Replace Certain Unicode Characters for the input (#23)

* allow setting return_type in predict_page

* replace certain unicode characters

* Update the unicode-char-searching examples

* rename char file

* Add tests

* Add arguments for `replace_empty_unicode`

examples/find-empty-unicode-chars/unicode-char-search.ipynb

@@ -0,0 +1,216 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import pandas as pd \n",
+    "df = pd.read_csv(\"unicode-char-categories.csv\")\n",
+    "# From https://www.fileformat.info/info/unicode/category/index.htm"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from transformers import AutoTokenizer"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "tokenizer = AutoTokenizer.from_pretrained(\"allenai/ivila-row-layoutlm-finetuned-s2vl-v2\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "UNICODE_CATEGORIES_TO_REPLACE = [\"Cc\", \"Cf\", \"Co\", \"Cs\", \"Mn\", \"Zl\", \"Zp\", \"Zs\"]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def calculate_tokenization_len(row):    \n",
+    "    uni_code = r'\\u' + row['Character'][2:]\n",
+    "    s = uni_code.encode().decode('unicode_escape')\n",
+    "    return len(tokenizer(s, add_special_tokens=False)['input_ids'])"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "df['Tokenization Length'] = df.apply(calculate_tokenization_len, axis=1)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 7,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>count</th>\n",
+       "      <th>mean</th>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>Code</th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>Cc</th>\n",
+       "      <td>65</td>\n",
+       "      <td>0.000000</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>Cf</th>\n",
+       "      <td>163</td>\n",
+       "      <td>0.766871</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>Co</th>\n",
+       "      <td>6</td>\n",
+       "      <td>0.666667</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>Cs</th>\n",
+       "      <td>6</td>\n",
+       "      <td>0.000000</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>Mn</th>\n",
+       "      <td>1950</td>\n",
+       "      <td>0.549744</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>Zl</th>\n",
+       "      <td>1</td>\n",
+       "      <td>0.000000</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>Zp</th>\n",
+       "      <td>1</td>\n",
+       "      <td>0.000000</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>Zs</th>\n",
+       "      <td>17</td>\n",
+       "      <td>0.000000</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "      count      mean\n",
+       "Code                 \n",
+       "Cc       65  0.000000\n",
+       "Cf      163  0.766871\n",
+       "Co        6  0.666667\n",
+       "Cs        6  0.000000\n",
+       "Mn     1950  0.549744\n",
+       "Zl        1  0.000000\n",
+       "Zp        1  0.000000\n",
+       "Zs       17  0.000000"
+      ]
+     },
+     "execution_count": 7,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "df.groupby('Code')['Tokenization Length'].agg(['count', 'mean']).loc[UNICODE_CATEGORIES_TO_REPLACE]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 8,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "array(['Cc', 'Cf', 'Co', 'Cs', 'Mc', 'Mn', 'So', 'Zl', 'Zp', 'Zs'],\n",
+       "      dtype=object)"
+      ]
+     },
+     "execution_count": 8,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "df[df['Tokenization Length']==0].Code.unique()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 9,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "df[df['Tokenization Length']==0]['Character'].to_csv(\"zero-length-unicode-chars.txt\", index=None, header=None)"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.8.3"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 4
+}