| 1 | /* File: ureadseq.h |
|---|
| 2 | * |
|---|
| 3 | * Header for module UReadSeq |
|---|
| 4 | */ |
|---|
| 5 | |
|---|
| 6 | #ifndef UREADSEQ_H |
|---|
| 7 | #define UREADSEQ_H |
|---|
| 8 | #define short int |
|---|
| 9 | |
|---|
| 10 | |
|---|
| 11 | typedef char boolean; |
|---|
| 12 | #define NEWLINE '\n' |
|---|
| 13 | #define false 0 |
|---|
| 14 | #define true 1 |
|---|
| 15 | #define min(a,b) (a<b)?a:b |
|---|
| 16 | #define max(a,b) (a>b)?a:b |
|---|
| 17 | #define skipwhitespace(string) {while (*string <= ' ' && *string != 0) string++;} |
|---|
| 18 | |
|---|
| 19 | /* NLM strings */ |
|---|
| 20 | #define is_upper(c) ('A'<=(c) && (c)<='Z') |
|---|
| 21 | #define is_lower(c) ('a'<=(c) && (c)<='z') |
|---|
| 22 | #define to_lower(c) ((char)(is_upper(c) ? (c)+' ' : (c))) |
|---|
| 23 | #define to_upper(c) ((char)(is_lower(c) ? (c)-' ' : (c))) |
|---|
| 24 | |
|---|
| 25 | |
|---|
| 26 | /* readSeq errors */ |
|---|
| 27 | #define eFileNotFound -1 |
|---|
| 28 | #define eNoData -2 |
|---|
| 29 | #define eMemFull -3 |
|---|
| 30 | #define eItemNotFound -4 |
|---|
| 31 | #define eOneFormat -5 |
|---|
| 32 | #define eUnequalSize -6 |
|---|
| 33 | #define eFileCreate -7 |
|---|
| 34 | #define eUnknownFormat -8 |
|---|
| 35 | #define eOptionBad -9 |
|---|
| 36 | #define eASNerr -10 |
|---|
| 37 | #define ePipeStdin -11 |
|---|
| 38 | |
|---|
| 39 | /* magic number for readSeq(whichEntry) to give seq list */ |
|---|
| 40 | #define kListSequences -1 |
|---|
| 41 | |
|---|
| 42 | /* sequence types parsed by getseqtype */ |
|---|
| 43 | #define kOtherSeq 0 |
|---|
| 44 | #define kDNA 1 |
|---|
| 45 | #define kRNA 2 |
|---|
| 46 | #define kNucleic 3 |
|---|
| 47 | #define kAmino 4 |
|---|
| 48 | |
|---|
| 49 | /* formats known to readSeq */ |
|---|
| 50 | #define kIG 1 |
|---|
| 51 | #define kGenBank 2 |
|---|
| 52 | #define kNBRF 3 |
|---|
| 53 | #define kEMBL 4 |
|---|
| 54 | #define kGCG 5 |
|---|
| 55 | #define kStrider 6 |
|---|
| 56 | #define kFitch 7 |
|---|
| 57 | #define kPearson 8 |
|---|
| 58 | #define kZuker 9 |
|---|
| 59 | #define kOlsen 10 |
|---|
| 60 | #define kPhylip2 11 |
|---|
| 61 | #define kPhylip4 12 |
|---|
| 62 | #define kPhylip3 kPhylip4 |
|---|
| 63 | #define kPhylip kPhylip4 |
|---|
| 64 | #define kPlain 13 /* keep this at #13 */ |
|---|
| 65 | #define kPIR 14 |
|---|
| 66 | #define kMSF 15 |
|---|
| 67 | #define kASN1 16 |
|---|
| 68 | #define kPAUP 17 |
|---|
| 69 | #define kPretty 18 |
|---|
| 70 | |
|---|
| 71 | #define kMaxFormat 18 |
|---|
| 72 | #define kMinFormat 1 |
|---|
| 73 | #define kNoformat -1 /* format not tested */ |
|---|
| 74 | #define kUnknown 0 /* format not determinable */ |
|---|
| 75 | |
|---|
| 76 | /* subsidiary types */ |
|---|
| 77 | #define kASNseqentry 51 |
|---|
| 78 | #define kASNseqset 52 |
|---|
| 79 | |
|---|
| 80 | #define kPhylipInterleave 61 |
|---|
| 81 | #define kPhylipSequential 62 |
|---|
| 82 | |
|---|
| 83 | |
|---|
| 84 | typedef struct { |
|---|
| 85 | boolean isactive, baseonlynum; |
|---|
| 86 | boolean numright, numleft, numtop, numbot; |
|---|
| 87 | boolean nameright, nameleft, nametop; |
|---|
| 88 | boolean noleaves, domatch, degap; |
|---|
| 89 | char matchchar, gapchar; |
|---|
| 90 | short numline, atseq; |
|---|
| 91 | short namewidth, numwidth; |
|---|
| 92 | short interline, spacer, seqwidth, tab; |
|---|
| 93 | } prettyopts; |
|---|
| 94 | |
|---|
| 95 | #define gPrettyInit(p) { \ |
|---|
| 96 | p.isactive=false;\ |
|---|
| 97 | p.baseonlynum=true;\ |
|---|
| 98 | p.numline= p.atseq= 0;\ |
|---|
| 99 | p.numright= p.numleft= p.numtop= p.numbot= false;\ |
|---|
| 100 | p.nameright= p.nameleft= p.nametop= false;\ |
|---|
| 101 | p.noleaves= p.domatch= p.degap= false;\ |
|---|
| 102 | p.matchchar='.';\ |
|---|
| 103 | p.gapchar='-';\ |
|---|
| 104 | p.namewidth=10;\ |
|---|
| 105 | p.numwidth=5;\ |
|---|
| 106 | p.interline=1;\ |
|---|
| 107 | p.spacer=10;\ |
|---|
| 108 | p.seqwidth=50;\ |
|---|
| 109 | p.tab=0; } |
|---|
| 110 | |
|---|
| 111 | extern prettyopts gPretty; |
|---|
| 112 | |
|---|
| 113 | |
|---|
| 114 | #ifdef __cplusplus |
|---|
| 115 | extern "C" { |
|---|
| 116 | #endif |
|---|
| 117 | |
|---|
| 118 | extern short seqFileFormat(const char *filename, long *skiplines, short *error ); |
|---|
| 119 | extern short seqFileFormatFp(FILE *fseq, long *skiplines, short *error ); |
|---|
| 120 | |
|---|
| 121 | extern char *listSeqs(const char *filename, const long skiplines, |
|---|
| 122 | const short format, short *nseq, short *error ); |
|---|
| 123 | |
|---|
| 124 | extern char *readSeq(const short whichEntry, const char *filename, |
|---|
| 125 | const long skiplines, const short format, |
|---|
| 126 | long *seqlen, short *nseq, short *error, char *seqid ); |
|---|
| 127 | |
|---|
| 128 | extern char *readSeqFp(const short whichEntry_, FILE *fp_, |
|---|
| 129 | const long skiplines_, const short format_, |
|---|
| 130 | long *seqlen_, short *nseq_, short *error_, char *seqid_ ); |
|---|
| 131 | |
|---|
| 132 | extern short writeSeq(FILE *outf, const char *seq, const long seqlen, |
|---|
| 133 | const short outform, const char *seqid ); |
|---|
| 134 | |
|---|
| 135 | extern unsigned long CRC32checksum(const char *seq, const long seqlen, unsigned long *checktotal); |
|---|
| 136 | extern unsigned long GCGchecksum(const char *seq, const long seqlen, unsigned long *checktotal); |
|---|
| 137 | #ifdef SMALLCHECKSUM |
|---|
| 138 | #define seqchecksum GCGchecksum |
|---|
| 139 | #else |
|---|
| 140 | #define seqchecksum CRC32checksum |
|---|
| 141 | #endif |
|---|
| 142 | |
|---|
| 143 | extern short getseqtype(const char *seq, const long seqlen ); |
|---|
| 144 | extern char *compressSeq( const char gapc, const char *seq, const long seqlen, long *newlen); |
|---|
| 145 | |
|---|
| 146 | #ifdef NCBI |
|---|
| 147 | |
|---|
| 148 | extern char *listASNSeqs(const char *filename, const long skiplines, |
|---|
| 149 | const short format, short *nseq, short *error ); |
|---|
| 150 | |
|---|
| 151 | extern char *readASNSeq(const short whichEntry, const char *filename, |
|---|
| 152 | const long skiplines, const short format, |
|---|
| 153 | long *seqlen, short *nseq, short *error, char **seqid ); |
|---|
| 154 | #endif |
|---|
| 155 | |
|---|
| 156 | |
|---|
| 157 | /* patches for some missing string.h stuff */ |
|---|
| 158 | extern int Strcasecmp(const char *a, const char *b); |
|---|
| 159 | extern int Strncasecmp(const char *a, const char *b, long maxn); |
|---|
| 160 | |
|---|
| 161 | #ifdef __cplusplus |
|---|
| 162 | } |
|---|
| 163 | #endif |
|---|
| 164 | |
|---|
| 165 | #endif /*UREADSEQ_H*/ |
|---|
| 166 | |
|---|