Adding wikipedia cleanup

clean.sh

@@ -0,0 +1,6 @@
+# Remove strings in parentheses
+sed -i -e 's|([^)]*)||g' lv.txt
+sed -i -e 's|\[[^)]*\]||g' lv.txt
+
+# Delete lines with foreign characters
+sed -i -e '/[óéàðæíúłßáśęινςü]/d' lv.txt

tools/wikipedia/GetWikipedia.ipynb

@@ -27,13 +27,22 @@
     "from tqdm import tqdm\n",
     "\n",
     "DATE = \"20221120\"\n",
+    "CUTOFF_SECTIONS = ['Atsauces un piezīmes', 'Atsauces', 'Ārējās saites', 'Literatūra', 'Skatīt arī',\n",
+    "                   ' Atsauces un piezīmes', ' Atsauces', ' Ārējās saites', ' Literatūra', ' Skatīt arī']\n",
     "\n",
     "dataset = load_dataset('joelito/EU_Wikipedias', date=DATE, language=\"lv\", split='train')\n",
     "\n",
     "with open(f'wikipedia_{DATE}.txt', 'w') as file:\n",
-    "    for entry in tqdm(dataset): \n",
+    "    for entry in tqdm(dataset):\n",
+    "        # Will cut off reference sections\n",
+    "        cutoffs = [len(entry[\"text\"])]\n",
+    "        for section in CUTOFF_SECTIONS:\n",
+    "            if entry[\"text\"].find('\\n\\n' + section) != -1:\n",
+    "                cutoffs.append(entry[\"text\"].find('\\n\\n' + section));\n",
+    "            \n",
     "        file.write(f'{entry[\"title\"]}\\n\\n')\n",
-    "        file.write(f'{entry[\"text\"]}\\n\\n\\n')\n",
+    "        file.write(f'{entry[\"text\"][0:min(cutoffs)]}\\n\\n\\n')\n",
+    "    \n",
     "    file.close()\n",
     "\n",
     "print('Done')    "

tools/wikipedia/GetWikipedia.py

@@ -2,13 +2,22 @@ from datasets import load_dataset
 from tqdm import tqdm
 
 DATE = "20221120"
+CUTOFF_SECTIONS = ['Atsauces un piezīmes', 'Atsauces', 'Ārējās saites', 'Literatūra', 'Skatīt arī',
+                   ' Atsauces un piezīmes', ' Atsauces', ' Ārējās saites', ' Literatūra', ' Skatīt arī']
 
 dataset = load_dataset('joelito/EU_Wikipedias', date=DATE, language="lv", split='train')
 
 with open(f'wikipedia_{DATE}.txt', 'w') as file:
-    for entry in tqdm(dataset): 
+    for entry in tqdm(dataset):
+        # Will cut off reference sections
+        cutoffs = [len(entry["text"])]
+        for section in CUTOFF_SECTIONS:
+            if entry["text"].find('\n\n' + section) != -1:
+                cutoffs.append(entry["text"].find('\n\n' + section));
+            
         file.write(f'{entry["title"]}\n\n')
-        file.write(f'{entry["text"]}\n\n\n')
+        file.write(f'{entry["text"][0:min(cutoffs)]}\n\n\n')
+    
     file.close()
 
-print('Done')  
+print('Done')