natural-pdf 0.1.33__py3-none-any.whl → 0.1.35__py3-none-any.whl

tools/bad_pdf_eval/llm_enrich_with_retry.py

@@ -0,0 +1,289 @@
+"""Enhanced LLM enrichment with automatic retry for low-scoring suggestions.
+
+Usage
+-----
+python -m tools.bad_pdf_eval.llm_enrich_with_retry --submission ja6EqV1 --model gpt-4o
+
+Environment
+-----------
+OPENAI_API_KEY must be set or passed via --api-key.
+"""
+
+import argparse
+import concurrent.futures as _futures
+import json
+import os
+from pathlib import Path
+from typing import Any, Dict, Iterable, List
+
+from openai import OpenAI
+from pydantic import BaseModel, Field
+
+# Import quality evaluation
+from tools.bad_pdf_eval.evaluate_quality import analyze_code_quality
+
+# Import existing functions and classes
+from tools.bad_pdf_eval.llm_enrich import (
+    CHEATSHEET_PATH,
+    DECISION_TREE_PATH,
+    EVAL_DIR,
+    WORKFLOWS_PATH,
+    DocOutput,
+    build_pdf_prompt,
+    build_prompt,
+    img_to_b64_jpeg,
+    read_md,
+)
+
+# Global variable
+FORCE = False
+
+
+class RetryOutput(BaseModel):
+    """Improved version after feedback."""
+
+    thought_process: str = Field(
+        ..., description="Revised reasoning addressing the specific feedback"
+    )
+    code_suggestion: str = Field(
+        ..., description="Improved Python snippet addressing all feedback points"
+    )
+
+
+def build_retry_prompt(
+    original_code: str, quality_analysis: Dict[str, Any], context: Dict[str, Any]
+) -> List[Dict[str, Any]]:
+    """Build a retry prompt with specific feedback."""
+
+    feedback_points = []
+
+    # Build specific feedback based on what's missing
+    if not quality_analysis["uses_guides"] and "table" in context.get("goal", "").lower():
+        feedback_points.append(
+            "• Your code doesn't use the Guides API for table extraction. "
+            "Use `Guides.from_content()` with actual column headers from the PDF, "
+            "then `snap_to_whitespace()` for better results."
+        )
+
+    if not quality_analysis["uses_real_text"]:
+        feedback_points.append(
+            "• Use actual text from the inspect/describe data as anchors. "
+            "Look for real headers, labels, or unique text in the evidence."
+        )
+
+    if not quality_analysis["uses_until"]:
+        feedback_points.append(
+            "• Use the `until=` parameter in `.below()` or `.above()` calls "
+            "to define region boundaries based on content, not pixels."
+        )
+
+    if quality_analysis["uses_tatr"] and quality_analysis["score"] < 6:
+        feedback_points.append(
+            "• Consider if TATR is really necessary. Can you use Guides or "
+            "direct region extraction instead?"
+        )
+
+    if not quality_analysis["uses_snap_to_whitespace"] and quality_analysis["uses_guides"]:
+        feedback_points.append(
+            "• Add `.snap_to_whitespace()` after creating guides to auto-align "
+            "to natural gaps in the content."
+        )
+
+    retry_prompt = f"""
+Your previous code suggestion scored {quality_analysis['score']}/12 in our quality evaluation.
+Here's specific feedback to improve it:
+
+{chr(10).join(feedback_points)}
+
+Original code:
+```python
+{original_code}
+```
+
+Please provide an improved version that addresses all the feedback points.
+Focus on using modern Natural PDF features and patterns.
+"""
+
+    messages = [
+        {
+            "role": "system",
+            "content": "You are a Natural PDF expert. Improve the code based on specific feedback.",
+        },
+        {"role": "user", "content": retry_prompt},
+    ]
+
+    return messages
+
+
+def enrich_with_retry(
+    summary_path: Path,
+    api_key: str,
+    model: str = "gpt-4o",
+    retry_threshold: int = 6,
+    max_retries: int = 2,
+):
+    """Enrich with automatic retry for low-quality suggestions.
+
+    The function will keep *all* attempts (initial + retries) in `attempts` list so we can
+    analyse which feedback helped.  The highest-scoring version becomes the primary
+    `thought_process` / `code_suggestion` stored at the root level.
+    """
+
+    summary = json.loads(summary_path.read_text())
+
+    # Skip if already enriched (unless forced)
+    if (
+        not FORCE
+        and summary.get("thought_process")
+        and summary.get("code_suggestion")
+        and summary.get("attempts")
+    ):
+        print(f"[skip] {summary_path.parent.name}: already enriched with attempts")
+        return
+
+    print(f"[send] {summary_path.parent.name}: requesting initial enrichment")
+
+    client = OpenAI(api_key=api_key)
+    msgs = build_pdf_prompt(summary)
+
+    attempts: List[Dict[str, Any]] = []  # keep all versions
+
+    # Initial attempt
+    completion = client.beta.chat.completions.parse(
+        model=model, messages=msgs, response_format=DocOutput
+    )
+
+    doc_out = completion.choices[0].message.parsed
+    quality = analyze_code_quality(doc_out.code_suggestion)
+    print(f"Initial quality score: {quality['score']}/12")
+    attempts.append(
+        {
+            "attempt": 0,
+            "score": quality["score"],
+            "thought_process": doc_out.thought_process,
+            "code_suggestion": doc_out.code_suggestion,
+        }
+    )
+
+    best_doc = doc_out
+    best_score = quality["score"]
+
+    # Retry loop
+    retry_count = 0
+    while quality["score"] < retry_threshold and retry_count < max_retries:
+        retry_count += 1
+        print(f"[retry {retry_count}] Score below threshold, requesting improvement...")
+
+        retry_msgs = build_retry_prompt(
+            doc_out.code_suggestion, quality, {"goal": summary.get("goal", "")}
+        )
+
+        retry_completion = client.beta.chat.completions.parse(
+            model=model, messages=retry_msgs, response_format=RetryOutput
+        )
+        retry_out = retry_completion.choices[0].message.parsed
+
+        # Evaluate new version
+        new_quality = analyze_code_quality(retry_out.code_suggestion)
+        print(f"Retry {retry_count} quality score: {new_quality['score']}/12")
+
+        # Record attempt details
+        attempts.append(
+            {
+                "attempt": retry_count,
+                "score": new_quality["score"],
+                "thought_process": retry_out.thought_process,
+                "code_suggestion": retry_out.code_suggestion,
+            }
+        )
+
+        # Update best if improved
+        if new_quality["score"] > best_score:
+            best_score = new_quality["score"]
+            best_doc.thought_process = retry_out.thought_process
+            best_doc.code_suggestion = retry_out.code_suggestion
+
+        # Prepare for next iteration
+        quality = new_quality
+        doc_out.code_suggestion = retry_out.code_suggestion
+        doc_out.thought_process = retry_out.thought_process
+
+    # Save results – keep best version at root, all attempts nested
+    summary["thought_process"] = best_doc.thought_process
+    summary["code_suggestion"] = best_doc.code_suggestion
+    summary["difficult_elements"] = getattr(
+        best_doc, "difficult_elements", summary.get("difficult_elements")
+    )
+    summary["test_case"] = getattr(best_doc, "test_case", summary.get("test_case"))
+    summary["quality_score"] = best_score
+    summary["retry_count"] = retry_count
+    summary["attempts"] = attempts  # new key for analysis
+
+    summary_path.write_text(json.dumps(summary, indent=2, ensure_ascii=False))
+    print(
+        f"[update] Best score: {best_score}/12 after {retry_count} retries (kept all {len(attempts)} attempts)"
+    )
+
+
+def main():
+    ap = argparse.ArgumentParser()
+    ap.add_argument("--submission", help="Submission ID to enrich")
+    ap.add_argument("--model", default="gpt-4o")
+    ap.add_argument("--api-key", default=os.getenv("OPENAI_API_KEY"))
+    ap.add_argument("--force", action="store_true")
+    ap.add_argument(
+        "--retry-threshold", type=int, default=6, help="Minimum quality score before retry"
+    )
+    ap.add_argument("--max-retries", type=int, default=2, help="Maximum number of retry attempts")
+    ap.add_argument(
+        "--workers",
+        type=int,
+        default=1,
+        help="Number of parallel workers (use 1 to disable parallelism)",
+    )
+    args = ap.parse_args()
+
+    if not args.api_key:
+        raise SystemExit("OPENAI_API_KEY not provided")
+
+    global FORCE
+    FORCE = args.force
+
+    def _get_paths() -> Iterable[Path]:
+        if args.submission:
+            p = EVAL_DIR / args.submission / "summary.json"
+            if not p.exists():
+                raise SystemExit("summary.json not found")
+            return [p]
+        else:
+            return list(EVAL_DIR.glob("*/summary.json"))
+
+    paths = _get_paths()
+
+    if args.workers <= 1:
+        # Sequential processing
+        for p in paths:
+            try:
+                enrich_with_retry(
+                    p, args.api_key, args.model, args.retry_threshold, args.max_retries
+                )
+            except Exception as e:
+                print(f"[error] {p.parent.name}: {e}")
+    else:
+        # Parallel processing with thread pool (IO-bound)
+        print(f"Running with {args.workers} parallel workers…")
+
+        def _safe_process(p: Path):
+            try:
+                enrich_with_retry(
+                    p, args.api_key, args.model, args.retry_threshold, args.max_retries
+                )
+            except Exception as exc:
+                print(f"[error] {p.parent.name}: {exc}")
+
+        with _futures.ThreadPoolExecutor(max_workers=args.workers) as ex:
+            list(ex.map(_safe_process, paths))
+
+
+if __name__ == "__main__":
+    main()

tools/bad_pdf_eval/reporter.py

@@ -14,4 +14,4 @@ def save_json(data: Dict[str, Any], path: Path):
 
 
 def log_section(title: str):
-    console.rule(f"[bold cyan]{title}") 
+    console.rule(f"[bold cyan]{title}")

tools/bad_pdf_eval/utils.py

@@ -1,8 +1,9 @@
 import re
+import ssl
+import urllib.request
 from pathlib import Path
 from typing import Optional
-import urllib.request
-import ssl
+
 from rich.console import Console
 
 ROOT_DIR = Path(__file__).resolve().parent.parent.parent  # project root
@@ -103,7 +104,9 @@ def find_local_pdf(submission_id: str, pdf_url: Optional[str] = None) -> Optiona
                     req = urllib.request.Request(pdf_url, headers={"User-Agent": "Mozilla/5.0"})
                     # Disable SSL verification edge-cases the storage host sometimes triggers
                     ctx = ssl.create_default_context()
-                    with urllib.request.urlopen(req, context=ctx, timeout=30) as resp, open(dest_path, "wb") as f:
+                    with urllib.request.urlopen(req, context=ctx, timeout=30) as resp, open(
+                        dest_path, "wb"
+                    ) as f:
                         f.write(resp.read())
                 except Exception as e:
                     # Fallback: try requests if available (venv usually has it)
@@ -124,4 +127,4 @@ def find_local_pdf(submission_id: str, pdf_url: Optional[str] = None) -> Optiona
             return None
 
     # None found
-    return None 
+    return None

natural_pdf-0.1.33.dist-info/RECORD

@@ -1,118 +0,0 @@
-natural_pdf/__init__.py,sha256=qDFJNF8sbEDO-2WSFAxoWEM8updOUP6dB-ckya0kxfs,3275
-natural_pdf/cli.py,sha256=IXrP2lCHihr-ed-CFiDbMTnSsutQa1j1PYALOLGbpsc,4019
-natural_pdf/analyzers/__init__.py,sha256=MQRctn4i5Q7u8pb8vQVHKEXUiVGpKyPZUECrlDH4AuU,673
-natural_pdf/analyzers/guides.py,sha256=tzyViSBDdM66mT0niwFTDIJ16UzRCZ18Iqv8wA5DYAk,90302
-natural_pdf/analyzers/shape_detection_mixin.py,sha256=q7gDM-z2t7bSTxjfV2aaW3533CySu1qsEpu4wb5Rp-I,62688
-natural_pdf/analyzers/text_options.py,sha256=qEkDaYWla0rIM_gszEOsu52q7C_dAfV81P2HLJZM2sw,3333
-natural_pdf/analyzers/text_structure.py,sha256=VfKTsTFrK877sC0grsis9jK3rrgp0Mbp13VWEbukTcs,28437
-natural_pdf/analyzers/utils.py,sha256=PYbzJzSAHZ7JsMes84WIrSbA0zkjJGs0CLvIeINsf_k,2100
-natural_pdf/analyzers/layout/__init__.py,sha256=oq1uJ5UkGGMbBKGirV1aRKK3hxAUyjTLywYkPCQH1f0,33
-natural_pdf/analyzers/layout/base.py,sha256=bYawhmc_0xqKG-xbxUSiazIU1om-aBox5Jh8qDqv-eM,6451
-natural_pdf/analyzers/layout/docling.py,sha256=4BJYyNVR6VegZGxyisvNIBBRvVk6YKPyDVs7ZdVfzEU,12676
-natural_pdf/analyzers/layout/gemini.py,sha256=ldECVCQ5HNQA3Omjg2NOsTrJXslyYb0vErDncmLIiuE,10510
-natural_pdf/analyzers/layout/layout_analyzer.py,sha256=1v23FVCIGzkoiyRqiLZBwGZssBFKphtMossMENMuMxE,15519
-natural_pdf/analyzers/layout/layout_manager.py,sha256=vDXBAaNwvp68CRcEPH58MGLxx01OdVgzOh7Uv53L6fs,10319
-natural_pdf/analyzers/layout/layout_options.py,sha256=-Nv6bcu4_pqSCN6uNhCZ9mvoCBtRDZIUkO6kjkuLXsg,7703
-natural_pdf/analyzers/layout/paddle.py,sha256=tX2bI1yayAdmRhvsfZ_Ygs7zAG5e9eW-pLJkw4NUpBQ,21325
-natural_pdf/analyzers/layout/pdfplumber_table_finder.py,sha256=Tk0Q7wv7nGYPo69lh6RoezjdepTnMl90SaNIrP29Pwc,5902
-natural_pdf/analyzers/layout/surya.py,sha256=ugRXPIHiLoh65lfbbiXO317TbgdtQ-5kVN1nonEf4ws,9778
-natural_pdf/analyzers/layout/table_structure_utils.py,sha256=nISZDBd46RPYkFHxbQyIHwg9WweG4DslpoYJ31OMJYA,2768
-natural_pdf/analyzers/layout/tatr.py,sha256=cVr0ZyhY2mNLAKZ4DGMm-b7XNJpILKh8x8ZpyDeUhLk,15032
-natural_pdf/analyzers/layout/yolo.py,sha256=ruchj28sxar0DWDALwUz1j30z0CLIEp2QAs0gLVvC4E,8346
-natural_pdf/classification/manager.py,sha256=pOP2LvJpTBGItvdIODnk735DXq7F2qqxN4AKmBORM3c,21775
-natural_pdf/classification/mixin.py,sha256=nYpmHQ4BlrealdPtIJt-_idME5o-xKLKNuAdIHzWL6c,7580
-natural_pdf/classification/results.py,sha256=Mcay-xLBHbYoZ8U7f4gMj2IhhH_yORNEkZHWdWJzmsU,3239
-natural_pdf/collections/mixins.py,sha256=sj76Cn6EdBtb5f-bdAV-1qpdixX8tI4BzPccPiYLI1w,5117
-natural_pdf/collections/pdf_collection.py,sha256=HLlyakM--23ZOeHDPucoM6Tw3yUyMXm0SSoqJwxRc2E,30744
-natural_pdf/core/__init__.py,sha256=QC8H4M3KbXwMFiQORZ0pdPlzx1Ix6oKKQSS7Ib2KEaA,38
-natural_pdf/core/element_manager.py,sha256=DbRzAKD3to5NpKc73Q-TXZIZkhx8zZtbi_UNu5K7AAU,52766
-natural_pdf/core/highlighting_service.py,sha256=WKDqRpex1yS8CWhkNitWtKhxbyRRCLu3Xsct_HTPsD4,40774
-natural_pdf/core/page.py,sha256=k4jezvsLqL07Raglc-rZmMnsVwBMo_A_OerklpBIejY,129477
-natural_pdf/core/pdf.py,sha256=u0ZCPuIijNecU-AJHLvqfAYVCr9h7MgUKnlEtH6RoZI,75969
-natural_pdf/describe/__init__.py,sha256=B3zjuHjFI_dFuBLgXR1Q4v7c72fVDyk84d2hs0H4KV8,561
-natural_pdf/describe/base.py,sha256=HaWlHltb-dw6ug4mfR_iBLHWxr1OdPwLaUshXRxO7gg,18462
-natural_pdf/describe/elements.py,sha256=COvKF3B_RbAxXl5ORJDubV4C5PsiuSfuzD0ufPIJTFM,12983
-natural_pdf/describe/mixin.py,sha256=U0x6v8r57KQb8qC3VVo64hvhfXQWsti8vdKBM7AXnMo,3116
-natural_pdf/describe/summary.py,sha256=7FIF3zF6bzNx-gx4pCJr2XQFKiVzOEDnWsAYQ_mr9L0,7982
-natural_pdf/elements/__init__.py,sha256=S8XeiNWJ1WcgnyYKdYV1yxQlAxCCO3FfITT8MQwNbyk,41
-natural_pdf/elements/base.py,sha256=VshU4RstdzONJFq_8UVIjT_lVOai0MwMFsSFrCN-IO8,47299
-natural_pdf/elements/collections.py,sha256=1E2MSg2NNcEcoRM2rumrv_CqIdO7DgbRHYEtfw35FaQ,128457
-natural_pdf/elements/image.py,sha256=UjHNzCgDzOseQmLpkKshcxg51DPmWNIAVYxZ0TAMyUI,1423
-natural_pdf/elements/line.py,sha256=aQm4pDdlQSDAAXqrdg4AU-oTl9JCXgYuaJN0EYls6E0,4920
-natural_pdf/elements/rect.py,sha256=kiVa3e377ZnqIOXc89d9ZSY4EcmDxtccdtUw-HOQzpw,3796
-natural_pdf/elements/region.py,sha256=23J5Tv7ffAgz3IBgDXPq9Ab_lLg2Sog7elFRb6nvvZE,140541
-natural_pdf/elements/text.py,sha256=kw7u2KfHtDB905YawP7Hs89kcR8XnbtpkYQGEk6LNyk,18860
-natural_pdf/export/mixin.py,sha256=L1q3MIEFWuvie4j4_EmW7GT3NerbZ1as0XMUoqTS7gM,5083
-natural_pdf/exporters/__init__.py,sha256=g1WRPCDVzceaUUsm8dchPhzdHFSjYM0NfRyc8iN0mtE,644
-natural_pdf/exporters/base.py,sha256=XhR1xlkHOh7suOuX7mWbsj1h2o1pZNet-OAS5YCJyeI,2115
-natural_pdf/exporters/hocr.py,sha256=wksvJvWLSxuAfhYzg_0T2_W8eqDoMgAVC-gwZ9FoO_k,19969
-natural_pdf/exporters/hocr_font.py,sha256=1wsGOMj6zoaRN2rxCwrv4MMLGawpNz984WgXpmWekgw,4574
-natural_pdf/exporters/original_pdf.py,sha256=dtvC4er6TWOfqq-n24Pejw3mlAuPd8IVyihggJtcf0s,6634
-natural_pdf/exporters/paddleocr.py,sha256=RBP03GCk0mLeC7tWtuti8AIUHlpOrtvbWkE2n7Ja7k8,19484
-natural_pdf/exporters/searchable_pdf.py,sha256=G2Tc4tpDXSYIufXJlkA8ppW_3DuzHAaweYKae33pI_c,16290
-natural_pdf/exporters/data/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-natural_pdf/exporters/data/pdf.ttf,sha256=x4RUIJJaI9iO2DCmOVe4r4Wmao2vjZ_JDoQ2c7LvGlk,572
-natural_pdf/exporters/data/sRGB.icc,sha256=KpLUuuRQt22LCqQhk9-XTXX2Jzjs6_dPAcXnWxKpV5Y,6922
-natural_pdf/extraction/manager.py,sha256=sASPJZ5cWFsl8A4PyTjg2yqkyC00tRl6glfoFA6HcsM,4979
-natural_pdf/extraction/mixin.py,sha256=_5wGnzOCEuRWhqdSUV1Lqo9HIi56YC4MWzbBxOkOEKU,23160
-natural_pdf/extraction/result.py,sha256=D5DhjxLW7IvhEkvsAP7Zs2YA8K4hyuoTg681CSn5qA0,1825
-natural_pdf/flows/__init__.py,sha256=cUN4A8hTDLZSRr4PO2W_lR4z6hWpbNG8Seox-IIcrLU,277
-natural_pdf/flows/collections.py,sha256=iF8SsfKKb-YVIGi3m-yMRnfKgo_0n_EGhojnYK24h-Q,28493
-natural_pdf/flows/element.py,sha256=mKzk3B7A7sWNvu4CDvAjLr3_ZFLt--ktrSNoLfLpFxU,23940
-natural_pdf/flows/flow.py,sha256=I61BpFVDQyo6ORsmoqoYiOEP1DBRp0vgDJjm_V8frhc,10562
-natural_pdf/flows/region.py,sha256=s_YAT_0KsrwUs73hhU9xr_35Ufr__XNhRjHSQkxcfYU,27647
-natural_pdf/ocr/__init__.py,sha256=VY8hhvDPf7Gh2lB-d2QRmghLLyTy6ydxlgo1cS4dOSk,2482
-natural_pdf/ocr/engine.py,sha256=ZBC1tZNM5EDbGDJJmZI9mNHr4nCMLEZvUFhiJq8GdF4,8741
-natural_pdf/ocr/engine_doctr.py,sha256=ptKrupMWoulZb-R93zr9btoe94JPWU7vlJuN7OBJEIM,17740
-natural_pdf/ocr/engine_easyocr.py,sha256=bWz6kHUgAJfe3rqdnZBAF-IPvw3B35DlvX5KDdFUtzo,9888
-natural_pdf/ocr/engine_paddle.py,sha256=9tQZl1VqN6d_KEWUY_S9tfrDLiR4FCHMjgSRNwPlsu8,16152
-natural_pdf/ocr/engine_surya.py,sha256=lOvSbZk53VKFVxRmqcQzM_0dHVdwTkRGiDZ9AWCgL1Q,5951
-natural_pdf/ocr/ocr_factory.py,sha256=qjGL3hm_nTzxjwYWP0JE7dCFXZjKN8Z7f9c0oqasb9M,5262
-natural_pdf/ocr/ocr_manager.py,sha256=jFJI8v3coapKpERoUlP-ptwguZG_Dl4VlclD0xQ6Us8,16192
-natural_pdf/ocr/ocr_options.py,sha256=l33QKu_93r-uwi3t_v8UH8pEgHo6HTVzP4tfmQFRF1w,5488
-natural_pdf/ocr/utils.py,sha256=OxuHwDbHWj6setvnC0QYwMHrAjxGkhmLzWHpMqqGupA,4397
-natural_pdf/qa/__init__.py,sha256=2u2KJcA71g1I0HnLD-j6yvDw1moAjo9kkLhhfoYRURM,166
-natural_pdf/qa/document_qa.py,sha256=cli1E9NBSVtT5Qo6n7ZRd7BpstnbpZfkljX69LGTYU8,19608
-natural_pdf/qa/qa_result.py,sha256=_q4dlSqsjtgomcI8-pqbOT69lqQKnEMkhZNydoxEkkE,2227
-natural_pdf/search/__init__.py,sha256=0Xa7tT_2q57wHObFMQLQLd4gd9AV0oyS-svV6BmmdMI,4276
-natural_pdf/search/lancedb_search_service.py,sha256=6dz2IEZUWk3hFW28C-LF_85pWohd7Sr5k44bM0pBdm4,14472
-natural_pdf/search/numpy_search_service.py,sha256=MoPBlyHTDqah1IrwBzyglEyiXlF4wqaU_5mml_ngvGc,10328
-natural_pdf/search/search_options.py,sha256=sq_e8_jSROicD94b_xtDtLnjEr_Zsy4icjzPkK0a8QA,3566
-natural_pdf/search/search_service_protocol.py,sha256=Dl-Q-CrutkhZwI69scbW9EWPeYM63qxB60_EA7YqIYo,6699
-natural_pdf/search/searchable_mixin.py,sha256=hqQ_AuID5eTGRCtKYdFLZ1zF35y73uk3x1M1VW9Il8U,23514
-natural_pdf/selectors/__init__.py,sha256=oZGeqSv53EqmIZOhcnawuaGGlRg1h79vArXuZCWKm4A,123
-natural_pdf/selectors/parser.py,sha256=T9r7XZhM1cGSYQrc9amUHbFtX-zBqd9_YPK0scwCjAQ,34231
-natural_pdf/tables/__init__.py,sha256=y65LM2wnu81yzvOX-J_5NXiIK4vEUtHa3EM1xv-0ttQ,105
-natural_pdf/tables/result.py,sha256=OYc-MjnP-VRTVaY-pBt84E-d8N3AaqzwAud0hHt5sVY,3979
-natural_pdf/templates/__init__.py,sha256=jYBxzfi73vew0f6yhIh1MlRxw4F_TVN2hKQR0YXOFe0,20
-natural_pdf/utils/__init__.py,sha256=s3M8FggaK1P3EBYn6R_-HgSDjNc9C73gyKe1hihtNWg,43
-natural_pdf/utils/bidi_mirror.py,sha256=SAe5SnL-xG5Wyo3LtkMttLdsnQqZhzAebLc7BAe6LhQ,1150
-natural_pdf/utils/debug.py,sha256=RN7H3E6ph-GtxubCW6psW7TO8o2BxcNLiEzByTVR9fk,995
-natural_pdf/utils/highlighting.py,sha256=EIY6ihVGtUTS_DjWyxpnr_UXpcR4btC1KhSGQ9VUfKg,698
-natural_pdf/utils/identifiers.py,sha256=P7n6owcubnF8oAMa_UfYtENmIaJQdH_AMC9Jbs2bWXo,1117
-natural_pdf/utils/locks.py,sha256=7HJqV0VsNcOfISnbw8goCKWP5ck11uSJo6T_x9XIPKI,215
-natural_pdf/utils/packaging.py,sha256=e7U2wWvpunlAWpPFexNkD_c4dYbPp5LcKo7og4bNGvk,22411
-natural_pdf/utils/reading_order.py,sha256=s3DsYq_3g_1YA07qhd4BGEjeIRTeyGtnwc_hNtSzwBY,7290
-natural_pdf/utils/text_extraction.py,sha256=mDeN1_VevNi3RwvFe48PM5vBh-A5WeBlYgP6lSjBaOk,10854
-natural_pdf/utils/visualization.py,sha256=n3IZpbY5cf9LItzGavBcNyVZZrrUVxjYnmqZHYPa7NU,9386
-natural_pdf/widgets/__init__.py,sha256=QTVaUmsw__FCweFYZebwPssQxxUFUMd0wpm_cUbGZJY,181
-natural_pdf/widgets/viewer.py,sha256=2VUY1TzWMDe9I-IVNOosKZ2LaqpjLB62ftMAdk-s6_8,24952
-natural_pdf-0.1.33.dist-info/licenses/LICENSE,sha256=9zfwINwJlarbDmdh6iJV4QUG54QSJlSAUcnC1YiC_Ns,1074
-optimization/memory_comparison.py,sha256=XEHtjduSmzXzxnsJMvemTcq-OAlvGUBAm5wwnOXq8TY,6524
-optimization/pdf_analyzer.py,sha256=G3XWhsEqIYbohEgTqz6wzxkAnOx4MkbvbSspx577-8w,19145
-optimization/performance_analysis.py,sha256=vVlFDywEXxhJLd9n2KVVqqQnS6rwWoHV_jlogboGF2k,13784
-optimization/test_cleanup_methods.py,sha256=B_zHiJr1hI8q-tdfBoFi0Jf5lj2PURjA_6teRBGoz8o,6277
-optimization/test_memory_fix.py,sha256=CWc0OSvFfKE0-nxqJOi_HAQc0GXUPKzkQbTeJp5UqxU,6364
-tools/bad_pdf_eval/__init__.py,sha256=Nqnn8clbgv-5l0PgxcTOldg8mkMKrFn4TvPL-rYUUGg,1
-tools/bad_pdf_eval/analyser.py,sha256=sR31aVVmTXRHS8uwLZXlPefTH2_lskxtAzuZwlhsyOo,13391
-tools/bad_pdf_eval/collate_summaries.py,sha256=Mcmf1OvVn0S0efj5ypk0syXKSrfUf6L5dowoGvOTgjU,5047
-tools/bad_pdf_eval/eval_suite.py,sha256=-MK-XLqBo1025sccwYL6tnf7mZ1ZEpxu6EsTYv2ppmU,4294
-tools/bad_pdf_eval/export_enrichment_csv.py,sha256=SMEm9WxFUN_RIf8AGfZfjGEmvBvrOw1q6ARMl-EazIU,1906
-tools/bad_pdf_eval/llm_enrich.py,sha256=PsFMymPc8BNck21T3vupTN18pLdum-A_OLoJEKr6f80,12234
-tools/bad_pdf_eval/reporter.py,sha256=LIhcguDZ5XKgb0WeJsyA7m0kcliebOohzveShvt_KmY,400
-tools/bad_pdf_eval/utils.py,sha256=FuxaPX6f26IjQXu1vP0a2i9h1jgJNbASb8mRyj5-elE,4849
-natural_pdf-0.1.33.dist-info/METADATA,sha256=mSAwh3vuD9aRvO_AC_XBZG5sw9SeiuidC86a7kuV--I,6711
-natural_pdf-0.1.33.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-natural_pdf-0.1.33.dist-info/entry_points.txt,sha256=1R_KMv7g60UBBpRqGfw7bppsMNGdayR-iJlb9ohEk_8,81
-natural_pdf-0.1.33.dist-info/top_level.txt,sha256=oZlRzSc3nZ9sV3L6kD_Di734Pp62ANrm46imFVa51qQ,58
-natural_pdf-0.1.33.dist-info/RECORD,,