HowdenParser 6.0.0__tar.gz → 7.0.0__tar.gz

{howdenparser-6.0.0 → howdenparser-7.0.0}/HowdenParser/parser.py

@@ -1,12 +1,13 @@
 from abc import ABC, abstractmethod
 import logging
 import dotenv
-from typing import overload, Literal
+from typing import overload, Literal, Any, cast
 from inspect import signature, Signature
-from typing import Any
 import json
 import hashlib
 from pathlib import Path
+import re
+
 dotenv.load_dotenv()
 logger = logging.getLogger(__name__)
 
@@ -26,7 +27,7 @@ class BaseParser(ABC):
         BaseParser._registry.pop("", None)
 
     @abstractmethod
-    def parse(self, text: str, include_pagenumbers: bool = False):
+    def parse(self, text: Path, include_pagenumbers: bool = False):
         pass
 
     def make_serializable(self, obj: Any) -> Any:
@@ -70,12 +71,52 @@ class BaseParser(ABC):
 class Parser(BaseParser):
     """Factory + registry interface for all parsers."""
 
-    def __new__(cls, config_or_dict: "Parameter | dict | str" = None, **kwargs) -> BaseParser:
-        """
-        Factory entrypoint.
-        Allows calling Parser(...) directly to create the correct subclass.
-        """
-        return cls._create(config_or_dict, **kwargs)
+    def __new__(cls, *args, **kwargs) -> "Parser":
+        """Factory entrypoint — returns the correct concrete parser subclass."""
+        return cls._create(*args, **kwargs)
+
+    # ---- Overloads so IDE validates kwargs per provider ----
+
+    @overload
+    def __init__(
+        self,
+        *,
+        provider_and_model: Literal["llamaparser:"],
+        tier: str,
+        version: str,
+        result_type: str,
+        preserve_layout_alignment_across_pages: bool,
+        merge_continued_tables: bool,
+    ) -> None: ...
+
+    @overload
+    def __init__(
+        self,
+        config_or_dict: Literal["mistralocr:ocr-large", "mistralocr:ocr-small"],
+    ) -> None: ...
+
+    @overload
+    def __init__(
+        self,
+        *,
+        provider_and_model: Literal["mistralocr:ocr-large", "mistralocr:ocr-small"],
+    ) -> None: ...
+
+    @overload
+    def __init__(
+        self,
+        config_or_dict: Literal["langchain:gpt-3.5-turbo", "langchain:gpt-4"],
+    ) -> None: ...
+
+    @overload
+    def __init__(
+        self,
+        *,
+        provider_and_model: Literal["langchain:gpt-3.5-turbo", "langchain:gpt-4"],
+    ) -> None: ...
+
+    def __init__(self, *_args, **_kwargs) -> None:
+        super().__init__()  # Never called — __new__ always returns a subclass instance
 
     @classmethod
     def available_parsers(cls) -> None:
@@ -89,39 +130,17 @@ class Parser(BaseParser):
         for key, values in result.items():
             print(f"{key} with parameters: {values}")
 
-    # ---- Overloads for IDE autocomplete ----
-    @overload
-    @classmethod
-    def _create(
-            cls,
-            *,
-            provider_and_model: Literal["llamaparser:"],
-            tier: str,
-            version: str,
-            result_type: str,
-            preserve_layout_alignment_across_pages: bool,
-            merge_continued_tables: bool,
-    ) -> "LlamaParser": ...
-
-    @overload
     @classmethod
-    def _create(
-            cls,
-            *,
-            provider_and_model: Literal["mistralocr:ocr-large", "mistralocr:ocr-small"],
-    ) -> "MistralOCRParser": ...
+    def _validate_provider_and_model(cls,provider_and_model: str) -> None:
+        if not re.match(r"^.+:", provider_and_model):
+            raise ValueError(
+                "provider_and_model must include a provider before the colon, e.g. 'llamaparser:'"
+        )
 
-    @overload
-    @classmethod
-    def _create(
-            cls,
-            *,
-            provider_and_model: Literal["langchain:gpt-3.5-turbo", "langchain:gpt-4"],
-    ) -> "LangChainParser": ...
 
     # ---- Implementation ----
     @classmethod
-    def _create(cls, config_or_dict: "Parameter | dict | str" = None, **kwargs) -> BaseParser:
+    def _create(cls, config_or_dict: dict | str = None, **kwargs) -> "Parser":
         """
         Dynamically create parser instances.
         Supports: Parameter, dict, str (provider_and_model), or kwargs.
@@ -145,6 +164,8 @@ class Parser(BaseParser):
 
         provider_and_model = str(merged_args.get("provider_and_model")).strip()
 
+        Parser._validate_provider_and_model(provider_and_model)
+
         if ":" not in provider_and_model:
             raise ValueError("provider_and_model must include a colon, e.g. 'llamaparser:'")
 
@@ -197,8 +218,7 @@ class Parser(BaseParser):
                 f"{parser_cls.__name__} cannot be created because it is missing required argument(s): {', '.join(missing)}. {example}"
             )
 
-        return parser_cls(**valid_args)
+        return cast("Parser", cast(Any, parser_cls)(**valid_args))
 
-    @abstractmethod
-    def parse(self, text: str, include_pagenumbers: bool = False):
-        pass
+    def parse(self, text: Path, include_pagenumbers: bool = False):
+        raise NotImplementedError

{howdenparser-6.0.0 → howdenparser-7.0.0}/HowdenParser/parsers/llama_parser.py

@@ -1,5 +1,7 @@
 import os
 import logging
+from HowdenCommonObjects.howden_result import HowdenResult
+from HowdenCommonObjects.usage import Usage
 from pathlib import Path
 from ..parser import BaseParser
 from typing import Any
@@ -7,8 +9,8 @@ from typing import Any
 class LlamaParser(BaseParser, name="llamaparser"):
     def __init__(self,
                  result_type: str,
-                 tier: str, # One of: ["fast", "cost_effective", "agentic", "agentic_plus"]
-                 version: str, # one of: ["latest","2026-03-12","2026-03-11"...] (many not listed)
+                 tier: str,  # One of: ["fast", "cost_effective", "agentic", "agentic_plus"]
+                 version: str,  # one of: ["latest","2026-03-12","2026-03-11"...] (many not listed)
                  provider_and_model: str,
                  merge_continued_tables: bool,
                  preserve_layout_alignment_across_pages: bool,
@@ -48,12 +50,15 @@ class LlamaParser(BaseParser, name="llamaparser"):
         if not api_key:
             raise EnvironmentError("Missing LLAMA-PARSER-API-TOKEN in .env file.")
 
-        rt_lower = str(self.result_type).lower() # Possible values for LlamaCloud expand: ["markdown", "text", "metadata", "items"]
-        if rt_lower in ("md","markdown"):
+        rt_lower = str(
+            self.result_type).lower()  # Possible values for LlamaCloud expand: ["markdown", "markdown_full", "text", "metadata", "items"]
+        if rt_lower in ("md", "markdown"):
             self.rt = 'markdown'
-        elif rt_lower in ("txt","text"):
+        elif rt_lower in ("md_full", "markdown_full"):
+            self.rt = 'markdown_full'
+        elif rt_lower in ("txt", "text"):
             self.rt = 'text'
-        elif rt_lower in ("json"):
+        elif rt_lower == "json":
             self.rt = 'items'
         else:
             raise NotImplementedError(f"Result type {self.result_type} is not supported.")
@@ -62,9 +67,10 @@ class LlamaParser(BaseParser, name="llamaparser"):
 
         logging.info("LlamaParser initialized inside worker thread")
 
-    def parse(self, file_path: Path, include_pagenumbers: bool = False) -> Any:
+    def parse(self, file_path: Path, include_pagenumbers: bool = False) -> HowdenResult:
         # Ensure parser is created inside the thread event loop
         self._lazy_init()
+
         def page_break(text: str, idx: int) -> str:
             return f"<PAGE_NUMBER {idx}>{text}</PAGE_NUMBER {idx}>"
 
@@ -83,22 +89,37 @@ class LlamaParser(BaseParser, name="llamaparser"):
                     "preserve_layout_alignment_across_pages": self.preserve_layout_alignment_across_pages,
                 }
             },
-            expand=[self.rt],
+            expand=[self.rt,"job_metadata","usage"],
+        )
+
+        usage = Usage(
+            type="parser",
+            provider=self.provider_and_model,
+            model=self.version + " " + self.tier,
+            operation=self.name,
+            pages = documents.job_metadata["pdf-pages"]
         )
 
         if self.rt == 'markdown':
-            if include_pagenumbers: return "\n".join(page_break(page.markdown, idx) for idx, page in enumerate(documents.markdown.pages, start=1))
-            else: return "\n".join(page.markdown for page in documents.markdown.pages)
+            if include_pagenumbers:
+                result = "\n".join(page_break(page.markdown, idx) for idx, page in enumerate(documents.markdown.pages, start=1))
+            else:
+                result = "\n".join(page.markdown for page in documents.markdown.pages)
+        elif self.rt == 'markdown_full':
+            result = documents.markdown_full
         elif self.rt == 'text':
-            if include_pagenumbers: return "\n".join(page_break(page.text, idx) for idx, page in enumerate(documents.text.pages, start=1))
-            else: return "\n".join(page.text for page in documents.text.pages)
+            if include_pagenumbers:
+                result = "\n".join(page_break(page.text, idx) for idx, page in enumerate(documents.text.pages, start=1))
+            else:
+                result = "\n".join(page.text for page in documents.text.pages)
         elif self.rt == 'items':
-            return documents.model_dump()
+            result = documents.model_dump()
         else:
             raise NotImplementedError(f"Result type {self.result_type} is not supported.")
+        return HowdenResult(result,usage)
 
-    def __call__(self, file_path: Path) -> str:
+    def __call__(self, file_path: Path) -> HowdenResult:
         return self.parse(file_path)
 
     def write_json_hyperparameter(self, folder_file_path: Path) -> None:
-        self.write_parameters(self._input_params, folder_file_path)
+        self.write_parameters(self._input_params, folder_file_path)

{howdenparser-6.0.0 → howdenparser-7.0.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: HowdenParser
-Version: 6.0.0
+Version: 7.0.0
 Summary: A simple configuration manager with Pydantic and JSON export.
 License: MIT
 Keywords: config,configuration,pydantic,json
@@ -12,8 +12,9 @@ Classifier: Programming Language :: Python :: 3
 Classifier: Programming Language :: Python :: 3.12
 Classifier: Programming Language :: Python :: 3.13
 Requires-Dist: dotenv (>=0.9.9,<0.10.0)
+Requires-Dist: howdencommonobjects (>=1.1.0)
 Requires-Dist: langchain (>=1.1.3,<2.0.0)
-Requires-Dist: llama-cloud (>=1.0)
+Requires-Dist: llama-cloud (==2.16.0)
 Requires-Dist: mistralai (>=1.9.3,<2.0.0)
 Requires-Dist: pdf2image (>=1.17.0,<2.0.0)
 Requires-Dist: pypdf2 (>=3.0.1,<4.0.0)

{howdenparser-6.0.0 → howdenparser-7.0.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "HowdenParser"
-version = "6.0.0"
+version = "7.0.0"
 description = "A simple configuration manager with Pydantic and JSON export."
 readme = "README.md"
 requires-python = ">=3.12,<3.14"
@@ -18,13 +18,14 @@ repository = "https://github.com/yourusername/config"
 documentation = "https://github.com/yourusername/config"
 dependencies = [
     "mistralai (>=1.9.3,<2.0.0)",
-    "llama-cloud (>=1.0)",
+    "llama-cloud==2.16.0",
     "pypdf2 (>=3.0.1,<4.0.0)",
     "pdf2image (>=1.17.0,<2.0.0)",
     "langchain (>=1.1.3,<2.0.0)",
     "pytest (>=9.0.2,<10.0.0)",
     "tomli-w (>=1.2.0,<2.0.0)",
     "dotenv (>=0.9.9,<0.10.0)",
+    "howdencommonobjects>=1.1.0",
 ]
 
 [project.license]
@@ -41,4 +42,5 @@ dev = [
     "toml (>=0.10.2,<0.11.0)",
     "tomli-w (>=1.2.0,<2.0.0)",
     "howdenconfig (>=1.0.6,<2.0.0)",
+    "howdencommonobjects",
 ]