docvortex 0.2.1__py3-none-any.whl

docvortex/__init__.py

@@ -0,0 +1,21 @@
+"""快速多格式文档解析与转换引擎。"""
+
+from .version import __version__
+from .api import analyze, parse, convert
+from .api import postprocess as postprocess_document, render as render_artifact
+from .export.bundle import load_bundle
+from .result import AnalysisResult, DocumentResult, RenderArtifact, ExportResult
+
+__all__ = [
+    "__version__",
+    "analyze",
+    "postprocess_document",
+    "parse",
+    "render_artifact",
+    "convert",
+    "load_bundle",
+    "AnalysisResult",
+    "DocumentResult",
+    "RenderArtifact",
+    "ExportResult",
+]

docvortex/analyzers/__init__.py

@@ -0,0 +1,3 @@
+"""DocVortex 模块边界。"""
+
+__all__ = []

docvortex/analyzers/native/__init__.py

@@ -0,0 +1,37 @@
+"""Flash PDF、EPUB、HTML、OFD、CSV 与 Office 模型公开入口。"""
+
+from .models import (
+    CsvModel,
+    DocModel,
+    DocxModel,
+    EpubModel,
+    HtmlModel,
+    OfdModel,
+    OdpModel,
+    OdsModel,
+    OdtModel,
+    PdfModel,
+    PptModel,
+    PptxModel,
+    RtfModel,
+    XlsModel,
+    XlsxModel,
+)
+
+__all__ = [
+    "PdfModel",
+    "CsvModel",
+    "EpubModel",
+    "HtmlModel",
+    "OfdModel",
+    "RtfModel",
+    "DocModel",
+    "DocxModel",
+    "PptModel",
+    "PptxModel",
+    "XlsModel",
+    "XlsxModel",
+    "OdtModel",
+    "OdsModel",
+    "OdpModel",
+]

docvortex/analyzers/native/_shared/__init__.py

@@ -0,0 +1,3 @@
+"""Flash 各文档格式复用的轻量叶子能力。"""
+
+__all__: list[str] = []

docvortex/analyzers/native/_shared/hyperlink.py

@@ -0,0 +1,16 @@
+"""兼容导出位于 utils 层的共享超链接安全策略。"""
+
+from __future__ import annotations
+
+from ....foundation.hyperlink import (
+    DEFAULT_EXTERNAL_HYPERLINK_SCHEMES,
+    OFFICE_EXTERNAL_HYPERLINK_SCHEMES,
+    sanitize_hyperlink_target,
+)
+
+
+__all__ = [
+    "DEFAULT_EXTERNAL_HYPERLINK_SCHEMES",
+    "OFFICE_EXTERNAL_HYPERLINK_SCHEMES",
+    "sanitize_hyperlink_target",
+]

docvortex/analyzers/native/_shared/image.py

@@ -0,0 +1,8 @@
+"""保留原有导入入口；共享实现由下层模块唯一维护。"""
+
+from docvortex.foundation.image_encoding import (
+    image_to_b64str as image_to_b64str,
+    image_to_bytes as image_to_bytes,
+)
+
+__all__ = ["image_to_b64str", "image_to_bytes"]

docvortex/analyzers/native/_shared/markup/__init__.py

@@ -0,0 +1,47 @@
+"""保留原有导入入口；共享实现由下层模块唯一维护。"""
+
+from docvortex.content.markup import (
+    AnchorTextNormalization as AnchorTextNormalization,
+    AnchorVisibilityScope as AnchorVisibilityScope,
+    ElementStyle as ElementStyle,
+    FormulaDisplay as FormulaDisplay,
+    FormulaExtraction as FormulaExtraction,
+    FormulaSourceKind as FormulaSourceKind,
+    MarkupAnchorDocument as MarkupAnchorDocument,
+    MarkupAnchorPolicy as MarkupAnchorPolicy,
+    MarkupAnchorRegistry as MarkupAnchorRegistry,
+    MarkupContext as MarkupContext,
+    MarkupProjector as MarkupProjector,
+    MarkupStylesheet as MarkupStylesheet,
+    ResolvedMarkupImage as ResolvedMarkupImage,
+    TextStyle as TextStyle,
+    TextStyleDelta as TextStyleDelta,
+    canonical_anchor as canonical_anchor,
+    element_id as element_id,
+    extract_formula as extract_formula,
+    strip_formula_delimiters as strip_formula_delimiters,
+    visible_element_text as visible_element_text,
+)
+
+__all__ = [
+    "AnchorTextNormalization",
+    "AnchorVisibilityScope",
+    "ElementStyle",
+    "FormulaDisplay",
+    "FormulaExtraction",
+    "FormulaSourceKind",
+    "MarkupAnchorDocument",
+    "MarkupAnchorPolicy",
+    "MarkupAnchorRegistry",
+    "MarkupContext",
+    "MarkupProjector",
+    "MarkupStylesheet",
+    "ResolvedMarkupImage",
+    "TextStyle",
+    "TextStyleDelta",
+    "canonical_anchor",
+    "element_id",
+    "extract_formula",
+    "strip_formula_delimiters",
+    "visible_element_text",
+]

docvortex/analyzers/native/_shared/markup/anchors.py

@@ -0,0 +1,30 @@
+"""保留原有导入入口；共享实现由下层模块唯一维护。"""
+
+from docvortex.content.markup.anchors import (
+    AnchorTextNormalization as AnchorTextNormalization,
+    AnchorVisibilityScope as AnchorVisibilityScope,
+    MarkupAnchorDocument as MarkupAnchorDocument,
+    MarkupAnchorPolicy as MarkupAnchorPolicy,
+    MarkupAnchorRegistry as MarkupAnchorRegistry,
+    MarkupStylesheet as MarkupStylesheet,
+    TextStyle as TextStyle,
+    _HEADING_TAGS as _HEADING_TAGS,
+    _XHTML_WHITESPACE_RE as _XHTML_WHITESPACE_RE,
+    _XML_ID as _XML_ID,
+    canonical_anchor as canonical_anchor,
+    element_id as element_id,
+    local_name as local_name,
+    visible_element_text as visible_element_text,
+    visible_raw_text_with_style as visible_raw_text_with_style,
+)
+
+__all__ = [
+    "AnchorTextNormalization",
+    "AnchorVisibilityScope",
+    "MarkupAnchorDocument",
+    "MarkupAnchorPolicy",
+    "MarkupAnchorRegistry",
+    "canonical_anchor",
+    "element_id",
+    "visible_element_text",
+]

docvortex/analyzers/native/_shared/markup/formula.py

@@ -0,0 +1,37 @@
+"""保留原有导入入口；共享实现由下层模块唯一维护。"""
+
+from docvortex.content.markup.formula import (
+    FormulaDisplay as FormulaDisplay,
+    FormulaExtraction as FormulaExtraction,
+    FormulaSourceKind as FormulaSourceKind,
+    _DISPLAY_FORMULA_TOKENS as _DISPLAY_FORMULA_TOKENS,
+    _MATH_SCRIPT_TYPE_RE as _MATH_SCRIPT_TYPE_RE,
+    _PRESENTATION_ARITIES as _PRESENTATION_ARITIES,
+    _PRESENTATION_MATHML_TAGS as _PRESENTATION_MATHML_TAGS,
+    _TEX_ANNOTATION_ENCODINGS as _TEX_ANNOTATION_ENCODINGS,
+    _class_tokens as _class_tokens,
+    _data_formula as _data_formula,
+    _formula_display as _formula_display,
+    _is_supported_presentation_mathml as _is_supported_presentation_mathml,
+    _mathml_alttext as _mathml_alttext,
+    _normalized_attribute as _normalized_attribute,
+    _presentation_mathml as _presentation_mathml,
+    _script_display as _script_display,
+    _subtree_attribute as _subtree_attribute,
+    _tex_annotation as _tex_annotation,
+    _tex_script as _tex_script,
+    extract_formula as extract_formula,
+    is_tex_script as is_tex_script,
+    local_name as local_name,
+    mathml_to_latex as mathml_to_latex,
+    strip_formula_delimiters as strip_formula_delimiters,
+)
+
+__all__ = [
+    "FormulaDisplay",
+    "FormulaExtraction",
+    "FormulaSourceKind",
+    "extract_formula",
+    "is_tex_script",
+    "strip_formula_delimiters",
+]

docvortex/analyzers/native/_shared/markup/projector.py

@@ -0,0 +1,57 @@
+"""保留原有导入入口；共享实现由下层模块唯一维护。"""
+
+from docvortex.content.markup.projector import (
+    BLOCK_TAGS as BLOCK_TAGS,
+    BlockType as BlockType,
+    FormulaExtraction as FormulaExtraction,
+    MarkupContext as MarkupContext,
+    MarkupProjector as MarkupProjector,
+    MarkupStylesheet as MarkupStylesheet,
+    RAW_ALGORITHM as RAW_ALGORITHM,
+    ResolvedMarkupImage as ResolvedMarkupImage,
+    SKIPPED_TAGS as SKIPPED_TAGS,
+    TextStyle as TextStyle,
+    VISUAL_TYPE_MAPPING as VISUAL_TYPE_MAPPING,
+    _CAPTION_TOKENS as _CAPTION_TOKENS,
+    _FOOTNOTE_TOKENS as _FOOTNOTE_TOKENS,
+    _InlineProjectionSegment as _InlineProjectionSegment,
+    _InlineSpanDict as _InlineSpanDict,
+    _LIST_PAGE_BLOCK_TAGS as _LIST_PAGE_BLOCK_TAGS,
+    _MAX_TABLE_SPAN as _MAX_TABLE_SPAN,
+    _VISUAL_ELEMENT_TAGS as _VISUAL_ELEMENT_TAGS,
+    _WHITESPACE_RE as _WHITESPACE_RE,
+    _XLINK_HREF as _XLINK_HREF,
+    _append_inline_segment as _append_inline_segment,
+    _append_list_block_content as _append_list_block_content,
+    _raw_visual_type as _raw_visual_type,
+    _semantic_tokens as _semantic_tokens,
+    append_code_span as append_code_span,
+    append_equation_span as append_equation_span,
+    append_hyperlink_span as append_hyperlink_span,
+    append_text_span as append_text_span,
+    bounded_table_span as bounded_table_span,
+    clean_text_node as clean_text_node,
+    entity_text as entity_text,
+    extend_inline_spans as extend_inline_spans,
+    extract_formula as extract_formula,
+    inline_span_plain_text as inline_span_plain_text,
+    local_name as local_name,
+    strip_span_dicts as strip_span_dicts,
+    text_spans as text_spans,
+    visible_raw_text_with_style as visible_raw_text_with_style,
+    visible_text as visible_text,
+)
+
+__all__ = [
+    "BLOCK_TAGS",
+    "MarkupContext",
+    "MarkupProjector",
+    "ResolvedMarkupImage",
+    "SKIPPED_TAGS",
+    "bounded_table_span",
+    "clean_text_node",
+    "entity_text",
+    "local_name",
+    "visible_raw_text_with_style",
+    "visible_text",
+]

docvortex/analyzers/native/_shared/markup/styles.py

@@ -0,0 +1,19 @@
+"""保留原有导入入口；共享实现由下层模块唯一维护。"""
+
+from docvortex.content.markup.styles import (
+    ElementStyle as ElementStyle,
+    MarkupStylesheet as MarkupStylesheet,
+    TextStyle as TextStyle,
+    TextStyleDelta as TextStyleDelta,
+    _CSS_COMMENT_RE as _CSS_COMMENT_RE,
+    _CSS_IMPORTANT_RE as _CSS_IMPORTANT_RE,
+    _ParsedDeclarations as _ParsedDeclarations,
+    _SelectorCascade as _SelectorCascade,
+    _TEXT_STYLE_FIELDS as _TEXT_STYLE_FIELDS,
+    _VISIBILITY_FIELDS as _VISIBILITY_FIELDS,
+    _numeric_font_weight as _numeric_font_weight,
+    _parse_declarations as _parse_declarations,
+    local_name as local_name,
+)
+
+__all__ = ["ElementStyle", "MarkupStylesheet", "TextStyle", "TextStyleDelta"]

docvortex/analyzers/native/_shared/mathml.py

@@ -0,0 +1,17 @@
+"""保留原有导入入口；共享实现由下层模块唯一维护。"""
+
+from docvortex.content.mathml import (
+    _GREEK_MAP as _GREEK_MAP,
+    _LATEX_ESCAPE_RE as _LATEX_ESCAPE_RE,
+    _LATEX_MATH_TOKEN_ESCAPES as _LATEX_MATH_TOKEN_ESCAPES,
+    _OPERATOR_MAP as _OPERATOR_MAP,
+    _children as _children,
+    _convert as _convert,
+    _escape_math_token as _escape_math_token,
+    _escape_text as _escape_text,
+    _join_children as _join_children,
+    local_name as local_name,
+    mathml_to_latex as mathml_to_latex,
+)
+
+__all__ = ["mathml_to_latex"]

docvortex/analyzers/native/_shared/names.py

@@ -0,0 +1,7 @@
+"""保留原有导入入口；共享实现由下层模块唯一维护。"""
+
+from docvortex.foundation.xml_names import (
+    local_name as local_name,
+)
+
+__all__ = ["local_name"]