Added parameter to store removed prepositions. Fixed an issue with li…

…nebreaks.
mille-s · Feb 22, 2023 · 8223a6a · 8223a6a
1 parent bafe975
commit 8223a6a
Showing 1 changed file with 24 additions and 18 deletions.
diff --git a/UD_Converter_release.ipynb b/UD_Converter_release.ipynb
@@ -31,7 +31,7 @@
         "# Download resources for UD to PredArg conversion"
       ],
       "metadata": {
-        "id": "dHQBcKE6CCIC"
+        "id": "clb0W-W2e9x0"
       }
     },
     {
@@ -58,7 +58,7 @@
         "# Install Stanza for UD parsing"
       ],
       "metadata": {
-        "id": "3bvg9S78CJ7q"
+        "id": "no-bubGqfBMD"
       }
     },
     {
@@ -77,7 +77,7 @@
         "  os.makedirs(output_folder)"
       ],
       "metadata": {
-        "id": "VnFqJeByB7om"
+        "id": "2IBkOS3ue89k"
       },
       "execution_count": null,
       "outputs": []
@@ -88,7 +88,7 @@
         "#Run Stanza (upload text file first)"
       ],
       "metadata": {
-        "id": "BiEFSiDSPeX8"
+        "id": "BjzmcIp9fK0c"
       }
     },
     {
@@ -142,7 +142,7 @@
         "    #print('\\n')\n",
         "    for sent in doc.sentences:\n",
         "      #UD_sentence = ''\n",
-        "      UD_sentence = '# text = '+doc.text\n",
+        "      UD_sentence = '# text = '+doc.text+'\\n'\n",
         "      for word in sent.words:\n",
         "        columns = []\n",
         "        columns.append(str(word.id))          #(0)\n",
@@ -167,11 +167,11 @@
         "    with open(os.path.join(output_folder, language+'_'+filename+'_STZ.conllu'), \"w\", encoding=\"utf-8\") as f:\n",
         "      f.write(UD_text)\n",
         "\n",
-        "  copy_files(output_folder, '/content/UD_Converter/inputs')\n"
+        "  copy_files(output_folder, '/content/UD_Converter/inputs')"
       ],
       "metadata": {
         "cellView": "form",
-        "id": "xKy5hZEVCBJW"
+        "id": "NtespU2ufO7L"
       },
       "execution_count": null,
       "outputs": []
@@ -182,7 +182,7 @@
         "# Run conversion to Predicate-argument structures"
       ],
       "metadata": {
-        "id": "O3KjrNHdOpfh"
+        "id": "S-Lba9RkfSoz"
       }
     },
     {
@@ -227,6 +227,8 @@
         "strPerFile = '10000'\n",
         "# perform structure well-formedness and file alignment checks and create debug files ('yes'/'no')\n",
         "debug = 'yes'\n",
+        "# Keep or erase intermediate files produced by the different components\n",
+        "keep_intermediate_files = 'no'\n",
         "\n",
         "#============================================================================================================\n",
         "# CONVERSION PARAMETERS ('yes'/'no')\n",
@@ -265,6 +267,8 @@
         "parentheses = default\n",
         "# keep quotation marks in the deep structure (SRST: 'no')\n",
         "quotationMarks = default\n",
+        "# keep label of adpositions in the deep structure on the node of the word it was attached to in UD (SRST: 'no')\n",
+        "adposition = 'yes'\n",
         "\n",
         "#============================================================================================================\n",
         "# DO NOT EDIT BELOW\n",
@@ -365,7 +369,7 @@
         "  originalID = 'no'\n",
         "convertFolder = tmpIn\n",
         "path_conllu2conll = os.path.join(path_jars, 'conllu2conll.py')\n",
-        "!python {path_conllu2conll} {inputFormat} {convertFolder} {originalID} {originalForm} {originalXpos} {parentheses} {quotationMarks} {orderPunc} {orderConj} {orderMWE} {track} {dt} {sentOutTmp} {reduce_tree}\n",
+        "!python {path_conllu2conll} {inputFormat} {convertFolder} {originalID} {originalForm} {originalXpos} {parentheses} {quotationMarks} {orderPunc} {orderConj} {orderMWE} {track} {dt} {sentOutTmp} {reduce_tree} {adposition}\n",
         "# Code for offline usage; replace previous line by following:\n",
         "# subprocess.call(['python', 'conllu2conll.py', inputFormat, convertFolder, originalID, originalForm, originalXpos, parentheses, quotationMarks, orderPunc, orderConj, orderMWE, track, dt, sentOutTmp, reduce_tree])\n",
         "\n",
@@ -418,11 +422,12 @@
         "        for line in proc.stdout:\n",
         "          sys.stdout.write(line)\n",
         "          logfile.write(line)\n",
-        "      \n",
-        "try:\n",
-        "  shutil.rmtree(tmpIn)\n",
-        "except Exception as e:\n",
-        "  print(e)\n",
+        "\n",
+        "if keep_intermediate_files == 'no':      \n",
+        "  try:\n",
+        "    shutil.rmtree(tmpIn)\n",
+        "  except Exception as e:\n",
+        "    print(e)\n",
         "\n",
         "print('\\n==============================\\nConcatenating output files...\\n==============================')\n",
         "# File concatenation: the big files that had been split in smaller files are brought back together.\n",
@@ -458,10 +463,11 @@
         "    # Code for offline usage; replace previous line by following:\n",
         "    # subprocess.call(['python', 'concatenateFiles.py', sentOutTmp, sentOut, 'utf-8', 'utf-8', 'txt', 'sent', track, dt])\n",
         "\n",
-        "try:\n",
-        "  shutil.rmtree(tmpOut)\n",
-        "except Exception as e:\n",
-        "  print(e)\n",
+        "if keep_intermediate_files == 'no':     \n",
+        "  try:\n",
+        "    shutil.rmtree(tmpOut)\n",
+        "  except Exception as e:\n",
+        "    print(e)\n",
         "\n",
         "if debug == 'yes':\n",
         "  print('\\n==============================\\nChecking outputs...\\n==============================')\n",