mistocr 0.2.10__py3-none-any.whl → 0.2.12__py3-none-any.whl

mistocr/__init__.py

@@ -1 +1 @@
-__version__ = "0.2.10"
+__version__ = "0.2.12"

mistocr/_modidx.py

@@ -22,6 +22,7 @@ d = { 'settings': { 'branch': 'main',
                               '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.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/core.py

@@ -79,11 +79,17 @@ def submit_batch(
 def wait_for_job(
     job:dict, # Job dict, 
     c:Mistral=None, # Mistral client, 
-    poll_interval:int=1 # Poll interval in seconds
+    poll_interval:int=1, # Poll interval in seconds
+    queued_timeout:int=300 # Timeout for QUEUED status in seconds
     ) -> dict: # Job dict (with status)  
     "Poll job until completion and return final job status"
+    queued_time = 0
     while job.status in ["QUEUED", "RUNNING"]:
         print(f'Mistral batch job status: {job.status}')
+        if job.status == "QUEUED":
+            queued_time += poll_interval
+            if queued_time >= queued_timeout:
+                raise TimeoutError(f"Job stayed in QUEUED status for {queued_time}s, exceeding timeout of {queued_timeout}s. Check your balance or Mistral Status.")
         time.sleep(poll_interval)
         job = c.batch.jobs.get(job_id=job.id)
     return job

mistocr/refine.py

@@ -4,9 +4,9 @@
 
 # %% auto 0
 __all__ = ['prompt_fix_hdgs', 'describe_img_prompt', 'get_hdgs', 'add_pg_hdgs', 'read_pgs_pg', 'fmt_hdgs_idx',
-           'HeadingCorrection', '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']
+           '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']
 
 # %% ../nbs/01_refine.ipynb 3
 from fastcore.all import *
@@ -64,6 +64,11 @@ class HeadingCorrection(BaseModel):
     index: int
     corrected: str
 
+# %% ../nbs/01_refine.ipynb 19
+class HeadingCorrections(BaseModel):
+    "Collection of heading corrections returned by the LLM"
+    corrections: list[HeadingCorrection]
+
 # %% ../nbs/01_refine.ipynb 21
 prompt_fix_hdgs = """Fix markdown heading hierarchy errors while preserving the document's intended structure.
 

{mistocr-0.2.10.dist-info → mistocr-0.2.12.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mistocr
-Version: 0.2.10
+Version: 0.2.12
 Summary: Batch OCR for PDFs with heading restoration and visual content integration
 Home-page: https://github.com/franckalbinet/mistocr
 Author: Solveit

mistocr-0.2.12.dist-info/RECORD

@@ -0,0 +1,11 @@
+mistocr/__init__.py,sha256=X4KG3FscE5AhbGbcdDDgdDC550CVpxNMwdNLcx6EQ7M,23
+mistocr/_modidx.py,sha256=55k_EmMPrRmLa04qCE-_8lukJPCax66q5MALfpa8T4A,4154
+mistocr/core.py,sha256=ohh2ru05gUKbIQCRHPMz_hw4ui39FtpoV3_S3n4bl_c,7592
+mistocr/pipeline.py,sha256=k5rtr9ccxTgwQAzw533iMIL9qwiCofugWv0fhotLzaI,1639
+mistocr/refine.py,sha256=hQg4ZYRTkPFEiCNNBU2ykgXFfVLCNAp2IuwOwHBVQ2k,12983
+mistocr-0.2.12.dist-info/licenses/LICENSE,sha256=xV8xoN4VOL0uw9X8RSs2IMuD_Ss_a9yAbtGNeBWZwnw,11337
+mistocr-0.2.12.dist-info/METADATA,sha256=MB_1D5vpqGUIA_jl-jNHj7E5LHJol6YGU8lw-ZCFj_4,8417
+mistocr-0.2.12.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+mistocr-0.2.12.dist-info/entry_points.txt,sha256=HjAiHozobM-alm_6bTF-ehRr2DD3KYE9PgRngelONOY,36
+mistocr-0.2.12.dist-info/top_level.txt,sha256=LelTYnSpSXLy1Hb1m2YP3gt8luwP-I8KV0NjP_ucdSs,8
+mistocr-0.2.12.dist-info/RECORD,,

mistocr-0.2.10.dist-info/RECORD

@@ -1,11 +0,0 @@
-mistocr/__init__.py,sha256=waXgc7p-jgGCsUjdVfO_KjlVZblnCvrzf4A0dsBj_lg,23
-mistocr/_modidx.py,sha256=WTS9JpZdbrp2LghjhOV-CK0JYChHE4PzttgKfh7pTy4,4028
-mistocr/core.py,sha256=KH5ND6ZMmETj9m9A-Ivtw5N2L-qm9vGwYgvHAzg-BsA,7218
-mistocr/pipeline.py,sha256=k5rtr9ccxTgwQAzw533iMIL9qwiCofugWv0fhotLzaI,1639
-mistocr/refine.py,sha256=H_IAF02k6CwBQdDJm9txknzUcTlz245zXitaHELX-P4,12791
-mistocr-0.2.10.dist-info/licenses/LICENSE,sha256=xV8xoN4VOL0uw9X8RSs2IMuD_Ss_a9yAbtGNeBWZwnw,11337
-mistocr-0.2.10.dist-info/METADATA,sha256=mkMu_9nYAXZ5jFdJd01AZqK3t93_Rt0xkkD0rRnl9Ew,8417
-mistocr-0.2.10.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-mistocr-0.2.10.dist-info/entry_points.txt,sha256=HjAiHozobM-alm_6bTF-ehRr2DD3KYE9PgRngelONOY,36
-mistocr-0.2.10.dist-info/top_level.txt,sha256=LelTYnSpSXLy1Hb1m2YP3gt8luwP-I8KV0NjP_ucdSs,8
-mistocr-0.2.10.dist-info/RECORD,,