Update Readme for Text-to-code codesearchnet
Browse files
README.md
CHANGED
|
@@ -1,3 +1,94 @@
|
|
| 1 |
---
|
| 2 |
license: mit
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
license: mit
|
| 3 |
+
tags:
|
| 4 |
+
- CodeSearchNet
|
| 5 |
+
- CodeXGLUE
|
| 6 |
+
size_categories:
|
| 7 |
+
- n<1K
|
| 8 |
---
|
| 9 |
+
|
| 10 |
+
### Dataset is imported from CodeXGLUE and pre-processed using their script.
|
| 11 |
+
|
| 12 |
+
# Where to find in Semeru:
|
| 13 |
+
The dataset can be found at /nfs/semeru/semeru_datasets/code_xglue/text-to-code/codesearchnet/python in Semeru
|
| 14 |
+
|
| 15 |
+
# CodeXGLUE -- Code Search (AdvTest)
|
| 16 |
+
|
| 17 |
+
## Task Definition
|
| 18 |
+
|
| 19 |
+
Given a natural language, the task is to search source code that matches the natural language. To test the generalization ability of a model, function names and variables in test sets are replaced by special tokens.
|
| 20 |
+
|
| 21 |
+
## Dataset
|
| 22 |
+
|
| 23 |
+
The dataset we use comes from [CodeSearchNet](https://arxiv.org/pdf/1909.09436.pdf) and we filter the dataset as the following:
|
| 24 |
+
|
| 25 |
+
- Remove examples that codes cannot be parsed into an abstract syntax tree.
|
| 26 |
+
- Remove examples that #tokens of documents is < 3 or >256
|
| 27 |
+
- Remove examples that documents contain special tokens (e.g. <img ...> or https:...)
|
| 28 |
+
- Remove examples that documents are not English.
|
| 29 |
+
|
| 30 |
+
Besides, to test the generalization ability of a model, function names and variables in test sets are replaced by special tokens.
|
| 31 |
+
|
| 32 |
+
|
| 33 |
+
### Data Format
|
| 34 |
+
|
| 35 |
+
After preprocessing dataset, you can obtain three .jsonl files, i.e. train.jsonl, valid.jsonl, test.jsonl
|
| 36 |
+
|
| 37 |
+
For each file, each line in the uncompressed file represents one function. One row is illustrated below.
|
| 38 |
+
|
| 39 |
+
- **repo:** the owner/repo
|
| 40 |
+
- **path:** the full path to the original file
|
| 41 |
+
- **func_name:** the function or method name
|
| 42 |
+
- **original_string:** the raw string before tokenization or parsing
|
| 43 |
+
- **language:** the programming language
|
| 44 |
+
- **code/function:** the part of the `original_string` that is code
|
| 45 |
+
- **code_tokens/function_tokens:** tokenized version of `code`
|
| 46 |
+
- **docstring:** the top-level comment or docstring, if it exists in the original string
|
| 47 |
+
- **docstring_tokens:** tokenized version of `docstring`
|
| 48 |
+
- **url:** the url for the example (identify natural language)
|
| 49 |
+
- **idx**: the index of code (identify code)
|
| 50 |
+
|
| 51 |
+
### Data Statistics
|
| 52 |
+
|
| 53 |
+
Data statistics of the dataset are shown in the below table:
|
| 54 |
+
|
| 55 |
+
| | #Examples |
|
| 56 |
+
| ----- | :-------: |
|
| 57 |
+
| Train | 251,820 |
|
| 58 |
+
| Dev | 9,604 |
|
| 59 |
+
| Test | 19,210 |
|
| 60 |
+
|
| 61 |
+
|
| 62 |
+
### Example
|
| 63 |
+
|
| 64 |
+
Given a text-code file evaluator/test.jsonl:
|
| 65 |
+
|
| 66 |
+
```json
|
| 67 |
+
{"url": "url0", "docstring": "doc0","function": "fun0", "idx": 10}
|
| 68 |
+
{"url": "url1", "docstring": "doc1","function": "fun1", "idx": 11}
|
| 69 |
+
{"url": "url2", "docstring": "doc2","function": "fun2", "idx": 12}
|
| 70 |
+
{"url": "url3", "docstring": "doc3","function": "fun3", "idx": 13}
|
| 71 |
+
{"url": "url4", "docstring": "doc4","function": "fun4", "idx": 14}
|
| 72 |
+
```
|
| 73 |
+
|
| 74 |
+
|
| 75 |
+
### Input Predictions
|
| 76 |
+
|
| 77 |
+
For each url for natural language, descending sort candidate codes and return their idx in order. For example:
|
| 78 |
+
|
| 79 |
+
```json
|
| 80 |
+
{"url": "url0", "answers": [10,11,12,13,14]}
|
| 81 |
+
{"url": "url1", "answers": [10,12,11,13,14]}
|
| 82 |
+
{"url": "url2", "answers": [13,11,12,10,14]}
|
| 83 |
+
{"url": "url3", "answers": [10,14,12,13,11]}
|
| 84 |
+
{"url": "url4", "answers": [10,11,12,13,14]}
|
| 85 |
+
```
|
| 86 |
+
|
| 87 |
+
|
| 88 |
+
## Reference
|
| 89 |
+
<pre><code>@article{husain2019codesearchnet,
|
| 90 |
+
title={Codesearchnet challenge: Evaluating the state of semantic code search},
|
| 91 |
+
author={Husain, Hamel and Wu, Ho-Hsiang and Gazit, Tiferet and Allamanis, Miltiadis and Brockschmidt, Marc},
|
| 92 |
+
journal={arXiv preprint arXiv:1909.09436},
|
| 93 |
+
year={2019}
|
| 94 |
+
}</code></pre>
|