Added a code that generates lexical selection rules from given file w…

…ith dix-like formatted stems.

texts/lexical_selection_generate.py

@@ -0,0 +1,89 @@
+# lexical_selection_generate.py
+# Reads manually revieved dix elements
+# Creates lexical selection rules for them in a new file.
+
+from argparse import ArgumentParser
+from collections import defaultdict
+
+# importing element tree
+# under the alias of ET
+import xml.etree.ElementTree as ET
+
+
+def write_variants(file_name,side_list):
+    with open(file_name,'w') as out_file:
+        previous_el = side_list[0].split(" ")[0]
+        for el in side_list:
+            tmp = el.split(" ")[0]
+            if previous_el != tmp:
+                out_file.write("\n")
+                previous_el = tmp
+            out_file.write(el)
+            out_file.write("\n")
+    out_file.close()
+
+if __name__ == "__main__":
+    # START
+    print("------------------------------------------------------------------------------------------------------------------------------------")
+
+    # Parsing arguments:
+    arg_parser = ArgumentParser()
+
+    arg_parser.add_argument("--input_file", help="Path to the dix-like formatted file you want rules tobe generated from.", required=True)
+    args = arg_parser.parse_args()
+
+    file_in = args.input_file
+    print("Starting analysis with file:", file_in)
+
+    # file_in = "tmp.txt"
+    file_out = file_in + "_result.txt"
+
+    output_file = open(file_out,"w")
+
+    # Reading file line by line:
+    with open(file_in, "r") as input_file:
+        for line in input_file.readlines():
+            line = line.strip()
+
+            # Skipping blank line:
+            if len(line) == 0:
+                continue
+
+            # Converting line into a proper xml format:
+            line = "<e>" + line + "</e>"
+
+            # Reading line as an XML Tree:
+            tree = ET.fromstring(line)
+            l = tree.find("l")
+            l_word = l.text
+            l_tag = l.find("s").get("n")
+            r = tree.find("r")
+            r_word = r.text
+            r_tag = r.find("s").get("n")
+            # print(l_word, l_tag,r_word, r_tag)
+
+            # Making LRX format:
+
+            # Multiline option:
+            #   <rule weight ="1">  <!-- default translation -->
+            #     <match lemma="ақырын" tags="adj">
+            #       <select lemma="sekin" tags="adj"/>
+            #     </match>
+            #   </rule>
+
+            # or, a better alternative: 
+            # <rule weight ="1"> <match lemma="ақырын" tags="adj"> <select lemma="sekin" tags="adj"/> </match> </rule>
+
+            output_file.write("<!-- default translation for the word \"{}\"-->\n".format(l_word))
+            rule1 = "<match lemma=\"{}\" tags=\"{}\"><select lemma=\"{}\" tags=\"{}\"/></match></rule>\n".format(l_word,l_tag,r_word,r_tag)
+            rule2 = "<match lemma=\"{}\" tags=\"{}.*\"><select lemma=\"{}\" tags=\"{}.*\"/></match></rule>\n".format(l_word,l_tag,r_word,r_tag)
+            print("Rule: ", rule2)
+            output_file.write(rule1)
+            output_file.write(rule2)
+            output_file.write("\n")
+
+            # print(ET.tostring(r, encoding="unicode"))
+
+    input_file.close()
+    output_file.close()
+    print("Finished. Resulting rules saved in a file: ", file_out)