HowdenParser 5.0.0__tar.gz → 5.2.0__tar.gz

{howdenparser-5.0.0 → howdenparser-5.2.0}/HowdenParser/parser.py

@@ -3,6 +3,9 @@ import logging
 import dotenv
 from typing import overload, Literal
 from inspect import signature, Signature
+from typing import Any
+import json
+import hashlib
 
 dotenv.load_dotenv()
 logger = logging.getLogger(__name__)
@@ -26,6 +29,30 @@ class BaseParser(ABC):
     def parse(self, text: str, include_pagenumbers: bool = False):
         pass
 
+    def make_serializable(self, obj: Any) -> Any:
+        if obj is None or isinstance(obj, (str, int, float, bool)):
+            return obj
+        if isinstance(obj, (list, tuple, set)):
+            return [self.make_serializable(v) for v in obj]
+        if isinstance(obj, dict):
+            return {k: self.make_serializable(v) for k, v in sorted(obj.items())}
+        if hasattr(obj, "__dict__"):
+            return self.make_serializable(vars(obj))
+        return f"{type(obj).__name__}:{repr(obj)}"
+
+    def compute_hash(self, parameter:dict) -> str:
+
+        attrs = {
+            k: v
+            for k, v in parameter.items()
+            if k != "hash"
+               and not k.startswith("_")
+               and k not in ("client", "provider")
+        }
+        serializable = self.make_serializable(attrs)
+        attrs_str = json.dumps(serializable, sort_keys=True, ensure_ascii=True)
+        hashed = hashlib.sha256(attrs_str.encode()).hexdigest()[:8]
+        return hashed
 
 class Parser(BaseParser):
     """Factory + registry interface for all parsers."""

howdenparser-5.2.0/HowdenParser/parsers/llama_parser.py

@@ -0,0 +1,88 @@
+import os
+import logging
+from pathlib import Path
+from ..parser import BaseParser
+
+class LlamaParser(BaseParser, name="llamaparser"):
+    def __init__(self, result_type: str, model: str, parse_mode: str, provider_and_model: str,
+                 merge_tables_across_pages_in_markdown: bool, preserve_layout_alignment_across_pages: bool,
+                 hide_footers: bool, hide_headers: bool) -> None:
+
+        super().__init__()
+        # Capture only input parameters
+        self._input_params = {
+            k: v
+            for k, v in locals().items()
+            if k not in ("self", "__class__", "__len__")
+        }
+
+        logging.info("Configuring LlamaParser (lazy init enabled)...")
+
+        # Store configuration only
+        self.result_type = result_type
+        self.model = model
+        self.parse_mode = parse_mode
+        self.provider_and_model = provider_and_model
+        self.merge_tables_across_pages_in_markdown = merge_tables_across_pages_in_markdown
+        self.preserve_layout_alignment_across_pages = preserve_layout_alignment_across_pages
+        self.hide_footers = hide_footers
+        self.hide_headers = hide_headers
+        self.hashed = self.compute_hash(self._input_params)
+        # DO NOT CREATE LlamaParse HERE
+        # It creates asyncio objects bound to the wrong loop
+        self._parser = None
+
+    def _lazy_init(self):
+        """Initialize LlamaParse inside the worker thread event loop."""
+        if self._parser is not None:
+            return
+
+        from llama_parse import LlamaParse, ResultType
+
+        api_key = os.getenv("LLAMA-PARSER-API-TOKEN")
+        if not api_key:
+            raise EnvironmentError("Missing LLAMA-PARSER-API-TOKEN in .env file.")
+
+        rt = self.result_type
+        if rt.lower() in ("md", "markdown"):
+            rt = ResultType.MD
+
+        # Construct the parser INSIDE the worker thread
+        self._parser = LlamaParse(
+            api_key=api_key,
+            result_type=rt,
+            model=self.model,
+            parse_mode=self.parse_mode,
+            merge_tables_across_pages_in_markdown=self.merge_tables_across_pages_in_markdown,
+            preserve_layout_alignment_across_pages=self.preserve_layout_alignment_across_pages,
+            hide_footers=self.hide_footers,
+            hide_headers=self.hide_headers,
+        )
+
+        logging.info("LlamaParser initialized inside worker thread")
+
+    def parse(self, file_path: Path, include_pagenumbers: bool = False) -> str:
+        # Ensure parser is created inside the thread event loop
+        self._lazy_init()
+
+        # Standard calls
+        if not include_pagenumbers:
+            documents = self._parser.load_data(str(file_path))
+            return "\n".join(doc.text for doc in documents)
+
+        # With page numbers
+        documents = self._parser.parse(file_path)
+
+        if self.result_type.lower() in ("md", "markdown"):
+            return "\n".join(
+                f"<PAGE_NUMBER {idx}>{page.md}</PAGE_NUMBER {idx}>"
+                for idx, page in enumerate(documents.pages, start=1)
+            )
+
+        return "\n".join(
+            f"<PAGE_NUMBER {idx}>{page.text}</PAGE_NUMBER {idx}>"
+            for idx, page in enumerate(documents.pages, start=1)
+        )
+
+    def __call__(self, file_path: Path) -> str:
+        return self.parse(file_path)

{howdenparser-5.0.0 → howdenparser-5.2.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: HowdenParser
-Version: 5.0.0
+Version: 5.2.0
 Summary: A simple configuration manager with Pydantic and JSON export.
 License: MIT
 Keywords: config,configuration,pydantic,json

{howdenparser-5.0.0 → howdenparser-5.2.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "HowdenParser"
-version = "5.0.0"
+version = "5.2.0"
 description = "A simple configuration manager with Pydantic and JSON export."
 readme = "README.md"
 requires-python = ">=3.12,<3.14"

howdenparser-5.0.0/HowdenParser/parsers/llama_parser.py

@@ -1,64 +0,0 @@
-import os
-import logging
-from pathlib import Path
-from ..parser import BaseParser
-
-class LlamaParser(BaseParser, name="llamaparser"):
-    def __init__(self, result_type: str,
-                 model: str,
-                 parse_mode: str,
-                 provider_and_model: str,
-                 merge_tables_across_pages_in_markdown: bool,
-                 preserve_layout_alignment_across_pages: bool,
-                 hide_footers: bool,
-                 hide_headers: bool
-                 ) -> None:
-        logging.info("Initializing LlamaParser...")
-        self.result_type = result_type
-        self.model = model
-        self.parse_mode = parse_mode
-        self.provider_and_model = provider_and_model
-        self.merge_tables_across_pages_in_markdown = merge_tables_across_pages_in_markdown
-        self.preserve_layout_alignment_across_pages = preserve_layout_alignment_across_pages
-        self.hide_footers=hide_footers
-        self.hide_headers=hide_headers
-
-        from llama_parse import LlamaParse, ResultType
-        if result_type.lower() in ("md", "markdown"):
-            result_type = ResultType.MD
-        api_key = os.getenv("LLAMA-PARSER-API-TOKEN")
-        if not api_key:
-            raise EnvironmentError("Missing LLAMA-PARSER-API-TOKEN in .env file.")
-        self._parser = LlamaParse(api_key=api_key,
-                                  result_type=result_type,
-                                  model=self.model,
-                                  parse_mode=self.parse_mode,
-                                  merge_tables_across_pages_in_markdown=self.merge_tables_across_pages_in_markdown,
-                                  preserve_layout_alignment_across_pages=self.preserve_layout_alignment_across_pages,
-                                  hide_footers=self.hide_footers,
-                                  hide_headers=self.hide_headers)
-
-    def parse(self, file_path: Path, include_pagenumbers: bool=False) -> str:
-        """
-        Parse the document at file_path. 
-        Args:
-            file_path (Path): Path to the document to be parsed.
-            include_pagenumbers (bool): Whether to include page numbers in the output. If True, page numbers will be wrapped around each page's content, like so: <PAGE_NUMBER 1>...content...</PAGE_NUMBER 1>
-        """
-
-        if not include_pagenumbers:
-            documents = self._parser.load_data(str(file_path))
-
-            result = "\n".join(doc.text for doc in documents)
-        else:
-            documents = self._parser.parse(file_path)
-
-            if self.result_type.lower() in ("md", "markdown"):
-                result = "\n".join(f"<PAGE_NUMBER {idx}>{page.md}</PAGE_NUMBER {idx}>" for idx, page in enumerate(documents.pages, start=1))
-            else: 
-                result = "\n".join(f"<PAGE_NUMBER {idx}>{page.text}</PAGE_NUMBER {idx}>" for idx, page in enumerate(documents.pages, start=1))
-
-        return result
-
-    def __call__(self, file_path: Path) -> str:
-        return self.parse(file_path)