms, 0 ) ) {
|
2773
|
+
char szErrCode[16];
|
2774
|
+
sprintf( szErrCode, "%d", k);
|
2775
|
+
AddMOLfileError( pStrErr, "0D Parities Reconciliation failed:" );
|
2776
|
+
AddMOLfileError( pStrErr, szErrCode );
|
2777
|
+
}
|
2778
|
+
|
2779
|
+
#endif
|
2780
|
+
|
2781
|
+
}
|
2782
|
+
return 0;
|
2783
|
+
}
|
2784
|
+
|
2785
|
+
#endif
|
2786
|
+
|