fix/library-versions

#5
by Paulo - opened
Files changed (4) hide show
  1. CHANGES.md +10 -0
  2. app.py +19 -11
  3. config.json +1 -2
  4. requirements.txt +5 -9
CHANGES.md ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ # ChangeLog
2
+
3
+ ## v. 1.1
4
+ * updated library versions
5
+ * cache directory defined as subdirectory of home dir
6
+ * cache dir for Transformers set as environment variable, instead of
7
+ in config file
8
+
9
+ ## v. 1.0
10
+ * initial version
app.py CHANGED
@@ -1,8 +1,22 @@
1
- import gradio as gr
2
  import os
3
- from pii_transform.api.e2e import PiiTextProcessor
 
 
 
 
 
 
 
4
  from pii_extract.defs import FMT_CONFIG_PLUGIN
 
5
 
 
 
 
 
 
 
 
6
  examples = []
7
  with open("examples.txt", "r") as f:
8
  examples = f.readlines()
@@ -15,19 +29,14 @@ language_choices = {
15
  "German": "de",
16
  "French": "fr",
17
  }
 
18
  language_code = "en"
19
- cache_dir = "/home/user/app/cache"
20
- os.makedirs(cache_dir, exist_ok=True)
21
- if os.path.isdir(cache_dir):
22
- gr.Info("Cache directory created at "+cache_dir)
23
- else:
24
- gr.Warning("Cache directory creation error")
25
 
26
  policy_help_string = """
27
  Policies are defined as follows:
28
 
29
- 1. **Annotate** - replace the PII instance by a \<TYPE:VALUE\> string, i.e. include both the PII type and its value
30
- 2. **Redact** - all PII instances are replaced by a \<PII\> generic string
31
  3. **Placeholder** - replace with a prototypical value
32
  4. **Synthetic** - substitute with synthetic data
33
 
@@ -170,4 +179,3 @@ with gr.Blocks() as demo:
170
  with gr.Accordion(label="Help Panel", open=False):
171
  gr.Markdown(value=policy_help_string)
172
  demo.queue().launch()
173
-
 
 
1
  import os
2
+ from pathlib import Path
3
+
4
+ # Set the HF cache directory for the Transformers plugin
5
+ cache_dir = Path.home() / "app"/ "cache"
6
+ cache_dir.mkdir(parents=True, exist_ok=True)
7
+ os.environ["HUGGINGFACE_HUB_CACHE"] = str(cache_dir)
8
+
9
+ from pii_process.api import PiiTextProcessor
10
  from pii_extract.defs import FMT_CONFIG_PLUGIN
11
+ import gradio as gr
12
 
13
+ # Test if the cache directory exists
14
+ if cache_dir.is_dir():
15
+ gr.Info("Cache directory created at "+str(cache_dir))
16
+ else:
17
+ gr.Warning("Cache directory creation error")
18
+
19
+ # Read examples
20
  examples = []
21
  with open("examples.txt", "r") as f:
22
  examples = f.readlines()
 
29
  "German": "de",
30
  "French": "fr",
31
  }
32
+
33
  language_code = "en"
 
 
 
 
 
 
34
 
35
  policy_help_string = """
36
  Policies are defined as follows:
37
 
38
+ 1. **Annotate** - replace the PII instance by a <TYPE:VALUE> string, i.e. include both the PII type and its value
39
+ 2. **Redact** - all PII instances are replaced by a <PII> generic string
40
  3. **Placeholder** - replace with a prototypical value
41
  4. **Synthetic** - substitute with synthetic data
42
 
 
179
  with gr.Accordion(label="Help Panel", open=False):
180
  gr.Markdown(value=policy_help_string)
181
  demo.queue().launch()
 
config.json CHANGED
@@ -4,7 +4,6 @@
4
  {
5
  "format": "piisa:config:pii-extract-plg-transformers:main:v1",
6
  "task_config": {
7
- "cachedir": "/home/user/app/cache",
8
  "reuse_engine": true,
9
  "aggregation": "max",
10
  "models": [
@@ -74,4 +73,4 @@
74
  }
75
  }
76
  ]
77
- }
 
4
  {
5
  "format": "piisa:config:pii-extract-plg-transformers:main:v1",
6
  "task_config": {
 
7
  "reuse_engine": true,
8
  "aggregation": "max",
9
  "models": [
 
73
  }
74
  }
75
  ]
76
+ }
requirements.txt CHANGED
@@ -1,9 +1,5 @@
1
- pii-data==0.4.0
2
- pii-extract-base==0.5.0
3
- pii-extract-plg-presidio==0.3.2
4
- pii-extract-plg-regex==0.4.1
5
- pii-extract-plg-transformers==0.1.1
6
- pii-preprocess==0.0.4
7
- pii-transform==0.5.1
8
- torch==2.1.0
9
- transformers==4.34.0
 
1
+ wheel
2
+ pii-process[transformers] >= 0.1.1
3
+
4
+ torch>=2.2
5
+ transformers>=4.34.0