root / tmp / org.txm.core / src / java / org / txm / scripts / importer / WriteIdAndNAttributes.groovy @ 2473
History | View | Annotate | Download (5.7 kB)
| 1 |
package org.txm.scripts.importer
|
|---|---|
| 2 |
|
| 3 |
import org.txm.importer.StaxIdentityParser |
| 4 |
|
| 5 |
class WriteIdAndNAttributes extends StaxIdentityParser { |
| 6 |
|
| 7 |
String textname
|
| 8 |
|
| 9 |
int nMileStone = 1, nPb = 1, nCb = 1, nLb = 1, nW = 1, nSeg = 1, nRejet = 1; |
| 10 |
String previousMileStone, previousPb, previousCb, previousW;
|
| 11 |
|
| 12 |
|
| 13 |
String PB = "pb", CB = "cb", LB = "lb", ID = "id", TYPE = "type", |
| 14 |
N = "n", CORRESP = "corresp", FACS="facs", W="w", PC="pc", SEG="seg", |
| 15 |
UNIT="unit", XML="xml", WP="wp", SURFACE="surface", POINT = ".", REJET ="rejet"; |
| 16 |
|
| 17 |
public WriteIdAndNAttributes(File xmlFile, String textname) { |
| 18 |
super(xmlFile);
|
| 19 |
|
| 20 |
this.textname = textname
|
| 21 |
} |
| 22 |
|
| 23 |
protected void writeAttributes() { |
| 24 |
// do nothing
|
| 25 |
} |
| 26 |
|
| 27 |
protected void processStartElement() { |
| 28 |
String id = null |
| 29 |
String n = null |
| 30 |
String type = null |
| 31 |
String corresp = null |
| 32 |
String facs = null |
| 33 |
|
| 34 |
for (int i = 0 ; i < parser.getAttributeCount() ; i++) { |
| 35 |
if (parser.getAttributeLocalName(i) == ID) {
|
| 36 |
id = parser.getAttributeValue(i) |
| 37 |
} else if (parser.getAttributeLocalName(i) == N) { |
| 38 |
n = parser.getAttributeValue(i) |
| 39 |
} else if (parser.getAttributeLocalName(i) == TYPE) { |
| 40 |
type = parser.getAttributeValue(i) |
| 41 |
} else if (parser.getAttributeLocalName(i) == CORRESP) { |
| 42 |
corresp = parser.getAttributeValue(i) |
| 43 |
} else if (parser.getAttributeLocalName(i) == FACS) { |
| 44 |
facs = parser.getAttributeValue(i) |
| 45 |
} |
| 46 |
} |
| 47 |
|
| 48 |
super.processStartElement(); // attributes are not written because super.writeAttributes() is not called |
| 49 |
|
| 50 |
if (localname == "milestone" && parser.getAttributeValue(null, UNIT) == SURFACE) { |
| 51 |
if (n == null) { |
| 52 |
n = nMileStone |
| 53 |
} else {
|
| 54 |
int tmp = Integer.parseInt(n) |
| 55 |
if (nMileStone < tmp) {
|
| 56 |
nMileStone = tmp |
| 57 |
n = tmp |
| 58 |
} else {
|
| 59 |
n = nMileStone; |
| 60 |
} |
| 61 |
} |
| 62 |
|
| 63 |
if (id == null) { // don't rewrite "id" |
| 64 |
if (facs == null) { |
| 65 |
println "Error: found milestone@type=\"surface\" with no @facs at "+parser.getLocation()
|
| 66 |
throw new Exception("no facs attribute") |
| 67 |
} |
| 68 |
if (facs.lastIndexOf(POINT) > 0) facs = facs.substring(0, facs.lastIndexOf(POINT)) |
| 69 |
previousMileStone = "surf_$textname"+"_"+facs |
| 70 |
id = previousMileStone |
| 71 |
} else {
|
| 72 |
previousMileStone = id |
| 73 |
} |
| 74 |
|
| 75 |
nMileStone++ |
| 76 |
} else if (localname == PB) { |
| 77 |
if (n == null) { |
| 78 |
n = nPb |
| 79 |
} else {
|
| 80 |
try {
|
| 81 |
int tmp = Integer.parseInt(n) |
| 82 |
if (nPb < tmp) {
|
| 83 |
nPb = tmp |
| 84 |
} else {
|
| 85 |
n = nPb; |
| 86 |
} |
| 87 |
} catch(Exception e) { } |
| 88 |
} |
| 89 |
|
| 90 |
if (id == null) { // don't rewrite "id" |
| 91 |
id = "page_${textname}_"+nPb
|
| 92 |
previousPb = "${textname}_"+nPb
|
| 93 |
} else {
|
| 94 |
previousPb = id |
| 95 |
} |
| 96 |
|
| 97 |
nRejet = nLb = nCb = 1
|
| 98 |
nPb++ |
| 99 |
} else if (localname == CB) { |
| 100 |
if (n == null) { |
| 101 |
n = nCb |
| 102 |
} else {
|
| 103 |
try {
|
| 104 |
int tmp = Integer.parseInt(n) |
| 105 |
if (nCb < tmp) {
|
| 106 |
nCb = tmp |
| 107 |
} else {
|
| 108 |
n = nCb; |
| 109 |
} |
| 110 |
} catch(Exception e) { } |
| 111 |
} |
| 112 |
|
| 113 |
if (id == null) { // don't rewrite "id" |
| 114 |
previousCb = "${previousPb}_"+nCb
|
| 115 |
id = "col_${previousPb}_"+nCb
|
| 116 |
} else {
|
| 117 |
previousCb = id |
| 118 |
} |
| 119 |
|
| 120 |
nLb = nRejet = 1
|
| 121 |
nCb++ |
| 122 |
} else if (localname == LB) { |
| 123 |
if (n == null) { |
| 124 |
if (REJET == type) {
|
| 125 |
println "Warning: no 'n' attribute provided for 'rejet' line break at "+parser.getLocation()
|
| 126 |
} |
| 127 |
|
| 128 |
n = nLb |
| 129 |
} else {
|
| 130 |
if (REJET != type) {
|
| 131 |
try { // if n is numeric update nLgRelative |
| 132 |
int tmp = Integer.parseInt(n) |
| 133 |
nLb = tmp; |
| 134 |
} catch(Exception e) { |
| 135 |
// n is not numeric use it
|
| 136 |
} |
| 137 |
} |
| 138 |
} |
| 139 |
|
| 140 |
if (id == null) { // don't rewrite "id" |
| 141 |
if (REJET == type) {
|
| 142 |
id = "line_${previousCb}_"+n+"_r"+nRejet |
| 143 |
} else {
|
| 144 |
id = "line_${previousCb}_"+n
|
| 145 |
} |
| 146 |
} |
| 147 |
|
| 148 |
if (corresp == null && REJET == type) { |
| 149 |
//println "Warning: no 'corresp' attribute provided for 'rejet' line break at "+parser.getLocation()
|
| 150 |
corresp = "#line_${previousCb}_"+n; // next word id in the same column |
| 151 |
} |
| 152 |
|
| 153 |
if (REJET != type) {
|
| 154 |
nLb++; |
| 155 |
} else {
|
| 156 |
nRejet++; |
| 157 |
} |
| 158 |
|
| 159 |
} else if (localname == W || localname == PC) { |
| 160 |
if (n == null) { |
| 161 |
n = nW |
| 162 |
} else {
|
| 163 |
try {
|
| 164 |
int tmp = Integer.parseInt(n) |
| 165 |
if (nW < tmp) {
|
| 166 |
nW = tmp |
| 167 |
} else {
|
| 168 |
n = nW; |
| 169 |
} |
| 170 |
} catch(Exception e) { } |
| 171 |
} |
| 172 |
|
| 173 |
if (id == null) { // don't rewrite "id" |
| 174 |
//previousW = "${previousPb}_"+nW
|
| 175 |
id = localname+"_${textname}_"+nW
|
| 176 |
} |
| 177 |
previousW = id |
| 178 |
nSeg = 1
|
| 179 |
nW++ |
| 180 |
} else if (localname == SEG && WP == type) { |
| 181 |
if (n == null) { |
| 182 |
n = nSeg |
| 183 |
} else {
|
| 184 |
try {
|
| 185 |
int tmp = Integer.parseInt(n) |
| 186 |
if (nSeg < tmp) {
|
| 187 |
nSeg = tmp |
| 188 |
} else {
|
| 189 |
n = nSeg; |
| 190 |
} |
| 191 |
} catch(Exception e) { } |
| 192 |
} |
| 193 |
|
| 194 |
if (id == null) { // don't rewrite "id" |
| 195 |
id = "w_p_"+previousW+"_"+nSeg |
| 196 |
} |
| 197 |
nSeg++ |
| 198 |
} |
| 199 |
|
| 200 |
// write attributes except ID, N and CORRESP
|
| 201 |
for (int i = 0 ; i < parser.getAttributeCount() ; i++) { |
| 202 |
if (parser.getAttributeLocalName(i) == ID) {
|
| 203 |
|
| 204 |
} else if (parser.getAttributeLocalName(i) == N) { |
| 205 |
|
| 206 |
} else if (parser.getAttributeLocalName(i) == CORRESP) { |
| 207 |
|
| 208 |
} else {
|
| 209 |
writeAttribute(parser.getAttributePrefix(i), parser.getAttributeLocalName(i), parser.getAttributeValue(i)) |
| 210 |
} |
| 211 |
} |
| 212 |
//if (localname == "lb") println "write id: "+["xml", "id", id]
|
| 213 |
if (id != null) |
| 214 |
writeAttribute(XML, ID, id) |
| 215 |
if (n != null) |
| 216 |
writeAttribute(null, N, n)
|
| 217 |
if (corresp != null) |
| 218 |
writeAttribute(null, CORRESP, corresp)
|
| 219 |
} |
| 220 |
|
| 221 |
public static void main(String[] args) { |
| 222 |
File xmlFile = new File("/home/mdecorde/xml/bugrejet/Psautier5-or28.xml") |
| 223 |
File outFile = new File("/home/mdecorde/xml/bugrejet/Psautier5-or28-o.xml") |
| 224 |
|
| 225 |
WriteIdAndNAttributes wiana = new WriteIdAndNAttributes(xmlFile, "qgraal_cm") |
| 226 |
println wiana.process(outFile) |
| 227 |
|
| 228 |
String text = xmlFile.getText().replaceAll(">", ">\n"); |
| 229 |
new File("/home/mdecorde/xml/bugrejet/Psautier5-or28-p.xml").write(text); |
| 230 |
text = outFile.getText().replaceAll(">", ">\n"); |
| 231 |
new File("/home/mdecorde/xml/bugrejet/Psautier5-or28-o-p.xml").write(text); |
| 232 |
} |
| 233 |
} |