質問編集履歴
1
コードの記載を修正
test
CHANGED
|
File without changes
|
test
CHANGED
|
@@ -5,29 +5,54 @@
|
|
|
5
5
|
|
|
6
6
|
[https://hituji-ws.com/code/python/python-gcp-ocr/](https://hituji-ws.com/code/python/python-gcp-ocr/)
|
|
7
7
|
```
|
|
8
|
+
import requests
|
|
9
|
+
import base64
|
|
8
|
-
|
|
10
|
+
import json
|
|
9
|
-
|
|
11
|
+
API_KEY = ''
|
|
12
|
+
GOOGLE_CLOUD_VISION_API_URL = 'https://vision.googleapis.com/v1/images:annotate?key='
|
|
10
13
|
|
|
14
|
+
# TEXT_DETECTION:比較的短い文字
|
|
11
|
-
|
|
15
|
+
# DOCUMENT_TEXT_DETECTION:文章
|
|
16
|
+
DETECTION_TYPE = "DOCUMENT_TEXT_DETECTION"
|
|
12
17
|
|
|
18
|
+
def request_cloud_vison_api(image_base64, type="DOCUMENT_TEXT_DETECTION"):
|
|
19
|
+
""" http のリクエストでVisionAPIにアクセス """
|
|
20
|
+
api_url = GOOGLE_CLOUD_VISION_API_URL + API_KEY
|
|
21
|
+
req_body = json.dumps({
|
|
22
|
+
'requests': [{
|
|
23
|
+
'image': {
|
|
24
|
+
'content': image_base64.decode('utf-8')
|
|
25
|
+
},
|
|
26
|
+
'features': [{
|
|
27
|
+
'type': type,
|
|
28
|
+
'maxResults': 10,
|
|
29
|
+
}]
|
|
30
|
+
}]
|
|
31
|
+
})
|
|
13
|
-
|
|
32
|
+
res = requests.post(api_url, data=req_body)
|
|
33
|
+
return res.json()
|
|
14
34
|
|
|
15
|
-
p = Path(__file__).parent / filein
|
|
16
|
-
with p.open('rb') as image_file:
|
|
17
|
-
content = image_file.read()
|
|
18
35
|
|
|
36
|
+
def img_to_base64(filepath):
|
|
37
|
+
""" 画像データをエンコード """
|
|
38
|
+
with open(filepath, 'rb') as img:
|
|
39
|
+
img_byte = img.read()
|
|
19
|
-
|
|
40
|
+
return base64.b64encode(img_byte)
|
|
41
|
+
|
|
42
|
+
|
|
43
|
+
def render_doc_text(file_path):
|
|
44
|
+
|
|
45
|
+
result = request_cloud_vison_api(image_base64=img_to_base64(file_path),
|
|
46
|
+
type=DETECTION_TYPE)
|
|
20
47
|
|
|
21
48
|
data_list = []
|
|
22
|
-
response = client.document_text_detection(image=image)
|
|
23
|
-
|
|
49
|
+
# データの取得 textAnnotationsに座標とテキスト fullTextAnnotationにテキスト
|
|
24
|
-
|
|
50
|
+
result_list = result["responses"][0]["textAnnotations"]
|
|
25
|
-
|
|
51
|
+
for d in result_list:
|
|
26
|
-
for paragraph in block.paragraphs:
|
|
27
|
-
for word in paragraph.words:
|
|
28
|
-
|
|
52
|
+
data_list.append([d['boundingPoly']['vertices'], d['description']])
|
|
53
|
+
|
|
29
|
-
|
|
54
|
+
# 1つ目除外
|
|
30
|
-
|
|
55
|
+
data_list = data_list[1:len(data_list)]
|
|
31
56
|
return data_list
|
|
32
57
|
```
|
|
33
58
|
|