my-aws-helpers 2.6.2__tar.gz → 2.6.4__tar.gz

my_aws_helpers-2.6.4/MANIFEST.in

@@ -0,0 +1 @@
+recursive-include my_aws_helpers *.txt

{my_aws_helpers-2.6.2 → my_aws_helpers-2.6.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: my_aws_helpers
-Version: 2.6.2
+Version: 2.6.4
 Summary: AWS Helpers
 Home-page: https://github.com/JarrodMccarthy/aws_helpers.git
 Author: Jarrod McCarthy

{my_aws_helpers-2.6.2 → my_aws_helpers-2.6.4}/my_aws_helpers/bedrock.py

@@ -10,12 +10,6 @@ from enum import Enum
 import pymupdf
 import concurrent.futures
 from dataclasses import dataclass
-from my_aws_helpers.logging import select_powertools_logger
-
-
-logger = select_powertools_logger('bedrock-boy')
-
-logger.info("Got logger")
 
 class PromptType(str, Enum):
     json = "json_system_prompt.txt"
@@ -56,9 +50,11 @@ class Bedrock:
     def __init__(
         self,
         model_id: str = "apac.anthropic.claude-3-5-sonnet-20241022-v2:0",
+        logger = None,
     ):
 
         self.session = Bedrock._set_session_params()
+        self.logger = logger
         region_name = "ap-southeast-2"
         if self.session is None:
             self.session = boto3.Session(region_name = region_name)
@@ -87,7 +83,7 @@ class Bedrock:
                 region_name=region_name
             )
         except Exception as e:
-            logger.exception(e)
+            print(e)
             return None
 
     def _get_prompt(self, prompt_type: str) -> Optional[str]:
@@ -100,7 +96,7 @@ class Bedrock:
                 prompt = f.read()
                 return prompt
         except Exception as e:
-            logger.exception(f"Failed to get {prompt_type} prompt due to {e}")
+            self.logger.exception(f"Failed to get {prompt_type} prompt due to {e}")
         return None
 
     def _ocr(
@@ -127,13 +123,13 @@ class Bedrock:
         ]
         retries = 3
         for i in range(retries):
-            logger.info(f"Attempt number {i} for {self.model_id} converse")
+            self.logger.info(f"Attempt number {i} for {self.model_id} converse")
             try:
                 response = self.client.converse(modelId = self.model_id, messages = message, system = system_prompt)
                 if response['ResponseMetadata']['HTTPStatusCode'] == 200:
                     break
             except Exception as e:
-                logger.exception(f"Error during conversation due to {e}")
+                self.logger.exception(f"Error during conversation due to {e}")
                 if i >= len(retries) - 1: raise Exception(e)
                 continue
         
@@ -152,7 +148,7 @@ class Bedrock:
         results = list()
         prompt = self._get_prompt(prompt_type=prompt_type)
         with concurrent.futures.ThreadPoolExecutor(max_workers=max_workers) as executor:
-            logger.info("Some log")
+            self.logger.info("Some log")
             execution_futures = {
                 executor.submit(
                     self._ocr, 
@@ -172,28 +168,27 @@ class Bedrock:
         prompt_type: str,
         zoom: int = 7,
     ) -> List[OCRResult]:
-        logger.info("Getting OCR Results")
         try:
+            self.logger.info("Getting OCR Results")
             document = pymupdf.open(stream=pdf_bytes, filetype="pdf")
+            pages: List[pymupdf.Page] = [p for p in document]
+
+            image_bytes_list: List[bytes] = list()
+            for i, p in enumerate(pages):
+                try:
+                    image_bytes: bytes = p.get_pixmap(matrix = pymupdf.Matrix(zoom, zoom)).tobytes("png")
+                    image_bytes_list.append(image_bytes)
+                except Exception as e:
+                    self.logger.error(f"Could not get pix map for page {i}")
+                    continue
+            prompt = self._get_prompt(prompt_type=prompt_type)
+            self.logger.info("Got Prompt")
+            results = list()
+            for i, image_bytes in enumerate(image_bytes_list):
+                self.logger.info(f"Starting OCR for page: {i}")
+                results.append(self._ocr(image_bytes=image_bytes, prompt=prompt))
+            return results
         except Exception as e:
-            logger.exception(f"Failed to open pdf due to {e}")
+            self.logger.exception(e)
             return []
-
-        pages: List[pymupdf.Page] = [p for p in document]
-
-        image_bytes_list: List[bytes] = list()
-        for i, p in enumerate(pages):
-            try:
-                image_bytes: bytes = p.get_pixmap(matrix = pymupdf.Matrix(zoom, zoom)).tobytes("png")
-                image_bytes_list.append(image_bytes)
-            except Exception as e:
-                logger.error(f"Could not get pix map for page {i}")
-                continue
-        prompt = self._get_prompt(prompt_type=prompt_type)
-        logger.info("Got Prompt")
-        results = list()
-        for i, image_bytes in enumerate(image_bytes_list):
-            logger.info(f"Starting OCR for page: {i}")
-            results.append(self._ocr(image_bytes=image_bytes, prompt=prompt))
-        return results
             

my_aws_helpers-2.6.4/my_aws_helpers/prompts/json_system_prompt.txt

@@ -0,0 +1,36 @@
+You are an intelligent document and image interpreter.
+
+Your task is to analyze the provided image and extract all meaningful data as structured **JSON**.
+
+## Output Requirements
+
+- Return **only valid JSON**.
+- If the image contains one or more **tables**, represent each row as a JSON object.
+- Use **the table headers as keys**.
+- if there are multiple sections, only include the row data in transactions
+- Every row should have the consistent headers
+- If content is unclear, use: `"[Unclear]"`.
+- **Do not fabricate** values not visible in the image.
+- if a descripion contains 2 lines of text, only include the most important text and the text should reside in 1 key in json response, do not split it
+- Do not return anything except the json content
+
+## Example
+
+If the image contains this table:
+
+|  Date  |          | Price |
+| ------ |----------|-------|
+| June 6 | desc 1   | $2.00 |
+| June 5 | misc 2   | $1.70 |
+|        | item x   | $1.50 |
+
+Return:
+
+```json
+{
+  "transactions": [
+    { "date": "June 6", "description": "desc 1", "price": "$2.00" },
+    { "date": "June 5", "description": "desc 2", "price": "$1.70" },
+    { "date": "June 5", "description": "item x", "price": "$1.50" },
+  ]
+}

my_aws_helpers-2.6.4/my_aws_helpers/prompts/markdown_system_prompt.txt

@@ -0,0 +1,35 @@
+You are an intelligent document and visual layout interpreter.
+
+Your task is to:
+1. Analyze the image provided a bank statement with a list of transactions.
+2. Convert the contents into **well-formatted Markdown**, preserving:
+   - Headings
+   - Lists
+   - Tables
+   - Emphasis (bold, italic)
+   - Line breaks and whitespace if needed
+
+## Output Format
+
+Please return only valid Markdown, and structure it clearly.
+
+Use this format:
+
+```markdown
+# [Main Title of Document or Topic]
+
+## Section 1 Title
+
+- Bullet point 1
+- Bullet point 2
+
+### Table
+
+| Column A | Column B |
+|----------|----------|
+| Row 1A   | Row 1B   |
+| Row 2A   | Row 2B   |
+
+3. The table data should be copied exactly as it appears without no exceptions
+
+4. Do not return anything except the markdown content

{my_aws_helpers-2.6.2 → my_aws_helpers-2.6.4}/my_aws_helpers.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: my-aws-helpers
-Version: 2.6.2
+Version: 2.6.4
 Summary: AWS Helpers
 Home-page: https://github.com/JarrodMccarthy/aws_helpers.git
 Author: Jarrod McCarthy

{my_aws_helpers-2.6.2 → my_aws_helpers-2.6.4}/my_aws_helpers.egg-info/SOURCES.txt

@@ -1,3 +1,4 @@
+MANIFEST.in
 README.md
 setup.cfg
 setup.py
@@ -18,4 +19,6 @@ my_aws_helpers.egg-info/requires.txt
 my_aws_helpers.egg-info/top_level.txt
 my_aws_helpers.egg-info/zip-safe
 my_aws_helpers/prompts/__init__.py
+my_aws_helpers/prompts/json_system_prompt.txt
+my_aws_helpers/prompts/markdown_system_prompt.txt
 tests/test_event.py

{my_aws_helpers-2.6.2 → my_aws_helpers-2.6.4}/setup.py

@@ -3,7 +3,7 @@ from setuptools import find_namespace_packages, setup
 
 base_path = os.path.abspath(os.path.dirname(__file__))
 
-version = "2.6.2"
+version = "2.6.4"
 
 setup(
     name="my_aws_helpers",