Révision 2170
| tmp/org.txm.groovy.core/src/groovy/org/txm/macro/prototypes/commands/CoocRecMacro.groovy (revision 2170) | ||
|---|---|---|
| 17 | 17 |
return false; |
| 18 | 18 |
} |
| 19 | 19 |
|
| 20 |
@Field @Option(name="word", usage="an example query", widget="String", required=true, def='Life') |
|
| 21 |
def word |
|
| 22 |
@Field @Option(name="ensure_transitive", usage="an example query", widget="Boolean", required=true, def='true') |
|
| 23 |
def ensure_transitive |
|
| 20 |
@Field @Option(name="starting_word", usage="The starting word", widget="String", required=true, def='Life') |
|
| 21 |
def starting_word |
|
| 24 | 22 |
@Field @Option(name="property", usage="an example query", widget="String", required=true, def='lemma') |
| 25 |
def property |
|
| 23 |
def property |
|
| 24 |
@Field @Option(name="mode", usage="an example query", widget="StringArray", metaVar="Poly Reccursive", required=true, def='Poly') |
|
| 25 |
def mode |
|
| 26 | 26 |
@Field @Option(name="distance_max", usage="an example query", widget="Integer", required=true, def='9') |
| 27 |
def distance_max |
|
| 27 |
def distance_max
|
|
| 28 | 28 |
@Field @Option(name="score_minimum", usage="an example query", widget="Float", required=true, def='2.5') |
| 29 |
def score_minimum |
|
| 29 |
def score_minimum
|
|
| 30 | 30 |
@Field @Option(name="profondeur_max", usage="an example query", widget="Integer", required=true, def='3') |
| 31 |
def profondeur_max |
|
| 32 |
@Field @Option(name="debug", usage="an example query", widget="Boolean", required=true, def='false')
|
|
| 33 |
def debug
|
|
| 31 |
def profondeur_max
|
|
| 32 |
@Field @Option(name="debug", usage="Show internal variable content", widget="StringArray", metaVar="OFF ON ALL REALLY ALL", required=true, def="OFF")
|
|
| 33 |
debug
|
|
| 34 | 34 |
|
| 35 | 35 |
// Open the parameters input dialog box |
| 36 | 36 |
if (!ParametersDialog.open(this)) return; |
| 37 |
if (debug == "OFF") debug = 0; else if (debug == "ON") debug = 1; else if (debug == "ALL") debug = 2 else if (debug == "REALLY ALL") debug = 3 |
|
| 38 |
mode = mode == "Poly" |
|
| 37 | 39 |
|
| 38 | 40 |
lmax = "" |
| 39 | 41 |
for (int i = 0 ; i < profondeur_max; i++) lmax += "\t" |
| ... | ... | |
| 45 | 47 |
return false |
| 46 | 48 |
} |
| 47 | 49 |
|
| 48 |
|
|
| 49 |
|
|
| 50 | 50 |
def rightCooc = new Cooccurrence(corpus) |
| 51 | 51 |
r = new HashSet() |
| 52 | 52 |
def printCoocs(def pivots, def level, def cooc, def right_dir) {
|
| 53 | 53 |
if (level.length() > lmax.length()) return; |
| 54 |
if (debug) println "#"+level+pivots |
|
| 55 |
|
|
| 54 |
if (debug > 1) println "#"+level+pivots
|
|
| 55 |
|
|
| 56 | 56 |
def bouts = [] |
| 57 |
|
|
| 57 |
|
|
| 58 | 58 |
for (def s : pivots) {
|
| 59 | 59 |
bouts << "[$property=\"$s\"%cd]" |
| 60 | 60 |
} |
| 61 | 61 |
String qs = bouts.join(" []{0, ${distance_max}} ")
|
| 62 |
|
|
| 62 |
|
|
| 63 | 63 |
def q = new CQLQuery(qs) |
| 64 |
if (debug) println level+q |
|
| 65 |
|
|
| 64 |
if (debug > 0) println level+q
|
|
| 65 |
|
|
| 66 | 66 |
r.addAll(pivots) |
| 67 |
|
|
| 68 |
if (right_dir) |
|
| 69 |
cooc.setParameters(q, [wordProperty], null, 0, 0, 1, distance_max+1, 1, score_minimum, 1, false, false); |
|
| 70 |
else |
|
| 71 |
cooc.setParameters(q, [wordProperty], null, distance_max+1, 1, 0, 0, 1, score_minimum, 1, false, false); |
|
| 72 |
|
|
| 67 |
|
|
| 68 |
if (mode) {
|
|
| 69 |
if (right_dir) |
|
| 70 |
cooc.setParameters(q, [wordProperty], null, 0, 0, 1, distance_max+1, 1, score_minimum, 1, false, false); |
|
| 71 |
else |
|
| 72 |
cooc.setParameters(q, [wordProperty], null, distance_max+1, 1, 0, 0, 1, score_minimum, 1, false, false); |
|
| 73 |
} else {
|
|
| 74 |
cooc.setParameters(q, [wordProperty], null, distance_max+1, 1, 1, distance_max+1, 1, score_minimum, 1, false, false); |
|
| 75 |
} |
|
| 73 | 76 |
cooc.setDirty(true) |
| 74 |
|
|
| 77 |
|
|
| 75 | 78 |
cooc.compute() |
| 79 |
if (debug > 0) println "cooc="+cooc.getLastParametersFromHistory() |
|
| 76 | 80 |
def lines = cooc.getLines() |
| 77 | 81 |
def words = [:] |
| 78 | 82 |
for (def line : lines) {
|
| 79 | 83 |
if (!r.contains(line.props[0])) {
|
| 80 | 84 |
words[line.props[0]] = line |
| 81 |
}
|
|
| 85 |
} |
|
| 82 | 86 |
} |
| 83 |
if (debug) println level+lines.size() |
|
| 84 |
|
|
| 87 |
if (debug > 1) println level+lines.size()
|
|
| 88 |
|
|
| 85 | 89 |
for (def word : words.keySet()) {
|
| 86 | 90 |
int size = 10+words[word].score |
| 87 | 91 |
if (size > 30) size = 30 |
| 88 | 92 |
|
| 89 | 93 |
def new_pivots = [] |
| 90 | 94 |
|
| 91 |
|
|
| 92 |
if (right_dir) {
|
|
| 93 |
def pivot = pivots[-1] |
|
| 94 |
println "$level\"$pivot\" -> \"$word\" [label=\"${words[word].score}\",fontsize=$size,weight=${words[word].score}];"
|
|
| 95 |
if (ensure_transitive) new_pivots.addAll(pivots) |
|
| 96 |
new_pivots << word |
|
| 97 |
} else {
|
|
| 95 |
if (mode) {
|
|
| 96 |
if (right_dir) {
|
|
| 97 |
def pivot = pivots[-1] |
|
| 98 |
println "$level\"$pivot\" -> \"$word\" [label=\"${words[word].score}\",fontsize=$size,weight=${words[word].score}];"
|
|
| 99 |
if (mode) new_pivots.addAll(pivots) |
|
| 100 |
new_pivots << word |
|
| 101 |
} else {
|
|
| 102 |
def pivot = pivots[0] |
|
| 103 |
println "$level\"$word\" -> \"$pivot\" [label=\"${words[word].score}\",fontsize=$size,weight=${words[word].score}];"
|
|
| 104 |
|
|
| 105 |
new_pivots << word |
|
| 106 |
|
|
| 107 |
if (mode) new_pivots.addAll(pivots) |
|
| 108 |
} |
|
| 109 |
printCoocs(new_pivots, level+" ", cooc, right_dir) |
|
| 110 |
} else {
|
|
| 98 | 111 |
def pivot = pivots[0] |
| 99 |
println "$level\"$word\" -> \"$pivot\" [label=\"${words[word].score}\",fontsize=$size,weight=${words[word].score}];"
|
|
| 100 |
|
|
| 101 |
new_pivots << word |
|
| 102 |
if (ensure_transitive) new_pivots.addAll(pivots) |
|
| 112 |
println "$level\"$pivot\" -- \"$word\" [label=\"${words[word].score}\",fontsize=$size,weight=${words[word].score}];"
|
|
| 113 |
printCoocs([word], level+" ", cooc, true) |
|
| 103 | 114 |
} |
| 104 |
|
|
| 105 |
printCoocs(new_pivots, level+" ", cooc, right_dir) |
|
| 106 | 115 |
} |
| 107 | 116 |
} |
| 108 | 117 |
|
| 109 |
println "digraph {"
|
|
| 110 |
println "rankdir=LR;" |
|
| 111 |
println "\"$word\" [shape=box];" |
|
| 112 |
printCoocs([word], "", rightCooc, false) |
|
| 113 |
r = new HashSet() |
|
| 114 |
printCoocs([word], "", rightCooc, true) |
|
| 118 |
|
|
| 119 |
if (mode) {
|
|
| 120 |
println "digraph {"
|
|
| 121 |
println "rankdir=LR;" |
|
| 122 |
} else {
|
|
| 123 |
println "graph {"
|
|
| 124 |
} |
|
| 125 |
|
|
| 126 |
println "\"$starting_word\" [shape=box];" |
|
| 127 |
if (mode) {
|
|
| 128 |
printCoocs([starting_word], "", rightCooc, true) |
|
| 129 |
r = new HashSet() |
|
| 130 |
printCoocs([starting_word], "", rightCooc, false) |
|
| 131 |
} else {
|
|
| 132 |
printCoocs([starting_word], "", rightCooc, true) |
|
| 133 |
} |
|
| 134 |
|
|
| 115 | 135 |
println "}" |
Formats disponibles : Unified diff