fabricatio 0.3.14.dev1__cp312-cp312-win_amd64.whl → 0.3.14.dev4__cp312-cp312-win_amd64.whl

fabricatio/parser.py

@@ -1,152 +1,112 @@
-"""A module to parse text using regular expressions."""
+"""A module for capturing patterns in text using regular expressions."""
 
 import re
+from dataclasses import dataclass, field
 from functools import lru_cache
-from re import Pattern, compile
-from typing import Any, Callable, Iterable, List, Optional, Self, Tuple, Type
+from typing import Any, Callable, Iterable, List, Optional, Self, Tuple, Type, Union
 
 import ujson
-from fabricatio.rust import CONFIG
 from json_repair import repair_json
-from pydantic import BaseModel, ConfigDict, Field, PositiveInt, PrivateAttr, ValidationError
 
 from fabricatio.journal import logger
+from fabricatio.rust import CONFIG
 
 
-class Capture(BaseModel):
+@dataclass(frozen=True)
+class Capture:
     """A class to capture patterns in text using regular expressions.
 
     Attributes:
-        pattern (str): The regular expression pattern to search for.
-        _compiled (Pattern): The compiled regular expression pattern.
+        target_groups (Tuple[int, ...]): The target groups to extract from the match.
+        pattern (str): The regex pattern to search for.
+        flags (int): Flags to apply when compiling the regex.
+        capture_type (Optional[str]): Optional hint for post-processing (e.g., 'json').
     """
 
-    model_config = ConfigDict(use_attribute_docstrings=True)
-    target_groups: Tuple[int, ...] = Field(default_factory=tuple)
-    """The target groups to capture from the pattern."""
-    pattern: str = Field(frozen=True)
+    pattern: str = field()
     """The regular expression pattern to search for."""
-    flags: PositiveInt = Field(default=re.DOTALL | re.MULTILINE | re.IGNORECASE, frozen=True)
-    """The flags to use when compiling the regular expression pattern."""
+    flags: int = re.DOTALL | re.MULTILINE | re.IGNORECASE
+    """Flags to control regex behavior (DOTALL, MULTILINE, IGNORECASE by default)."""
     capture_type: Optional[str] = None
-    """The type of capture to perform, e.g., 'json', which is used to dispatch the fixer accordingly."""
-    _compiled: Pattern = PrivateAttr()
-
-    def model_post_init(self, __context: Any) -> None:
-        """Initialize the compiled pattern."""
-        self._compiled = compile(self.pattern, self.flags)
-
-    def fix[T](self, text: str | Iterable[str] | T) -> str | List[str] | T:
-        """Fix the text using the pattern.
+    """Optional type identifier for post-processing (e.g., 'json' for JSON repair)."""
+    target_groups: Tuple[int, ...] = field(default_factory=tuple)
+    """Tuple of group indices to extract from the match (1-based indexing)."""
 
-        Args:
-            text (str | List[str]): The text to fix.
-
-        Returns:
-            str | List[str]: The fixed text with the same type as input.
-        """
+    def fix(self, text: Union[str, Iterable[str], Any]) -> Union[str, List[str], Any]:
+        """Fix the text based on capture_type (e.g., JSON repair)."""
         match self.capture_type:
             case "json" if CONFIG.general.use_json_repair:
-                logger.debug("Applying json repair to text.")
+                logger.debug("Applying JSON repair to text.")
                 if isinstance(text, str):
-                    return repair_json(text, ensure_ascii=False)  # pyright: ignore [reportReturnType]
-                return [repair_json(item, ensure_ascii=False) for item in
-                        text]  # pyright: ignore [reportReturnType, reportGeneralTypeIssues]
+                    return repair_json(text, ensure_ascii=False)
+                return [repair_json(item, ensure_ascii=False) for item in text]
             case _:
-                return text  # pyright: ignore [reportReturnType]
-
-    def capture(self, text: str) -> Tuple[str, ...] | str | None:
-        """Capture the first occurrence of the pattern in the given text.
-
-        Args:
-            text (str): The text to search the pattern in.
-
-        Returns:
-            str | None: The captured text if the pattern is found, otherwise None.
-
-        """
-        if (match := self._compiled.match(text) or self._compiled.search(text)) is None:
-            logger.debug(f"Capture Failed {type(text)}: \n{text}")
+                return text
+
+    def capture(self, text: str) -> Optional[Union[str, Tuple[str, ...]]]:
+        """Capture the first match of the pattern in the text."""
+        compiled = re.compile(self.pattern, self.flags)
+        match = compiled.match(text) or compiled.search(text)
+        if match is None:
+            logger.debug(f"Capture Failed: {text}")
             return None
+
         groups = self.fix(match.groups())
         if self.target_groups:
             cap = tuple(groups[g - 1] for g in self.target_groups)
-            logger.debug(f"Captured text: {'\n\n'.join(cap)}")
+            logger.debug(f"Captured texts: {'\n==\n'.join(cap)}")
             return cap
         cap = groups[0]
         logger.debug(f"Captured text: \n{cap}")
         return cap
 
-    def convert_with[T](self, text: str, convertor: Callable[[Tuple[str, ...]], T] | Callable[[str], T]) -> T | None:
-        """Convert the given text using the pattern.
-
-        Args:
-            text (str): The text to search the pattern in.
-            convertor (Callable[[Tuple[str, ...]], T] | Callable[[str], T]): The function to convert the captured text.
-
-        Returns:
-            str | None: The converted text if the pattern is found, otherwise None.
-        """
+    def convert_with(
+        self,
+        text: str,
+        convertor: Callable[[Union[str, Tuple[str, ...]]], Any],
+    ) -> Optional[Any]:
+        """Convert captured text using a provided function."""
         if (cap := self.capture(text)) is None:
             return None
         try:
-            return convertor(cap)  # pyright: ignore [reportArgumentType]
-        except (ValueError, SyntaxError, ValidationError) as e:
-            logger.error(f"Failed to convert text using {convertor.__name__} to convert.\nerror: {e}\n {cap}")
+            return convertor(cap)
+        except Exception as e:  # noqa: BLE001
+            logger.error(f"Failed to convert text using {convertor.__name__}: {e}\n{cap}")
             return None
 
-    def validate_with[K, T, E](
-            self,
-            text: str,
-            target_type: Type[T],
-            elements_type: Optional[Type[E]] = None,
-            length: Optional[int] = None,
-            deserializer: Callable[[Tuple[str, ...]], K] | Callable[[str], K] = ujson.loads,
-    ) -> T | None:
-        """Validate the given text using the pattern.
-
-        Args:
-            text (str): The text to search the pattern in.
-            target_type (Type[T]): The expected type of the output, dict or list.
-            elements_type (Optional[Type[E]]): The expected type of the elements in the output dict keys or list elements.
-            length (Optional[int]): The expected length of the output, bool(length)==False means no length validation.
-            deserializer (Callable[[Tuple[str, ...]], K] | Callable[[str], K]): The function to deserialize the captured text.
-
-        Returns:
-            T | None: The validated text if the pattern is found and the output is of the expected type, otherwise None.
-        """
-        judges = [lambda output_obj: isinstance(output_obj, target_type)]
+    def validate_with[T, K, E](
+        self,
+        text: str,
+        target_type: Type[T],
+        elements_type: Optional[Type[E]] = None,
+        length: Optional[int] = None,
+        deserializer: Callable[[Union[str, Tuple[str, ...]]], K] = lambda x: ujson.loads(x) if isinstance(x, str) else ujson.loads(x[0]),
+    ) -> Optional[T]:
+        """Deserialize and validate the captured text against expected types."""
+        judges = [lambda obj: isinstance(obj, target_type)]
         if elements_type:
-            judges.append(lambda output_obj: all(isinstance(e, elements_type) for e in output_obj))
+            judges.append(lambda obj: all(isinstance(e, elements_type) for e in obj))
         if length:
-            judges.append(lambda output_obj: len(output_obj) == length)
+            judges.append(lambda obj: len(obj) == length)
 
         if (out := self.convert_with(text, deserializer)) and all(j(out) for j in judges):
-            return out  # pyright: ignore [reportReturnType]
+            return out  # type: ignore
         return None
 
     @classmethod
     @lru_cache(32)
     def capture_code_block(cls, language: str) -> Self:
-        """Capture the first occurrence of a code block in the given text.
-
-        Args:
-            language (str): The text containing the code block.
-
-        Returns:
-            Self: The instance of the class with the captured code block.
-        """
+        """Capture a code block of the given language."""
         return cls(pattern=f"```{language}(.*?)```", capture_type=language)
 
     @classmethod
     @lru_cache(32)
     def capture_generic_block(cls, language: str) -> Self:
-        """Capture the first occurrence of a generic code block in the given text.
-
-        Returns:
-            Self: The instance of the class with the captured code block.
-        """
-        return cls(pattern=f"--- Start of {language} ---(.*?)--- end of {language} ---", capture_type=language)
+        """Capture a generic block of the given language."""
+        return cls(
+            pattern=f"--- Start of {language} ---(.*?)--- End of {language} ---",
+            capture_type=language,
+        )
 
 
 JsonCapture = Capture.capture_code_block("json")

fabricatio/rust.pyi

@@ -10,12 +10,12 @@ Key Features:
 - Cryptographic utilities: BLAKE3 hashing.
 - Text utilities: Word boundary splitting and word counting.
 """
+
 from enum import StrEnum
-from typing import Any, Dict, List, Optional, Self, Tuple, Union, overload
+from typing import Any, Dict, List, Literal, Optional, Self, Tuple, Union, overload
 
 from pydantic import JsonValue
 
-
 class TemplateManager:
     """Template rendering engine using Handlebars templates.
 
@@ -47,10 +47,8 @@ class TemplateManager:
 
     @overload
     def render_template(self, name: str, data: Dict[str, Any]) -> str: ...
-
     @overload
     def render_template(self, name: str, data: List[Dict[str, Any]]) -> List[str]: ...
-
     def render_template(self, name: str, data: Dict[str, Any] | List[Dict[str, Any]]) -> str | List[str]:
         """Render a template with context data.
 
@@ -67,10 +65,8 @@ class TemplateManager:
 
     @overload
     def render_template_raw(self, template: str, data: Dict[str, Any]) -> str: ...
-
     @overload
     def render_template_raw(self, template: str, data: List[Dict[str, Any]]) -> List[str]: ...
-
     def render_template_raw(self, template: str, data: Dict[str, Any] | List[Dict[str, Any]]) -> str | List[str]:
         """Render a template with context data.
 
@@ -82,7 +78,6 @@ class TemplateManager:
             Rendered template content as string or list of strings
         """
 
-
 class BibManager:
     """BibTeX bibliography manager for parsing and querying citation data."""
 
@@ -191,7 +186,6 @@ class BibManager:
             Field value if found, None otherwise
         """
 
-
 def blake3_hash(content: bytes) -> str:
     """Calculate the BLAKE3 cryptographic hash of data.
 
@@ -202,11 +196,9 @@ def blake3_hash(content: bytes) -> str:
         Hex-encoded BLAKE3 hash string
     """
 
-
 def detect_language(string: str) -> str:
     """Detect the language of a given string."""
 
-
 def split_word_bounds(string: str) -> List[str]:
     """Split the string into words based on word boundaries.
 
@@ -217,7 +209,6 @@ def split_word_bounds(string: str) -> List[str]:
         A list of words extracted from the string.
     """
 
-
 def split_sentence_bounds(string: str) -> List[str]:
     """Split the string into sentences based on sentence boundaries.
 
@@ -228,7 +219,6 @@ def split_sentence_bounds(string: str) -> List[str]:
         A list of sentences extracted from the string.
     """
 
-
 def split_into_chunks(string: str, max_chunk_size: int, max_overlapping_rate: float = 0.3) -> List[str]:
     """Split the string into chunks of a specified size.
 
@@ -241,7 +231,6 @@ def split_into_chunks(string: str, max_chunk_size: int, max_overlapping_rate: fl
         A list of chunks extracted from the string.
     """
 
-
 def word_count(string: str) -> int:
     """Count the number of words in the string.
 
@@ -252,67 +241,51 @@ def word_count(string: str) -> int:
         The number of words in the string.
     """
 
-
 def is_chinese(string: str) -> bool:
     """Check if the given string is in Chinese."""
 
-
 def is_english(string: str) -> bool:
     """Check if the given string is in English."""
 
-
 def is_japanese(string: str) -> bool:
     """Check if the given string is in Japanese."""
 
-
 def is_korean(string: str) -> bool:
     """Check if the given string is in Korean."""
 
-
 def is_arabic(string: str) -> bool:
     """Check if the given string is in Arabic."""
 
-
 def is_russian(string: str) -> bool:
     """Check if the given string is in Russian."""
 
-
 def is_german(string: str) -> bool:
     """Check if the given string is in German."""
 
-
 def is_french(string: str) -> bool:
     """Check if the given string is in French."""
 
-
 def is_hindi(string: str) -> bool:
     """Check if the given string is in Hindi."""
 
-
 def is_italian(string: str) -> bool:
     """Check if the given string is in Italian."""
 
-
 def is_dutch(string: str) -> bool:
     """Check if the given string is in Dutch."""
 
-
 def is_portuguese(string: str) -> bool:
     """Check if the given string is in Portuguese."""
 
-
 def is_swedish(string: str) -> bool:
     """Check if the given string is in Swedish."""
 
-
 def is_turkish(string: str) -> bool:
     """Check if the given string is in Turkish."""
 
-
 def is_vietnamese(string: str) -> bool:
     """Check if the given string is in Vietnamese."""
 
-
 def tex_to_typst(string: str) -> str:
     """Convert TeX to Typst.
 
@@ -323,7 +296,6 @@ def tex_to_typst(string: str) -> str:
         The converted Typst string.
     """
 
-
 def convert_all_inline_tex(string: str) -> str:
     """Convert all inline TeX code in the string.
 
@@ -334,7 +306,6 @@ def convert_all_inline_tex(string: str) -> str:
         The converted string with inline TeX code replaced.
     """
 
-
 def convert_all_block_tex(string: str) -> str:
     """Convert all block TeX code in the string.
 
@@ -345,7 +316,6 @@ def convert_all_block_tex(string: str) -> str:
         The converted string with block TeX code replaced.
     """
 
-
 def fix_misplaced_labels(string: str) -> str:
     """A func to fix labels in a string.
 
@@ -356,7 +326,6 @@ def fix_misplaced_labels(string: str) -> str:
         The fixed string with labels properly placed.
     """
 
-
 def comment(string: str) -> str:
     """Add comment to the string.
 
@@ -367,7 +336,6 @@ def comment(string: str) -> str:
         The string with each line prefixed by '// '.
     """
 
-
 def uncomment(string: str) -> str:
     """Remove comment from the string.
 
@@ -378,7 +346,6 @@ def uncomment(string: str) -> str:
         The string with comments (lines starting with '// ' or '//') removed.
     """
 
-
 def split_out_metadata(string: str) -> Tuple[Optional[JsonValue], str]:
     """Split out metadata from a string.
 
@@ -389,7 +356,6 @@ def split_out_metadata(string: str) -> Tuple[Optional[JsonValue], str]:
         A tuple containing the metadata as a Python object (if parseable) and the remaining string.
     """
 
-
 def to_metadata(data: JsonValue) -> str:
     """Convert a Python object to a YAML string.
 
@@ -400,15 +366,12 @@ def to_metadata(data: JsonValue) -> str:
         The YAML string representation of the input data.
     """
 
-
 def convert_to_inline_formula(string: str) -> str:
     r"""Convert `$...$` to inline formula `\(...\)` and trim spaces."""
 
-
 def convert_to_block_formula(string: str) -> str:
     r"""Convert `$$...$$` to block formula `\[...\]` and trim spaces."""
 
-
 def inplace_update(string: str, wrapper: str, new_body: str) -> Optional[str]:
     """Replace content between wrapper strings.
 
@@ -422,7 +385,6 @@ def inplace_update(string: str, wrapper: str, new_body: str) -> Optional[str]:
 
     """
 
-
 def extract_body(string: str, wrapper: str) -> Optional[str]:
     """Extract the content between two occurrences of a wrapper string.
 
@@ -434,7 +396,6 @@ def extract_body(string: str, wrapper: str) -> Optional[str]:
         The content between the first two occurrences of the wrapper string if found, otherwise None.
     """
 
-
 class LLMConfig:
     """LLM configuration structure.
 
@@ -486,7 +447,6 @@ class LLMConfig:
     frequency_penalty: Optional[float]
     """Penalizes new tokens based on their frequency in text so far (-2.0-2.0)."""
 
-
 class EmbeddingConfig:
     """Embedding configuration structure."""
 
@@ -511,7 +471,6 @@ class EmbeddingConfig:
     api_key: Optional[SecretStr]
     """The API key."""
 
-
 class RagConfig:
     """RAG (Retrieval Augmented Generation) configuration structure."""
 
@@ -527,14 +486,12 @@ class RagConfig:
     milvus_dimensions: Optional[int]
     """The dimensions for Milvus vectors."""
 
-
 class DebugConfig:
     """Debug configuration structure."""
 
     log_level: Optional[str]
     """The logging level to use."""
 
-
 class TemplateManagerConfig:
     """Template manager configuration structure."""
 
@@ -547,7 +504,6 @@ class TemplateManagerConfig:
     template_suffix: Optional[str]
     """The suffix of the templates."""
 
-
 class TemplateConfig:
     """Template configuration structure."""
 
@@ -632,7 +588,6 @@ class TemplateConfig:
     chap_summary_template: str
     """The name of the chap summary template which will be used to generate a chapter summary."""
 
-
 class RoutingConfig:
     """Routing configuration structure for controlling request dispatching behavior."""
 
@@ -648,7 +603,6 @@ class RoutingConfig:
     cooldown_time: Optional[int]
     """Time to cooldown a deployment after failure in seconds."""
 
-
 class GeneralConfig:
     """General configuration structure for application-wide settings."""
 
@@ -658,7 +612,6 @@ class GeneralConfig:
     use_json_repair: bool
     """Whether to automatically repair malformed JSON."""
 
-
 class ToolBoxConfig:
     """Configuration for toolbox functionality."""
 
@@ -668,7 +621,6 @@ class ToolBoxConfig:
     data_module_name: str
     """The name of the module containing the data."""
 
-
 class PymitterConfig:
     """Pymitter configuration structure for controlling event emission and listener behavior."""
 
@@ -681,7 +633,6 @@ class PymitterConfig:
     max_listeners: int
     """The maximum number of listeners per event. -1 means unlimited."""
 
-
 class Config:
     """Configuration structure containing all system components."""
 
@@ -715,27 +666,23 @@ class Config:
     pymitter: PymitterConfig
     """Pymitter configuration."""
 
-
 CONFIG: Config
 
-
 class SecretStr:
     """A string that should not be exposed."""
 
     def __init__(self, source: str) -> None: ...
-
     def get_secret_value(self) -> str:
         """Expose the secret string."""
 
-
 TEMPLATE_MANAGER: TemplateManager
 
-
 class Event:
     """Event class that represents a hierarchical event with segments.
 
     Events can be constructed from strings, lists of strings, or other Events.
     """
+
     segments: List[str]
 
     def __init__(self, segments: Optional[List[str]] = None) -> None:
@@ -841,12 +788,9 @@ class Event:
         """
 
     def __hash__(self) -> int: ...
-
     def __eq__(self, other: object) -> bool: ...
-
     def __ne__(self, other: object) -> bool: ...
 
-
 class TaskStatus(StrEnum, str):
     """Enumeration of possible task statuses."""
 
@@ -864,3 +808,39 @@ class TaskStatus(StrEnum, str):
 
     Cancelled: TaskStatus
     """Task has been cancelled."""
+
+class TEIClient:
+    """Client for TEI reranking service.
+
+    Handles communication with a TEI reranking service to reorder text snippets
+    based on their relevance to a query.
+    """
+
+    def __init__(self, base_url: str) -> None:
+        """Initialize the TEI client.
+
+        Args:
+            base_url: URL to the TEI reranking service
+        """
+
+    async def arerank(
+        self,
+        query: str,
+        texts: List[str],
+        truncate: bool = False,
+        truncation_direction: Literal["Left", "Right"] = "Left",
+    ) -> List[Tuple[int, float]]:
+        """Rerank texts based on relevance to query.
+
+        Args:
+            query: The query to match texts against
+            texts: List of text snippets to rerank
+            truncate: Whether to truncate texts to fit model context
+            truncation_direction: Direction to truncate from ("Left" or "Right")
+
+        Returns:
+            List of tuples containing (original_index, relevance_score)
+
+        Raises:
+            RuntimeError: If reranking fails or truncation_direction is invalid
+        """