rasgapdf 0.1.0__tar.gz → 0.1.2__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 (13) hide show
  1. rasgapdf-0.1.2/PKG-INFO +157 -0
  2. rasgapdf-0.1.2/README.md +141 -0
  3. {rasgapdf-0.1.0 → rasgapdf-0.1.2}/pyproject.toml +4 -3
  4. rasgapdf-0.1.2/rasgapdf.egg-info/PKG-INFO +157 -0
  5. rasgapdf-0.1.0/PKG-INFO +0 -203
  6. rasgapdf-0.1.0/README.md +0 -188
  7. rasgapdf-0.1.0/rasgapdf.egg-info/PKG-INFO +0 -203
  8. {rasgapdf-0.1.0 → rasgapdf-0.1.2}/rasgapdf/__init__.py +0 -0
  9. {rasgapdf-0.1.0 → rasgapdf-0.1.2}/rasgapdf/librasgapdf.so +0 -0
  10. {rasgapdf-0.1.0 → rasgapdf-0.1.2}/rasgapdf.egg-info/SOURCES.txt +0 -0
  11. {rasgapdf-0.1.0 → rasgapdf-0.1.2}/rasgapdf.egg-info/dependency_links.txt +0 -0
  12. {rasgapdf-0.1.0 → rasgapdf-0.1.2}/rasgapdf.egg-info/top_level.txt +0 -0
  13. {rasgapdf-0.1.0 → rasgapdf-0.1.2}/setup.cfg +0 -0
rasgapdf-0.1.2/PKG-INFO ADDED
@@ -0,0 +1,157 @@
1
+ Metadata-Version: 2.4
2
+ Name: rasgapdf
3
+ Version: 0.1.2
4
+ Summary: Ultra-fast 2D spatial PDF parser powered by Rust
5
+ Author: Messias Martins
6
+ Project-URL: Homepage, https://huggingface.co/martinsmessias/rasgapdf
7
+ Project-URL: Repository, https://huggingface.co/martinsmessias/rasgapdf
8
+ Project-URL: Discussions, https://huggingface.co/martinsmessias/rasgapdf/discussions
9
+ Classifier: Programming Language :: Python :: 3
10
+ Classifier: Programming Language :: Rust
11
+ Classifier: License :: OSI Approved :: MIT License
12
+ Classifier: Operating System :: POSIX :: Linux
13
+ Classifier: Topic :: Text Processing :: General
14
+ Requires-Python: >=3.8
15
+ Description-Content-Type: text/markdown
16
+
17
+ # ⚡ RasgaPDF
18
+
19
+ <div align="center">
20
+
21
+ **The Blazing-Fast 2D Spatial PDF Parser for Python.**
22
+ *Extract exact physical coordinates `[x0, y0, x1, y1]`, word spans, text blocks, and tabular structures at native hardware speeds.*
23
+
24
+ [![PyPI version](https://img.shields.io/pypi/v/rasgapdf.svg?color=blue&logo=pypi&logoColor=white)](https://pypi.org/project/rasgapdf/)
25
+ [![Python versions](https://img.shields.io/pypi/pyversions/rasgapdf.svg?logo=python&logoColor=white)](https://pypi.org/project/rasgapdf/)
26
+ [![License: MIT](https://img.shields.io/badge/License-MIT-green.svg)](https://opensource.org/licenses/MIT)
27
+
28
+ </div>
29
+
30
+ ---
31
+
32
+ ## 🚀 Why RasgaPDF?
33
+
34
+ Most Python PDF extraction libraries make fragile guesses: they flatten multi-column layouts into corrupted text streams, mangle tables, and strip away 2D positional geometry.
35
+
36
+ **RasgaPDF does one thing with uncompromising precision:**
37
+ It returns the **exact 2D physical spatial geometry** of the PDF directly to Python:
38
+
39
+ - **Exact Bounding Boxes:** Every character, word span, line, block, and table has exact coordinates `[x0, y0, x1, y1]`.
40
+ - **Zero Lossy Heuristics:** No broken markdown assumptions. Downstream LLMs, RAG pipelines, or vision-language models receive the true physical layout.
41
+ - **Pure Native Speed:** Powered by a SIMD-parallelized Rust core. Extracts typical pages in **~3 to 8 ms**.
42
+ - **Self-Contained Wheel:** Pre-compiled native binaries bundled directly in the wheel. Zero external runtime dependencies.
43
+
44
+ ---
45
+
46
+ ## ⚡ Quick Benchmarks (Python)
47
+
48
+ Measured on Intel Core i7-1365U (Single-thread latency, 1-page complex PDF):
49
+
50
+ | Library | Parse Time | Output Type | Spatial Accuracy |
51
+ | :--- | :--- | :--- | :--- |
52
+ | **⚡ RasgaPDF (`parse_raw`)** | **~7.7 ms** | **Flat Coordinates `[x0, y0, x1, y1]`** | **Exact Physical Canvas** |
53
+ | **🚀 RasgaPDF (`parse_pdf`)** | **~8.5 ms** | **Hierarchical Blocks & Tables** | **Exact Physical Canvas** |
54
+ | PyMuPDF (`fitz`) | ~28.0 ms | Text / Pixmaps | Bounding Boxes |
55
+ | `pdfplumber` | ~410.0 ms | Python Dicts | High Overhead |
56
+ | `pypdf` | ~95.0 ms | Plain String | No Geometry |
57
+
58
+ ---
59
+
60
+ ## 📦 Installation
61
+
62
+ ```bash
63
+ pip install --upgrade rasgapdf
64
+ ```
65
+
66
+ ---
67
+
68
+ ## 💡 Usage
69
+
70
+ RasgaPDF offers two distinct modes for Python engineers:
71
+
72
+ ### ⚡ Mode 1: Raw Physical Coordinates (`parse_raw`)
73
+ The fastest and most direct extraction possible. Returns a flat list of physical 2D coordinates for every text span on the page, with zero block/table overhead (~3-8 ms):
74
+
75
+ ```python
76
+ import rasgapdf
77
+
78
+ # Parse raw physical coordinates
79
+ doc = rasgapdf.parse_raw("document.pdf")
80
+
81
+ print(f"Extracted {doc['total_spans']} spans in {doc['parse_time_ms']:.2f} ms")
82
+
83
+ for span in doc["spans"]:
84
+ print(
85
+ f"[Page {span['page']}] "
86
+ f"bbox=({span['x0']:.2f}, {span['y0']:.2f}, {span['x1']:.2f}, {span['y1']:.2f}) "
87
+ f"font={span['font_name']} ({span['font_size']}pt) "
88
+ f"text=\"{span['text']}\""
89
+ )
90
+ ```
91
+
92
+ #### Output Structure:
93
+ ```json
94
+ {
95
+ "file_name": "document.pdf",
96
+ "total_pages": 1,
97
+ "total_spans": 20,
98
+ "parse_time_ms": 7.78,
99
+ "spans": [
100
+ {
101
+ "page": 1,
102
+ "x0": 50.0,
103
+ "y0": 67.0,
104
+ "x1": 244.0,
105
+ "y1": 83.0,
106
+ "font_size": 12.0,
107
+ "font_name": "Helvetica",
108
+ "text": "The Future of Artificial Intelligence"
109
+ }
110
+ ]
111
+ }
112
+ ```
113
+
114
+ ---
115
+
116
+ ### 🎯 Mode 2: Hierarchical Structured Document (`parse_pdf`)
117
+ Parses the visual hierarchy: paragraphs (`text_blocks`), text lines (`lines`), spans, and auto-detects matrix tables via column alignment:
118
+
119
+ ```python
120
+ import rasgapdf
121
+
122
+ doc = rasgapdf.parse_pdf("invoice.pdf", detect_tables=True)
123
+
124
+ for page in doc["pages"]:
125
+ print(f"--- Page {page['page_number']} ({page['width']}x{page['height']} pt) ---")
126
+
127
+ # 1. Inspect Text Blocks
128
+ for block in page["text_blocks"]:
129
+ print(f"Block #{block['block_id']} at {block['bbox']}:")
130
+ for line in block["lines"]:
131
+ print(f" Line: {line['text']}")
132
+
133
+ # 2. Inspect Detected Tables
134
+ for table in page["tables"]:
135
+ print(f"Table ({table['col_count']} cols x {table['row_count']} rows):")
136
+ print("Headers:", table["headers"])
137
+ for row in table["rows"]:
138
+ print("Row:", row)
139
+ ```
140
+
141
+ ---
142
+
143
+ ## ⚙️ Parameters for `parse_pdf`
144
+
145
+ | Parameter | Type | Default | Description |
146
+ | :--- | :--- | :--- | :--- |
147
+ | `file_path` | `str` | *required* | Path to the PDF file on disk |
148
+ | `detect_tables` | `bool` | `True` | Automatically detect and reconstruct tables |
149
+ | `min_table_cols` | `int` | `3` | Minimum number of columns required to classify as a table |
150
+ | `col_tolerance` | `float` | `16.0` | Horizontal alignment threshold in typographical points |
151
+ | `row_tolerance` | `float` | `6.0` | Vertical line clustering threshold in points |
152
+
153
+ ---
154
+
155
+ ## 📄 License
156
+
157
+ MIT License. Copyright (c) 2026 Messias Martins.
rasgapdf-0.1.2/README.md ADDED
@@ -0,0 +1,141 @@
1
+ # ⚡ RasgaPDF
2
+
3
+ <div align="center">
4
+
5
+ **The Blazing-Fast 2D Spatial PDF Parser for Python.**
6
+ *Extract exact physical coordinates `[x0, y0, x1, y1]`, word spans, text blocks, and tabular structures at native hardware speeds.*
7
+
8
+ [![PyPI version](https://img.shields.io/pypi/v/rasgapdf.svg?color=blue&logo=pypi&logoColor=white)](https://pypi.org/project/rasgapdf/)
9
+ [![Python versions](https://img.shields.io/pypi/pyversions/rasgapdf.svg?logo=python&logoColor=white)](https://pypi.org/project/rasgapdf/)
10
+ [![License: MIT](https://img.shields.io/badge/License-MIT-green.svg)](https://opensource.org/licenses/MIT)
11
+
12
+ </div>
13
+
14
+ ---
15
+
16
+ ## 🚀 Why RasgaPDF?
17
+
18
+ Most Python PDF extraction libraries make fragile guesses: they flatten multi-column layouts into corrupted text streams, mangle tables, and strip away 2D positional geometry.
19
+
20
+ **RasgaPDF does one thing with uncompromising precision:**
21
+ It returns the **exact 2D physical spatial geometry** of the PDF directly to Python:
22
+
23
+ - **Exact Bounding Boxes:** Every character, word span, line, block, and table has exact coordinates `[x0, y0, x1, y1]`.
24
+ - **Zero Lossy Heuristics:** No broken markdown assumptions. Downstream LLMs, RAG pipelines, or vision-language models receive the true physical layout.
25
+ - **Pure Native Speed:** Powered by a SIMD-parallelized Rust core. Extracts typical pages in **~3 to 8 ms**.
26
+ - **Self-Contained Wheel:** Pre-compiled native binaries bundled directly in the wheel. Zero external runtime dependencies.
27
+
28
+ ---
29
+
30
+ ## ⚡ Quick Benchmarks (Python)
31
+
32
+ Measured on Intel Core i7-1365U (Single-thread latency, 1-page complex PDF):
33
+
34
+ | Library | Parse Time | Output Type | Spatial Accuracy |
35
+ | :--- | :--- | :--- | :--- |
36
+ | **⚡ RasgaPDF (`parse_raw`)** | **~7.7 ms** | **Flat Coordinates `[x0, y0, x1, y1]`** | **Exact Physical Canvas** |
37
+ | **🚀 RasgaPDF (`parse_pdf`)** | **~8.5 ms** | **Hierarchical Blocks & Tables** | **Exact Physical Canvas** |
38
+ | PyMuPDF (`fitz`) | ~28.0 ms | Text / Pixmaps | Bounding Boxes |
39
+ | `pdfplumber` | ~410.0 ms | Python Dicts | High Overhead |
40
+ | `pypdf` | ~95.0 ms | Plain String | No Geometry |
41
+
42
+ ---
43
+
44
+ ## 📦 Installation
45
+
46
+ ```bash
47
+ pip install --upgrade rasgapdf
48
+ ```
49
+
50
+ ---
51
+
52
+ ## 💡 Usage
53
+
54
+ RasgaPDF offers two distinct modes for Python engineers:
55
+
56
+ ### ⚡ Mode 1: Raw Physical Coordinates (`parse_raw`)
57
+ The fastest and most direct extraction possible. Returns a flat list of physical 2D coordinates for every text span on the page, with zero block/table overhead (~3-8 ms):
58
+
59
+ ```python
60
+ import rasgapdf
61
+
62
+ # Parse raw physical coordinates
63
+ doc = rasgapdf.parse_raw("document.pdf")
64
+
65
+ print(f"Extracted {doc['total_spans']} spans in {doc['parse_time_ms']:.2f} ms")
66
+
67
+ for span in doc["spans"]:
68
+ print(
69
+ f"[Page {span['page']}] "
70
+ f"bbox=({span['x0']:.2f}, {span['y0']:.2f}, {span['x1']:.2f}, {span['y1']:.2f}) "
71
+ f"font={span['font_name']} ({span['font_size']}pt) "
72
+ f"text=\"{span['text']}\""
73
+ )
74
+ ```
75
+
76
+ #### Output Structure:
77
+ ```json
78
+ {
79
+ "file_name": "document.pdf",
80
+ "total_pages": 1,
81
+ "total_spans": 20,
82
+ "parse_time_ms": 7.78,
83
+ "spans": [
84
+ {
85
+ "page": 1,
86
+ "x0": 50.0,
87
+ "y0": 67.0,
88
+ "x1": 244.0,
89
+ "y1": 83.0,
90
+ "font_size": 12.0,
91
+ "font_name": "Helvetica",
92
+ "text": "The Future of Artificial Intelligence"
93
+ }
94
+ ]
95
+ }
96
+ ```
97
+
98
+ ---
99
+
100
+ ### 🎯 Mode 2: Hierarchical Structured Document (`parse_pdf`)
101
+ Parses the visual hierarchy: paragraphs (`text_blocks`), text lines (`lines`), spans, and auto-detects matrix tables via column alignment:
102
+
103
+ ```python
104
+ import rasgapdf
105
+
106
+ doc = rasgapdf.parse_pdf("invoice.pdf", detect_tables=True)
107
+
108
+ for page in doc["pages"]:
109
+ print(f"--- Page {page['page_number']} ({page['width']}x{page['height']} pt) ---")
110
+
111
+ # 1. Inspect Text Blocks
112
+ for block in page["text_blocks"]:
113
+ print(f"Block #{block['block_id']} at {block['bbox']}:")
114
+ for line in block["lines"]:
115
+ print(f" Line: {line['text']}")
116
+
117
+ # 2. Inspect Detected Tables
118
+ for table in page["tables"]:
119
+ print(f"Table ({table['col_count']} cols x {table['row_count']} rows):")
120
+ print("Headers:", table["headers"])
121
+ for row in table["rows"]:
122
+ print("Row:", row)
123
+ ```
124
+
125
+ ---
126
+
127
+ ## ⚙️ Parameters for `parse_pdf`
128
+
129
+ | Parameter | Type | Default | Description |
130
+ | :--- | :--- | :--- | :--- |
131
+ | `file_path` | `str` | *required* | Path to the PDF file on disk |
132
+ | `detect_tables` | `bool` | `True` | Automatically detect and reconstruct tables |
133
+ | `min_table_cols` | `int` | `3` | Minimum number of columns required to classify as a table |
134
+ | `col_tolerance` | `float` | `16.0` | Horizontal alignment threshold in typographical points |
135
+ | `row_tolerance` | `float` | `6.0` | Vertical line clustering threshold in points |
136
+
137
+ ---
138
+
139
+ ## 📄 License
140
+
141
+ MIT License. Copyright (c) 2026 Messias Martins.
{rasgapdf-0.1.0 → rasgapdf-0.1.2}/pyproject.toml RENAMED
@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
4
4
 
5
5
  [project]
6
6
  name = "rasgapdf"
7
- version = "0.1.0"
7
+ version = "0.1.2"
8
8
  authors = [
9
9
  { name="Messias Martins" },
10
10
  ]
@@ -21,8 +21,9 @@ classifiers = [
21
21
  dependencies = []
22
22
 
23
23
  [project.urls]
24
- "Homepage" = "https://github.com/messiasmartins/rasgapdf"
25
- "Bug Tracker" = "https://github.com/messiasmartins/rasgapdf/issues"
24
+ "Homepage" = "https://huggingface.co/martinsmessias/rasgapdf"
25
+ "Repository" = "https://huggingface.co/martinsmessias/rasgapdf"
26
+ "Discussions" = "https://huggingface.co/martinsmessias/rasgapdf/discussions"
26
27
 
27
28
  [tool.setuptools.packages.find]
28
29
  where = ["."]
rasgapdf-0.1.2/rasgapdf.egg-info/PKG-INFO ADDED
@@ -0,0 +1,157 @@
1
+ Metadata-Version: 2.4
2
+ Name: rasgapdf
3
+ Version: 0.1.2
4
+ Summary: Ultra-fast 2D spatial PDF parser powered by Rust
5
+ Author: Messias Martins
6
+ Project-URL: Homepage, https://huggingface.co/martinsmessias/rasgapdf
7
+ Project-URL: Repository, https://huggingface.co/martinsmessias/rasgapdf
8
+ Project-URL: Discussions, https://huggingface.co/martinsmessias/rasgapdf/discussions
9
+ Classifier: Programming Language :: Python :: 3
10
+ Classifier: Programming Language :: Rust
11
+ Classifier: License :: OSI Approved :: MIT License
12
+ Classifier: Operating System :: POSIX :: Linux
13
+ Classifier: Topic :: Text Processing :: General
14
+ Requires-Python: >=3.8
15
+ Description-Content-Type: text/markdown
16
+
17
+ # ⚡ RasgaPDF
18
+
19
+ <div align="center">
20
+
21
+ **The Blazing-Fast 2D Spatial PDF Parser for Python.**
22
+ *Extract exact physical coordinates `[x0, y0, x1, y1]`, word spans, text blocks, and tabular structures at native hardware speeds.*
23
+
24
+ [![PyPI version](https://img.shields.io/pypi/v/rasgapdf.svg?color=blue&logo=pypi&logoColor=white)](https://pypi.org/project/rasgapdf/)
25
+ [![Python versions](https://img.shields.io/pypi/pyversions/rasgapdf.svg?logo=python&logoColor=white)](https://pypi.org/project/rasgapdf/)
26
+ [![License: MIT](https://img.shields.io/badge/License-MIT-green.svg)](https://opensource.org/licenses/MIT)
27
+
28
+ </div>
29
+
30
+ ---
31
+
32
+ ## 🚀 Why RasgaPDF?
33
+
34
+ Most Python PDF extraction libraries make fragile guesses: they flatten multi-column layouts into corrupted text streams, mangle tables, and strip away 2D positional geometry.
35
+
36
+ **RasgaPDF does one thing with uncompromising precision:**
37
+ It returns the **exact 2D physical spatial geometry** of the PDF directly to Python:
38
+
39
+ - **Exact Bounding Boxes:** Every character, word span, line, block, and table has exact coordinates `[x0, y0, x1, y1]`.
40
+ - **Zero Lossy Heuristics:** No broken markdown assumptions. Downstream LLMs, RAG pipelines, or vision-language models receive the true physical layout.
41
+ - **Pure Native Speed:** Powered by a SIMD-parallelized Rust core. Extracts typical pages in **~3 to 8 ms**.
42
+ - **Self-Contained Wheel:** Pre-compiled native binaries bundled directly in the wheel. Zero external runtime dependencies.
43
+
44
+ ---
45
+
46
+ ## ⚡ Quick Benchmarks (Python)
47
+
48
+ Measured on Intel Core i7-1365U (Single-thread latency, 1-page complex PDF):
49
+
50
+ | Library | Parse Time | Output Type | Spatial Accuracy |
51
+ | :--- | :--- | :--- | :--- |
52
+ | **⚡ RasgaPDF (`parse_raw`)** | **~7.7 ms** | **Flat Coordinates `[x0, y0, x1, y1]`** | **Exact Physical Canvas** |
53
+ | **🚀 RasgaPDF (`parse_pdf`)** | **~8.5 ms** | **Hierarchical Blocks & Tables** | **Exact Physical Canvas** |
54
+ | PyMuPDF (`fitz`) | ~28.0 ms | Text / Pixmaps | Bounding Boxes |
55
+ | `pdfplumber` | ~410.0 ms | Python Dicts | High Overhead |
56
+ | `pypdf` | ~95.0 ms | Plain String | No Geometry |
57
+
58
+ ---
59
+
60
+ ## 📦 Installation
61
+
62
+ ```bash
63
+ pip install --upgrade rasgapdf
64
+ ```
65
+
66
+ ---
67
+
68
+ ## 💡 Usage
69
+
70
+ RasgaPDF offers two distinct modes for Python engineers:
71
+
72
+ ### ⚡ Mode 1: Raw Physical Coordinates (`parse_raw`)
73
+ The fastest and most direct extraction possible. Returns a flat list of physical 2D coordinates for every text span on the page, with zero block/table overhead (~3-8 ms):
74
+
75
+ ```python
76
+ import rasgapdf
77
+
78
+ # Parse raw physical coordinates
79
+ doc = rasgapdf.parse_raw("document.pdf")
80
+
81
+ print(f"Extracted {doc['total_spans']} spans in {doc['parse_time_ms']:.2f} ms")
82
+
83
+ for span in doc["spans"]:
84
+ print(
85
+ f"[Page {span['page']}] "
86
+ f"bbox=({span['x0']:.2f}, {span['y0']:.2f}, {span['x1']:.2f}, {span['y1']:.2f}) "
87
+ f"font={span['font_name']} ({span['font_size']}pt) "
88
+ f"text=\"{span['text']}\""
89
+ )
90
+ ```
91
+
92
+ #### Output Structure:
93
+ ```json
94
+ {
95
+ "file_name": "document.pdf",
96
+ "total_pages": 1,
97
+ "total_spans": 20,
98
+ "parse_time_ms": 7.78,
99
+ "spans": [
100
+ {
101
+ "page": 1,
102
+ "x0": 50.0,
103
+ "y0": 67.0,
104
+ "x1": 244.0,
105
+ "y1": 83.0,
106
+ "font_size": 12.0,
107
+ "font_name": "Helvetica",
108
+ "text": "The Future of Artificial Intelligence"
109
+ }
110
+ ]
111
+ }
112
+ ```
113
+
114
+ ---
115
+
116
+ ### 🎯 Mode 2: Hierarchical Structured Document (`parse_pdf`)
117
+ Parses the visual hierarchy: paragraphs (`text_blocks`), text lines (`lines`), spans, and auto-detects matrix tables via column alignment:
118
+
119
+ ```python
120
+ import rasgapdf
121
+
122
+ doc = rasgapdf.parse_pdf("invoice.pdf", detect_tables=True)
123
+
124
+ for page in doc["pages"]:
125
+ print(f"--- Page {page['page_number']} ({page['width']}x{page['height']} pt) ---")
126
+
127
+ # 1. Inspect Text Blocks
128
+ for block in page["text_blocks"]:
129
+ print(f"Block #{block['block_id']} at {block['bbox']}:")
130
+ for line in block["lines"]:
131
+ print(f" Line: {line['text']}")
132
+
133
+ # 2. Inspect Detected Tables
134
+ for table in page["tables"]:
135
+ print(f"Table ({table['col_count']} cols x {table['row_count']} rows):")
136
+ print("Headers:", table["headers"])
137
+ for row in table["rows"]:
138
+ print("Row:", row)
139
+ ```
140
+
141
+ ---
142
+
143
+ ## ⚙️ Parameters for `parse_pdf`
144
+
145
+ | Parameter | Type | Default | Description |
146
+ | :--- | :--- | :--- | :--- |
147
+ | `file_path` | `str` | *required* | Path to the PDF file on disk |
148
+ | `detect_tables` | `bool` | `True` | Automatically detect and reconstruct tables |
149
+ | `min_table_cols` | `int` | `3` | Minimum number of columns required to classify as a table |
150
+ | `col_tolerance` | `float` | `16.0` | Horizontal alignment threshold in typographical points |
151
+ | `row_tolerance` | `float` | `6.0` | Vertical line clustering threshold in points |
152
+
153
+ ---
154
+
155
+ ## 📄 License
156
+
157
+ MIT License. Copyright (c) 2026 Messias Martins.
rasgapdf-0.1.0/PKG-INFO DELETED
@@ -1,203 +0,0 @@
1
- Metadata-Version: 2.4
2
- Name: rasgapdf
3
- Version: 0.1.0
4
- Summary: Ultra-fast 2D spatial PDF parser powered by Rust
5
- Author: Messias Martins
6
- Project-URL: Homepage, https://github.com/messiasmartins/rasgapdf
7
- Project-URL: Bug Tracker, https://github.com/messiasmartins/rasgapdf/issues
8
- Classifier: Programming Language :: Python :: 3
9
- Classifier: Programming Language :: Rust
10
- Classifier: License :: OSI Approved :: MIT License
11
- Classifier: Operating System :: POSIX :: Linux
12
- Classifier: Topic :: Text Processing :: General
13
- Requires-Python: >=3.8
14
- Description-Content-Type: text/markdown
15
-
16
- # ⚡ RasgaPDF
17
-
18
- <div align="center">
19
-
20
- **The Blazing-Fast 2D Spatial PDF Parser powered by Pure Rust.**
21
- *Extract exact physical coordinates (Raw X/Y) or structured blocks & tables at native hardware speeds.*
22
-
23
- [![License: MIT](https://img.shields.io/badge/License-MIT-blue.svg)](LICENSE)
24
- [![Rust](https://img.shields.io/badge/Rust-2021%2B-orange?logo=rust)](Cargo.toml)
25
- [![Python](https://img.shields.io/badge/Python-3.8%2B-yellow?logo=python)](bindings/python)
26
- [![Node.js](https://img.shields.io/badge/Node.js-18%2B-green?logo=node.js)](bindings/nodejs)
27
- [![Go](https://img.shields.io/badge/Go-1.20%2B-cyan?logo=go)](bindings/go)
28
- [![.NET](https://img.shields.io/badge/.NET-8.0%2B-purple?logo=dotnet)](bindings/csharp)
29
- [![Hugging Face](https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-martinsmessias%2Frasgapdf-blue)](https://huggingface.co/martinsmessias/rasgapdf)
30
-
31
- [**Live Gradio Demo**](#-interactive-gradio-demo) • [**Quickstart**](#-quickstart) • [**Two Modes**](#-two-parsing-modes) • [**Architecture**](docs/ARCHITECTURE.md) • [**Spatial Schema**](docs/SCHEMA.md) • [**Contributing**](CONTRIBUTING.md)
32
-
33
- </div>
34
-
35
- ---
36
-
37
- ## 🎯 The Philosophy: Pure Spatial Parsing
38
-
39
- Most PDF libraries make fragile guesses: they convert multi-column documents into flat, corrupted text streams, mangle tables, and strip away spatial geometry.
40
-
41
- **RasgaPDF does one thing with uncompromising precision:**
42
- It returns the **exact 2D physical spatial geometry** of the PDF.
43
-
44
- - **Bounding Boxes Everywhere:** Every character, word span, line, block, and table has exact coordinates `[x0, y0, x1, y1]`.
45
- - **Zero Lossy Guesswork:** No broken heuristics pretending to be Markdown. Your downstream LLMs, RAG pipelines, or OCR validators receive the true physical layout.
46
- - **Microsecond Rayon Parallelism:** Multi-threaded page processing in pure Rust utilizing AVX2/AVX-512 SIMD.
47
- - **Polyglot First:** Native C-ABI FFI bindings for **Python**, **Node.js**, **Go**, and **C# (.NET)**.
48
-
49
- ---
50
-
51
- ## ⚡ Two Parsing Modes
52
-
53
- RasgaPDF offers two distinct modes for different engineering needs:
54
-
55
- ### ⚡ Modo 1: Coordenadas Brutas (Raw X/Y Puro)
56
- O retorno mais direto, bruto e rápido possível. Retorna exclusivamente a lista plana de coordenadas físicas de cada elemento desenhado na página, sem criar hierarquias de blocos ou overhead de tabelas:
57
- - **Latência:** ~3 a 8 ms por página
58
- - **Saídas suportadas:** JSON plano, Text formatado, TSV ou CSV
59
- - **Uso ideal:** RAG espacial, visualizadores 2D, indexação vetorial, pipelines Unix com `awk`/`grep`.
60
-
61
- ```
62
- [Pág. 1] (X0: 50.00, Y0: 67.00, X1: 244.00, Y1: 83.00) [12.0pt] A Revolução da Inteligência Artificial
63
- [Pág. 1] (X0: 50.00, Y0: 103.00, X1: 478.00, Y1: 119.00) [12.0pt] A Inteligência Artificial (IA) tem transformado...
64
- ```
65
-
66
- ### 🎯 Modo 2: Estruturado Hierárquico
67
- Processa a geometria e agrupa elementos em parágrafos visuais (`blocks`), linhas (`lines`), spans de fonte e detecta matrizes de tabelas automaticamente via projeção horizontal de colunas:
68
- - **Latência:** ~8 a 15 ms por página
69
- - **Saída:** Árvore hierárquica `SpatialDocument` -> `pages` -> `text_blocks` -> `lines` -> `tables`.
70
-
71
- ---
72
-
73
- ## ⚡ Performance Benchmark
74
-
75
- Measured on Intel Core i7-1365U (10 Cores / 12 Threads, AVX2):
76
-
77
- | Library / Tool | Language | Single Page | 50 Pages | Memory Footprint | Output Type |
78
- | :--- | :--- | :--- | :--- | :--- | :--- |
79
- | **⚡ RasgaPDF (Raw X/Y)** | **Rust (Native)** | **7.7 ms** | **95 ms** | **~14 MB** | **Flat Raw Coordinates** |
80
- | **🚀 RasgaPDF (Structured)**| **Rust (Native)** | **8.5 ms** | **142 ms** | **~18 MB** | **Spatial Blocks & Tables** |
81
- | PyMuPDF (`fitz`) | C / Python | 28.0 ms | 620 ms | ~45 MB | Text / Pixmap |
82
- | `pdfplumber` | Python | 410.0 ms | 9,800 ms | ~140 MB | Python dicts |
83
- | `pypdf` | Python | 95.0 ms | 2,100 ms | ~85 MB | Flat text string |
84
- | Apache Tika | Java | 650.0 ms | 12,400 ms | ~320 MB | XHTML |
85
-
86
- ---
87
-
88
- ## 💻 Standalone CLI Tool
89
-
90
- ```bash
91
- cargo build --release
92
- ./target/release/rasgapdf --help
93
-
94
- # ⚡ MODO 1: Coordenadas Brutas (Raw X/Y)
95
- ./target/release/rasgapdf sample.pdf --raw
96
- ./target/release/rasgapdf sample.pdf --raw --format text
97
- ./target/release/rasgapdf sample.pdf --raw --format csv
98
- ./target/release/rasgapdf sample.pdf --raw --format tsv
99
-
100
- # 🎯 MODO 2: Estruturado (Hierárquico com Blocos e Tabelas)
101
- ./target/release/rasgapdf sample.pdf --pretty
102
- ./target/release/rasgapdf sample.pdf -o output.json
103
- ```
104
-
105
- ---
106
-
107
- ## 🚀 Polyglot Quickstart
108
-
109
- ### 🐍 Python
110
- ```python
111
- import rasgapdf
112
-
113
- # ⚡ Modo 1: Coordenadas Brutas (Raw X/Y)
114
- raw = rasgapdf.parse_raw("document.pdf")
115
- print(f"Extraídos {raw['total_spans']} spans em {raw['parse_time_ms']:.2f} ms")
116
- for s in raw["spans"]:
117
- print(f"Pág {s['page']} | ({s['x0']}, {s['y0']}) -> {s['text']}")
118
-
119
- # 🎯 Modo 2: Estruturado (Hierárquico)
120
- doc = rasgapdf.parse_pdf("document.pdf", detect_tables=True)
121
- for page in doc["pages"]:
122
- for table in page["tables"]:
123
- print("Tabela detectada:", table["rows"])
124
- ```
125
-
126
- ### 🟢 Node.js
127
- ```javascript
128
- const rasgapdf = require('rasgapdf');
129
-
130
- // ⚡ Modo 1: Coordenadas Brutas
131
- const raw = rasgapdf.parseRaw('document.pdf');
132
- console.log(`Raw: ${raw.total_spans} spans em ${raw.parse_time_ms} ms`);
133
-
134
- // 🎯 Modo 2: Estruturado
135
- const doc = rasgapdf.parsePdf('document.pdf');
136
- console.log(`Blocos de texto: ${doc.pages[0].text_blocks.length}`);
137
- ```
138
-
139
- ### 🔵 Go
140
- ```go
141
- package main
142
-
143
- import (
144
- "fmt"
145
- "github.com/messiasmartins/rasgapdf/bindings/go"
146
- )
147
-
148
- func main() {
149
- // ⚡ Modo 1: Coordenadas Brutas
150
- raw, _ := rasgapdf.ParseRaw("document.pdf")
151
- fmt.Printf("Raw: %d spans em %.2f ms\n", raw.TotalSpans, raw.ParseTimeMs)
152
-
153
- // 🎯 Modo 2: Estruturado
154
- opts := rasgapdf.DefaultOptions()
155
- doc, _ := rasgapdf.ParseFile("document.pdf", &opts)
156
- fmt.Printf("Estruturado: %d páginas\n", doc.TotalPages)
157
- }
158
- ```
159
-
160
- ### 🟣 C# / .NET 8
161
- ```csharp
162
- using RasgaPdf;
163
-
164
- // ⚡ Modo 1: Coordenadas Brutas
165
- var raw = SpatialPdf.ParseRaw("document.pdf");
166
- Console.WriteLine($"Raw: {raw.TotalSpans} spans em {raw.ParseTimeMs} ms");
167
-
168
- // 🎯 Modo 2: Estruturado
169
- var doc = SpatialPdf.Parse("document.pdf");
170
- Console.WriteLine($"Estruturado: {doc.TotalPages} páginas");
171
- ```
172
-
173
- ---
174
-
175
- ## 🎨 Interactive Gradio Demo
176
-
177
- We provide an interactive **visual spatial explorer** built with Gradio:
178
-
179
- ```bash
180
- cd demo_gradio
181
- pip install -r requirements.txt
182
- python app.py
183
- ```
184
- Acesse `http://localhost:7860` para:
185
- - Inspecionar a **tabela interativa de Coordenadas Físicas (Raw X/Y)**.
186
- - Visualizar os **bounding boxes translúcidos** sobrepostos à página do PDF.
187
- - Analisar **tabelas detectadas** renderizadas em DataFrames.
188
- - Exportar JSONs brutos ou estruturados.
189
-
190
- ---
191
-
192
- ## 🤝 Contributing & Community
193
-
194
- Contributions are what make the open-source community thrive!
195
- - Read our [Contributing Guide](CONTRIBUTING.md) to get started.
196
- - Adhere to our [Code of Conduct](CODE_OF_CONDUCT.md).
197
-
198
- ---
199
-
200
- ## 📄 License
201
-
202
- RasgaPDF is licensed under the [MIT License](LICENSE).
203
- Copyright (c) 2026 Messias Martins.
rasgapdf-0.1.0/README.md DELETED
@@ -1,188 +0,0 @@
1
- # ⚡ RasgaPDF
2
-
3
- <div align="center">
4
-
5
- **The Blazing-Fast 2D Spatial PDF Parser powered by Pure Rust.**
6
- *Extract exact physical coordinates (Raw X/Y) or structured blocks & tables at native hardware speeds.*
7
-
8
- [![License: MIT](https://img.shields.io/badge/License-MIT-blue.svg)](LICENSE)
9
- [![Rust](https://img.shields.io/badge/Rust-2021%2B-orange?logo=rust)](Cargo.toml)
10
- [![Python](https://img.shields.io/badge/Python-3.8%2B-yellow?logo=python)](bindings/python)
11
- [![Node.js](https://img.shields.io/badge/Node.js-18%2B-green?logo=node.js)](bindings/nodejs)
12
- [![Go](https://img.shields.io/badge/Go-1.20%2B-cyan?logo=go)](bindings/go)
13
- [![.NET](https://img.shields.io/badge/.NET-8.0%2B-purple?logo=dotnet)](bindings/csharp)
14
- [![Hugging Face](https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-martinsmessias%2Frasgapdf-blue)](https://huggingface.co/martinsmessias/rasgapdf)
15
-
16
- [**Live Gradio Demo**](#-interactive-gradio-demo) • [**Quickstart**](#-quickstart) • [**Two Modes**](#-two-parsing-modes) • [**Architecture**](docs/ARCHITECTURE.md) • [**Spatial Schema**](docs/SCHEMA.md) • [**Contributing**](CONTRIBUTING.md)
17
-
18
- </div>
19
-
20
- ---
21
-
22
- ## 🎯 The Philosophy: Pure Spatial Parsing
23
-
24
- Most PDF libraries make fragile guesses: they convert multi-column documents into flat, corrupted text streams, mangle tables, and strip away spatial geometry.
25
-
26
- **RasgaPDF does one thing with uncompromising precision:**
27
- It returns the **exact 2D physical spatial geometry** of the PDF.
28
-
29
- - **Bounding Boxes Everywhere:** Every character, word span, line, block, and table has exact coordinates `[x0, y0, x1, y1]`.
30
- - **Zero Lossy Guesswork:** No broken heuristics pretending to be Markdown. Your downstream LLMs, RAG pipelines, or OCR validators receive the true physical layout.
31
- - **Microsecond Rayon Parallelism:** Multi-threaded page processing in pure Rust utilizing AVX2/AVX-512 SIMD.
32
- - **Polyglot First:** Native C-ABI FFI bindings for **Python**, **Node.js**, **Go**, and **C# (.NET)**.
33
-
34
- ---
35
-
36
- ## ⚡ Two Parsing Modes
37
-
38
- RasgaPDF offers two distinct modes for different engineering needs:
39
-
40
- ### ⚡ Modo 1: Coordenadas Brutas (Raw X/Y Puro)
41
- O retorno mais direto, bruto e rápido possível. Retorna exclusivamente a lista plana de coordenadas físicas de cada elemento desenhado na página, sem criar hierarquias de blocos ou overhead de tabelas:
42
- - **Latência:** ~3 a 8 ms por página
43
- - **Saídas suportadas:** JSON plano, Text formatado, TSV ou CSV
44
- - **Uso ideal:** RAG espacial, visualizadores 2D, indexação vetorial, pipelines Unix com `awk`/`grep`.
45
-
46
- ```
47
- [Pág. 1] (X0: 50.00, Y0: 67.00, X1: 244.00, Y1: 83.00) [12.0pt] A Revolução da Inteligência Artificial
48
- [Pág. 1] (X0: 50.00, Y0: 103.00, X1: 478.00, Y1: 119.00) [12.0pt] A Inteligência Artificial (IA) tem transformado...
49
- ```
50
-
51
- ### 🎯 Modo 2: Estruturado Hierárquico
52
- Processa a geometria e agrupa elementos em parágrafos visuais (`blocks`), linhas (`lines`), spans de fonte e detecta matrizes de tabelas automaticamente via projeção horizontal de colunas:
53
- - **Latência:** ~8 a 15 ms por página
54
- - **Saída:** Árvore hierárquica `SpatialDocument` -> `pages` -> `text_blocks` -> `lines` -> `tables`.
55
-
56
- ---
57
-
58
- ## ⚡ Performance Benchmark
59
-
60
- Measured on Intel Core i7-1365U (10 Cores / 12 Threads, AVX2):
61
-
62
- | Library / Tool | Language | Single Page | 50 Pages | Memory Footprint | Output Type |
63
- | :--- | :--- | :--- | :--- | :--- | :--- |
64
- | **⚡ RasgaPDF (Raw X/Y)** | **Rust (Native)** | **7.7 ms** | **95 ms** | **~14 MB** | **Flat Raw Coordinates** |
65
- | **🚀 RasgaPDF (Structured)**| **Rust (Native)** | **8.5 ms** | **142 ms** | **~18 MB** | **Spatial Blocks & Tables** |
66
- | PyMuPDF (`fitz`) | C / Python | 28.0 ms | 620 ms | ~45 MB | Text / Pixmap |
67
- | `pdfplumber` | Python | 410.0 ms | 9,800 ms | ~140 MB | Python dicts |
68
- | `pypdf` | Python | 95.0 ms | 2,100 ms | ~85 MB | Flat text string |
69
- | Apache Tika | Java | 650.0 ms | 12,400 ms | ~320 MB | XHTML |
70
-
71
- ---
72
-
73
- ## 💻 Standalone CLI Tool
74
-
75
- ```bash
76
- cargo build --release
77
- ./target/release/rasgapdf --help
78
-
79
- # ⚡ MODO 1: Coordenadas Brutas (Raw X/Y)
80
- ./target/release/rasgapdf sample.pdf --raw
81
- ./target/release/rasgapdf sample.pdf --raw --format text
82
- ./target/release/rasgapdf sample.pdf --raw --format csv
83
- ./target/release/rasgapdf sample.pdf --raw --format tsv
84
-
85
- # 🎯 MODO 2: Estruturado (Hierárquico com Blocos e Tabelas)
86
- ./target/release/rasgapdf sample.pdf --pretty
87
- ./target/release/rasgapdf sample.pdf -o output.json
88
- ```
89
-
90
- ---
91
-
92
- ## 🚀 Polyglot Quickstart
93
-
94
- ### 🐍 Python
95
- ```python
96
- import rasgapdf
97
-
98
- # ⚡ Modo 1: Coordenadas Brutas (Raw X/Y)
99
- raw = rasgapdf.parse_raw("document.pdf")
100
- print(f"Extraídos {raw['total_spans']} spans em {raw['parse_time_ms']:.2f} ms")
101
- for s in raw["spans"]:
102
- print(f"Pág {s['page']} | ({s['x0']}, {s['y0']}) -> {s['text']}")
103
-
104
- # 🎯 Modo 2: Estruturado (Hierárquico)
105
- doc = rasgapdf.parse_pdf("document.pdf", detect_tables=True)
106
- for page in doc["pages"]:
107
- for table in page["tables"]:
108
- print("Tabela detectada:", table["rows"])
109
- ```
110
-
111
- ### 🟢 Node.js
112
- ```javascript
113
- const rasgapdf = require('rasgapdf');
114
-
115
- // ⚡ Modo 1: Coordenadas Brutas
116
- const raw = rasgapdf.parseRaw('document.pdf');
117
- console.log(`Raw: ${raw.total_spans} spans em ${raw.parse_time_ms} ms`);
118
-
119
- // 🎯 Modo 2: Estruturado
120
- const doc = rasgapdf.parsePdf('document.pdf');
121
- console.log(`Blocos de texto: ${doc.pages[0].text_blocks.length}`);
122
- ```
123
-
124
- ### 🔵 Go
125
- ```go
126
- package main
127
-
128
- import (
129
- "fmt"
130
- "github.com/messiasmartins/rasgapdf/bindings/go"
131
- )
132
-
133
- func main() {
134
- // ⚡ Modo 1: Coordenadas Brutas
135
- raw, _ := rasgapdf.ParseRaw("document.pdf")
136
- fmt.Printf("Raw: %d spans em %.2f ms\n", raw.TotalSpans, raw.ParseTimeMs)
137
-
138
- // 🎯 Modo 2: Estruturado
139
- opts := rasgapdf.DefaultOptions()
140
- doc, _ := rasgapdf.ParseFile("document.pdf", &opts)
141
- fmt.Printf("Estruturado: %d páginas\n", doc.TotalPages)
142
- }
143
- ```
144
-
145
- ### 🟣 C# / .NET 8
146
- ```csharp
147
- using RasgaPdf;
148
-
149
- // ⚡ Modo 1: Coordenadas Brutas
150
- var raw = SpatialPdf.ParseRaw("document.pdf");
151
- Console.WriteLine($"Raw: {raw.TotalSpans} spans em {raw.ParseTimeMs} ms");
152
-
153
- // 🎯 Modo 2: Estruturado
154
- var doc = SpatialPdf.Parse("document.pdf");
155
- Console.WriteLine($"Estruturado: {doc.TotalPages} páginas");
156
- ```
157
-
158
- ---
159
-
160
- ## 🎨 Interactive Gradio Demo
161
-
162
- We provide an interactive **visual spatial explorer** built with Gradio:
163
-
164
- ```bash
165
- cd demo_gradio
166
- pip install -r requirements.txt
167
- python app.py
168
- ```
169
- Acesse `http://localhost:7860` para:
170
- - Inspecionar a **tabela interativa de Coordenadas Físicas (Raw X/Y)**.
171
- - Visualizar os **bounding boxes translúcidos** sobrepostos à página do PDF.
172
- - Analisar **tabelas detectadas** renderizadas em DataFrames.
173
- - Exportar JSONs brutos ou estruturados.
174
-
175
- ---
176
-
177
- ## 🤝 Contributing & Community
178
-
179
- Contributions are what make the open-source community thrive!
180
- - Read our [Contributing Guide](CONTRIBUTING.md) to get started.
181
- - Adhere to our [Code of Conduct](CODE_OF_CONDUCT.md).
182
-
183
- ---
184
-
185
- ## 📄 License
186
-
187
- RasgaPDF is licensed under the [MIT License](LICENSE).
188
- Copyright (c) 2026 Messias Martins.
rasgapdf-0.1.0/rasgapdf.egg-info/PKG-INFO DELETED
@@ -1,203 +0,0 @@
1
- Metadata-Version: 2.4
2
- Name: rasgapdf
3
- Version: 0.1.0
4
- Summary: Ultra-fast 2D spatial PDF parser powered by Rust
5
- Author: Messias Martins
6
- Project-URL: Homepage, https://github.com/messiasmartins/rasgapdf
7
- Project-URL: Bug Tracker, https://github.com/messiasmartins/rasgapdf/issues
8
- Classifier: Programming Language :: Python :: 3
9
- Classifier: Programming Language :: Rust
10
- Classifier: License :: OSI Approved :: MIT License
11
- Classifier: Operating System :: POSIX :: Linux
12
- Classifier: Topic :: Text Processing :: General
13
- Requires-Python: >=3.8
14
- Description-Content-Type: text/markdown
15
-
16
- # ⚡ RasgaPDF
17
-
18
- <div align="center">
19
-
20
- **The Blazing-Fast 2D Spatial PDF Parser powered by Pure Rust.**
21
- *Extract exact physical coordinates (Raw X/Y) or structured blocks & tables at native hardware speeds.*
22
-
23
- [![License: MIT](https://img.shields.io/badge/License-MIT-blue.svg)](LICENSE)
24
- [![Rust](https://img.shields.io/badge/Rust-2021%2B-orange?logo=rust)](Cargo.toml)
25
- [![Python](https://img.shields.io/badge/Python-3.8%2B-yellow?logo=python)](bindings/python)
26
- [![Node.js](https://img.shields.io/badge/Node.js-18%2B-green?logo=node.js)](bindings/nodejs)
27
- [![Go](https://img.shields.io/badge/Go-1.20%2B-cyan?logo=go)](bindings/go)
28
- [![.NET](https://img.shields.io/badge/.NET-8.0%2B-purple?logo=dotnet)](bindings/csharp)
29
- [![Hugging Face](https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-martinsmessias%2Frasgapdf-blue)](https://huggingface.co/martinsmessias/rasgapdf)
30
-
31
- [**Live Gradio Demo**](#-interactive-gradio-demo) • [**Quickstart**](#-quickstart) • [**Two Modes**](#-two-parsing-modes) • [**Architecture**](docs/ARCHITECTURE.md) • [**Spatial Schema**](docs/SCHEMA.md) • [**Contributing**](CONTRIBUTING.md)
32
-
33
- </div>
34
-
35
- ---
36
-
37
- ## 🎯 The Philosophy: Pure Spatial Parsing
38
-
39
- Most PDF libraries make fragile guesses: they convert multi-column documents into flat, corrupted text streams, mangle tables, and strip away spatial geometry.
40
-
41
- **RasgaPDF does one thing with uncompromising precision:**
42
- It returns the **exact 2D physical spatial geometry** of the PDF.
43
-
44
- - **Bounding Boxes Everywhere:** Every character, word span, line, block, and table has exact coordinates `[x0, y0, x1, y1]`.
45
- - **Zero Lossy Guesswork:** No broken heuristics pretending to be Markdown. Your downstream LLMs, RAG pipelines, or OCR validators receive the true physical layout.
46
- - **Microsecond Rayon Parallelism:** Multi-threaded page processing in pure Rust utilizing AVX2/AVX-512 SIMD.
47
- - **Polyglot First:** Native C-ABI FFI bindings for **Python**, **Node.js**, **Go**, and **C# (.NET)**.
48
-
49
- ---
50
-
51
- ## ⚡ Two Parsing Modes
52
-
53
- RasgaPDF offers two distinct modes for different engineering needs:
54
-
55
- ### ⚡ Modo 1: Coordenadas Brutas (Raw X/Y Puro)
56
- O retorno mais direto, bruto e rápido possível. Retorna exclusivamente a lista plana de coordenadas físicas de cada elemento desenhado na página, sem criar hierarquias de blocos ou overhead de tabelas:
57
- - **Latência:** ~3 a 8 ms por página
58
- - **Saídas suportadas:** JSON plano, Text formatado, TSV ou CSV
59
- - **Uso ideal:** RAG espacial, visualizadores 2D, indexação vetorial, pipelines Unix com `awk`/`grep`.
60
-
61
- ```
62
- [Pág. 1] (X0: 50.00, Y0: 67.00, X1: 244.00, Y1: 83.00) [12.0pt] A Revolução da Inteligência Artificial
63
- [Pág. 1] (X0: 50.00, Y0: 103.00, X1: 478.00, Y1: 119.00) [12.0pt] A Inteligência Artificial (IA) tem transformado...
64
- ```
65
-
66
- ### 🎯 Modo 2: Estruturado Hierárquico
67
- Processa a geometria e agrupa elementos em parágrafos visuais (`blocks`), linhas (`lines`), spans de fonte e detecta matrizes de tabelas automaticamente via projeção horizontal de colunas:
68
- - **Latência:** ~8 a 15 ms por página
69
- - **Saída:** Árvore hierárquica `SpatialDocument` -> `pages` -> `text_blocks` -> `lines` -> `tables`.
70
-
71
- ---
72
-
73
- ## ⚡ Performance Benchmark
74
-
75
- Measured on Intel Core i7-1365U (10 Cores / 12 Threads, AVX2):
76
-
77
- | Library / Tool | Language | Single Page | 50 Pages | Memory Footprint | Output Type |
78
- | :--- | :--- | :--- | :--- | :--- | :--- |
79
- | **⚡ RasgaPDF (Raw X/Y)** | **Rust (Native)** | **7.7 ms** | **95 ms** | **~14 MB** | **Flat Raw Coordinates** |
80
- | **🚀 RasgaPDF (Structured)**| **Rust (Native)** | **8.5 ms** | **142 ms** | **~18 MB** | **Spatial Blocks & Tables** |
81
- | PyMuPDF (`fitz`) | C / Python | 28.0 ms | 620 ms | ~45 MB | Text / Pixmap |
82
- | `pdfplumber` | Python | 410.0 ms | 9,800 ms | ~140 MB | Python dicts |
83
- | `pypdf` | Python | 95.0 ms | 2,100 ms | ~85 MB | Flat text string |
84
- | Apache Tika | Java | 650.0 ms | 12,400 ms | ~320 MB | XHTML |
85
-
86
- ---
87
-
88
- ## 💻 Standalone CLI Tool
89
-
90
- ```bash
91
- cargo build --release
92
- ./target/release/rasgapdf --help
93
-
94
- # ⚡ MODO 1: Coordenadas Brutas (Raw X/Y)
95
- ./target/release/rasgapdf sample.pdf --raw
96
- ./target/release/rasgapdf sample.pdf --raw --format text
97
- ./target/release/rasgapdf sample.pdf --raw --format csv
98
- ./target/release/rasgapdf sample.pdf --raw --format tsv
99
-
100
- # 🎯 MODO 2: Estruturado (Hierárquico com Blocos e Tabelas)
101
- ./target/release/rasgapdf sample.pdf --pretty
102
- ./target/release/rasgapdf sample.pdf -o output.json
103
- ```
104
-
105
- ---
106
-
107
- ## 🚀 Polyglot Quickstart
108
-
109
- ### 🐍 Python
110
- ```python
111
- import rasgapdf
112
-
113
- # ⚡ Modo 1: Coordenadas Brutas (Raw X/Y)
114
- raw = rasgapdf.parse_raw("document.pdf")
115
- print(f"Extraídos {raw['total_spans']} spans em {raw['parse_time_ms']:.2f} ms")
116
- for s in raw["spans"]:
117
- print(f"Pág {s['page']} | ({s['x0']}, {s['y0']}) -> {s['text']}")
118
-
119
- # 🎯 Modo 2: Estruturado (Hierárquico)
120
- doc = rasgapdf.parse_pdf("document.pdf", detect_tables=True)
121
- for page in doc["pages"]:
122
- for table in page["tables"]:
123
- print("Tabela detectada:", table["rows"])
124
- ```
125
-
126
- ### 🟢 Node.js
127
- ```javascript
128
- const rasgapdf = require('rasgapdf');
129
-
130
- // ⚡ Modo 1: Coordenadas Brutas
131
- const raw = rasgapdf.parseRaw('document.pdf');
132
- console.log(`Raw: ${raw.total_spans} spans em ${raw.parse_time_ms} ms`);
133
-
134
- // 🎯 Modo 2: Estruturado
135
- const doc = rasgapdf.parsePdf('document.pdf');
136
- console.log(`Blocos de texto: ${doc.pages[0].text_blocks.length}`);
137
- ```
138
-
139
- ### 🔵 Go
140
- ```go
141
- package main
142
-
143
- import (
144
- "fmt"
145
- "github.com/messiasmartins/rasgapdf/bindings/go"
146
- )
147
-
148
- func main() {
149
- // ⚡ Modo 1: Coordenadas Brutas
150
- raw, _ := rasgapdf.ParseRaw("document.pdf")
151
- fmt.Printf("Raw: %d spans em %.2f ms\n", raw.TotalSpans, raw.ParseTimeMs)
152
-
153
- // 🎯 Modo 2: Estruturado
154
- opts := rasgapdf.DefaultOptions()
155
- doc, _ := rasgapdf.ParseFile("document.pdf", &opts)
156
- fmt.Printf("Estruturado: %d páginas\n", doc.TotalPages)
157
- }
158
- ```
159
-
160
- ### 🟣 C# / .NET 8
161
- ```csharp
162
- using RasgaPdf;
163
-
164
- // ⚡ Modo 1: Coordenadas Brutas
165
- var raw = SpatialPdf.ParseRaw("document.pdf");
166
- Console.WriteLine($"Raw: {raw.TotalSpans} spans em {raw.ParseTimeMs} ms");
167
-
168
- // 🎯 Modo 2: Estruturado
169
- var doc = SpatialPdf.Parse("document.pdf");
170
- Console.WriteLine($"Estruturado: {doc.TotalPages} páginas");
171
- ```
172
-
173
- ---
174
-
175
- ## 🎨 Interactive Gradio Demo
176
-
177
- We provide an interactive **visual spatial explorer** built with Gradio:
178
-
179
- ```bash
180
- cd demo_gradio
181
- pip install -r requirements.txt
182
- python app.py
183
- ```
184
- Acesse `http://localhost:7860` para:
185
- - Inspecionar a **tabela interativa de Coordenadas Físicas (Raw X/Y)**.
186
- - Visualizar os **bounding boxes translúcidos** sobrepostos à página do PDF.
187
- - Analisar **tabelas detectadas** renderizadas em DataFrames.
188
- - Exportar JSONs brutos ou estruturados.
189
-
190
- ---
191
-
192
- ## 🤝 Contributing & Community
193
-
194
- Contributions are what make the open-source community thrive!
195
- - Read our [Contributing Guide](CONTRIBUTING.md) to get started.
196
- - Adhere to our [Code of Conduct](CODE_OF_CONDUCT.md).
197
-
198
- ---
199
-
200
- ## 📄 License
201
-
202
- RasgaPDF is licensed under the [MIT License](LICENSE).
203
- Copyright (c) 2026 Messias Martins.
{rasgapdf-0.1.0 → rasgapdf-0.1.2}/rasgapdf/__init__.py RENAMED
File without changes
{rasgapdf-0.1.0 → rasgapdf-0.1.2}/setup.cfg RENAMED
File without changes