Upload embedding & notebook
Browse files- instruction-text.ipynb +800 -0
- node_type_embedding.pth +3 -0
instruction-text.ipynb
ADDED
@@ -0,0 +1,800 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"cells": [
|
3 |
+
{
|
4 |
+
"cell_type": "code",
|
5 |
+
"execution_count": null,
|
6 |
+
"metadata": {},
|
7 |
+
"outputs": [],
|
8 |
+
"source": [
|
9 |
+
"from transformers import set_seed\n",
|
10 |
+
"import pandas as pd\n",
|
11 |
+
"import matplotlib.pyplot as plt\n",
|
12 |
+
"from collections import Counter\n",
|
13 |
+
"import numpy as np\n",
|
14 |
+
"import random\n",
|
15 |
+
"import torch"
|
16 |
+
]
|
17 |
+
},
|
18 |
+
{
|
19 |
+
"cell_type": "markdown",
|
20 |
+
"metadata": {},
|
21 |
+
"source": [
|
22 |
+
"three fields in each prompt: question, bot, task\n",
|
23 |
+
"\n",
|
24 |
+
"input to the model is:\n",
|
25 |
+
"```\n",
|
26 |
+
"<s>human\n",
|
27 |
+
"[question]\n",
|
28 |
+
"<s>bot\n",
|
29 |
+
"[bot]\n",
|
30 |
+
"```\n",
|
31 |
+
"where a question is\n",
|
32 |
+
"```\n",
|
33 |
+
"[program]\n",
|
34 |
+
"\n",
|
35 |
+
"[question]\n",
|
36 |
+
"```"
|
37 |
+
]
|
38 |
+
},
|
39 |
+
{
|
40 |
+
"cell_type": "code",
|
41 |
+
"execution_count": null,
|
42 |
+
"metadata": {},
|
43 |
+
"outputs": [],
|
44 |
+
"source": [
|
45 |
+
"debug = ''\n",
|
46 |
+
"in_dir = f\"/Users/zzy/Documents/graph{debug}\"\n",
|
47 |
+
"out_dir = f\"/Users/zzy/Documents/graph{debug}/instruction\"\n",
|
48 |
+
"no_return_sample_num = 20 if len(debug) > 0 else 40000\n",
|
49 |
+
"\n",
|
50 |
+
"figsize = (24, 16)\n",
|
51 |
+
"fontsize = 28\n",
|
52 |
+
"fontsize_tick = 16\n",
|
53 |
+
"\n",
|
54 |
+
"def filter_df(df, n=None):\n",
|
55 |
+
" try:\n",
|
56 |
+
" n = n if n is not None else no_return_sample_num\n",
|
57 |
+
" return pd.concat([df[df.source.apply(lambda x: 'return ' in x)], df[df.source.apply(lambda x: 'return ' not in x)].sample(n)]).reset_index(drop=True)\n",
|
58 |
+
" except:\n",
|
59 |
+
" return df\n",
|
60 |
+
"\n",
|
61 |
+
"def capitalize(s: str):\n",
|
62 |
+
" return s[0].upper() + s[1:]\n",
|
63 |
+
"\n",
|
64 |
+
"def replace_digit(s: str):\n",
|
65 |
+
" return s.replace('10', 'ten').replace('1', 'one').replace('2', 'two').replace('3', 'three').replace('4', 'four').replace('5', 'five').replace('6', 'six').replace('7', 'seven').replace('8', 'eight').replace('9', 'nine')\n",
|
66 |
+
"\n",
|
67 |
+
"def print_df(df, n=10):\n",
|
68 |
+
" for i in range(n):\n",
|
69 |
+
" print(df.loc[i].question)\n",
|
70 |
+
" print(df.loc[i].bot)\n",
|
71 |
+
" print('---'*10)\n",
|
72 |
+
"\n",
|
73 |
+
"graph_type_map = {'AST': 'abstract syntax tree', 'DFG': 'data flow graph'}\n",
|
74 |
+
"NODE_TYPES = [\n",
|
75 |
+
" 'assignment expression',\n",
|
76 |
+
" 'basic block',\n",
|
77 |
+
" 'binary expression',\n",
|
78 |
+
" 'break statement',\n",
|
79 |
+
" 'call expression',\n",
|
80 |
+
" 'catch clause',\n",
|
81 |
+
" 'class expression',\n",
|
82 |
+
" 'compile unit',\n",
|
83 |
+
" 'conditional expression',\n",
|
84 |
+
" 'continue statement',\n",
|
85 |
+
" 'export statement',\n",
|
86 |
+
" 'for statement',\n",
|
87 |
+
" 'function expression',\n",
|
88 |
+
" 'identifier expression', \n",
|
89 |
+
" 'if statement',\n",
|
90 |
+
" 'import expression',\n",
|
91 |
+
" 'key value parameter',\n",
|
92 |
+
" 'literal expression',\n",
|
93 |
+
" 'member access',\n",
|
94 |
+
" 'new expression',\n",
|
95 |
+
" 'new with type expression',\n",
|
96 |
+
" 'object expression',\n",
|
97 |
+
" 'object property',\n",
|
98 |
+
" 'parameter',\n",
|
99 |
+
" 'Python delete',\n",
|
100 |
+
" 'Python with',\n",
|
101 |
+
" 'Python with expression clause',\n",
|
102 |
+
" 'Python yield expression',\n",
|
103 |
+
" 'range statement',\n",
|
104 |
+
" 'return statement',\n",
|
105 |
+
" 'scope',\n",
|
106 |
+
" 'spread collection expression',\n",
|
107 |
+
" 'spread dictionary expression',\n",
|
108 |
+
" 'super expression',\n",
|
109 |
+
" 'switch case',\n",
|
110 |
+
" 'switch statement',\n",
|
111 |
+
" 'this expression',\n",
|
112 |
+
" 'throw statement',\n",
|
113 |
+
" 'try statement',\n",
|
114 |
+
" 'tuple expression',\n",
|
115 |
+
" 'unary expression',\n",
|
116 |
+
" 'variable declaration',\n",
|
117 |
+
" 'while statement'\n",
|
118 |
+
"]"
|
119 |
+
]
|
120 |
+
},
|
121 |
+
{
|
122 |
+
"cell_type": "code",
|
123 |
+
"execution_count": null,
|
124 |
+
"metadata": {},
|
125 |
+
"outputs": [],
|
126 |
+
"source": [
|
127 |
+
"# ph stands for place holder\n",
|
128 |
+
"ph1 = 'aohg981thgboir2bnjosi1839r8g9udnfv,mqwfo'\n",
|
129 |
+
"ph2 = 'io12i3ru9ginal90109ja-efi1-3gasd130gn0wa9'\n",
|
130 |
+
"ph3 = '2091rng09wegnb2p09jojmpzf,k[2e00-jmaa]'\n",
|
131 |
+
"ph4 = '0391gnea-g0-jr0aegbm[afk0-249jgps]waeg0'\n",
|
132 |
+
"ph5 = 'io1hngi0enriqgpgv]139gonpiamofj10onem;alf'\n",
|
133 |
+
"ph_list = [ph1, ph2, ph3, ph4, ph5]\n",
|
134 |
+
"punc_list = [\",\", \"?\", \".\", \";\", \"'s\"]\n",
|
135 |
+
"\n",
|
136 |
+
"def replace_place_holder(s, node_text, placeholder=\"{node}\"):\n",
|
137 |
+
" # this function injects a multi-line code snippet into the template\n",
|
138 |
+
"\n",
|
139 |
+
" if placeholder not in s:\n",
|
140 |
+
" return s\n",
|
141 |
+
"\n",
|
142 |
+
" # 1. remove the white spaces around {node} placeholder\n",
|
143 |
+
" s = s.replace(f\"{placeholder} \", f\"{placeholder}\").replace(f\" {placeholder}\", f\"{placeholder}\")\n",
|
144 |
+
" for punc in punc_list:\n",
|
145 |
+
" s = s.replace(f\"{placeholder}{punc} \", f\"{placeholder}{punc}\")\n",
|
146 |
+
" \n",
|
147 |
+
" # 2. injects the code, but first replace patterns like '\\n.' in both the code and template (the template may contain previously injected code)\n",
|
148 |
+
" for ph, punc in zip(ph_list, punc_list):\n",
|
149 |
+
" node_text = node_text.replace(f\"\\n{punc}\", ph)\n",
|
150 |
+
" s = s.replace(f\"\\n{punc}\", ph)\n",
|
151 |
+
" s = s.replace(placeholder, node_text)\n",
|
152 |
+
"\n",
|
153 |
+
" # 3. replace patterns like \"\\n.\" caused by the template\n",
|
154 |
+
" for punc in punc_list:\n",
|
155 |
+
" s = s.replace(f\"\\n{punc}\", f\"{punc}\\n\")\n",
|
156 |
+
"\n",
|
157 |
+
" # 4. replace the placerholders inserted in step 2\n",
|
158 |
+
" for ph, punc in zip(ph_list, punc_list):\n",
|
159 |
+
" s = s.replace(ph, f\"\\n{punc}\")\n",
|
160 |
+
" return s"
|
161 |
+
]
|
162 |
+
},
|
163 |
+
{
|
164 |
+
"cell_type": "markdown",
|
165 |
+
"metadata": {},
|
166 |
+
"source": [
|
167 |
+
"## node classification"
|
168 |
+
]
|
169 |
+
},
|
170 |
+
{
|
171 |
+
"cell_type": "code",
|
172 |
+
"execution_count": null,
|
173 |
+
"metadata": {},
|
174 |
+
"outputs": [],
|
175 |
+
"source": [
|
176 |
+
"questions = [\n",
|
177 |
+
" 'In the {graph} of this {lang} program, what is the type of this node: {node}.',\n",
|
178 |
+
" 'Tell me the node type of {node} in the {graph} of this {lang} program.',\n",
|
179 |
+
" 'What is the node type of {node} in the {graph} of this {lang} program',\n",
|
180 |
+
" \"In the {graph} of the provided {lang} program, could you identify the type of the node {node}?\",\n",
|
181 |
+
" \"What kind of node is {node} in the {graph} of this {lang} program?\",\n",
|
182 |
+
" \"Can you classify the node {node} in the {graph} of this {lang} program?\",\n",
|
183 |
+
" \"What category does the node {node} fall under in the {graph} of this {lang} program?\",\n",
|
184 |
+
" \"Regarding the {graph} of this {lang} program, what is the classification of the node {node}?\",\n",
|
185 |
+
" \"In the context of the {graph} of this {lang} program, what is the nature of the node identified as {node}?\",\n",
|
186 |
+
" \"Could you tell me what the node {node} represents in the {graph} of this {lang} program?\",\n",
|
187 |
+
" \"I'm curious, what type of node is {node} in the {graph} of the {lang} program presented?\",\n",
|
188 |
+
" \"What is the designation of the node {node} within the {graph} of this {lang} program?\",\n",
|
189 |
+
" \"Could you specify the node type for {node} in the {graph} of this particular {lang} program?\",\n",
|
190 |
+
" 'Determine the node type of {node} in the {graph} of this {lang} program.',\n",
|
191 |
+
"]\n",
|
192 |
+
"answers = [\n",
|
193 |
+
" \"This node, {node}, is classified as a {answer}.\",\n",
|
194 |
+
" \"The node {node} is a {answer}.\",\n",
|
195 |
+
" \"This node is identified as a {answer}.\",\n",
|
196 |
+
" \"It's a {answer}.\",\n",
|
197 |
+
" \"Regarding the node {node}, it falls under the category of a {answer}.\",\n",
|
198 |
+
" \"{node} is classified as a {answer} in the {graph} of this program.\",\n",
|
199 |
+
" \"The classification of the node {node} is a {answer}.\",\n",
|
200 |
+
" \"Within the {graph} of this program, {node} is a {answer} type of node.\",\n",
|
201 |
+
" \"As for the node identified as {node}, it's considered a {answer}.\",\n",
|
202 |
+
" \"The node {node} is of the {answer} variety.\",\n",
|
203 |
+
" 'The type of this node is {answer}.',\n",
|
204 |
+
" 'The given node is a {answer}.',\n",
|
205 |
+
" \"{answer}.\",\n",
|
206 |
+
" \"{answer}\",\n",
|
207 |
+
"]\n",
|
208 |
+
"print(len(questions), len(set(questions)))\n",
|
209 |
+
"print(len(answers), len(set(answers)))\n",
|
210 |
+
"assert all(a.count('{answer}') == 1 for a in answers)\n",
|
211 |
+
"assert all(a.count('{graph}') == 1 for a in questions)\n",
|
212 |
+
"assert all(a.count('{lang}') == 1 for a in questions)\n",
|
213 |
+
"\n",
|
214 |
+
"bots = [a for a in answers]\n",
|
215 |
+
"prompts = [[q, b] for q in questions for b in bots]\n",
|
216 |
+
"print(len(prompts))"
|
217 |
+
]
|
218 |
+
},
|
219 |
+
{
|
220 |
+
"cell_type": "code",
|
221 |
+
"execution_count": null,
|
222 |
+
"metadata": {},
|
223 |
+
"outputs": [],
|
224 |
+
"source": [
|
225 |
+
"set_seed(0)\n",
|
226 |
+
"results = {}\n",
|
227 |
+
"for lang in ['Java', 'Python']:\n",
|
228 |
+
" for graph in ['DFG', 'AST']:\n",
|
229 |
+
" result = []\n",
|
230 |
+
" df = pd.read_json(f\"{in_dir}/{lang}_{graph}.jsonl\", lines=True)\n",
|
231 |
+
" if lang == 'Python':\n",
|
232 |
+
" df = filter_df(df)\n",
|
233 |
+
"\n",
|
234 |
+
" question, bot = [], []\n",
|
235 |
+
" for i in range(len(df)):\n",
|
236 |
+
"\n",
|
237 |
+
" # filter out nodes that occur multiple times in the source code\n",
|
238 |
+
" node_texts = df.loc[i]['text']\n",
|
239 |
+
" node_ids = df.loc[i]['node_ids']\n",
|
240 |
+
" source = df.loc[i]['source']\n",
|
241 |
+
" assert len(node_ids) == len(node_texts)\n",
|
242 |
+
"\n",
|
243 |
+
" occurrences = np.array([source.count(t) for t in node_texts])\n",
|
244 |
+
" nodes_single_occurrence = np.where(occurrences == 1)[0]\n",
|
245 |
+
"\n",
|
246 |
+
" # we sample 1 node from each program\n",
|
247 |
+
" nodes = np.random.choice(nodes_single_occurrence, 1)\n",
|
248 |
+
" for node in nodes:\n",
|
249 |
+
" node_text = node_texts[node]\n",
|
250 |
+
" node_id = node_ids[node]\n",
|
251 |
+
" node_type = NODE_TYPES[node_id]\n",
|
252 |
+
"\n",
|
253 |
+
" p = random.sample(prompts, 1)[0]\n",
|
254 |
+
" p = [s.replace('{graph}', f\"{graph_type_map[graph]}\").replace('{lang}', f\"{lang}\") for s in p]\n",
|
255 |
+
" \n",
|
256 |
+
" response = p[1]\n",
|
257 |
+
" # deal with answer first in the response before plugging in the node text to avoid replacing something in the code\n",
|
258 |
+
" if any(node_type.startswith(l) for l in 'aeio'):\n",
|
259 |
+
" response = response.replace(' a ', ' an ')\n",
|
260 |
+
" response = response.replace('{answer}', node_type)\n",
|
261 |
+
" response = capitalize(response)\n",
|
262 |
+
" \n",
|
263 |
+
" if '\\n' in node_text:\n",
|
264 |
+
" node_text = f\"\\n```\\n{node_text}\\n```\\n\"\n",
|
265 |
+
" assert ph1 not in node_text and ph2 not in node_text and ph3 not in node_text and ph4 not in node_text and ph5 not in node_text\n",
|
266 |
+
" q = replace_place_holder(p[0], node_text)\n",
|
267 |
+
" response = replace_place_holder(response, node_text)\n",
|
268 |
+
" else:\n",
|
269 |
+
" node_text = f\"`{node_text}`\"\n",
|
270 |
+
" q = p[0].replace('{node}', node_text)\n",
|
271 |
+
" response = response.replace('{node}', node_text)\n",
|
272 |
+
"\n",
|
273 |
+
" source = source.strip('\\n')\n",
|
274 |
+
" q = f\"```\\n{source}\\n```\\n\\n{q}\"\n",
|
275 |
+
" question.append(q)\n",
|
276 |
+
" bot.append(response)\n",
|
277 |
+
" result.append(node_id)\n",
|
278 |
+
"\n",
|
279 |
+
" df['question'] = question\n",
|
280 |
+
" df['bot'] = bot\n",
|
281 |
+
" df = df.drop(columns={'text', 'source', 'node_ids', 'edge_index'})\n",
|
282 |
+
" df.to_json(f\"{out_dir}/Node_Classification_{lang}_{graph}.jsonl\", orient='records', lines=True)\n",
|
283 |
+
" results[f\"{lang}-{graph}\"] = result"
|
284 |
+
]
|
285 |
+
},
|
286 |
+
{
|
287 |
+
"cell_type": "markdown",
|
288 |
+
"metadata": {},
|
289 |
+
"source": [
|
290 |
+
"## parent node"
|
291 |
+
]
|
292 |
+
},
|
293 |
+
{
|
294 |
+
"cell_type": "code",
|
295 |
+
"execution_count": null,
|
296 |
+
"metadata": {},
|
297 |
+
"outputs": [],
|
298 |
+
"source": [
|
299 |
+
"questions = [\n",
|
300 |
+
" 'In the {graph} of this {lang} program, what is the parent node of this {node_type}: {node}.',\n",
|
301 |
+
" 'What is the parent of {node_type} {node} in the {graph} of this {lang} program?',\n",
|
302 |
+
" 'What is the parent node of {node_type} {node} in the {graph} of this {lang} program?',\n",
|
303 |
+
" 'Based on the {graph} of this {lang} program, identify the parent of {node_type} {node}.',\n",
|
304 |
+
" 'Based on the {graph} of this {lang} program, identify the parent of this {node_type}: {node}.',\n",
|
305 |
+
" 'Identify the parent of {node_type} {node} in the {graph} of this {lang} program.',\n",
|
306 |
+
" \"In the {graph} of the {lang} program presented, what is the predecessor of {node_type} {node}?\",\n",
|
307 |
+
" \"What node acts as the parent to {node_type} {node} in the {graph} of the displayed {lang} program?\",\n",
|
308 |
+
" \"Can you determine the parent node of {node_type} {node} in the {graph} of this {lang} program?\",\n",
|
309 |
+
" \"Which node is directly above {node_type} {node} in the hierarchy of the {graph} of the provided {lang} program?\",\n",
|
310 |
+
" \"Whose child is {node_type} {node} within the {graph} of this {lang} program?\",\n",
|
311 |
+
" \"What is the immediate ancestor of the {node_type} {node} in the {graph} of this {lang} program?\",\n",
|
312 |
+
" \"Regarding the {graph} of this {lang} program, can you point out the parent of {node_type} {node}?\",\n",
|
313 |
+
" \"In terms of graph theory, what is the parent of the {node_type} {node} in the {graph} of this {lang} program?\",\n",
|
314 |
+
" \"Who has the parental role for {node_type} {node} in the {graph}'s topology of this {lang} program?\",\n",
|
315 |
+
" \"For {node_type} {node} in the {graph} of the given {lang} program, which node supplies the incoming edge?\",\n",
|
316 |
+
"]\n",
|
317 |
+
"answers1 = [\n",
|
318 |
+
" \"In the {graph} of the given {lang} program, the parent of the given {node_type} is {parent}, which is a {parent_type}.\",\n",
|
319 |
+
" \"This {node_type}'s parent is the {parent_type} {parent}.\",\n",
|
320 |
+
" \"The given {node_type}'s parent in the {graph} of this {lang} program is the {parent_type} {parent}.\",\n",
|
321 |
+
" \"The parent of {node_type} {node} in the {graph} of this {lang} program is identified as {parent}, categorized as a {parent_type}.\",\n",
|
322 |
+
" \"In the structure of the {graph} of this {lang} program, {node_type} {node} finds its parent in node {parent}, which is a {parent_type}.\",\n",
|
323 |
+
" \"Node {parent}, a {parent_type}, serves as the parent to {node_type} {node} in the {graph} of this {lang} program.\",\n",
|
324 |
+
" \"As per the hierarchy in the {graph}, the {parent_type} node {parent} is the direct predecessor to {node_type} {node}.\",\n",
|
325 |
+
" \"Upon inspection, it is clear that the parent of {node_type} {node} is the {parent_type} {parent}.\",\n",
|
326 |
+
" \"The {node_type} {node} is immediately descended from {parent}, a {parent_type} in the {graph} of this {lang} program.\",\n",
|
327 |
+
" \"{node_type} {node}'s parental node is determined to be {parent}, which falls into the category of {parent_type}.\",\n",
|
328 |
+
" \"For {node_type} {node}, its lineage traces back to the {parent_type} node {parent} as its parent.\",\n",
|
329 |
+
" \"Within the nodal arrangement of the {graph}, {parent} is the progenitor to {node_type} {node}, having the classification of a {parent_type}.\",\n",
|
330 |
+
" \"Tracing the edges leads to confirming {parent}, a {parent_type}, as the parent of {node_type} {node}.\"\n",
|
331 |
+
"]\n",
|
332 |
+
"answers2 = [\n",
|
333 |
+
" 'This {node_type} has no parent in the {graph} of this {lang} program.',\n",
|
334 |
+
" 'This {node_type} has no parent in the {graph} of the given {lang} program.',\n",
|
335 |
+
" 'There is no edge pointing to this {node_type} in the {graph}. Therefore it does not have any parent.',\n",
|
336 |
+
" 'There is no edge pointing to this {node_type} in the {graph} of the given {lang} program. Therefore it does not have any parent.',\n",
|
337 |
+
" \"Within the confines of the {graph} of this {lang} program, {node_type} {node} does not have a parent node.\",\n",
|
338 |
+
" \"{node_type} {node} stands without a parent in the {graph}'s structure.\",\n",
|
339 |
+
" \"No parent node is associated with {node_type} {node} in the {graph} of the provided {lang} program.\",\n",
|
340 |
+
" \"A review of the code establishes that there is no preceding node to {node_type} {node} in the {graph}; it has no parent.\",\n",
|
341 |
+
" \"The {node_type} designated as {node} appears to lack a parental connection within the {graph} of this code.\",\n",
|
342 |
+
" \"In terms of graph topology, {node_type} {node} is an orphan node with no parent.\",\n",
|
343 |
+
" \"There is no edge incoming to {node_type} {node}, indicating the absence of a parent in the {graph} of this {lang} program.\",\n",
|
344 |
+
" \"After analyzing the code, it becomes evident that {node_type} {node} lacks a directly linked parent node in the {graph}.\",\n",
|
345 |
+
" \"The {graph} denotes that {node_type} {node} is disconnected from any parental lineage.\",\n",
|
346 |
+
" \"As depicted in the code, {node_type} {node} exists without a parent node to claim in the {graph}.\",\n",
|
347 |
+
"]\n",
|
348 |
+
"answers3 = [\n",
|
349 |
+
" \"There are multiple parents of this {node_type} in the {graph}:\\n\",\n",
|
350 |
+
" \"This {node_type} has more than one parent in the {graph}:\\n\"\n",
|
351 |
+
"]\n",
|
352 |
+
"print(len(questions), len(set(questions)))\n",
|
353 |
+
"print(len(answers1), len(set(answers1)))\n",
|
354 |
+
"print(len(answers2), len(set(answers2)))\n",
|
355 |
+
"print(len(answers3), len(set(answers3)))\n",
|
356 |
+
"assert all(a.count('{graph}') == 1 for a in questions)\n",
|
357 |
+
"assert all(a.count('{lang}') == 1 for a in questions)\n",
|
358 |
+
"\n",
|
359 |
+
"bots = [a for a in answers1]\n",
|
360 |
+
"bots_none = [a for a in answers2]\n",
|
361 |
+
"bots_multiple = [a for a in answers3]\n",
|
362 |
+
"prompts = [[q, b] for q in questions for b in bots]\n",
|
363 |
+
"print(len(prompts))"
|
364 |
+
]
|
365 |
+
},
|
366 |
+
{
|
367 |
+
"cell_type": "code",
|
368 |
+
"execution_count": null,
|
369 |
+
"metadata": {},
|
370 |
+
"outputs": [],
|
371 |
+
"source": [
|
372 |
+
"set_seed(1)\n",
|
373 |
+
"\n",
|
374 |
+
"results = {}\n",
|
375 |
+
"for lang in ['Java', 'Python']:\n",
|
376 |
+
" for graph in ['DFG', 'AST']:\n",
|
377 |
+
" result = []\n",
|
378 |
+
" df = pd.read_json(f\"{in_dir}/{lang}_{graph}.jsonl\", lines=True)\n",
|
379 |
+
" if lang == 'Python':\n",
|
380 |
+
" df = filter_df(df)\n",
|
381 |
+
" \n",
|
382 |
+
" question, bot = [], []\n",
|
383 |
+
" for i in range(len(df)):\n",
|
384 |
+
"\n",
|
385 |
+
" # filter out nodes that occur multiple times in the source code\n",
|
386 |
+
" node_texts = df.loc[i]['text']\n",
|
387 |
+
" node_ids = df.loc[i]['node_ids']\n",
|
388 |
+
" source = df.loc[i]['source']\n",
|
389 |
+
" edge_index = torch.tensor(df.loc[i]['edge_index'])\n",
|
390 |
+
" assert len(node_ids) == len(node_texts)\n",
|
391 |
+
"\n",
|
392 |
+
" occurrences = np.array([source.count(t) for t in node_texts])\n",
|
393 |
+
" nodes_single_occurrence = np.where(occurrences == 1)[0]\n",
|
394 |
+
" nodes_with_parents = [n for n in nodes_single_occurrence if n in edge_index[:, 1]]\n",
|
395 |
+
" nodes_without_parents = [n for n in nodes_single_occurrence if n not in edge_index[:, 1]]\n",
|
396 |
+
" \n",
|
397 |
+
" # we roughly maintain a balanced distribution\n",
|
398 |
+
" if random.random() < 0.75 and len(nodes_with_parents) > 0:\n",
|
399 |
+
" node = random.sample(nodes_with_parents, 1)[0]\n",
|
400 |
+
" elif len(nodes_without_parents) > 0:\n",
|
401 |
+
" node = random.sample(nodes_without_parents, 1)[0]\n",
|
402 |
+
" else:\n",
|
403 |
+
" node = np.random.choice(nodes_single_occurrence, 1)[0]\n",
|
404 |
+
" \n",
|
405 |
+
" node_text = node_texts[node]\n",
|
406 |
+
" node_id = node_ids[node]\n",
|
407 |
+
" node_type = NODE_TYPES[node_id]\n",
|
408 |
+
" edge_to_node = [edge for edge in edge_index if edge[1] == node]\n",
|
409 |
+
"\n",
|
410 |
+
" p = (random.sample(prompts, 1)[0] + random.sample(bots_none, 1) + random.sample(bots_multiple, 1)).copy()\n",
|
411 |
+
" assert p[2] in bots_none\n",
|
412 |
+
" p = [s.replace('{graph}', f\"{graph_type_map[graph]}\").replace('{lang}', f\"{lang}\") for s in p]\n",
|
413 |
+
"\n",
|
414 |
+
" # deal with answer first in the response before plugging in the node text to avoid replacing something in the code\n",
|
415 |
+
" num_parents = len(edge_to_node)\n",
|
416 |
+
" response = p[2] if num_parents == 0 else (p[1] if num_parents == 1 else p[3])\n",
|
417 |
+
" response = capitalize(response.replace('{node_type}', node_type))\n",
|
418 |
+
"\n",
|
419 |
+
" if num_parents > 1:\n",
|
420 |
+
" # no problem here\n",
|
421 |
+
" for j in range(num_parents):\n",
|
422 |
+
" parent_node = edge_to_node[j][0]\n",
|
423 |
+
" parent_text = node_texts[parent_node]\n",
|
424 |
+
" parent_id = node_ids[parent_node]\n",
|
425 |
+
" parent_type = NODE_TYPES[parent_id]\n",
|
426 |
+
" if '\\n' in parent_text:\n",
|
427 |
+
" parent_text = f\"\\n```\\n{parent_text}\\n```\\n\"\n",
|
428 |
+
" response += f\"{parent_type}:{parent_text}\"\n",
|
429 |
+
" else:\n",
|
430 |
+
" parent_text = f\"`{parent_text}`\"\n",
|
431 |
+
" response += f\"{parent_type}: {parent_text}\\n\"\n",
|
432 |
+
" elif num_parents == 1:\n",
|
433 |
+
" parent_node = edge_to_node[0][0]\n",
|
434 |
+
" parent_text = node_texts[parent_node]\n",
|
435 |
+
" parent_id = node_ids[parent_node]\n",
|
436 |
+
" parent_type = NODE_TYPES[parent_id]\n",
|
437 |
+
" if any(parent_type.startswith(l) for l in 'aeio'):\n",
|
438 |
+
" response = response.replace(' a ', ' an ')\n",
|
439 |
+
" if '\\n' in parent_text:\n",
|
440 |
+
" parent_text = f\"\\n```\\n{parent_text}\\n```\\n\"\n",
|
441 |
+
" response = replace_place_holder(response, parent_text, \"{parent}\")\n",
|
442 |
+
" else:\n",
|
443 |
+
" parent_text = f\"`{parent_text}`\"\n",
|
444 |
+
" response = response.replace('{parent}', parent_text)\n",
|
445 |
+
" response = response.replace('{parent_type}', parent_type) \n",
|
446 |
+
" \n",
|
447 |
+
" # now deal with node text\n",
|
448 |
+
" assert response.count('{node}') <= 1\n",
|
449 |
+
" if '\\n' in node_text:\n",
|
450 |
+
" node_text = f\"\\n```\\n{node_text}\\n```\\n\"\n",
|
451 |
+
" # note that now response may contain \"\\n,\" patterns\n",
|
452 |
+
" q = replace_place_holder(p[0].replace('{node_type}', node_type), node_text)\n",
|
453 |
+
" response = replace_place_holder(response, node_text)\n",
|
454 |
+
" else:\n",
|
455 |
+
" node_text = f\"`{node_text}`\"\n",
|
456 |
+
" q = p[0].replace('{node_type}', node_type).replace('{node}', node_text)\n",
|
457 |
+
" response = response.replace('{node}', node_text)\n",
|
458 |
+
"\n",
|
459 |
+
" source = source.strip('\\n')\n",
|
460 |
+
" q = f\"```\\n{source}\\n```\\n\\n{q}\"\n",
|
461 |
+
" question.append(q)\n",
|
462 |
+
" bot.append(response)\n",
|
463 |
+
" result.append(num_parents)\n",
|
464 |
+
"\n",
|
465 |
+
" df['question'] = question\n",
|
466 |
+
" df['bot'] = bot\n",
|
467 |
+
" df = df.drop(columns={'text', 'source', 'node_ids', 'edge_index'})\n",
|
468 |
+
" df.to_json(f\"{out_dir}/Parent_{lang}_{graph}.jsonl\", orient='records', lines=True)\n",
|
469 |
+
" results[f\"{lang}-{graph}\"] = result"
|
470 |
+
]
|
471 |
+
},
|
472 |
+
{
|
473 |
+
"cell_type": "markdown",
|
474 |
+
"metadata": {},
|
475 |
+
"source": [
|
476 |
+
"## Children"
|
477 |
+
]
|
478 |
+
},
|
479 |
+
{
|
480 |
+
"cell_type": "code",
|
481 |
+
"execution_count": null,
|
482 |
+
"metadata": {},
|
483 |
+
"outputs": [],
|
484 |
+
"source": [
|
485 |
+
"questions = [\n",
|
486 |
+
" 'In the {graph} of this {lang} program, what are the children of this {node_type}: {node}.',\n",
|
487 |
+
" 'Identify all children of {node_type} {node} in the {graph} of this {lang} program.',\n",
|
488 |
+
" 'Find the child nodes of {node_type} {node} in the {graph} of this {lang} program.',\n",
|
489 |
+
" 'In the {graph} of this {lang} program, how many children does the {node_type} {node} have? What are they?',\n",
|
490 |
+
" \"How many children does {node_type} {node} have in the {graph} of this {lang} program? What are they?\",\n",
|
491 |
+
" 'Please find all children of {node_type} {node} in the {graph} of this {lang} program.',\n",
|
492 |
+
" 'Can you find all children of {node_type} {node} in the {graph} of this {lang} program?',\n",
|
493 |
+
" \"List all the descendant nodes of {node_type} {node} in the {graph} of this {lang} program.\",\n",
|
494 |
+
" \"What are the direct children of the {node_type} {node} in the {graph} of this {lang} program?\",\n",
|
495 |
+
" \"Can you enumerate the offspring of {node_type} {node} within the {graph} of this {lang} program?\",\n",
|
496 |
+
" \"Could you provide the list of child nodes attached to {node_type} {node} in the {graph} of this {lang} program?\",\n",
|
497 |
+
" \"Please identify the child nodes emanating from {node_type} {node} in the {graph} of this {lang} program.\",\n",
|
498 |
+
" \"Show me the child nodes of {node_type} {node} in the {graph} of this {lang} program.\",\n",
|
499 |
+
" \"What nodes are directly connected to {node_type} {node} as its children in the {graph} of this {lang} program?\",\n",
|
500 |
+
" \"I need to know all the child elements of {node_type} {node} in the {graph} of this {lang} program. Can you provide that?\",\n",
|
501 |
+
" \"Are there any nodes that directly derive from {node_type} {node} in the {graph} of this {lang} program?\",\n",
|
502 |
+
" \"Which nodes act as successors to the node tagged as {node_type} {node} in the {graph} of this {lang} program?\",\n",
|
503 |
+
" \"What are the adjacent nodes that are children of {node_type} {node} in the {graph} of this {lang} program?\",\n",
|
504 |
+
" \"Identify the nodes that are immediate successors of {node_type} {node} in the {graph} of this {lang} program.\",\n",
|
505 |
+
" \"Detail the nodes branching from {node_type} {node} in the {graph} of this {lang} program.\",\n",
|
506 |
+
" \"Reveal all nodes that are directly beneath {node_type} {node} in the topology of the {graph} of this {lang} program.\",\n",
|
507 |
+
"]\n",
|
508 |
+
"answers1 = [\n",
|
509 |
+
" \"The given {node_type} has {child_num} children in the {graph}, they are:\\n\",\n",
|
510 |
+
" \"This {node_type} has {child_num} children:\\n\",\n",
|
511 |
+
" \"{node_type} {node} has a total of {child_num} children in the {graph}, which are:\\n\",\n",
|
512 |
+
" \"There are {child_num} child nodes of {node_type} {node}, specifically:\\n\",\n",
|
513 |
+
" \"As for the children of {node_type} {node}, you will find {child_num} direct descendants:\\n\",\n",
|
514 |
+
" \"The count of {node_type} {node}'s children amounts to {child_num}. They include:\\n\",\n",
|
515 |
+
" \"Upon identification, {node_type} {node} appears to have {child_num} offspring, namely:\\n\",\n",
|
516 |
+
" \"{node_type} {node} is parent to the following {child_num} nodes:\\n\",\n",
|
517 |
+
" \"A list of the {child_num} children under {node_type} {node} is as follows:\\n\",\n",
|
518 |
+
" \"Directly under {node_type} {node}, there are {child_num} children listed as:\\n\",\n",
|
519 |
+
" \"{node_type} {node} holds the hierarchy over {child_num} child nodes, which are:\\n\",\n",
|
520 |
+
" \"{child_num} children spring from {node_type} {node}, which are given below:\\n\",\n",
|
521 |
+
"]\n",
|
522 |
+
"answers2 = [\n",
|
523 |
+
" \"This {node_type} does not have any child nodes in the {graph}.\",\n",
|
524 |
+
" \"This {node_type} does not have any children in the {graph}.\",\n",
|
525 |
+
" \"There are no children of this {node_type} in the {graph} of the given code.\",\n",
|
526 |
+
" \"The given {node_type} does not have any children in the {graph}.\",\n",
|
527 |
+
" \"After examining the code, it's determined that in the {graph} this {node_type} has no children.\",\n",
|
528 |
+
" \"{node_type} {node} stands alone with zero child nodes descending from it.\",\n",
|
529 |
+
" \"I've checked the {node_type} {node} and found it has no direct descendants.\",\n",
|
530 |
+
" \"There are no child nodes attached to {node_type} {node} in the {graph} of this program.\",\n",
|
531 |
+
" \"No descendants can be traced from this {node_type}.\",\n",
|
532 |
+
" \"The {node_type} {node} is devoid of child nodes within the {graph} of the code.\",\n",
|
533 |
+
" \"Upon inspection, no nodes emerge as children of {node_type} {node}.\",\n",
|
534 |
+
" \"{node_type} {node} exists without progeny in the hierarchical layout.\",\n",
|
535 |
+
" \"It appears {node_type} {node} has no children.\",\n",
|
536 |
+
"]\n",
|
537 |
+
"\n",
|
538 |
+
"print(len(questions), len(set(questions)))\n",
|
539 |
+
"print(len(answers1), len(set(answers1)))\n",
|
540 |
+
"print(len(answers2), len(set(answers2)))\n",
|
541 |
+
"assert all(a.count('{answer}') == 1 for a in answers)\n",
|
542 |
+
"assert all(a.count('{graph}') == 1 for a in questions)\n",
|
543 |
+
"assert all(a.count('{lang}') == 1 for a in questions)\n",
|
544 |
+
"\n",
|
545 |
+
"bots = [a for a in answers1]\n",
|
546 |
+
"bots_none = [a for a in answers2]\n",
|
547 |
+
"prompts = [[q, b] for q in questions for b in bots]\n",
|
548 |
+
"print(len(prompts))"
|
549 |
+
]
|
550 |
+
},
|
551 |
+
{
|
552 |
+
"cell_type": "code",
|
553 |
+
"execution_count": null,
|
554 |
+
"metadata": {},
|
555 |
+
"outputs": [],
|
556 |
+
"source": [
|
557 |
+
"set_seed(2)\n",
|
558 |
+
"\n",
|
559 |
+
"results = {}\n",
|
560 |
+
"for lang in ['Java', 'Python']:\n",
|
561 |
+
" for graph in ['DFG', 'AST']:\n",
|
562 |
+
" result = []\n",
|
563 |
+
" df = pd.read_json(f\"{in_dir}/{lang}_{graph}.jsonl\", lines=True)\n",
|
564 |
+
" if lang == 'Python':\n",
|
565 |
+
" df = filter_df(df)\n",
|
566 |
+
" \n",
|
567 |
+
" question, bot = [], []\n",
|
568 |
+
" selected_idx = []\n",
|
569 |
+
" for i in range(len(df)):\n",
|
570 |
+
"\n",
|
571 |
+
" # filter out nodes that occur multiple times in the source code\n",
|
572 |
+
" node_texts = df.loc[i]['text']\n",
|
573 |
+
" node_ids = df.loc[i]['node_ids']\n",
|
574 |
+
" source = df.loc[i]['source']\n",
|
575 |
+
" edge_index = torch.tensor(df.loc[i]['edge_index'])\n",
|
576 |
+
"\n",
|
577 |
+
" occurrences = np.array([source.count(t) for t in node_texts])\n",
|
578 |
+
" nodes_single_occurrence = np.where(occurrences == 1)[0]\n",
|
579 |
+
" nodes_with_children = [n for n in nodes_single_occurrence if n in edge_index[:, 0]]\n",
|
580 |
+
" nodes_without_children = [n for n in nodes_single_occurrence if n not in edge_index[:, 0]]\n",
|
581 |
+
" \n",
|
582 |
+
" # we roughly maintain a balanced distribution\n",
|
583 |
+
" if random.random() < 0.85 and len(nodes_with_children) > 0:\n",
|
584 |
+
" node = random.sample(nodes_with_children, 1)[0]\n",
|
585 |
+
" elif len(nodes_without_children) > 0:\n",
|
586 |
+
" node = random.sample(nodes_without_children, 1)[0]\n",
|
587 |
+
" else:\n",
|
588 |
+
" node = np.random.choice(nodes_single_occurrence, 1)[0]\n",
|
589 |
+
" \n",
|
590 |
+
" node_text = node_texts[node]\n",
|
591 |
+
" node_id = node_ids[node]\n",
|
592 |
+
" node_type = NODE_TYPES[node_id]\n",
|
593 |
+
" edge_from_node = [edge for edge in edge_index if edge[0] == node]\n",
|
594 |
+
"\n",
|
595 |
+
" p = (random.sample(prompts, 1)[0] + random.sample(bots_none, 1)).copy()\n",
|
596 |
+
" assert p[1] in bots\n",
|
597 |
+
" p = [s.replace('{graph}', f\"{graph_type_map[graph]}\").replace('{lang}', f\"{lang}\") for s in p]\n",
|
598 |
+
"\n",
|
599 |
+
" num_children = len(edge_from_node)\n",
|
600 |
+
" if num_children > 10:\n",
|
601 |
+
" continue\n",
|
602 |
+
" else:\n",
|
603 |
+
" selected_idx.append(i)\n",
|
604 |
+
" response = p[2] if num_children == 0 else p[1]\n",
|
605 |
+
" response = capitalize(response.replace('{node_type}', node_type))\n",
|
606 |
+
" if num_children == 1:\n",
|
607 |
+
" response = response.replace('{child_num}', \"1\").replace('children', 'child').replace('nodes', 'node').replace('they are', 'it is').replace(' are', ' is').replace('descendants', 'descendant').replace('They include', 'It is').replace(' spring ', ' springs ')\n",
|
608 |
+
" else:\n",
|
609 |
+
" response = response.replace('{child_num}', f\"{num_children}\")\n",
|
610 |
+
" \n",
|
611 |
+
" if '\\n' in node_text:\n",
|
612 |
+
" node_text = f\"\\n```\\n{node_text}\\n```\\n\"\n",
|
613 |
+
" q = replace_place_holder(p[0].replace('{node_type}', node_type), node_text)\n",
|
614 |
+
" response = replace_place_holder(response, node_text)\n",
|
615 |
+
" else:\n",
|
616 |
+
" node_text = f\"`{node_text}`\"\n",
|
617 |
+
" q = p[0].replace('{node_type}', node_type).replace('{node}', node_text)\n",
|
618 |
+
" response = response.replace('{node}', node_text)\n",
|
619 |
+
" \n",
|
620 |
+
" for j in range(num_children):\n",
|
621 |
+
" child_node = edge_from_node[j][1]\n",
|
622 |
+
" child_text = node_texts[child_node]\n",
|
623 |
+
" child_id = node_ids[child_node]\n",
|
624 |
+
" child_type = NODE_TYPES[child_id]\n",
|
625 |
+
"\n",
|
626 |
+
" if '\\n' in child_text:\n",
|
627 |
+
" child_text = f\"\\n```\\n{child_text}\\n```\\n\"\n",
|
628 |
+
" response += f\"{child_type}:{child_text}\"\n",
|
629 |
+
" else:\n",
|
630 |
+
" child_text = f\"`{child_text}`\"\n",
|
631 |
+
" response += f\"{child_type}: {child_text}\\n\"\n",
|
632 |
+
" if num_children != len(set((node_ids[e[1]], node_texts[e[1]]) for e in edge_from_node)):\n",
|
633 |
+
" response += \"Note that there are multiple children with the same node type and literal representation.\"\n",
|
634 |
+
"\n",
|
635 |
+
" source = source.strip('\\n')\n",
|
636 |
+
" q = f\"```\\n{source}\\n```\\n\\n{q}\"\n",
|
637 |
+
" question.append(q)\n",
|
638 |
+
" bot.append(response)\n",
|
639 |
+
" result.append(num_children)\n",
|
640 |
+
"\n",
|
641 |
+
" df = df.loc[selected_idx].reset_index(drop=True)\n",
|
642 |
+
" df['question'] = question\n",
|
643 |
+
" df['bot'] = bot\n",
|
644 |
+
" df = df.drop(columns={'text', 'source', 'node_ids', 'edge_index'})\n",
|
645 |
+
" df.to_json(f\"{out_dir}/Children_{lang}_{graph}.jsonl\", orient='records', lines=True)\n",
|
646 |
+
" results[f\"{lang}-{graph}\"] = result"
|
647 |
+
]
|
648 |
+
},
|
649 |
+
{
|
650 |
+
"cell_type": "markdown",
|
651 |
+
"metadata": {},
|
652 |
+
"source": [
|
653 |
+
"## Edge prediction"
|
654 |
+
]
|
655 |
+
},
|
656 |
+
{
|
657 |
+
"cell_type": "code",
|
658 |
+
"execution_count": null,
|
659 |
+
"metadata": {},
|
660 |
+
"outputs": [],
|
661 |
+
"source": [
|
662 |
+
"questions = [\n",
|
663 |
+
" \"In the {graph} of this {lang} program, is there {edge_or_link} from {node_type1} {node1} to {node_type2} {node2}?\",\n",
|
664 |
+
" 'In the {graph} of this {lang} program, is there {edge_or_link} pointing from {node_type1} {node1} to {node_type2} {node2}?',\n",
|
665 |
+
" \"Please tell me if there is {edge_or_link} pointing from {node_type1} {node1} to {node_type2} {node2} in the {graph} of this {lang} program.\",\n",
|
666 |
+
" 'Is there {edge_or_link} from {node_type1} {node1} to {node_type2} {node2} in the {graph} of this {lang} program?',\n",
|
667 |
+
" \"Does a connection exist from {node_type1} {node1} to {node_type2} {node2} in the {graph} of this {lang} program?\",\n",
|
668 |
+
" \"In the {graph} of this {lang} program, do we have an arrow leading from {node_type1} {node1} to {node_type2} {node2}?\",\n",
|
669 |
+
" \"Is it true that {node_type1} {node1} is a predecessor of {node_type2} {node2} in the {graph} of this {lang} program?\",\n",
|
670 |
+
"]\n",
|
671 |
+
"answers1 = [\n",
|
672 |
+
" \"Yes, that is the case.\",\n",
|
673 |
+
" \"Yes, there is {edge_or_link} from {node_type1} {node1} to {node_type2} {node2}.\",\n",
|
674 |
+
" \"Yes, there is {edge_or_link} from {node_type1} {node1} to {node_type2} {node2} in the {graph} of this code.\",\n",
|
675 |
+
" \"Yes, there is {edge_or_link} pointing from {node_type1} {node1} to {node_type2} {node2} in the {graph}.\",\n",
|
676 |
+
" \"Affirmative, there exists {edge_or_link} from {node_type1} {node1} to {node_type2} {node2}.\",\n",
|
677 |
+
" \"Yes, that is the case. {node1} is directly connected to {node2}.\",\n",
|
678 |
+
"]\n",
|
679 |
+
"answers2 = [\n",
|
680 |
+
" \"No, that is not the case.\",\n",
|
681 |
+
" \"No, {node_type1} {node1} is not linked to {node_type2} {node2} by any edge in the {graph} of the given code.\",\n",
|
682 |
+
" \"No, there is no {edge_or_link} from {node_type1} {node1} to {node_type2} {node2}.\",\n",
|
683 |
+
" \"No, such {edge_or_link} is absent from the {graph}.\",\n",
|
684 |
+
" \"The code does not show {node_type1} {node1} as a predecessor to {node_type2} {node2} in the {graph}.\",\n",
|
685 |
+
"]\n",
|
686 |
+
"\n",
|
687 |
+
"print(len(questions), len(set(questions)))\n",
|
688 |
+
"print(len(answers1), len(set(answers1)))\n",
|
689 |
+
"print(len(answers2), len(set(answers2)))\n",
|
690 |
+
"assert all(a.count('{graph}') == 1 for a in questions)\n",
|
691 |
+
"assert all(a.count('{lang}') == 1 for a in questions)\n",
|
692 |
+
"\n",
|
693 |
+
"bots = [a for a in answers1]\n",
|
694 |
+
"bots_none = [a for a in answers2]\n",
|
695 |
+
"prompts = [[q, b] for q in questions for b in bots]\n",
|
696 |
+
"print(len(prompts))"
|
697 |
+
]
|
698 |
+
},
|
699 |
+
{
|
700 |
+
"cell_type": "code",
|
701 |
+
"execution_count": null,
|
702 |
+
"metadata": {},
|
703 |
+
"outputs": [],
|
704 |
+
"source": [
|
705 |
+
"set_seed(3)\n",
|
706 |
+
"results = {}\n",
|
707 |
+
"for lang in ['Java', 'Python']:\n",
|
708 |
+
" for graph in ['DFG', 'AST']:\n",
|
709 |
+
" result = []\n",
|
710 |
+
" df = pd.read_json(f\"{in_dir}/{lang}_{graph}.jsonl\", lines=True)\n",
|
711 |
+
" if lang == 'Python':\n",
|
712 |
+
" df = filter_df(df)\n",
|
713 |
+
" \n",
|
714 |
+
" question, bot = [], []\n",
|
715 |
+
" for i in range(len(df)):\n",
|
716 |
+
"\n",
|
717 |
+
" # filter out nodes that occur multiple times in the source code\n",
|
718 |
+
" node_texts = df.loc[i]['text']\n",
|
719 |
+
" node_ids = df.loc[i]['node_ids']\n",
|
720 |
+
" source = df.loc[i]['source']\n",
|
721 |
+
" edge_index = df.loc[i]['edge_index']\n",
|
722 |
+
"\n",
|
723 |
+
" occurrences = np.array([source.count(t) for t in node_texts])\n",
|
724 |
+
" nodes_single_occurrence = np.where(occurrences == 1)[0]\n",
|
725 |
+
" edge_index_elligible = [e for e in edge_index if (e[0] in nodes_single_occurrence and e[1] in nodes_single_occurrence)]\n",
|
726 |
+
" if graph == 'AST' and random.random() > 0.1:\n",
|
727 |
+
" edge_index_elligible = [e for e in edge_index_elligible if (node_texts[e[1]] not in node_texts[e[0]])]\n",
|
728 |
+
" \n",
|
729 |
+
" # we make sure at least half the problems have positive answer\n",
|
730 |
+
" if random.random() < 0.75 and len(edge_index_elligible) > 0:\n",
|
731 |
+
" n1, n2 = random.sample(edge_index_elligible, 1)[0]\n",
|
732 |
+
" else:\n",
|
733 |
+
" n1, n2 = np.random.choice(nodes_single_occurrence, 2)\n",
|
734 |
+
" \n",
|
735 |
+
" n1_text, n2_text = node_texts[n1], node_texts[n2]\n",
|
736 |
+
" n1_type, n2_type = NODE_TYPES[node_ids[n1]], NODE_TYPES[node_ids[n2]]\n",
|
737 |
+
"\n",
|
738 |
+
" p = random.sample(prompts, 1)[0] + random.sample(bots_none, 1)\n",
|
739 |
+
" p = [s.replace('{graph}', f\"{graph_type_map[graph]}\").replace('{lang}', f\"{lang}\") for s in p]\n",
|
740 |
+
" edge_or_link = 'an edge' if random.random() < 0.5 else 'a link'\n",
|
741 |
+
" p = [s.replace('{edge_or_link}', edge_or_link) for s in p[:-1]] + [p[-1].replace('such {edge_or_link}', f\"such {edge_or_link}\").replace('{edge_or_link}', edge_or_link.split()[-1])]\n",
|
742 |
+
" \n",
|
743 |
+
" q, b = '', ''\n",
|
744 |
+
" b = p[1] if [n1, n2] in edge_index else p[2]\n",
|
745 |
+
" b = b.replace('{node_type1}', n1_type).replace('{node_type2}', n2_type)\n",
|
746 |
+
"\n",
|
747 |
+
" if '\\n' in n1_text:\n",
|
748 |
+
" n1_text = f\"\\n```\\n{n1_text}\\n```\\n\"\n",
|
749 |
+
" q = replace_place_holder(p[0].replace('{node_type1}', n1_type).replace('{node_type2}', n2_type), n1_text, \"{node1}\")\n",
|
750 |
+
" b = replace_place_holder(b, n1_text, \"{node1}\")\n",
|
751 |
+
" else:\n",
|
752 |
+
" n1_text = f\"`{n1_text}`\"\n",
|
753 |
+
" q = p[0].replace('{node_type1}', n1_type).replace('{node_type2}', n2_type).replace('{node1}', n1_text)\n",
|
754 |
+
" b = b.replace('{node1}', n1_text)\n",
|
755 |
+
" \n",
|
756 |
+
" if '\\n' in n2_text:\n",
|
757 |
+
" n2_text = f\"\\n```\\n{n2_text}\\n```\\n\"\n",
|
758 |
+
" q = replace_place_holder(q, n2_text, \"{node2}\")\n",
|
759 |
+
" b = replace_place_holder(b, n2_text, \"{node2}\")\n",
|
760 |
+
" else:\n",
|
761 |
+
" n2_text = f\"`{n2_text}`\"\n",
|
762 |
+
" q = q.replace('{node2}', n2_text)\n",
|
763 |
+
" b = b.replace('{node2}', n2_text)\n",
|
764 |
+
" \n",
|
765 |
+
" source = source.strip('\\n')\n",
|
766 |
+
" q = f\"```\\n{source}\\n```\\n\\n{q}\"\n",
|
767 |
+
" question.append(q)\n",
|
768 |
+
" bot.append(b)\n",
|
769 |
+
" result.append(int([n1, n2] in edge_index))\n",
|
770 |
+
"\n",
|
771 |
+
" df['question'] = question\n",
|
772 |
+
" df['bot'] = bot\n",
|
773 |
+
" df = df.drop(columns={'text', 'source', 'node_ids', 'edge_index'})\n",
|
774 |
+
" df.to_json(f\"{out_dir}/Edge_Prediction_{lang}_{graph}.jsonl\", orient='records', lines=True)\n",
|
775 |
+
" results[f\"{lang}-{graph}\"] = result"
|
776 |
+
]
|
777 |
+
}
|
778 |
+
],
|
779 |
+
"metadata": {
|
780 |
+
"kernelspec": {
|
781 |
+
"display_name": "py39",
|
782 |
+
"language": "python",
|
783 |
+
"name": "python3"
|
784 |
+
},
|
785 |
+
"language_info": {
|
786 |
+
"codemirror_mode": {
|
787 |
+
"name": "ipython",
|
788 |
+
"version": 3
|
789 |
+
},
|
790 |
+
"file_extension": ".py",
|
791 |
+
"mimetype": "text/x-python",
|
792 |
+
"name": "python",
|
793 |
+
"nbconvert_exporter": "python",
|
794 |
+
"pygments_lexer": "ipython3",
|
795 |
+
"version": "3.9.17"
|
796 |
+
}
|
797 |
+
},
|
798 |
+
"nbformat": 4,
|
799 |
+
"nbformat_minor": 2
|
800 |
+
}
|
node_type_embedding.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:dafda5e624bc797d04774bf727448973daf5fa93f17478ba41892d5692d6e2e4
|
3 |
+
size 44815
|