mistocr 0.2.1__py3-none-any.whl → 0.2.7__py3-none-any.whl

mistocr/__init__.py

@@ -1 +1 @@
-__version__ = "0.2.1"
+__version__ = "0.2.7"

mistocr/_modidx.py

@@ -21,7 +21,8 @@ d = { 'settings': { 'branch': 'main',
                               'mistocr.core.upload_pdf': ('core.html#upload_pdf', 'mistocr/core.py'),
                               'mistocr.core.wait_for_job': ('core.html#wait_for_job', 'mistocr/core.py')},
             'mistocr.pipeline': {'mistocr.pipeline.pdf_to_md': ('pipeline.html#pdf_to_md', 'mistocr/pipeline.py')},
-            'mistocr.refine': { 'mistocr.refine.HeadingCorrections': ('refine.html#headingcorrections', 'mistocr/refine.py'),
+            'mistocr.refine': { 'mistocr.refine.HeadingCorrection': ('refine.html#headingcorrection', 'mistocr/refine.py'),
+                                'mistocr.refine.HeadingCorrections': ('refine.html#headingcorrections', 'mistocr/refine.py'),
                                 'mistocr.refine.ImgDescription': ('refine.html#imgdescription', 'mistocr/refine.py'),
                                 'mistocr.refine.add_descs_to_pg': ('refine.html#add_descs_to_pg', 'mistocr/refine.py'),
                                 'mistocr.refine.add_descs_to_pgs': ('refine.html#add_descs_to_pgs', 'mistocr/refine.py'),

mistocr/refine.py

@@ -3,7 +3,7 @@
 # AUTOGENERATED! DO NOT EDIT! File to edit: ../nbs/01_refine.ipynb.
 
 # %% auto 0
-__all__ = ['prompt_fix_hdgs', 'describe_img_prompt', 'get_hdgs', 'add_pg_hdgs', 'read_pgs_pg', 'fmt_hdgs_idx',
+__all__ = ['describe_img_prompt', 'get_hdgs', 'add_pg_hdgs', 'read_pgs_pg', 'fmt_hdgs_idx', 'HeadingCorrection',
            'HeadingCorrections', 'fix_hdg_hierarchy', 'mk_fixes_lut', 'apply_hdg_fixes', 'fix_hdgs', 'ImgDescription',
            'describe_img', 'limit', 'parse_r', 'describe_imgs', 'save_img_descs', 'add_descs_to_pg', 'add_descs_to_pgs',
            'add_img_descs']
@@ -59,40 +59,12 @@ def fmt_hdgs_idx(
 
 
 # %% ../nbs/01_refine.ipynb 18
-class HeadingCorrections(BaseModel):
-    corrections: dict[int, str]  # index → corrected heading
-
-# %% ../nbs/01_refine.ipynb 20
-prompt_fix_hdgs = """Fix markdown heading hierarchy errors while preserving the document's intended structure.
-
-INPUT FORMAT: Each heading is prefixed with its index number (e.g., "0. # Title ... page 1")
-
-RULES - Apply these fixes in order:
-
-1. **Single H1 rule**: Documents must have exactly ONE # heading (typically the document title at the top)
-   - If index 0 is already #, then all subsequent headings (index 1+) must be ## or deeper
-   - If no H1 exists, the first major heading should be #, and all others ## or deeper
-   - NO exceptions: appendices, references, and all sections are ## or deeper after the title
-
-2. **Infer depth from numbering patterns**: If headings contain section numbers, deeper nesting means deeper heading level
-   - Parent section (e.g., "1", "2", "A") should be shallower than child (e.g., "1.1", "2.a", "A.1")
-   - Child section should be one # deeper than parent
-   - Works with any numbering: "1/1.1/1.1.1", "A/A.1/A.1.a", "I/I.A/I.A.1", etc.
-
-3. **Level jumps**: Headings can only increase by one # at a time when moving deeper
-   - Wrong: ## Section → ##### Subsection
-   - Fixed: ## Section → ### Subsection
-
-4. **Decreasing levels is OK**: Moving back up the hierarchy (### to ##) is valid for new sections
-
-OUTPUT: Return a Python dictionary mapping index to corrected heading (without the index prefix).
-IMPORTANT: Preserve the " ... page N" suffix in all corrected headings.
-Only include entries that need changes.
-
-Headings to analyze:
-{headings_list}
-"""
+class HeadingCorrection(BaseModel):
+    index: int
+    corrected: str
 
+class HeadingCorrections(BaseModel):
+    corrections: list[HeadingCorrection]
 
 # %% ../nbs/01_refine.ipynb 22
 def fix_hdg_hierarchy(
@@ -106,7 +78,8 @@ def fix_hdg_hierarchy(
     if prompt is None: prompt = prompt_fix_hdgs
     prompt = prompt.format(headings_list=fmt_hdgs_idx(hdgs))
     r = completion(model=model, messages=[{"role": "user", "content": prompt}], response_format=HeadingCorrections, api_key=api_key)
-    return json.loads(r.choices[0].message.content)['corrections']
+    fixes =  json.loads(r.choices[0].message.content)['corrections']
+    return {o['index']: o['corrected'] for o in fixes}
 
 
 # %% ../nbs/01_refine.ipynb 25
@@ -120,7 +93,7 @@ def mk_fixes_lut(
     "Make a lookup table of fixes"
     if api_key is None: api_key = os.getenv('ANTHROPIC_API_KEY')
     fixes = fix_hdg_hierarchy(hdgs, model=model, api_key=api_key, **kwargs)
-    return {hdgs[int(k)]:v for k,v in fixes.items()}
+    return {hdgs[k]:v for k,v in fixes.items()}
 
 # %% ../nbs/01_refine.ipynb 28
 def apply_hdg_fixes(
@@ -197,8 +170,8 @@ async def describe_imgs(
     imgs: list[Path], # List of image file paths to describe
     model: str = 'claude-sonnet-4-5', # Model to use for image description
     prompt: str = describe_img_prompt, # Prompt template for description
-    semaphore: int = 2, # Max concurrent API requests
-    delay: float = 1 # Delay in seconds between requests
+    semaphore: int = 10, # Max concurrent API requests
+    delay: float = 0.1 # Delay in seconds between requests
 ) -> dict[str, dict]: # Dict mapping filename to parsed description
     "Describe multiple images in parallel with rate limiting"
     sem = Semaphore(semaphore)

{mistocr-0.2.1.dist-info → mistocr-0.2.7.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mistocr
-Version: 0.2.1
+Version: 0.2.7
 Summary: Batch OCR for PDFs with heading restoration and visual content integration
 Home-page: https://github.com/franckalbinet/mistocr
 Author: Solveit
@@ -38,7 +38,7 @@ Dynamic: requires-dist
 Dynamic: requires-python
 Dynamic: summary
 
-# mistocr
+# Mistocr
 
 
 <!-- WARNING: THIS FILE WAS AUTOGENERATED! DO NOT EDIT! -->
@@ -69,12 +69,20 @@ fundamental challenges that raw OCR output leaves unsolved:
   markdown. This makes visual information searchable and accessible for
   downstream applications.
 
-- **Cost-efficient batch processing**: By exclusively using Mistral’s
-  batch API, mistocr cuts costs by 50% (\$0.50 vs \$1.00 per 1000 pages)
-  while eliminating the boilerplate code typically required.
+- **Cost-efficient batch processing**: The OCR step exclusively uses
+  Mistral’s batch API, cutting costs by 50% (\$0.50 vs \$1.00 per 1000
+  pages) while eliminating the boilerplate code typically required.
 
-**In short**: Production-ready batch OCR with intelligent postprocessing
-that ensures your documents are actually usable for AI systems.
+**In short**: Complete PDF OCR with heading hierarchy fixes and image
+descriptions for RAG and LLM pipelines.
+
+> [!NOTE]
+>
+> **Want to see mistocr in action?** This
+> [tutorial](https://share.solve.it.com/d/97f75412ca949af76a5945b4dfc443c7)
+> demonstrates real-world PDF processing and shows how clean markdown
+> enables structure-aware navigation through long documents—letting you
+> find exactly what you need, fast.
 
 ## Get Started
 
@@ -94,7 +102,10 @@ os.environ['ANTHROPIC_API_KEY'] = 'your-key-here'  # for refine features (see Ad
 
 ### Complete Pipeline
 
-Full pipeline with all features:
+#### Single File Processing
+
+Process a single PDF with OCR (using Mistral’s batch API for cost
+efficiency), heading fixes, and image descriptions:
 
 ``` python
 from mistocr.pipeline import pdf_to_md
@@ -171,12 +182,12 @@ instead.
 
 ### Advanced Usage
 
-**Batch process entire folders:**
+**Batch OCR for entire folders:**
 
 ``` python
 from mistocr.core import ocr_pdf
 
-# Process all PDFs in a folder
+# OCR all PDFs in a folder using Mistral's batch API
 output_dirs = ocr_pdf('path/to/pdf_folder', dst='output_folder')
 ```
 

mistocr-0.2.7.dist-info/RECORD

@@ -0,0 +1,11 @@
+mistocr/__init__.py,sha256=XHypfHSPdgXFKmOdoewn7czU670gt8InhHhzlP5j_aA,22
+mistocr/_modidx.py,sha256=55k_EmMPrRmLa04qCE-_8lukJPCax66q5MALfpa8T4A,4154
+mistocr/core.py,sha256=KH5ND6ZMmETj9m9A-Ivtw5N2L-qm9vGwYgvHAzg-BsA,7218
+mistocr/pipeline.py,sha256=k5rtr9ccxTgwQAzw533iMIL9qwiCofugWv0fhotLzaI,1639
+mistocr/refine.py,sha256=zSCF0gOtEKhhQTQgVq4Jh5Ujk8l8CGSO_rURhsQ09P8,10351
+mistocr-0.2.7.dist-info/licenses/LICENSE,sha256=xV8xoN4VOL0uw9X8RSs2IMuD_Ss_a9yAbtGNeBWZwnw,11337
+mistocr-0.2.7.dist-info/METADATA,sha256=eyQ65s8HsoHUUINrGiijrC8e0RzO_Wvte3rk2OLU8QY,8416
+mistocr-0.2.7.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+mistocr-0.2.7.dist-info/entry_points.txt,sha256=HjAiHozobM-alm_6bTF-ehRr2DD3KYE9PgRngelONOY,36
+mistocr-0.2.7.dist-info/top_level.txt,sha256=LelTYnSpSXLy1Hb1m2YP3gt8luwP-I8KV0NjP_ucdSs,8
+mistocr-0.2.7.dist-info/RECORD,,

mistocr-0.2.1.dist-info/RECORD

@@ -1,11 +0,0 @@
-mistocr/__init__.py,sha256=HfjVOrpTnmZ-xVFCYSVmX50EXaBQeJteUHG-PD6iQs8,22
-mistocr/_modidx.py,sha256=JtXJb6ErVwyGFCWgeVz7N9JvybBp8Zyao4Ahn3smr90,4030
-mistocr/core.py,sha256=KH5ND6ZMmETj9m9A-Ivtw5N2L-qm9vGwYgvHAzg-BsA,7218
-mistocr/pipeline.py,sha256=k5rtr9ccxTgwQAzw533iMIL9qwiCofugWv0fhotLzaI,1639
-mistocr/refine.py,sha256=zsPoxWD63bk1rzRVO9OPsevWeMNORHgT_y8H7T7CxYs,11785
-mistocr-0.2.1.dist-info/licenses/LICENSE,sha256=xV8xoN4VOL0uw9X8RSs2IMuD_Ss_a9yAbtGNeBWZwnw,11337
-mistocr-0.2.1.dist-info/METADATA,sha256=-y9Ze92RygrKGCfHbBjlGXlv-5iRYVAOyHtC9MHnplw,7990
-mistocr-0.2.1.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-mistocr-0.2.1.dist-info/entry_points.txt,sha256=HjAiHozobM-alm_6bTF-ehRr2DD3KYE9PgRngelONOY,36
-mistocr-0.2.1.dist-info/top_level.txt,sha256=LelTYnSpSXLy1Hb1m2YP3gt8luwP-I8KV0NjP_ucdSs,8
-mistocr-0.2.1.dist-info/RECORD,,