Commit 681727718acccf4a63d6dca1cb4c682a05217abb
1 parent
a6478f31
Added Świgra
Showing
122 changed files
with
111222 additions
and
0 deletions
Too many changes to show.
To preserve performance only 2 of 122 files are displayed.
swigra/disambiguator-pcfg/README
0 → 100644
| 1 | +Katalog zawiera pliki źródłowe dezambiguatora dla lasów składniowych w formacie Świgra/Składnica oraz: | ||
| 2 | + | ||
| 3 | +pcfg-grammar.pkl - gramatyka pcfg slużąca do dezambiguacji | ||
| 4 | +skrypt make-model.py służy do stworzenia nowej gamatyki | ||
| 5 | + | ||
| 6 | +skrypt disamb-tree.py służy do automatycznej anotacji ujednoznacznionego drzewa | ||
| 7 | + | ||
| 8 | +sposób użycia: | ||
| 9 | + | ||
| 10 | + python disamb-tree.py sciezka/do/nazwa_pliku_wejsciowego.xml | ||
| 11 | + skrypt utworzy plik sciezka/do/nazwa_pliku_wejscioweg-disamb.xml | ||
| 0 | \ No newline at end of file | 12 | \ No newline at end of file |
swigra/disambiguator-pcfg/disamb-tree.py
0 → 100644
| 1 | +# -*- encoding: utf-8 -*- | ||
| 2 | +__author__ = 'nika' | ||
| 3 | + | ||
| 4 | +from xml.sax import make_parser | ||
| 5 | +from treeparser import TreeParser | ||
| 6 | +import pickle | ||
| 7 | +import sys | ||
| 8 | +import xml.dom.minidom | ||
| 9 | +import codecs | ||
| 10 | + | ||
| 11 | +try: | ||
| 12 | + f = sys.argv[1] | ||
| 13 | +except IndexError: | ||
| 14 | + print("ERROR: no filename given") | ||
| 15 | + exit() | ||
| 16 | + | ||
| 17 | +# parse xml file for future processing | ||
| 18 | +dom = xml.dom.minidom.parse(sys.argv[1]) | ||
| 19 | +nodes = dom.getElementsByTagName("node") | ||
| 20 | + | ||
| 21 | +if nodes: # if there is anything to disambiguate | ||
| 22 | + # load grammar | ||
| 23 | +# grammar = pickle.load(open("grammars/pcfg-tfw-130718.pkl")) | ||
| 24 | + grammar = pickle.load(open("grammars/pcfg-tfw-150326.pkl")) | ||
| 25 | + # make parser, parse tree | ||
| 26 | + parser = make_parser() | ||
| 27 | + handler = TreeParser(f) | ||
| 28 | + parser.setContentHandler(handler) | ||
| 29 | + parser.parse(f) | ||
| 30 | + tree = handler.getTree() | ||
| 31 | + | ||
| 32 | + # disambiguation | ||
| 33 | + tree.act_pcfg(grammar) | ||
| 34 | + disamb_nodes = tree.getDisambNodes() | ||
| 35 | + | ||
| 36 | +# update chosen nodes (if any) | ||
| 37 | +for node in nodes: | ||
| 38 | + if node.attributes["nid"].value in disamb_nodes: | ||
| 39 | + node.attributes["chosen"] = "true" | ||
| 40 | + children_all = node.getElementsByTagName("children") | ||
| 41 | + for children in children_all: | ||
| 42 | + chosen = True | ||
| 43 | + for child in children.getElementsByTagName("child"): | ||
| 44 | + if child.attributes["nid"].value not in disamb_nodes: | ||
| 45 | + chosen = False | ||
| 46 | + if chosen: | ||
| 47 | + children.attributes["chosen"] = "true" | ||
| 48 | + else: | ||
| 49 | + node.attributes["chosen"] = "false" | ||
| 50 | + | ||
| 51 | +new_f = sys.argv[1].rsplit(".", 1)[0] + "-disamb.xml" | ||
| 52 | +print "saving in :", new_f | ||
| 53 | +open(new_f, 'w').write(codecs.encode(dom.toxml(), 'utf-8')) |