pdf-anonymizer-cli 0.3.1__tar.gz → 0.3.2__tar.gz

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (14) hide show
  1. {pdf_anonymizer_cli-0.3.1 → pdf_anonymizer_cli-0.3.2}/PKG-INFO +44 -14
  2. pdf_anonymizer_cli-0.3.2/README.md +120 -0
  3. {pdf_anonymizer_cli-0.3.1 → pdf_anonymizer_cli-0.3.2}/pyproject.toml +1 -1
  4. {pdf_anonymizer_cli-0.3.1 → pdf_anonymizer_cli-0.3.2}/src/pdf_anonymizer_cli/cli.py +9 -10
  5. {pdf_anonymizer_cli-0.3.1 → pdf_anonymizer_cli-0.3.2}/src/pdf_anonymizer_cli.egg-info/PKG-INFO +44 -14
  6. pdf_anonymizer_cli-0.3.1/README.md +0 -90
  7. {pdf_anonymizer_cli-0.3.1 → pdf_anonymizer_cli-0.3.2}/setup.cfg +0 -0
  8. {pdf_anonymizer_cli-0.3.1 → pdf_anonymizer_cli-0.3.2}/src/pdf_anonymizer_cli/__init__.py +0 -0
  9. {pdf_anonymizer_cli-0.3.1 → pdf_anonymizer_cli-0.3.2}/src/pdf_anonymizer_cli/main.py +0 -0
  10. {pdf_anonymizer_cli-0.3.1 → pdf_anonymizer_cli-0.3.2}/src/pdf_anonymizer_cli.egg-info/SOURCES.txt +0 -0
  11. {pdf_anonymizer_cli-0.3.1 → pdf_anonymizer_cli-0.3.2}/src/pdf_anonymizer_cli.egg-info/dependency_links.txt +0 -0
  12. {pdf_anonymizer_cli-0.3.1 → pdf_anonymizer_cli-0.3.2}/src/pdf_anonymizer_cli.egg-info/entry_points.txt +0 -0
  13. {pdf_anonymizer_cli-0.3.1 → pdf_anonymizer_cli-0.3.2}/src/pdf_anonymizer_cli.egg-info/requires.txt +0 -0
  14. {pdf_anonymizer_cli-0.3.1 → pdf_anonymizer_cli-0.3.2}/src/pdf_anonymizer_cli.egg-info/top_level.txt +0 -0
{pdf_anonymizer_cli-0.3.1 → pdf_anonymizer_cli-0.3.2}/PKG-INFO RENAMED
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.4
2
2
  Name: pdf-anonymizer-cli
3
- Version: 0.3.1
3
+ Version: 0.3.2
4
4
  Summary: CLI for a tool to anonymize PDF, Markdown, and plain text files using LLMs.
5
5
  Author-email: Leonid Ganeline <leo.gan.57@gmail.com>
6
6
  License: MIT
@@ -15,30 +15,49 @@ Requires-Dist: pdf-anonymizer-core
15
15
 
16
16
  A command-line interface for anonymizing PDF, Markdown, and plain text files using LLMs.
17
17
 
18
+ - **High-Quality Anonymization**: Leverages LLMs to identify and replace Personally Identifiable Information (PII) with high accuracy.
19
+ - **Large File Support**: Consistently anonymizes large files (tested up to 1GB).
20
+ - **Multi-Provider & Cost-Effective**: Free to use with local [Ollama](https://ollama.com/) models. It also supports major providers like [OpenAI](https://openai.com/), [Anthropic](https://www.anthropic.com/), [Google](https://ai.google.com/), [Hugging Face](https://huggingface.co/), and [OpenRouter](https://openrouter.ai/).
21
+ - **Reversible**: Supports deanonymization to recover original data when needed.
22
+ - **Multi-Format**: Works with PDF, Markdown, and plain text files.
23
+
24
+
18
25
  ## Installation
19
26
 
20
- This project uses `uv` and is structured as a monorepo. The dependencies for the CLI and its core library are managed at the root of the project.
27
+ Install the CLI with your favorite package manager. To use a specific LLM provider, you must install the corresponding extra.
28
+
29
+ - **Google**: `pip install "pdf-anonymizer-cli[google]"`
30
+ - **Ollama**: `pip install "pdf-anonymizer-cli[ollama]"`
31
+ - **Hugging Face**: `pip install "pdf-anonymizer-cli[huggingface]"`
32
+ - **OpenRouter**: `pip install "pdf-anonymizer-cli[openrouter]"`
33
+ - **OpenAI**: `pip install "pdf-anonymizer-cli[openai]"`
34
+ - **Anthropic**: `pip install "pdf-anonymizer-cli[anthropic]"`
35
+
36
+ You can also install multiple extras at once:
37
+
38
+ ```bash
39
+ pip install "pdf-anonymizer-cli[google,openrouter]"
40
+ ```
41
+
42
+ This installs the `pdf-anonymizer` executable.
21
43
 
22
- 1. **Install `uv`**: Follow the [official installation instructions](https://astral.sh/docs/uv#installation).
23
- 2. **Install dependencies from the repository root**:
24
- ```bash
25
- # From the repository root
26
- uv sync
27
- ```
28
- This installs the `pdf-anonymizer` executable.
29
44
 
30
45
  ## Environment Variables
31
46
 
32
47
  The CLI will automatically load a `.env` file from the current directory or any parent directory. For consistency, it's recommended to place a single `.env` file at the root of the repository.
33
48
 
34
- - `GOOGLE_API_KEY`: Required when using Google's Gemini models.
35
- - `OLLAMA_HOST`: Optional, defaults to `http://localhost:11434` when using local Ollama models.
49
+ - `GOOGLE_API_KEY`: Required when using Google models.
36
50
  - `HUGGING_FACE_TOKEN`: Required when using Hugging Face models. You can get a token from [here](https://huggingface.co/docs/hub/security-tokens).
51
+ - `OPENROUTER_API_KEY`: Required when using OpenRouter models.
52
+ - `OPENAI_API_KEY`: Required when using OpenAI models.
53
+ - `ANTHROPIC_API_KEY`: Required when using Anthropic models.
54
+ - `OLLAMA_HOST`: Optional, defaults to `http://localhost:11434` when using Ollama models.
37
55
 
38
56
  Example `.env` file:
39
57
  ```env
40
58
  GOOGLE_API_KEY="YOUR_API_KEY_HERE"
41
59
  HUGGING_FACE_TOKEN="YOUR_HF_TOKEN_HERE"
60
+ OPENROUTER_API_KEY="YOUR_OPENROUTER_KEY"
42
61
  ```
43
62
 
44
63
  ## Usage
@@ -65,20 +84,31 @@ pdf-anonymizer run FILE_PATH [FILE_PATH ...] \
65
84
  - `--anonymized-entities PATH`: Path to a file with a list of entities to anonymize.
66
85
 
67
86
  **Models**:
87
+ You can use any of the predefined models below, or specify a new model using the format `"provider/model-name"`.
88
+ For example: `--model-name "google/gemini-flash-latest"`.
89
+
68
90
  - **Google**: `gemini-2.5-pro`, `gemini-2.5-flash` (default), `gemini-2.5-flash-lite`.
69
91
  - **Ollama**: `gemma:7b`, `phi4-mini`.
70
92
  - **Hugging Face**: `openai/gpt-oss-20b`, `mistralai/Mistral-7B-Instruct-v0.1`, `HuggingFaceH4/zephyr-7b-beta`.
93
+ - **OpenRouter**: `openai/gpt-4o`, `google/gemini-pro`.
94
+ - **OpenAI**: `gpt-4o`, `gpt-5`.
95
+ - **Anthropic**: `claude-4-sonet`, `claude-4.5-sonet`.
71
96
 
72
97
  ### Examples
73
98
 
74
- **Basic anonymization**:
99
+ **Basic anonymization with the default model (Google)**:
75
100
  ```bash
76
101
  pdf-anonymizer run document.pdf
77
102
  ```
78
103
 
79
- **Custom model and prompt**:
104
+ **A new model (Google) and a simple prompt**:
105
+ ```bash
106
+ pdf-anonymizer run notes.md --model-name "google/gemini-flash-latest" --prompt-name simple
107
+ ```
108
+
109
+ **Using an OpenRouter model**:
80
110
  ```bash
81
- pdf-anonymizer run notes.md --model-name phi4-mini --prompt-name simple
111
+ pdf-anonymizer run report.pdf --model-name "openai/gpt-4o"
82
112
  ```
83
113
 
84
114
  ### Deanonymize
pdf_anonymizer_cli-0.3.2/README.md ADDED
@@ -0,0 +1,120 @@
1
+ # 🦉🫥 PDF Anonymizer CLI
2
+
3
+ A command-line interface for anonymizing PDF, Markdown, and plain text files using LLMs.
4
+
5
+ - **High-Quality Anonymization**: Leverages LLMs to identify and replace Personally Identifiable Information (PII) with high accuracy.
6
+ - **Large File Support**: Consistently anonymizes large files (tested up to 1GB).
7
+ - **Multi-Provider & Cost-Effective**: Free to use with local [Ollama](https://ollama.com/) models. It also supports major providers like [OpenAI](https://openai.com/), [Anthropic](https://www.anthropic.com/), [Google](https://ai.google.com/), [Hugging Face](https://huggingface.co/), and [OpenRouter](https://openrouter.ai/).
8
+ - **Reversible**: Supports deanonymization to recover original data when needed.
9
+ - **Multi-Format**: Works with PDF, Markdown, and plain text files.
10
+
11
+
12
+ ## Installation
13
+
14
+ Install the CLI with your favorite package manager. To use a specific LLM provider, you must install the corresponding extra.
15
+
16
+ - **Google**: `pip install "pdf-anonymizer-cli[google]"`
17
+ - **Ollama**: `pip install "pdf-anonymizer-cli[ollama]"`
18
+ - **Hugging Face**: `pip install "pdf-anonymizer-cli[huggingface]"`
19
+ - **OpenRouter**: `pip install "pdf-anonymizer-cli[openrouter]"`
20
+ - **OpenAI**: `pip install "pdf-anonymizer-cli[openai]"`
21
+ - **Anthropic**: `pip install "pdf-anonymizer-cli[anthropic]"`
22
+
23
+ You can also install multiple extras at once:
24
+
25
+ ```bash
26
+ pip install "pdf-anonymizer-cli[google,openrouter]"
27
+ ```
28
+
29
+ This installs the `pdf-anonymizer` executable.
30
+
31
+
32
+ ## Environment Variables
33
+
34
+ The CLI will automatically load a `.env` file from the current directory or any parent directory. For consistency, it's recommended to place a single `.env` file at the root of the repository.
35
+
36
+ - `GOOGLE_API_KEY`: Required when using Google models.
37
+ - `HUGGING_FACE_TOKEN`: Required when using Hugging Face models. You can get a token from [here](https://huggingface.co/docs/hub/security-tokens).
38
+ - `OPENROUTER_API_KEY`: Required when using OpenRouter models.
39
+ - `OPENAI_API_KEY`: Required when using OpenAI models.
40
+ - `ANTHROPIC_API_KEY`: Required when using Anthropic models.
41
+ - `OLLAMA_HOST`: Optional, defaults to `http://localhost:11434` when using Ollama models.
42
+
43
+ Example `.env` file:
44
+ ```env
45
+ GOOGLE_API_KEY="YOUR_API_KEY_HERE"
46
+ HUGGING_FACE_TOKEN="YOUR_HF_TOKEN_HERE"
47
+ OPENROUTER_API_KEY="YOUR_OPENROUTER_KEY"
48
+ ```
49
+
50
+ ## Usage
51
+
52
+ ### Anonymize
53
+
54
+ The `run` command anonymizes one or more files.
55
+
56
+ ```bash
57
+ pdf-anonymizer run FILE_PATH [FILE_PATH ...] \
58
+ [--characters-to-anonymize INTEGER] \
59
+ [--prompt-name {simple|detailed}] \
60
+ [--model-name TEXT] \
61
+ [--anonymized-entities PATH]
62
+ ```
63
+
64
+ **Arguments**:
65
+ - `FILE_PATH`: Path to one or several PDF, Markdown, or text files for anonymization.
66
+
67
+ **Options**:
68
+ - `--characters-to-anonymize INTEGER`: Number of characters to process in each chunk (default: `100000`).
69
+ - `--prompt-name [simple|detailed]`: The prompt template to use (default: `detailed`).
70
+ - `--model-name TEXT`: The language model to use.
71
+ - `--anonymized-entities PATH`: Path to a file with a list of entities to anonymize.
72
+
73
+ **Models**:
74
+ You can use any of the predefined models below, or specify a new model using the format `"provider/model-name"`.
75
+ For example: `--model-name "google/gemini-flash-latest"`.
76
+
77
+ - **Google**: `gemini-2.5-pro`, `gemini-2.5-flash` (default), `gemini-2.5-flash-lite`.
78
+ - **Ollama**: `gemma:7b`, `phi4-mini`.
79
+ - **Hugging Face**: `openai/gpt-oss-20b`, `mistralai/Mistral-7B-Instruct-v0.1`, `HuggingFaceH4/zephyr-7b-beta`.
80
+ - **OpenRouter**: `openai/gpt-4o`, `google/gemini-pro`.
81
+ - **OpenAI**: `gpt-4o`, `gpt-5`.
82
+ - **Anthropic**: `claude-4-sonet`, `claude-4.5-sonet`.
83
+
84
+ ### Examples
85
+
86
+ **Basic anonymization with the default model (Google)**:
87
+ ```bash
88
+ pdf-anonymizer run document.pdf
89
+ ```
90
+
91
+ **A new model (Google) and a simple prompt**:
92
+ ```bash
93
+ pdf-anonymizer run notes.md --model-name "google/gemini-flash-latest" --prompt-name simple
94
+ ```
95
+
96
+ **Using an OpenRouter model**:
97
+ ```bash
98
+ pdf-anonymizer run report.pdf --model-name "openai/gpt-4o"
99
+ ```
100
+
101
+ ### Deanonymize
102
+
103
+ The `deanonymize` command reverts anonymization using a mapping file.
104
+
105
+ ```bash
106
+ pdf-anonymizer deanonymize ANONYMIZED_FILE MAPPING_FILE
107
+ ```
108
+
109
+ **Arguments**:
110
+ - `ANONYMIZED_FILE`: Path to the anonymized text file.
111
+ - `MAPPING_FILE`: Path to the JSON mapping file.
112
+
113
+ **Example**:
114
+ ```bash
115
+ pdf-anonymizer deanonymize \
116
+ data/anonymized/document.anonymized.md \
117
+ data/mappings/document.mapping.json
118
+ ```
119
+
120
+ This will create a deanonymized version of the file at `data/deanonymized/document.deanonymized.md`.
{pdf_anonymizer_cli-0.3.1 → pdf_anonymizer_cli-0.3.2}/pyproject.toml RENAMED
@@ -1,6 +1,6 @@
1
1
  [project]
2
2
  name = "pdf-anonymizer-cli"
3
- version = "0.3.1"
3
+ version = "0.3.2"
4
4
  description = "CLI for a tool to anonymize PDF, Markdown, and plain text files using LLMs."
5
5
  authors = [{ name = "Leonid Ganeline", email = "leo.gan.57@gmail.com" }]
6
6
  license = { text = "MIT" }
{pdf_anonymizer_cli-0.3.1 → pdf_anonymizer_cli-0.3.2}/src/pdf_anonymizer_cli/cli.py RENAMED
@@ -10,10 +10,9 @@ from pdf_anonymizer_core.conf import (
10
10
  DEFAULT_CHARACTERS_TO_ANONYMIZE,
11
11
  DEFAULT_MODEL_NAME,
12
12
  DEFAULT_PROMPT_NAME,
13
- ModelName,
14
- ModelProvider,
15
13
  PromptEnum,
16
14
  get_enum_value,
15
+ get_provider_and_model_name,
17
16
  )
18
17
  from pdf_anonymizer_core.core import anonymize_file
19
18
  from pdf_anonymizer_core.prompts import detailed, simple
@@ -66,12 +65,11 @@ def run(
66
65
  ),
67
66
  ] = get_enum_value(PromptEnum, DEFAULT_PROMPT_NAME),
68
67
  model_name: Annotated[
69
- ModelName,
68
+ str,
70
69
  typer.Option(
71
- help="The name of the model to use for anonymization.",
72
- case_sensitive=False,
70
+ help="The name of the model to use for anonymization (supports enum values or 'provider/model').",
73
71
  ),
74
- ] = get_enum_value(ModelName, DEFAULT_MODEL_NAME),
72
+ ] = DEFAULT_MODEL_NAME,
75
73
  anonymized_entities: Annotated[
76
74
  Optional[Path],
77
75
  typer.Option(
@@ -98,8 +96,9 @@ def run(
98
96
  """
99
97
  load_environment()
100
98
 
101
- if model_name.provider == ModelProvider.GOOGLE:
102
- if "gemini" in model_name.value and not os.getenv("GOOGLE_API_KEY"):
99
+ provider_name, _ = get_provider_and_model_name(model_name)
100
+ if provider_name == "google":
101
+ if "gemini" in model_name and not os.getenv("GOOGLE_API_KEY"):
103
102
  logging.error(
104
103
  "Error: GOOGLE_API_KEY not found. Please set it in the .env file."
105
104
  )
@@ -107,7 +106,7 @@ def run(
107
106
 
108
107
  logging.info(f" --file-paths: {file_paths}")
109
108
  logging.info(f" --characters-to-anonymize: {characters_to_anonymize}")
110
- logging.info(f" --model-name: {model_name.value}")
109
+ logging.info(f" --model-name: {model_name}")
111
110
 
112
111
  # Select the appropriate prompt template
113
112
  prompt_templates: Dict[str, str] = {
@@ -132,7 +131,7 @@ def run(
132
131
  str(file_path),
133
132
  characters_to_anonymize,
134
133
  prompt_template,
135
- model_name.value,
134
+ model_name,
136
135
  entities_to_anonymize,
137
136
  )
138
137
 
{pdf_anonymizer_cli-0.3.1 → pdf_anonymizer_cli-0.3.2}/src/pdf_anonymizer_cli.egg-info/PKG-INFO RENAMED
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.4
2
2
  Name: pdf-anonymizer-cli
3
- Version: 0.3.1
3
+ Version: 0.3.2
4
4
  Summary: CLI for a tool to anonymize PDF, Markdown, and plain text files using LLMs.
5
5
  Author-email: Leonid Ganeline <leo.gan.57@gmail.com>
6
6
  License: MIT
@@ -15,30 +15,49 @@ Requires-Dist: pdf-anonymizer-core
15
15
 
16
16
  A command-line interface for anonymizing PDF, Markdown, and plain text files using LLMs.
17
17
 
18
+ - **High-Quality Anonymization**: Leverages LLMs to identify and replace Personally Identifiable Information (PII) with high accuracy.
19
+ - **Large File Support**: Consistently anonymizes large files (tested up to 1GB).
20
+ - **Multi-Provider & Cost-Effective**: Free to use with local [Ollama](https://ollama.com/) models. It also supports major providers like [OpenAI](https://openai.com/), [Anthropic](https://www.anthropic.com/), [Google](https://ai.google.com/), [Hugging Face](https://huggingface.co/), and [OpenRouter](https://openrouter.ai/).
21
+ - **Reversible**: Supports deanonymization to recover original data when needed.
22
+ - **Multi-Format**: Works with PDF, Markdown, and plain text files.
23
+
24
+
18
25
  ## Installation
19
26
 
20
- This project uses `uv` and is structured as a monorepo. The dependencies for the CLI and its core library are managed at the root of the project.
27
+ Install the CLI with your favorite package manager. To use a specific LLM provider, you must install the corresponding extra.
28
+
29
+ - **Google**: `pip install "pdf-anonymizer-cli[google]"`
30
+ - **Ollama**: `pip install "pdf-anonymizer-cli[ollama]"`
31
+ - **Hugging Face**: `pip install "pdf-anonymizer-cli[huggingface]"`
32
+ - **OpenRouter**: `pip install "pdf-anonymizer-cli[openrouter]"`
33
+ - **OpenAI**: `pip install "pdf-anonymizer-cli[openai]"`
34
+ - **Anthropic**: `pip install "pdf-anonymizer-cli[anthropic]"`
35
+
36
+ You can also install multiple extras at once:
37
+
38
+ ```bash
39
+ pip install "pdf-anonymizer-cli[google,openrouter]"
40
+ ```
41
+
42
+ This installs the `pdf-anonymizer` executable.
21
43
 
22
- 1. **Install `uv`**: Follow the [official installation instructions](https://astral.sh/docs/uv#installation).
23
- 2. **Install dependencies from the repository root**:
24
- ```bash
25
- # From the repository root
26
- uv sync
27
- ```
28
- This installs the `pdf-anonymizer` executable.
29
44
 
30
45
  ## Environment Variables
31
46
 
32
47
  The CLI will automatically load a `.env` file from the current directory or any parent directory. For consistency, it's recommended to place a single `.env` file at the root of the repository.
33
48
 
34
- - `GOOGLE_API_KEY`: Required when using Google's Gemini models.
35
- - `OLLAMA_HOST`: Optional, defaults to `http://localhost:11434` when using local Ollama models.
49
+ - `GOOGLE_API_KEY`: Required when using Google models.
36
50
  - `HUGGING_FACE_TOKEN`: Required when using Hugging Face models. You can get a token from [here](https://huggingface.co/docs/hub/security-tokens).
51
+ - `OPENROUTER_API_KEY`: Required when using OpenRouter models.
52
+ - `OPENAI_API_KEY`: Required when using OpenAI models.
53
+ - `ANTHROPIC_API_KEY`: Required when using Anthropic models.
54
+ - `OLLAMA_HOST`: Optional, defaults to `http://localhost:11434` when using Ollama models.
37
55
 
38
56
  Example `.env` file:
39
57
  ```env
40
58
  GOOGLE_API_KEY="YOUR_API_KEY_HERE"
41
59
  HUGGING_FACE_TOKEN="YOUR_HF_TOKEN_HERE"
60
+ OPENROUTER_API_KEY="YOUR_OPENROUTER_KEY"
42
61
  ```
43
62
 
44
63
  ## Usage
@@ -65,20 +84,31 @@ pdf-anonymizer run FILE_PATH [FILE_PATH ...] \
65
84
  - `--anonymized-entities PATH`: Path to a file with a list of entities to anonymize.
66
85
 
67
86
  **Models**:
87
+ You can use any of the predefined models below, or specify a new model using the format `"provider/model-name"`.
88
+ For example: `--model-name "google/gemini-flash-latest"`.
89
+
68
90
  - **Google**: `gemini-2.5-pro`, `gemini-2.5-flash` (default), `gemini-2.5-flash-lite`.
69
91
  - **Ollama**: `gemma:7b`, `phi4-mini`.
70
92
  - **Hugging Face**: `openai/gpt-oss-20b`, `mistralai/Mistral-7B-Instruct-v0.1`, `HuggingFaceH4/zephyr-7b-beta`.
93
+ - **OpenRouter**: `openai/gpt-4o`, `google/gemini-pro`.
94
+ - **OpenAI**: `gpt-4o`, `gpt-5`.
95
+ - **Anthropic**: `claude-4-sonet`, `claude-4.5-sonet`.
71
96
 
72
97
  ### Examples
73
98
 
74
- **Basic anonymization**:
99
+ **Basic anonymization with the default model (Google)**:
75
100
  ```bash
76
101
  pdf-anonymizer run document.pdf
77
102
  ```
78
103
 
79
- **Custom model and prompt**:
104
+ **A new model (Google) and a simple prompt**:
105
+ ```bash
106
+ pdf-anonymizer run notes.md --model-name "google/gemini-flash-latest" --prompt-name simple
107
+ ```
108
+
109
+ **Using an OpenRouter model**:
80
110
  ```bash
81
- pdf-anonymizer run notes.md --model-name phi4-mini --prompt-name simple
111
+ pdf-anonymizer run report.pdf --model-name "openai/gpt-4o"
82
112
  ```
83
113
 
84
114
  ### Deanonymize
pdf_anonymizer_cli-0.3.1/README.md DELETED
@@ -1,90 +0,0 @@
1
- # 🦉🫥 PDF Anonymizer CLI
2
-
3
- A command-line interface for anonymizing PDF, Markdown, and plain text files using LLMs.
4
-
5
- ## Installation
6
-
7
- This project uses `uv` and is structured as a monorepo. The dependencies for the CLI and its core library are managed at the root of the project.
8
-
9
- 1. **Install `uv`**: Follow the [official installation instructions](https://astral.sh/docs/uv#installation).
10
- 2. **Install dependencies from the repository root**:
11
- ```bash
12
- # From the repository root
13
- uv sync
14
- ```
15
- This installs the `pdf-anonymizer` executable.
16
-
17
- ## Environment Variables
18
-
19
- The CLI will automatically load a `.env` file from the current directory or any parent directory. For consistency, it's recommended to place a single `.env` file at the root of the repository.
20
-
21
- - `GOOGLE_API_KEY`: Required when using Google's Gemini models.
22
- - `OLLAMA_HOST`: Optional, defaults to `http://localhost:11434` when using local Ollama models.
23
- - `HUGGING_FACE_TOKEN`: Required when using Hugging Face models. You can get a token from [here](https://huggingface.co/docs/hub/security-tokens).
24
-
25
- Example `.env` file:
26
- ```env
27
- GOOGLE_API_KEY="YOUR_API_KEY_HERE"
28
- HUGGING_FACE_TOKEN="YOUR_HF_TOKEN_HERE"
29
- ```
30
-
31
- ## Usage
32
-
33
- ### Anonymize
34
-
35
- The `run` command anonymizes one or more files.
36
-
37
- ```bash
38
- pdf-anonymizer run FILE_PATH [FILE_PATH ...] \
39
- [--characters-to-anonymize INTEGER] \
40
- [--prompt-name {simple|detailed}] \
41
- [--model-name TEXT] \
42
- [--anonymized-entities PATH]
43
- ```
44
-
45
- **Arguments**:
46
- - `FILE_PATH`: Path to one or several PDF, Markdown, or text files for anonymization.
47
-
48
- **Options**:
49
- - `--characters-to-anonymize INTEGER`: Number of characters to process in each chunk (default: `100000`).
50
- - `--prompt-name [simple|detailed]`: The prompt template to use (default: `detailed`).
51
- - `--model-name TEXT`: The language model to use.
52
- - `--anonymized-entities PATH`: Path to a file with a list of entities to anonymize.
53
-
54
- **Models**:
55
- - **Google**: `gemini-2.5-pro`, `gemini-2.5-flash` (default), `gemini-2.5-flash-lite`.
56
- - **Ollama**: `gemma:7b`, `phi4-mini`.
57
- - **Hugging Face**: `openai/gpt-oss-20b`, `mistralai/Mistral-7B-Instruct-v0.1`, `HuggingFaceH4/zephyr-7b-beta`.
58
-
59
- ### Examples
60
-
61
- **Basic anonymization**:
62
- ```bash
63
- pdf-anonymizer run document.pdf
64
- ```
65
-
66
- **Custom model and prompt**:
67
- ```bash
68
- pdf-anonymizer run notes.md --model-name phi4-mini --prompt-name simple
69
- ```
70
-
71
- ### Deanonymize
72
-
73
- The `deanonymize` command reverts anonymization using a mapping file.
74
-
75
- ```bash
76
- pdf-anonymizer deanonymize ANONYMIZED_FILE MAPPING_FILE
77
- ```
78
-
79
- **Arguments**:
80
- - `ANONYMIZED_FILE`: Path to the anonymized text file.
81
- - `MAPPING_FILE`: Path to the JSON mapping file.
82
-
83
- **Example**:
84
- ```bash
85
- pdf-anonymizer deanonymize \
86
- data/anonymized/document.anonymized.md \
87
- data/mappings/document.mapping.json
88
- ```
89
-
90
- This will create a deanonymized version of the file at `data/deanonymized/document.deanonymized.md`.