| 1 |
/* FormatParser.java */ |
| 2 |
|
| 3 |
|
| 4 |
/* |
| 5 |
This parser parses Fortran format strings. |
| 6 |
*/ |
| 7 |
|
| 8 |
|
| 9 |
options { |
| 10 |
STATIC = true; |
| 11 |
DEBUG_PARSER = false; |
| 12 |
DEBUG_TOKEN_MANAGER = false; |
| 13 |
DEBUG_LOOKAHEAD = false; |
| 14 |
} |
| 15 |
|
| 16 |
PARSER_BEGIN(FormatParser) |
| 17 |
package org.j_paine.formatter; |
| 18 |
|
| 19 |
class FormatParser |
| 20 |
{ |
| 21 |
} |
| 22 |
|
| 23 |
PARSER_END(FormatParser) |
| 24 |
|
| 25 |
|
| 26 |
SKIP : |
| 27 |
{ |
| 28 |
<(" ")+> |
| 29 |
} |
| 30 |
|
| 31 |
|
| 32 |
TOKEN : |
| 33 |
{ |
| 34 |
<INTEGER: (["0"-"9"])+ > |
| 35 |
} |
| 36 |
// An unsigned integer, for repetition factors, field widths, etc. |
| 37 |
// previously: <INTEGER: ["1"-"9"] (["0"-"9"])* > |
| 38 |
|
| 39 |
|
| 40 |
TOKEN : |
| 41 |
{ |
| 42 |
<STRING: "'" ( ~["'"] )* "'" > |
| 43 |
} |
| 44 |
// A string literal inside a format. We haven't implemented |
| 45 |
// embedded quotes yet. |
| 46 |
|
| 47 |
TOKEN : { < A_DESC : "A" | "a" > } |
| 48 |
TOKEN : { < P_DESC : "P" | "p" > } |
| 49 |
TOKEN : { < X_DESC : "X" | "x" > } |
| 50 |
TOKEN : { < I_DESC : "I" | "i" > } |
| 51 |
TOKEN : { < F_DESC : "F" | "f" > } |
| 52 |
TOKEN : { < D_DESC : "D" | "d" > } |
| 53 |
TOKEN : { < E_DESC : "E" | "e" > } |
| 54 |
TOKEN : { < G_DESC : "G" | "g" > } |
| 55 |
TOKEN : { < L_DESC : "L" | "l" > } |
| 56 |
|
| 57 |
int Integer(): |
| 58 |
{ Token t; |
| 59 |
} |
| 60 |
{ |
| 61 |
t=<INTEGER> |
| 62 |
{ return (Integer.valueOf(t.image)).intValue(); } |
| 63 |
} |
| 64 |
|
| 65 |
/* I split FormatIOElement into FormatIOElementFloat and |
| 66 |
* FormatIOElementNonFloat because a floating point edit |
| 67 |
* descriptor (F, E, D, or G) may follow a P edit descriptor |
| 68 |
* without a comma. --kgs |
| 69 |
*/ |
| 70 |
|
| 71 |
FormatElement FormatIOElementFloat(): |
| 72 |
{ FormatElement fe; |
| 73 |
int w, d, m; |
| 74 |
w = d = m = -1; |
| 75 |
} |
| 76 |
{ |
| 77 |
// for Iw.m, ignore the .m value |
| 78 |
/* added D and G edit descriptors, but just use the E implementation. --kgs */ |
| 79 |
( |
| 80 |
<F_DESC> w=Integer() "." d=Integer() { fe=new FormatF(w,d); } |
| 81 |
| <D_DESC> w=Integer() "." d=Integer() { fe=new FormatE(w,d); } |
| 82 |
| <E_DESC> w=Integer() "." d=Integer() { fe=new FormatE(w,d); } |
| 83 |
| <G_DESC> w=Integer() "." d=Integer() { fe=new FormatE(w,d); } |
| 84 |
) |
| 85 |
{ return fe; } |
| 86 |
} |
| 87 |
|
| 88 |
FormatElement FormatIOElementNonFloat(): |
| 89 |
{ FormatElement fe; |
| 90 |
int w, d, m; |
| 91 |
w = d = m = -1; |
| 92 |
} |
| 93 |
{ |
| 94 |
// for Iw.m, ignore the .m value |
| 95 |
/* added L edit descriptor. --kgs */ |
| 96 |
( |
| 97 |
<A_DESC> [w=Integer()] { fe=new FormatA(w); } |
| 98 |
| <I_DESC> w=Integer() ["." m=Integer()] { fe=new FormatI(w); } |
| 99 |
| <L_DESC> w=Integer() { fe=new FormatL(w); } |
| 100 |
) |
| 101 |
{ return fe; } |
| 102 |
} |
| 103 |
// This represents a format element that transfers one |
| 104 |
// data item. |
| 105 |
|
| 106 |
|
| 107 |
FormatElement FormatNonIOElement(): {} |
| 108 |
{ |
| 109 |
<X_DESC> { return new FormatX(); } |
| 110 |
} |
| 111 |
// This represents a format element that doesn't transfer |
| 112 |
// any data items. |
| 113 |
|
| 114 |
|
| 115 |
FormatElement FormatElement(): |
| 116 |
{ FormatElement fe; |
| 117 |
} |
| 118 |
{ |
| 119 |
( fe=FormatIOElementFloat() |
| 120 |
| fe=FormatIOElementNonFloat() |
| 121 |
| fe=FormatNonIOElement() |
| 122 |
| fe=FormatScale() |
| 123 |
) |
| 124 |
{ return fe; } |
| 125 |
} |
| 126 |
|
| 127 |
FormatElement FormatScale(): |
| 128 |
{ FormatElement fe = null; |
| 129 |
int r=1; |
| 130 |
} |
| 131 |
{ |
| 132 |
/* Commas may be omitted between a P edit descriptor and an |
| 133 |
* immediately following F, E, D, or G edit descriptor (13.5.9). |
| 134 |
* --kgs |
| 135 |
*/ |
| 136 |
|
| 137 |
<P_DESC> [ [r=Integer()] (fe=FormatIOElementFloat()) ] |
| 138 |
{ |
| 139 |
return new FormatP(r, fe); |
| 140 |
} |
| 141 |
} |
| 142 |
|
| 143 |
FormatSlash FormatSlash(): {} |
| 144 |
{ |
| 145 |
"/" { return new FormatSlash(); } |
| 146 |
} |
| 147 |
// These are a special case. Unlike other format elements, |
| 148 |
// Fortran permits several slashes to be concatenated without |
| 149 |
// commas to separate them, and you can't use a repetition |
| 150 |
// factor on them. |
| 151 |
|
| 152 |
|
| 153 |
FormatString FormatString(): |
| 154 |
{ Token t; |
| 155 |
String s; |
| 156 |
} |
| 157 |
{ |
| 158 |
( t=<STRING> ) |
| 159 |
{ s = t.image; |
| 160 |
s = s.substring(1,s.length()-1); // Remove the quotes. |
| 161 |
return new FormatString(s); |
| 162 |
} |
| 163 |
} |
| 164 |
// Another special case that can't be repeated, and can be |
| 165 |
// concatenated to other elements without commas. |
| 166 |
|
| 167 |
|
| 168 |
void OptionalFormatSlashesOrStrings( Format f ): |
| 169 |
{ FormatUniv fs; |
| 170 |
} |
| 171 |
{ |
| 172 |
( (fs=FormatSlash() | fs=FormatString()) { f.addElement(fs); } )* |
| 173 |
} |
| 174 |
|
| 175 |
|
| 176 |
FormatRepeatedItem FormatRepeatedItem(): |
| 177 |
{ int r=1; |
| 178 |
FormatUniv fu; |
| 179 |
} |
| 180 |
{ |
| 181 |
[ r=Integer() ] |
| 182 |
( "(" fu=Format() ")" |
| 183 |
| fu=FormatElement() |
| 184 |
) |
| 185 |
{ |
| 186 |
/* here we check whether the parsed format element is a P edit |
| 187 |
* descriptor. in that case, it may have parsed a floating point |
| 188 |
* edit descriptor along with it (if it followed without a comma) |
| 189 |
* so return that element here. --kgs |
| 190 |
*/ |
| 191 |
|
| 192 |
if(fu instanceof FormatP) { |
| 193 |
FormatRepeatedItem ritem; |
| 194 |
|
| 195 |
ritem = ((FormatP)fu).getRepeatedItem(); |
| 196 |
|
| 197 |
if(ritem != null) |
| 198 |
return ritem; |
| 199 |
else |
| 200 |
return new FormatRepeatedItem( r, fu ); |
| 201 |
} |
| 202 |
else |
| 203 |
return new FormatRepeatedItem( r, fu ); |
| 204 |
} |
| 205 |
} |
| 206 |
|
| 207 |
|
| 208 |
void FormatGroup( Format f ): |
| 209 |
{ FormatRepeatedItem fri; |
| 210 |
} |
| 211 |
{ |
| 212 |
( OptionalFormatSlashesOrStrings( f ) |
| 213 |
[ fri = FormatRepeatedItem() { if(fri != null) f.addElement(fri); } |
| 214 |
OptionalFormatSlashesOrStrings( f ) |
| 215 |
] |
| 216 |
) |
| 217 |
} |
| 218 |
// This rather messy syntax allows us to have slashes and/or |
| 219 |
// strings either side of a format element or repeated group |
| 220 |
// without needing to separate them from each other or the element |
| 221 |
// with commas. |
| 222 |
// It also means that we can have empty format groups and format |
| 223 |
// groups that don't transfer any data elements. So for example, |
| 224 |
// the format ,/, is valid under this grammar. |
| 225 |
|
| 226 |
|
| 227 |
Format Format(): |
| 228 |
{ FormatRepeatedItem fri; |
| 229 |
Format f = new Format(); |
| 230 |
} |
| 231 |
{ |
| 232 |
( FormatGroup(f) ) |
| 233 |
( "," ( FormatGroup(f) ) )* |
| 234 |
{ return f; } |
| 235 |
} |