fabricatio 0.2.12.dev1__cp312-cp312-win_amd64.whl → 0.2.12.dev3__cp312-cp312-win_amd64.whl

fabricatio/actions/article_rag.py

@@ -4,9 +4,10 @@ from asyncio import gather
 from pathlib import Path
 from typing import List, Optional
 
-from pydantic import PositiveInt
+from pydantic import Field, PositiveInt
 
 from fabricatio import BibManager
+from fabricatio.capabilities.advanced_rag import AdvancedRAG
 from fabricatio.capabilities.censor import Censor
 from fabricatio.capabilities.extract import Extract
 from fabricatio.capabilities.rag import RAG
@@ -18,6 +19,7 @@ from fabricatio.models.extra.article_essence import ArticleEssence
 from fabricatio.models.extra.article_main import Article, ArticleChapter, ArticleSection, ArticleSubsection
 from fabricatio.models.extra.article_outline import ArticleOutline
 from fabricatio.models.extra.rule import RuleSet
+from fabricatio.models.kwargs_types import ChooseKwargs, LLMKwargs
 from fabricatio.utils import ask_retain, ok
 
 TYPST_CITE_USAGE = (
@@ -27,17 +29,17 @@ TYPST_CITE_USAGE = (
     "Illegal citing syntax examples(seperated by |): [[1],[2],[3]]|[[1],[1-2]]\n"
     "Those reference mark shall not be omitted during the extraction\n"
     "It's recommended to cite multiple references that supports your conclusion at a time.\n"
-    "Wrapp inline expression using $ $,like '$>5m$' '$89%$' , and wrapp block equation using $$ $$. if you are using '$' as the money unit, you should add a '\\' before it to avoid being interpreted as a inline equation. For example 'The pants worths 5\\$.'\n"
+    "Wrap inline expression with '\\(' and '\\)',like '\\(>5m\\)' '\\(89%\\)', and wrap block equation with '\\[' and '\\]'.\n"
     "In addition to that, you can add a label outside the block equation which can be used as a cross reference identifier, the label is a string wrapped in `<` and `>` like `<energy-release-rate-equation>`.Note that the label string should be a summarizing title for the equation being labeled.\n"
     "you can refer to that label by using the syntax with prefix of `@eqt:`, which indicate that this notation is citing a label from the equations. For example ' @eqt:energy-release-rate-equation ' DO remember that the notation shall have both suffixed and prefixed space char which enable the compiler to distinguish the notation from the plaintext."
-    "Below is a usage example:\n"
+    "Below is two usage example:\n"
     "```typst\n"
     "See @eqt:mass-energy-equation , it's the foundation of physics.\n"
-    "$$\n"
+    "\\[\n"
     "E = m c^2\n"
-    "$$  <mass-energy-equation>\n\n\n"
-    "In @eqt:mass-energy-equation , $m$ stands for mass, $c$ stands for speed of light, and $E$ stands for energy. \n"
-    "```"
+    "\\]  <mass-energy-equation>\n\n\n"
+    "In @eqt:mass-energy-equation , \\(m\\) stands for mass, \\(c\\) stands for speed of light, and \\(E\\) stands for energy. \n"
+    "```\n"
 )
 
 
@@ -50,9 +52,9 @@ class WriteArticleContentRAG(Action, RAG, Extract):
     """The limit of references to be retrieved"""
     threshold: float = 0.62
     """The threshold of relevance"""
-    extractor_model: str
+    extractor_model: LLMKwargs
     """The model to use for extracting the content from the retrieved references."""
-    query_model: str
+    query_model: LLMKwargs
     """The model to use for querying the database"""
     supervisor: bool = False
     """Whether to use supervisor mode"""
@@ -154,12 +156,12 @@ class WriteArticleContentRAG(Action, RAG, Extract):
                 f"Above is the subsection titled `{subsec.title}`.\n"
                 f"I need you to extract the content to update my subsection obj provided below.\n{self.req}"
                 f"{subsec.display()}\n",
-                model=self.extractor_model,
+                **self.extractor_model,
             ),
             "Failed to propose new subsection.",
         )
         for p in new_subsec.paragraphs:
-            p.content = cm.apply(p.content).replace("$$", "\n$$\n")
+            p.content = cm.apply(p.content)
         subsec.update_from(new_subsec)
         logger.debug(f"{subsec.title}:rpl\n{subsec.display()}")
         return subsec
@@ -175,28 +177,16 @@ class WriteArticleContentRAG(Action, RAG, Extract):
         extra_instruction: str = "",
     ) -> str:
         """Write the raw paragraphs of the subsec."""
-        return (
-            (
-                await self.aask(
-                    f"{cm.as_prompt()}\nAbove is some related reference from other auther retrieved for you."
-                    f"{article_outline.finalized_dump()}\n\nAbove is my article outline, I m writing graduate thesis titled `{article.title}`. "
-                    f"More specifically, i m witting the Chapter `{chap.title}` >> Section `{sec.title}` >> Subsection `{subsec.title}`.\n"
-                    f"Please help me write the paragraphs of the subsec mentioned above, which is `{subsec.title}`.\n"
-                    f"{self.req}\n"
-                    f"You SHALL use `{article.language}` as writing language.\n{extra_instruction}\n"
-                    f"Do not use numbered list to display the outcome, you should regard you are writing the main text of the thesis.\n"
-                    f"You should not copy others' works from the references directly on to my thesis, we can only harness the conclusion they have drawn.\n"
-                    f"No extra explanation is allowed."
-                )
-            )
-            .replace(r" \( ", "$")
-            .replace(r" \) ", "$")
-            .replace(r"\(", "$")
-            .replace(r"\)", "$")
-            .replace("\\[\n", "$$\n")
-            .replace("\\[ ", "$$\n")
-            .replace("\n\\]", "\n$$")
-            .replace(" \\]", "\n$$")
+        return await self.aask(
+            f"{cm.as_prompt()}\nAbove is some related reference from other auther retrieved for you."
+            f"{article_outline.finalized_dump()}\n\nAbove is my article outline, I m writing graduate thesis titled `{article.title}`. "
+            f"More specifically, i m witting the Chapter `{chap.title}` >> Section `{sec.title}` >> Subsection `{subsec.title}`.\n"
+            f"Please help me write the paragraphs of the subsec mentioned above, which is `{subsec.title}`.\n"
+            f"{self.req}\n"
+            f"You SHALL use `{article.language}` as writing language.\n{extra_instruction}\n"
+            f"Do not use numbered list to display the outcome, you should regard you are writing the main text of the thesis.\n"
+            f"You should not copy others' works from the references directly on to my thesis, we can only harness the conclusion they have drawn.\n"
+            f"No extra explanation is allowed."
         )
 
     async def search_database(
@@ -222,7 +212,7 @@ class WriteArticleContentRAG(Action, RAG, Extract):
         ref_q = ok(
             await self.arefined_query(
                 search_req,
-                model=self.query_model,
+                **self.query_model,
             ),
             "Failed to refine query.",
         )
@@ -232,7 +222,7 @@ class WriteArticleContentRAG(Action, RAG, Extract):
         ret = await self.aretrieve(
             ref_q,
             ArticleChunk,
-            final_limit=self.ref_limit,
+            max_accepted=self.ref_limit,
             result_per_query=self.result_per_query,
             similarity_threshold=self.threshold,
         )
@@ -240,7 +230,7 @@ class WriteArticleContentRAG(Action, RAG, Extract):
         cm.add_chunks(ok(ret))
         ref_q = await self.arefined_query(
             f"{cm.as_prompt()}\n\nAbove is the retrieved references in the first RAG, now we need to perform the second RAG.\n\n{search_req}",
-            model=self.query_model,
+            **self.query_model,
         )
 
         if ref_q is None:
@@ -252,7 +242,7 @@ class WriteArticleContentRAG(Action, RAG, Extract):
         ret = await self.aretrieve(
             ref_q,
             ArticleChunk,
-            final_limit=int(self.ref_limit * self.search_increment_multiplier),
+            max_accepted=int(self.ref_limit * self.search_increment_multiplier),
             result_per_query=int(self.result_per_query * self.search_increment_multiplier),
             similarity_threshold=self.threshold,
         )
@@ -262,18 +252,19 @@ class WriteArticleContentRAG(Action, RAG, Extract):
         cm.add_chunks(ret)
 
 
-class ArticleConsultRAG(Action, RAG):
+class ArticleConsultRAG(Action, AdvancedRAG):
     """Write an article based on the provided outline."""
 
     output_key: str = "consult_count"
-
+    search_increment_multiplier: float = 1.6
+    """The multiplier to increase the limit of references to retrieve per query."""
     ref_limit: int = 20
     """The final limit of references."""
     ref_per_q: int = 3
     """The limit of references to retrieve per query."""
     similarity_threshold: float = 0.62
     """The similarity threshold of references to retrieve."""
-    ref_q_model: Optional[str] = None
+    ref_q_model: ChooseKwargs = Field(default_factory=ChooseKwargs)
     """The model to use for refining query."""
     req: str = TYPST_CITE_USAGE
     """The request for the rag model."""
@@ -285,7 +276,7 @@ class ArticleConsultRAG(Action, RAG):
         from questionary import confirm, text
         from rich import print as r_print
 
-        from fabricatio.rust import convert_all_block_tex, convert_all_inline_tex
+        from fabricatio.rust import convert_all_block_tex, convert_all_inline_tex, fix_misplaced_labels
 
         self.target_collection = collection_name or self.safe_target_collection
 
@@ -295,16 +286,19 @@ class ArticleConsultRAG(Action, RAG):
         while (req := await text("User: ").ask_async()) is not None:
             if await confirm("Empty the cm?").ask_async():
                 cm.empty()
-            ref_q = await self.arefined_query(req, model=self.ref_q_model)
-            refs = await self.aretrieve(
-                ok(ref_q, "Failed to refine query."),
-                ArticleChunk,
-                final_limit=self.ref_limit,
+            await self.clued_search(
+                req,
+                cm,
+                refinery_kwargs=self.ref_q_model,
+                expand_multiplier=self.search_increment_multiplier,
+                base_accepted=self.ref_limit,
                 result_per_query=self.ref_per_q,
                 similarity_threshold=self.similarity_threshold,
             )
 
-            ret = await self.aask(f"{cm.add_chunks(refs).as_prompt()}\n{self.req}\n{req}")
+            ret = await self.aask(f"{cm.as_prompt()}\n{self.req}\n{req}")
+
+            ret = fix_misplaced_labels(ret)
             ret = convert_all_inline_tex(ret)
             ret = convert_all_block_tex(ret)
             ret = cm.apply(ret)
@@ -396,7 +390,7 @@ class TweakArticleRAG(Action, RAG, Censor):
         await self.censor_obj_inplace(
             subsec,
             ruleset=ruleset,
-            reference=f"{'\n\n'.join(d.display() for d in await self.aretrieve(refind_q, document_model=ArticleEssence, final_limit=self.ref_limit))}\n\n"
+            reference=f"{'\n\n'.join(d.display() for d in await self.aretrieve(refind_q, document_model=ArticleEssence, max_accepted=self.ref_limit))}\n\n"
             f"You can use Reference above to rewrite the `{subsec.__class__.__name__}`.\n"
             f"You should Always use `{subsec.language}` as written language, "
             f"which is the original language of the `{subsec.title}`. "

fabricatio/actions/output.py

@@ -1,9 +1,10 @@
 """Dump the finalized output to a file."""
 
 from pathlib import Path
-from typing import Any, Dict, Iterable, List, Mapping, Optional, Type
+from typing import Any, Iterable, List, Mapping, Optional, Type
 
 from fabricatio import TEMPLATE_MANAGER
+from fabricatio.fs import dump_text
 from fabricatio.journal import logger
 from fabricatio.models.action import Action
 from fabricatio.models.generic import FinalizedDumpAble, FromMapping, PersistentAble
@@ -51,7 +52,7 @@ class RenderedDump(Action, LLMUsage):
 
     async def _execute(
         self,
-        to_dump: Dict[str, Any],
+        to_dump: FinalizedDumpAble,
         task_input: Optional[Task] = None,
         dump_path: Optional[str | Path] = None,
         **_,
@@ -66,8 +67,16 @@ class RenderedDump(Action, LLMUsage):
                 "Could not find the path of file to dump the data.",
             )
         )
+
         logger.info(f"Saving output to {dump_path.as_posix()}")
-        return TEMPLATE_MANAGER.render_template(self.template_name, to_dump)
+        dump_text(
+            dump_path,
+            TEMPLATE_MANAGER.render_template(
+                self.template_name, {to_dump.__class__.__name__: to_dump.finalized_dump()}
+            ),
+        )
+
+        return dump_path.as_posix()
 
 
 class PersistentAll(Action, LLMUsage):

fabricatio/capabilities/advanced_rag.py

@@ -0,0 +1,56 @@
+"""Advanced RAG (Retrieval Augmented Generation) model."""
+
+from typing import Optional, Unpack
+
+from fabricatio.capabilities.rag import RAG
+from fabricatio.journal import logger
+from fabricatio.models.adv_kwargs_types import FetchKwargs
+from fabricatio.models.extra.aricle_rag import ArticleChunk, CitationManager
+from fabricatio.models.kwargs_types import ChooseKwargs
+
+
+class AdvancedRAG(RAG):
+    """A class representing the Advanced RAG (Retrieval Augmented Generation) model."""
+
+    async def clued_search(
+        self,
+        requirement: str,
+        cm: CitationManager,
+        max_capacity: int = 40,
+        max_round: int = 3,
+        expand_multiplier: float = 1.4,
+        base_accepted: int = 12,
+        refinery_kwargs: Optional[ChooseKwargs] = None,
+        **kwargs: Unpack[FetchKwargs],
+    ) -> CitationManager:
+        """Asynchronously performs a clued search based on a given requirement and citation manager."""
+        if max_round<=0:
+            raise ValueError("max_round should be greater than 0")
+        if max_round == 1:
+            logger.warning(
+                "max_round should be greater than 1, otherwise it behaves nothing different from the `self.aretrieve`"
+            )
+
+        refinery_kwargs = refinery_kwargs or {}
+
+        for i in range(max_round + 1, 1):
+            logger.info(f"Round [{i + 1}/{max_round}] search started.")
+            ref_q = await self.arefined_query(
+                f"{cm.as_prompt()}\n\nAbove is the retrieved references in the {i - 1}th RAG, now we need to perform the {i}th RAG."
+                f"\n\n{requirement}",
+                **refinery_kwargs,
+            )
+            if ref_q is None:
+                logger.error(f"At round [{i + 1}/{max_round}] search, failed to refine the query, exit.")
+                return cm
+            refs = await self.aretrieve(ref_q, ArticleChunk, base_accepted, **kwargs)
+
+            if (max_capacity := max_capacity - len(refs)) < 0:
+                cm.add_chunks(refs[0:max_capacity])
+                logger.debug(f"At round [{i + 1}/{max_round}] search, the capacity is not enough, exit.")
+                return cm
+
+            cm.add_chunks(refs)
+            base_accepted = int(base_accepted * expand_multiplier)
+        logger.debug(f"Exceeded max_round: {max_round}, exit.")
+        return cm

fabricatio/capabilities/rag.py

@@ -189,7 +189,7 @@ class RAG(EmbeddingUsage):
         self,
         query: List[str] | str,
         document_model: Type[D],
-        final_limit: int = 20,
+        max_accepted: int = 20,
         **kwargs: Unpack[FetchKwargs],
     ) -> List[D]:
         """Retrieve data from the collection.
@@ -197,7 +197,7 @@ class RAG(EmbeddingUsage):
         Args:
             query (List[str] | str): The query to be used for retrieval.
             document_model (Type[D]): The model class used to convert retrieved data into document objects.
-            final_limit (int): The final limit on the number of results to return.
+            max_accepted (int): The final limit on the number of results to return.
             **kwargs (Unpack[FetchKwargs]): Additional keyword arguments for retrieval.
 
         Returns:
@@ -211,9 +211,9 @@ class RAG(EmbeddingUsage):
                 document_model=document_model,
                 **kwargs,
             )
-        )[:final_limit]
+        )[:max_accepted]
 
-    async def arefined_query(self, question: List[str] | str, **kwargs: Unpack[ChooseKwargs]) -> Optional[List[str]]:
+    async def arefined_query(self, question: List[str] | str, **kwargs: Unpack[ChooseKwargs[Optional[List[str]]]]) -> Optional[List[str]]:
         """Refines the given question using a template.
 
         Args:

fabricatio/models/extra/article_base.py

@@ -1,6 +1,6 @@
 """A foundation for hierarchical document components with dependency tracking."""
 
-from abc import ABC, abstractmethod
+from abc import ABC
 from enum import StrEnum
 from typing import Generator, List, Optional, Self, Tuple
 
@@ -18,7 +18,8 @@ from fabricatio.models.generic import (
     Titled,
     WordCount,
 )
-from fabricatio.rust import comment
+from fabricatio.rust import split_out_metadata, to_metadata, word_count
+from fabricatio.utils import fallback_kwargs
 from pydantic import Field
 
 
@@ -49,21 +50,46 @@ class ArticleMetaData(SketchedAble, Described, WordCount, Titled, Language):
     @property
     def typst_metadata_comment(self) -> str:
         """Generates a comment for the metadata of the article component."""
-        return comment(
-            (f"Desc:\n  {self.description}\n" if self.description else "")
-            + (f"Aims:\n  {'\n  '.join(self.aims)}\n" if self.aims else "")
-            + (f"Expected Word Count:{self.expected_word_count}" if self.expected_word_count else "")
-
+        return to_metadata(self.model_dump(include={"description", "aims", "expected_word_count"}, by_alias=True))
+
+
+class FromTypstCode(ArticleMetaData):
+    """Base class for article components that can be created from a Typst code snippet."""
+
+    @classmethod
+    def from_typst_code(cls, title: str, body: str, **kwargs) -> Self:
+        """Converts a Typst code snippet into an article component."""
+        data, body = split_out_metadata(body)
+
+        return cls(
+            heading=title,
+            **fallback_kwargs(
+                data or {},
+                elaboration="",
+                expected_word_count=word_count(body),
+                aims=[],
+            ),
+            **kwargs,
         )
 
 
+class ToTypstCode(ArticleMetaData):
+    """Base class for article components that can be converted to a Typst code snippet."""
+
+    def to_typst_code(self) -> str:
+        """Converts the component into a Typst code snippet for rendering."""
+        return f"{self.title}\n{self.typst_metadata_comment}\n"
+
+
 class ArticleOutlineBase(
-    ArticleMetaData,
     ResolveUpdateConflict,
     ProposedUpdateAble,
     PersistentAble,
     ModelHash,
     Introspect,
+    FromTypstCode,
+    ToTypstCode,
+    ABC,
 ):
     """Base class for article outlines."""
 
@@ -89,17 +115,13 @@ class ArticleOutlineBase(
         """Updates the current instance with the attributes of another instance."""
         return self.update_metadata(other)
 
-    @abstractmethod
-    def to_typst_code(self) -> str:
-        """Converts the component into a Typst code snippet for rendering."""
-
 
 class SubSectionBase(ArticleOutlineBase):
     """Base class for article sections and subsections."""
 
     def to_typst_code(self) -> str:
         """Converts the component into a Typst code snippet for rendering."""
-        return f"=== {self.title}\n{self.typst_metadata_comment}\n"
+        return f"=== {super().to_typst_code()}"
 
     def introspect(self) -> str:
         """Introspects the article subsection outline."""
@@ -124,9 +146,7 @@ class SectionBase[T: SubSectionBase](ArticleOutlineBase):
         Returns:
             str: The formatted Typst code snippet.
         """
-        return f"== {self.title}\n{self.typst_metadata_comment}\n" + "\n\n".join(
-            subsec.to_typst_code() for subsec in self.subsections
-        )
+        return f"== {super().to_typst_code()}" + "\n\n".join(subsec.to_typst_code() for subsec in self.subsections)
 
     def resolve_update_conflict(self, other: Self) -> str:
         """Resolve update errors in the article outline."""
@@ -169,9 +189,7 @@ class ChapterBase[T: SectionBase](ArticleOutlineBase):
 
     def to_typst_code(self) -> str:
         """Converts the chapter into a Typst formatted code snippet for rendering."""
-        return f"= {self.title}\n{self.typst_metadata_comment}\n" + "\n\n".join(
-            sec.to_typst_code() for sec in self.sections
-        )
+        return f"= {super().to_typst_code()}" + "\n\n".join(sec.to_typst_code() for sec in self.sections)
 
     def resolve_update_conflict(self, other: Self) -> str:
         """Resolve update errors in the article outline."""
@@ -203,12 +221,13 @@ class ChapterBase[T: SectionBase](ArticleOutlineBase):
         return ""
 
 
-class ArticleBase[T: ChapterBase](FinalizedDumpAble, AsPrompt, WordCount, Described, Titled, Language, ABC):
+class ArticleBase[T: ChapterBase](FinalizedDumpAble, AsPrompt, FromTypstCode, ToTypstCode, ABC):
     """Base class for article outlines."""
 
-    title: str = Field(alias="heading", description=Titled.model_fields["title"].description)
-    description: str = Field(alias="abstract")
-    """The abstract serves as a concise summary of an academic article, encapsulating its core purpose, methodologies, key results,
+    description: str = Field(
+        alias="elaboration",
+    )
+    """The abstract of this article, which serves as a concise summary of an academic article, encapsulating its core purpose, methodologies, key results,
     and conclusions while enabling readers to rapidly assess the relevance and significance of the study.
     Functioning as the article's distilled essence, it succinctly articulates the research problem, objectives,
     and scope, providing a roadmap for the full text while also facilitating database indexing, literature reviews,
@@ -293,6 +312,10 @@ class ArticleBase[T: ChapterBase](FinalizedDumpAble, AsPrompt, WordCount, Descri
         for _, _, subsec in self.iter_subsections():
             yield subsec.title
 
+    def to_typst_code(self) -> str:
+        """Generates the Typst code representation of the article."""
+        return f"// #{super().to_typst_code()}\n\n" + "\n\n".join(a.to_typst_code() for a in self.chapters)
+
     def finalized_dump(self) -> str:
         """Generates standardized hierarchical markup for academic publishing systems.
 
@@ -313,26 +336,16 @@ class ArticleBase[T: ChapterBase](FinalizedDumpAble, AsPrompt, WordCount, Descri
             === Implementation Details
             == Evaluation Protocol
         """
-        return (
-            comment(
-                f"Title:{self.title}\n"
-                + (f"Desc:\n{self.description}\n" if self.description else "")
-                + f"Word Count:{self.expected_word_count}"
-                if self.expected_word_count
-                else ""
-            )
-            + "\n\n"
-            + "\n\n".join(a.to_typst_code() for a in self.chapters)
-        )
+        return self.to_typst_code()
 
-    def avg_chap_wordcount[S](self: S) -> S:
+    def avg_chap_wordcount[S: "ArticleBase"](self: S) -> S:
         """Set all chap have same word count sum up to be `self.expected_word_count`."""
         avg = int(self.expected_word_count / len(self.chapters))
         for c in self.chapters:
             c.expected_word_count = avg
         return self
 
-    def avg_sec_wordcount[S](self: S) -> S:
+    def avg_sec_wordcount[S: "ArticleBase"](self: S) -> S:
         """Set all sec have same word count sum up to be `self.expected_word_count`."""
         for c in self.chapters:
             avg = int(c.expected_word_count / len(c.sections))
@@ -340,7 +353,7 @@ class ArticleBase[T: ChapterBase](FinalizedDumpAble, AsPrompt, WordCount, Descri
                 s.expected_word_count = avg
         return self
 
-    def avg_subsec_wordcount[S](self: S) -> S:
+    def avg_subsec_wordcount[S: "ArticleBase"](self: S) -> S:
         """Set all subsec have same word count sum up to be `self.expected_word_count`."""
         for _, s in self.iter_sections():
             avg = int(s.expected_word_count / len(s.subsections))
@@ -348,6 +361,6 @@ class ArticleBase[T: ChapterBase](FinalizedDumpAble, AsPrompt, WordCount, Descri
                 ss.expected_word_count = avg
         return self
 
-    def avg_wordcount_recursive(self) -> Self:
+    def avg_wordcount_recursive[S: "ArticleBase"](self: S) -> S:
         """Set all chap, sec, subsec have same word count sum up to be `self.expected_word_count`."""
-        return self.avg_chap_wordcount().avg_sec_wordcount().avg_sec_wordcount()
+        return self.avg_chap_wordcount().avg_sec_wordcount().avg_subsec_wordcount()

fabricatio/models/extra/article_main.py

@@ -18,11 +18,16 @@ from fabricatio.models.extra.article_outline import (
     ArticleSubsectionOutline,
 )
 from fabricatio.models.generic import Described, PersistentAble, SequencePatch, SketchedAble, WithRef, WordCount
-from fabricatio.rust import convert_all_block_tex, convert_all_inline_tex, word_count
-from fabricatio.utils import fallback_kwargs
+from fabricatio.rust import (
+    convert_all_block_tex,
+    convert_all_inline_tex,
+    fix_misplaced_labels,
+    split_out_metadata,
+    word_count,
+)
 from pydantic import Field, NonNegativeInt
 
-PARAGRAPH_SEP = "// - - -"
+PARAGRAPH_SEP = "\n\n// - - -\n\n"
 
 
 class Paragraph(SketchedAble, WordCount, Described):
@@ -93,17 +98,17 @@ class ArticleSubsection(SubSectionBase):
         Returns:
             str: Typst code snippet for rendering.
         """
-        return super().to_typst_code() + f"\n\n{PARAGRAPH_SEP}\n\n".join(p.content for p in self.paragraphs)
+        return super().to_typst_code() + PARAGRAPH_SEP.join(p.content for p in self.paragraphs)
 
     @classmethod
-    def from_typst_code(cls, title: str, body: str) -> Self:
+    def from_typst_code(cls, title: str, body: str, **kwargs) -> Self:
         """Creates an Article object from the given Typst code."""
-        return cls(
-            heading=title,
-            elaboration="",
-            paragraphs=[Paragraph.from_content(p) for p in body.split(PARAGRAPH_SEP)],
-            expected_word_count=word_count(body),
-            aims=[],
+        _, para_body = split_out_metadata(body)
+
+        return super().from_typst_code(
+            title,
+            body,
+            paragraphs=[Paragraph.from_content(p) for p in para_body.split(PARAGRAPH_SEP)],
         )
 
 
@@ -111,16 +116,14 @@ class ArticleSection(SectionBase[ArticleSubsection]):
     """Atomic argumentative unit with high-level specificity."""
 
     @classmethod
-    def from_typst_code(cls, title: str, body: str) -> Self:
+    def from_typst_code(cls, title: str, body: str, **kwargs) -> Self:
         """Creates an Article object from the given Typst code."""
-        return cls(
+        return super().from_typst_code(
+            title,
+            body,
             subsections=[
                 ArticleSubsection.from_typst_code(*pack) for pack in extract_sections(body, level=3, section_char="=")
             ],
-            heading=title,
-            elaboration="",
-            expected_word_count=word_count(body),
-            aims=[],
         )
 
 
@@ -128,21 +131,18 @@ class ArticleChapter(ChapterBase[ArticleSection]):
     """Thematic progression implementing research function."""
 
     @classmethod
-    def from_typst_code(cls, title: str, body: str) -> Self:
+    def from_typst_code(cls, title: str, body: str, **kwargs) -> Self:
         """Creates an Article object from the given Typst code."""
-        return cls(
+        return super().from_typst_code(
+            title,
+            body,
             sections=[
                 ArticleSection.from_typst_code(*pack) for pack in extract_sections(body, level=2, section_char="=")
             ],
-            heading=title,
-            elaboration="",
-            expected_word_count=word_count(body),
-            aims=[],
         )
 
 
 class Article(
-    SketchedAble,
     WithRef[ArticleOutline],
     PersistentAble,
     ArticleBase[ArticleChapter],
@@ -165,22 +165,11 @@ class Article(
         """Convert tex to typst code."""
         for _, _, subsec in self.iter_subsections():
             for p in subsec.paragraphs:
+                p.content = fix_misplaced_labels(p.content)
                 p.content = convert_all_inline_tex(p.content)
                 p.content = convert_all_block_tex(p.content)
         return self
 
-    def fix_wrapper(self) -> Self:
-        """Fix wrapper."""
-        for _, _, subsec in self.iter_subsections():
-            for p in subsec.paragraphs:
-                p.content = (
-                    p.content.replace(r" \( ", "$")
-                    .replace(r" \) ", "$")
-                    .replace("\\[\n", "$$\n")
-                    .replace("\n\\]", "\n$$")
-                )
-        return self
-
     @override
     def iter_subsections(self) -> Generator[Tuple[ArticleChapter, ArticleSection, ArticleSubsection], None, None]:
         return super().iter_subsections()  # pyright: ignore [reportReturnType]
@@ -268,16 +257,12 @@ class Article(
     @classmethod
     def from_typst_code(cls, title: str, body: str, **kwargs) -> Self:
         """Generates an article from the given Typst code."""
-        return cls(
+        return super().from_typst_code(
+            title,
+            body,
             chapters=[
                 ArticleChapter.from_typst_code(*pack) for pack in extract_sections(body, level=1, section_char="=")
             ],
-            heading=title,
-            **fallback_kwargs(
-                kwargs,
-                expected_word_count=word_count(body),
-                abstract="",
-            ),
         )
 
     @classmethod

fabricatio/models/extra/article_outline.py

@@ -1,7 +1,8 @@
 """A module containing the ArticleOutline class, which represents the outline of an academic paper."""
 
-from typing import Dict
+from typing import Dict, Self
 
+from fabricatio.fs.readers import extract_sections
 from fabricatio.models.extra.article_base import (
     ArticleBase,
     ChapterBase,
@@ -9,7 +10,7 @@ from fabricatio.models.extra.article_base import (
     SubSectionBase,
 )
 from fabricatio.models.extra.article_proposal import ArticleProposal
-from fabricatio.models.generic import PersistentAble, SketchedAble, WithRef
+from fabricatio.models.generic import PersistentAble, WithRef
 
 
 class ArticleSubsectionOutline(SubSectionBase):
@@ -18,14 +19,39 @@ class ArticleSubsectionOutline(SubSectionBase):
 
 class ArticleSectionOutline(SectionBase[ArticleSubsectionOutline]):
     """A slightly more detailed research component specification for academic paper generation, Must contain subsections."""
+    @classmethod
+    def from_typst_code(cls, title: str, body: str, **kwargs) -> Self:
+        """Parse the given Typst code into an ArticleSectionOutline instance."""
+        return super().from_typst_code(
+            title,
+            body,
+            subsections=[
+                ArticleSubsectionOutline.from_typst_code(*pack)
+                for pack in extract_sections(body, level=3, section_char="=")
+            ],
+        )
+
 
 
 class ArticleChapterOutline(ChapterBase[ArticleSectionOutline]):
     """Macro-structural unit implementing standard academic paper organization. Must contain sections."""
 
+    @classmethod
+    def from_typst_code(cls, title: str, body: str, **kwargs) -> Self:
+        """Parse the given Typst code into an ArticleChapterOutline instance."""
+        return super().from_typst_code(
+            title,
+            body,
+            sections=[
+                ArticleSectionOutline.from_typst_code(*pack)
+                for pack in extract_sections(body, level=2, section_char="=")
+            ],
+
+        )
+
+
 
 class ArticleOutline(
-    SketchedAble,
     WithRef[ArticleProposal],
     PersistentAble,
     ArticleBase[ArticleChapterOutline],
@@ -38,3 +64,15 @@ class ArticleOutline(
             "Original Article Proposal": self.referenced.display(),
             "Original Article Outline": self.display(),
         }
+
+    @classmethod
+    def from_typst_code(cls, title: str, body: str, **kwargs) -> Self:
+        """Parse the given Typst code into an ArticleOutline instance."""
+        return super().from_typst_code(
+            title,
+            body,
+            chapters=[
+                ArticleChapterOutline.from_typst_code(*pack)
+                for pack in extract_sections(body, level=1, section_char="=")
+            ],
+        )

fabricatio/rust.pyi

@@ -12,7 +12,9 @@ Key Features:
 """
 
 from pathlib import Path
-from typing import Any, Dict, List, Optional, overload
+from typing import Any, Dict, List, Optional, Tuple, overload
+
+from pydantic import JsonValue
 
 class TemplateManager:
     """Template rendering engine using Handlebars templates.
@@ -325,6 +327,16 @@ 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.
+
+    Args:
+        string: The input string containing misplaced labels.
+
+    Returns:
+        The fixed string with labels properly placed.
+    """
+
 def comment(string: str) -> str:
     """Add comment to the string.
 
@@ -344,3 +356,24 @@ def uncomment(string: str) -> str:
     Returns:
         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.
+
+    Args:
+        string: The input string containing metadata.
+
+    Returns:
+        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.
+
+    Args:
+        data: The Python object to be converted to YAML.
+
+    Returns:
+        The YAML string representation of the input data.
+    """
+

{fabricatio-0.2.12.dev1.dist-info → fabricatio-0.2.12.dev3.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: fabricatio
-Version: 0.2.12.dev1
+Version: 0.2.12.dev3
 Classifier: License :: OSI Approved :: MIT License
 Classifier: Programming Language :: Rust
 Classifier: Programming Language :: Python :: 3.12

{fabricatio-0.2.12.dev1.dist-info → fabricatio-0.2.12.dev3.dist-info}/RECORD

@@ -1,20 +1,21 @@
-fabricatio-0.2.12.dev1.dist-info/METADATA,sha256=rE72BRRmswpE0cWYnDFdoklrXGTLORE-lDpDJahO-0o,5263
-fabricatio-0.2.12.dev1.dist-info/WHEEL,sha256=jABKVkLC9kJr8mi_er5jOqpiQUjARSLXDUIIxDqsS50,96
-fabricatio-0.2.12.dev1.dist-info/licenses/LICENSE,sha256=do7J7EiCGbq0QPbMAL_FqLYufXpHnCnXBOuqVPwSV8Y,1088
+fabricatio-0.2.12.dev3.dist-info/METADATA,sha256=iMt48rBNujWl2lSSwiAHtT22G-K7zpXLPwnV8K_32mw,5263
+fabricatio-0.2.12.dev3.dist-info/WHEEL,sha256=jABKVkLC9kJr8mi_er5jOqpiQUjARSLXDUIIxDqsS50,96
+fabricatio-0.2.12.dev3.dist-info/licenses/LICENSE,sha256=do7J7EiCGbq0QPbMAL_FqLYufXpHnCnXBOuqVPwSV8Y,1088
 fabricatio/actions/article.py,sha256=SFl1zc0hz9vW2sW4VJm9-w8E7kLEty-2LzXi9wgWMmE,10905
-fabricatio/actions/article_rag.py,sha256=P3wAwLZK92_LDTtzODrmtIGdLGPiOejMKGY8ydLnnPU,18938
+fabricatio/actions/article_rag.py,sha256=dJJfPkfwnx8P4BebESOpPaeTfvER-1U68aPXpwMdDCQ,18723
 fabricatio/actions/fs.py,sha256=gJR14U4ln35nt8Z7OWLVAZpqGaLnED-r1Yi-lX22tkI,959
-fabricatio/actions/output.py,sha256=FoSEbvhPdWe8ARhgID7wkSH7kS8jb3BDqRUqpvIkPM0,8095
+fabricatio/actions/output.py,sha256=lX0HkDse3ypCzZgeF-8Dr-EnNFdBiE-WQ1iLPFlGM1g,8302
 fabricatio/actions/rag.py,sha256=KN-OWgcQjGmNgSZ-s5B8m4LpYKSGFJR8eq72mo2CP9k,3592
 fabricatio/actions/rules.py,sha256=dkvCgNDjt2KSO1VgPRsxT4YBmIIMeetZb5tiz-slYkU,3640
 fabricatio/actions/__init__.py,sha256=wVENCFtpVb1rLFxoOFJt9-8smLWXuJV7IwA8P3EfFz4,48
 fabricatio/capabilities/advanced_judge.py,sha256=selB0Gwf1F4gGJlwBiRo6gI4KOUROgh3WnzO3mZFEls,706
+fabricatio/capabilities/advanced_rag.py,sha256=gt4ccqyMSVwQLwsEyJrjPc-RXMZZBwObIjsh8A8ILbM,2386
 fabricatio/capabilities/censor.py,sha256=bBT5qy-kp7fh8g4Lz3labSwxwJ60gGd_vrkc6k1cZ1U,4719
 fabricatio/capabilities/check.py,sha256=kYqzohhv2bZfl1aKSUt7a8snT8YEl2zgha_ZdAdMMfQ,8622
 fabricatio/capabilities/correct.py,sha256=W_cInqlciNEhyMK0YI53jk4EvW9uAdge90IO9OElUmA,10420
 fabricatio/capabilities/extract.py,sha256=PMjkWvbsv57IYT7zzd_xbIu4eQqQjpcmBtJzqlWZhHY,2495
 fabricatio/capabilities/propose.py,sha256=hkBeSlmcTdfYWT-ph6nlbtHXBozi_JXqXlWcnBy3W78,2007
-fabricatio/capabilities/rag.py,sha256=kqcunWBC6oA4P1rzIG2Xu9zqSg73H3uKPF41JJQ1HVI,9595
+fabricatio/capabilities/rag.py,sha256=a48dEaWE-sniwLhNtGz4xlSmRi_-PdOgDzNaDwnam_g,9619
 fabricatio/capabilities/rating.py,sha256=iMtQs3H6vCjuEjiuuz4SRKMVaX7yff7MHWz-slYvi5g,17835
 fabricatio/capabilities/review.py,sha256=-EMZe0ADFPT6fPGmra16UPjJC1M3rAs6dPFdTZ88Fgg,5060
 fabricatio/capabilities/task.py,sha256=uks1U-4LNCUdwdRxAbJJjMc31hOw6jlrcYriuQQfb04,4475
@@ -32,10 +33,10 @@ fabricatio/models/adv_kwargs_types.py,sha256=kUO-SiZtFuz5cZCmMLnJJ9tjQ4-Zd_foo6R
 fabricatio/models/events.py,sha256=wiirk_ASg3iXDOZU_gIimci1VZVzWE1nDmxy-hQVJ9M,4150
 fabricatio/models/extra/advanced_judge.py,sha256=INUl_41C8jkausDekkjnEmTwNfLCJ23TwFjq2cM23Cw,1092
 fabricatio/models/extra/aricle_rag.py,sha256=p91JI8FmFSrHVg5KbhJq4w8vQdx4VUW75SrtQUi9ju4,10987
-fabricatio/models/extra/article_base.py,sha256=kUd9wSi1E0GcM1xBaHyrNyICzM0oYf_SyV6wjcTHmlM,14061
+fabricatio/models/extra/article_base.py,sha256=R8EOlQBs7hadjP9cxG_uc24DJDpiVpJW9Wb9DA6XmXE,14516
 fabricatio/models/extra/article_essence.py,sha256=mlIkkRMR3I1RtqiiOnmIE3Vy623L4eECumkRzryE1pw,2749
-fabricatio/models/extra/article_main.py,sha256=Oz6X7tHztb6zwnJQnw9XlgVwTp8kfI9ywkWUY3_fe_E,11824
-fabricatio/models/extra/article_outline.py,sha256=w7O0SHgC7exbptWVbR62FMHAueMgBpyWKVYMGGl_oj8,1427
+fabricatio/models/extra/article_main.py,sha256=zbWhXAvDPBfg3EHZRf7kVEChPZagqWbzeYFJdJtaDOY,11211
+fabricatio/models/extra/article_outline.py,sha256=C9WNZNSz6gyuw9lDp29qidPvHBTENaeqcHfCoE_Y7F4,2793
 fabricatio/models/extra/article_proposal.py,sha256=NbyjW-7UiFPtnVD9nte75re4xL2pD4qL29PpNV4Cg_M,1870
 fabricatio/models/extra/patches.py,sha256=_WNCxtYzzsVfUxI16vu4IqsLahLYRHdbQN9er9tqhC0,997
 fabricatio/models/extra/problem.py,sha256=8tTU-3giFHOi5j7NJsvH__JJyYcaGrcfsRnkzQNm0Ew,7216
@@ -50,7 +51,7 @@ fabricatio/models/tool.py,sha256=jQ51g4lwTPfsMF1nbreDJtBczbxIHoXcPuLSOqHliq8,125
 fabricatio/models/usages.py,sha256=0bzITf0vug9ZaN6qnjNfFB7T8BAvpXE0bvx0otFYLLA,33356
 fabricatio/parser.py,sha256=-RbW2yzfJiu2ARq-lZw4tfgsjY2rIZWtJpoUmaE6gJQ,6637
 fabricatio/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-fabricatio/rust.pyi,sha256=GJRLeeQ1UIaf5kOgJWX2GkwIacoTBk3yBKuD5cKW8i4,10489
+fabricatio/rust.pyi,sha256=kU-I0PD-tG0RGTy5xUobneXY-E06Qopjp4v1awpYmGw,11339
 fabricatio/rust_instances.py,sha256=Byeo8KHW_dJiXujJq7YPGDLBX5bHNDYbBc4sY3uubVY,313
 fabricatio/toolboxes/arithmetic.py,sha256=WLqhY-Pikv11Y_0SGajwZx3WhsLNpHKf9drzAqOf_nY,1369
 fabricatio/toolboxes/fs.py,sha256=l4L1CVxJmjw9Ld2XUpIlWfV0_Fu_2Og6d3E13I-S4aE,736
@@ -60,6 +61,6 @@ fabricatio/workflows/articles.py,sha256=ObYTFUqLUk_CzdmmnX6S7APfxcGmPFqnFr9pdjU7
 fabricatio/workflows/rag.py,sha256=-YYp2tlE9Vtfgpg6ROpu6QVO8j8yVSPa6yDzlN3qVxs,520
 fabricatio/workflows/__init__.py,sha256=5ScFSTA-bvhCesj3U9Mnmi6Law6N1fmh5UKyh58L3u8,51
 fabricatio/__init__.py,sha256=Rmvq2VgdS2u68vnOi2i5RbeWbAwrJDbk8D8D883PJWE,1022
-fabricatio/rust.cp312-win_amd64.pyd,sha256=GQijCHR-8G4LJIApA7L4158SZJCyzJHd9AGqi593EUs,4143616
-fabricatio-0.2.12.dev1.data/scripts/tdown.exe,sha256=HzNOjmihQ39UQc4s3Ko8tyZm9RJgwXruIiI8XU2Tk2o,3349504
-fabricatio-0.2.12.dev1.dist-info/RECORD,,
+fabricatio/rust.cp312-win_amd64.pyd,sha256=YARjZHId5jH51qUqMMIKn7RotIiwm8oT6UqBv8PotKI,4435456
+fabricatio-0.2.12.dev3.data/scripts/tdown.exe,sha256=rpmuLrxy-ZZiqVzrMYH_oFpanqkpE6dlI1aY9rCxCiA,3350016
+fabricatio-0.2.12.dev3.dist-info/RECORD,,