connl2009 output for kres

tools/parser/msd/msdmap.py

@@ -0,0 +1,60 @@
+import pkg_resources
+import pandas
+import sys
+
+# msd mappings from slo to ang
+class Msdmap():
+    def __init__(self):
+        # http://nl.ijs.si/ME/V4/msd/html/msd.categories-sl.html
+        self.pos_slo_ang = [
+            ("samostalnik", "S", "Noun", "N"),
+            ("glagol", "G", "Verb", "V"),
+            ("pridevnik", "P", "Adjective", "A"),
+            ("prislov", "R", "Adverb", "R"),
+            ("zaimek", "Z", "Pronoun", "P"),
+            ("števnik", "K", "Numeral", "M"),
+            ("predlog", "D", "Adposition", "S"),
+            ("veznik", "V", "Conjunction", "C"),
+            ("členek", "L", "Particle", "Q"),
+            ("medmet", "M", "Interjection", "I"),
+            ("okrajšava", "O", "Abbreviation", "Y"),
+            ("neuvrščeno", "N", "Residual", "X"),
+        ]
+
+        table_name = "msd-human-sl.tbl"
+        table_path = pkg_resources.resource_filename(__name__, "msd/" + table_name)
+        self.msd_table = pandas.read_csv(
+            table_name,
+            "\t",
+            names=["id", "eng_msd", "eng_long", "slo_msd", "slo_long", "slo_very_long"]
+        )
+
+    def slo_msd_to_eng_long(self, slo_msd):
+        return (self.msd_table[self.msd_table["slo_msd"] == slo_msd]["eng_long"]).values[0]
+
+    def pos_slo_ang_map(self, col, query):
+        for pos in self.pos_slo_ang:
+            if pos[col] == query:
+                return pos
+        raise ValueError("Wrong part of speech.")
+
+    def msd_from_slo(self, msd):
+        pos = self.pos_slo_ang_map(1, msd[0])
+        category = pos[2]
+        attr = [self.pos_val_map(category, 1, m)
+                for m in msd[1:] if m != "-"]
+        return (pos, attr)
+
+
+if __name__ == "__main__":
+    msdmap = Msdmap()
+    test_msds = [
+        "Soser",
+        "Ppnzmm",
+        "Gp-d-mz"
+    ]
+
+    for msd in test_msds:
+        print(msd)
+        print(msdmap.slo_msd_to_eng_long(msd))
+        print()