mistocr 0.2.5__tar.gz → 0.2.7__tar.gz

{mistocr-0.2.5/mistocr.egg-info → mistocr-0.2.7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mistocr
-Version: 0.2.5
+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

mistocr-0.2.7/mistocr/__init__.py

@@ -0,0 +1 @@
+__version__ = "0.2.7"

{mistocr-0.2.5 → mistocr-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-0.2.5 → mistocr-0.2.7}/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(

{mistocr-0.2.5 → mistocr-0.2.7/mistocr.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mistocr
-Version: 0.2.5
+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

{mistocr-0.2.5 → mistocr-0.2.7}/settings.ini

@@ -1,7 +1,7 @@
 [DEFAULT]
 repo = mistocr
 lib_name = mistocr
-version = 0.2.5
+version = 0.2.7
 min_python = 3.9
 license = apache2
 black_formatting = False

mistocr-0.2.5/mistocr/__init__.py

@@ -1 +0,0 @@
-__version__ = "0.2.5"