epub-translator 0.1.6__tar.gz → 0.1.7__tar.gz

{epub_translator-0.1.6 → epub_translator-0.1.7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: epub-translator
-Version: 0.1.6
+Version: 0.1.7
 Summary: Translate the epub book using LLM. The translated book will retain the original text and list the translated text side by side with the original text.
 License: MIT
 Keywords: epub,llm,translation,translator
@@ -24,6 +24,7 @@ Classifier: Topic :: Software Development :: Localization
 Classifier: Topic :: Text Processing :: Markup
 Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
 Requires-Dist: jinja2 (>=3.1.6,<4.0.0)
+Requires-Dist: mathml2latex (>=0.2.12,<0.3.0)
 Requires-Dist: openai (>=2.14.0,<3.0.0)
 Requires-Dist: resource-segmentation (>=0.0.7,<0.1.0)
 Requires-Dist: tiktoken (>=0.12.0,<1.0.0)
@@ -59,6 +60,13 @@ Translate EPUB books using Large Language Models while preserving the original t
 - **Flexible LLM Support**: Works with any OpenAI-compatible API endpoint
 - **Caching**: Built-in caching for progress recovery when translation fails
 
+## Use Cases
+
+- **Language Learning**: Read books in their original language with side-by-side translations
+- **Academic Research**: Access foreign literature with bilingual references
+- **Content Localization**: Prepare books for international audiences
+- **Cross-Cultural Reading**: Enjoy literature while understanding cultural nuances
+
 ## Installation
 
 ```bash
@@ -357,13 +365,6 @@ llm = LLM(
 )
 ```
 
-## Use Cases
-
-- **Language Learning**: Read books in their original language with side-by-side translations
-- **Academic Research**: Access foreign literature with bilingual references
-- **Content Localization**: Prepare books for international audiences
-- **Cross-Cultural Reading**: Enjoy literature while understanding cultural nuances
-
 ## Advanced Features
 
 ### Custom Translation Prompts

{epub_translator-0.1.6 → epub_translator-0.1.7}/epub_translator/data/translate.jinja

@@ -13,6 +13,9 @@ Translation rules:
 {% if user_prompt -%}
 User may provide additional requirements in <rules> tags before the source text. Follow them, but prioritize the rules above if conflicts arise.
 
+<rules>
+{{ user_prompt }}
+</rules>
 {% endif -%}
 
 Output only the translated text, nothing else.

{epub_translator-0.1.6 → epub_translator-0.1.7}/epub_translator/segment/__init__.py

@@ -21,6 +21,7 @@ from .text_segment import (
     TextPosition,
     TextSegment,
     combine_text_segments,
+    find_block_depth,
     incision_between,
     search_text_segments,
 )

{epub_translator-0.1.6 → epub_translator-0.1.7}/epub_translator/segment/text_segment.py

@@ -4,7 +4,7 @@ from enum import Enum, auto
 from typing import Self
 from xml.etree.ElementTree import Element
 
-from ..xml import expand_left_element_texts, expand_right_element_texts, is_inline_tag, normalize_text_in_element
+from ..xml import expand_left_element_texts, expand_right_element_texts, is_inline_element, normalize_text_in_element
 
 
 class TextPosition(Enum):
@@ -100,7 +100,7 @@ def search_text_segments(root: Element) -> Generator[TextSegment, None, None]:
 def _search_text_segments(stack: list[Element], element: Element) -> Generator[TextSegment, None, None]:
     text = normalize_text_in_element(element.text)
     next_stack = stack + [element]
-    next_block_depth = _find_block_depth(next_stack)
+    next_block_depth = find_block_depth(next_stack)
 
     if text is not None:
         yield TextSegment(
@@ -125,12 +125,11 @@ def _search_text_segments(stack: list[Element], element: Element) -> Generator[T
             )
 
 
-def _find_block_depth(parent_stack: list[Element]) -> int:
+def find_block_depth(parent_stack: list[Element]) -> int:
     index: int = 0
-    for i in range(len(parent_stack) - 1, -1, -1):
-        if not is_inline_tag(parent_stack[i].tag):
+    for i in range(len(parent_stack)):
+        if not is_inline_element(parent_stack[i]):
             index = i
-            break
     return index + 1  # depth is a count not index
 
 

{epub_translator-0.1.6 → epub_translator-0.1.7}/epub_translator/segment/utils.py

@@ -8,22 +8,6 @@ def element_fingerprint(element: Element) -> str:
     return f"<{element.tag} {' '.join(attrs)}/>"
 
 
-def unwrap_parents(element: Element) -> tuple[Element, list[Element]]:
-    parents: list[Element] = []
-    while True:
-        if len(element) != 1:
-            break
-        child = element[0]
-        if not element.text:
-            break
-        if not child.tail:
-            break
-        parents.append(element)
-        element = child
-        element.tail = None
-    return element, parents
-
-
 def id_in_element(element: Element) -> int | None:
     id_str = element.get(ID_KEY, None)
     if id_str is None:

{epub_translator-0.1.6 → epub_translator-0.1.7}/epub_translator/translation/xml_interrupter.py

@@ -1,9 +1,13 @@
 from collections.abc import Generator, Iterable
 from typing import cast
-from xml.etree.ElementTree import Element
+from xml.etree.ElementTree import Element, tostring
 
-from ..segment import TextSegment
-from ..utils import ensure_list, normalize_whitespace
+from bs4 import BeautifulSoup
+from mathml2latex.mathml import process_mathml
+
+from ..segment import TextSegment, combine_text_segments, find_block_depth
+from ..utils import ensure_list
+from ..xml import clone_element
 
 _ID_KEY = "__XML_INTERRUPTER_ID"
 _MATH_TAG = "math"
@@ -37,8 +41,10 @@ class XMLInterrupter:
     def interrupt_block_element(self, element: Element) -> Element:
         interrupted_element = self._placeholder2interrupted.pop(id(element), None)
         if interrupted_element is None:
+            element.attrib.pop(_ID_KEY, None)
             return element
         else:
+            interrupted_element.attrib.pop(_ID_KEY, None)
             return interrupted_element
 
     def _expand_source_text_segment(self, text_segment: TextSegment):
@@ -81,14 +87,18 @@ class XMLInterrupter:
                     _ID_KEY: cast(str, interrupted_element.get(_ID_KEY)),
                 },
             )
+            interrupted_display = interrupted_element.get("display", None)
+            if interrupted_display is not None:
+                placeholder_element.set("display", interrupted_display)
+
             raw_parent_stack = text_segment.parent_stack[:interrupted_index]
             parent_stack = raw_parent_stack + [placeholder_element]
             merged_text_segment = TextSegment(
-                text="".join(t.text for t in text_segments),
+                text=self._render_latex(text_segments),
                 parent_stack=parent_stack,
                 left_common_depth=text_segments[0].left_common_depth,
                 right_common_depth=text_segments[-1].right_common_depth,
-                block_depth=len(parent_stack),
+                block_depth=find_block_depth(parent_stack),
                 position=text_segments[0].position,
             )
             self._placeholder2interrupted[id(placeholder_element)] = interrupted_element
@@ -116,8 +126,8 @@ class XMLInterrupter:
                 # 原始栈退光，仅留下相对 interrupted 元素的栈，这种格式与 translated 要求一致
                 text_segment.left_common_depth = max(0, text_segment.left_common_depth - interrupted_index)
                 text_segment.right_common_depth = max(0, text_segment.right_common_depth - interrupted_index)
-                text_segment.block_depth = 1
                 text_segment.parent_stack = text_segment.parent_stack[interrupted_index:]
+                text_segment.block_depth = find_block_depth(text_segment.parent_stack)
 
         return merged_text_segment
 
@@ -129,37 +139,51 @@ class XMLInterrupter:
                 break
         return interrupted_index
 
+    def _render_latex(self, text_segments: list[TextSegment]) -> str:
+        math_element, _ = next(combine_text_segments(text_segments))
+        while math_element.tag != _MATH_TAG:
+            if len(math_element) == 0:
+                return ""
+            math_element = math_element[0]
+
+        math_element = clone_element(math_element)
+        math_element.attrib.pop(_ID_KEY, None)
+        math_element.tail = None
+        latex: str | None = None
+        try:
+            mathml_str = tostring(math_element, encoding="unicode")
+            soup = BeautifulSoup(mathml_str, "html.parser")
+            latex = process_mathml(soup)
+        except Exception:
+            pass
+
+        if latex is None:
+            latex = "".join(t.text for t in text_segments)
+        elif math_element.get("display", None) == "inline":
+            latex = f"${latex}$"
+        else:
+            latex = f"$${latex}$$"
+
+        return f" {latex} "
+
     def _expand_translated_text_segment(self, text_segment: TextSegment):
-        interrupted_id = text_segment.block_parent.attrib.pop(_ID_KEY, None)
+        parent_element = text_segment.parent_stack[-1]
+        interrupted_id = parent_element.attrib.pop(_ID_KEY, None)
         if interrupted_id is None:
             yield text_segment
             return
 
-        raw_text_segments = self._raw_text_segments.pop(interrupted_id, None)
-        if not raw_text_segments:
+        if parent_element is text_segment.block_parent:
+            # Block-level math， need to be hidden
             return
 
-        raw_block = raw_text_segments[0].parent_stack[0]
-        if not self._is_inline_math(raw_block):
+        raw_text_segments = self._raw_text_segments.pop(interrupted_id, None)
+        if not raw_text_segments:
+            yield text_segment
             return
 
         for raw_text_segment in raw_text_segments:
+            text_basic_parent_stack = text_segment.parent_stack[:-1]
             raw_text_segment.block_parent.attrib.pop(_ID_KEY, None)
+            raw_text_segment.parent_stack = text_basic_parent_stack + raw_text_segment.parent_stack
             yield raw_text_segment
-
-    def _has_no_math_texts(self, element: Element):
-        if element.tag == _MATH_TAG:
-            return True
-        if element.text and normalize_whitespace(element.text).strip():
-            return False
-        for child_element in element:
-            if not self._has_no_math_texts(child_element):
-                return False
-            if child_element.tail and normalize_whitespace(child_element.tail).strip():
-                return False
-        return True
-
-    def _is_inline_math(self, element: Element) -> bool:
-        if element.tag != _MATH_TAG:
-            return False
-        return element.get("display", "").lower() != "block"

epub_translator-0.1.7/epub_translator/xml/inline.py

@@ -0,0 +1,113 @@
+from xml.etree.ElementTree import Element
+
+# HTML inline-level elements
+# Reference: https://developer.mozilla.org/en-US/docs/Web/HTML/Inline_elements
+# Reference: https://developer.mozilla.org/en-US/docs/Glossary/Inline-level_content
+# Reference: https://developer.mozilla.org/en-US/docs/MathML/Element
+_HTML_INLINE_TAGS = frozenset(
+    (
+        # Inline text semantics
+        "a",
+        "abbr",
+        "b",
+        "bdi",
+        "bdo",
+        "br",
+        "cite",
+        "code",
+        "data",
+        "dfn",
+        "em",
+        "i",
+        "kbd",
+        "mark",
+        "q",
+        "rp",
+        "rt",
+        "ruby",
+        "s",
+        "samp",
+        "small",
+        "span",
+        "strong",
+        "sub",
+        "sup",
+        "time",
+        "u",
+        "var",
+        "wbr",
+        # Image and multimedia
+        "img",
+        "svg",
+        "canvas",
+        "audio",
+        "video",
+        "map",
+        "area",
+        # Form elements
+        "input",
+        "button",
+        "select",
+        "textarea",
+        "label",
+        "output",
+        "progress",
+        "meter",
+        # Embedded content
+        "iframe",
+        "embed",
+        "object",
+        # Other inline elements
+        "script",
+        "del",
+        "ins",
+        "slot",
+        # MathML elements
+        # Token elements
+        "mi",  # identifier
+        "mn",  # number
+        "mo",  # operator
+        "ms",  # string literal
+        "mspace",  # space
+        "mtext",  # text
+        # General layout
+        "menclose",  # enclosed content
+        "merror",  # syntax error message
+        "mfenced",  # parentheses (deprecated)
+        "mfrac",  # fraction
+        "mpadded",  # space around content
+        "mphantom",  # invisible content
+        "mroot",  # radical with index
+        "mrow",  # grouped sub-expressions
+        "msqrt",  # square root
+        "mstyle",  # style change
+        # Scripts and limits
+        "mmultiscripts",  # prescripts and tensor indices
+        "mover",  # overscript
+        "mprescripts",  # prescripts separator
+        "msub",  # subscript
+        "msubsup",  # subscript-superscript pair
+        "msup",  # superscript
+        "munder",  # underscript
+        "munderover",  # underscript-overscript pair
+        # Table math
+        "mtable",  # table or matrix
+        "mtr",  # row in table or matrix
+        "mtd",  # cell in table or matrix
+        # Semantic annotations
+        "annotation",  # data annotation
+        "annotation-xml",  # XML annotation
+        "semantics",  # semantic annotation container
+        # Other
+        "maction",  # bind actions to sub-expressions (deprecated)
+    )
+)
+
+
+def is_inline_element(element: Element) -> bool:
+    if element.tag.lower() in _HTML_INLINE_TAGS:
+        return True
+    display = element.get("display", None)
+    if display is not None and display.lower() == "inline":
+        return True
+    return False

{epub_translator-0.1.6 → epub_translator-0.1.7}/epub_translator/xml_translator/submitter.py

@@ -4,7 +4,7 @@ from enum import Enum, auto
 from xml.etree.ElementTree import Element
 
 from ..segment import TextSegment, combine_text_segments
-from ..xml import index_of_parent, is_inline_tag, iter_with_stack
+from ..xml import index_of_parent, is_inline_element, iter_with_stack
 from .stream_mapper import InlineSegmentMapping
 
 
@@ -78,7 +78,7 @@ class _Submitter:
         preserved_elements: list[Element] = []
         if self._action == SubmitKind.REPLACE:
             for child in list(node.raw_element):
-                if not is_inline_tag(child.tag):
+                if not is_inline_element(child):
                     child.tail = None
                     preserved_elements.append(child)
 
@@ -87,7 +87,7 @@ class _Submitter:
 
         if combined is not None:
             # 在 APPEND_BLOCK 模式下，如果是 inline tag，则在文本前面加空格
-            if self._action == SubmitKind.APPEND_BLOCK and is_inline_tag(combined.tag) and combined.text:
+            if self._action == SubmitKind.APPEND_BLOCK and is_inline_element(combined) and combined.text:
                 combined.text = " " + combined.text
             parent.insert(index + 1, combined)
             index += 1
@@ -200,7 +200,7 @@ class _Submitter:
         preserved_elements: list[Element] = []
         for i in range(start_index, end_index):
             elem = node_element[i]
-            if not is_inline_tag(elem.tag):
+            if not is_inline_element(elem):
                 elem.tail = None
                 preserved_elements.append(elem)
 
@@ -223,7 +223,7 @@ class _Submitter:
 
         if combined.text:
             will_inject_space = self._action == SubmitKind.APPEND_TEXT or (
-                is_inline_tag(combined.tag) and self._action == SubmitKind.APPEND_BLOCK
+                is_inline_element(combined) and self._action == SubmitKind.APPEND_BLOCK
             )
             if tail_element is not None:
                 tail_element.tail = self._append_text_in_element(

{epub_translator-0.1.6 → epub_translator-0.1.7}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "epub-translator"
-version = "0.1.6"
+version = "0.1.7"
 description = "Translate the epub book using LLM. The translated book will retain the original text and list the translated text side by side with the original text."
 keywords = ["epub", "llm", "translation", "translator"]
 authors = [
@@ -33,6 +33,7 @@ dependencies = [
     "jinja2>=3.1.6,<4.0.0",
     "resource-segmentation>=0.0.7,<0.1.0",
     "openai>=2.14.0,<3.0.0",
+    "mathml2latex (>=0.2.12,<0.3.0)",
 ]
 
 [project.urls]

epub_translator-0.1.6/epub_translator/data/mmltex/README.md

@@ -1,67 +0,0 @@
-# XSLT MathML Library
-
-This directory contains XSLT stylesheets from the **XSLT MathML Library 2.1.2**, a set of XSLT stylesheets to transform MathML 2.0 to LaTeX.
-
-**Note**: These files are included for reference purposes. Our project uses a custom Python implementation to convert MathML to LaTeX, but we keep these XSLT files as a reference for understanding MathML element mappings and conversion rules.
-
-## File Manifest
-
-- `mmltex.xsl` - Main stylesheet
-- `tokens.xsl` - Token elements (mi, mn, mo, etc.)
-- `glayout.xsl` - Layout elements (mfrac, msqrt, etc.)
-- `scripts.xsl` - Script elements (msub, msup, etc.)
-- `tables.xsl` - Table elements (mtable, mtr, mtd)
-- `entities.xsl` - Entity definitions
-- `cmarkup.xsl` - Content markup elements
-
-## Original Project Information
-
-**Original Author**: Vasil Yaroshevich
-
-**Original Website**: http://www.raleigh.ru/MathML/mmltex/
-
-**Archived Links**:
-- Sourceforge Project: https://sourceforge.net/projects/xsltml/files/xsltml/
-- Archived Documentation: https://web.archive.org/web/20160109063934/http://www.raleigh.ru/MathML/mmltex/index.php
-- Google Translated (English): https://translate.google.com/translate?sl=ru&tl=en&u=https%3A%2F%2Fweb.archive.org%2Fweb%2F20160114170851%2Fhttp%3A%2F%2Fwww.raleigh.ru%2FMathML%2Fmmltex%2Findex.php
-
----
-
-## Copyright
-
-Copyright (C) 2001-2003 Vasil Yaroshevich
-
-Permission is hereby granted, free of charge, to any person
-obtaining a copy of this software and associated documentation
-files (the "Software"), to deal in the Software without
-restriction, including without limitation the rights to use,
-copy, modify, merge, publish, distribute, sublicense, and/or
-sell copies of the Software, and to permit persons to whom the
-Software is furnished to do so, subject to the following
-conditions:
-
-The above copyright notice and this permission notice shall be
-included in all copies or substantial portions of the Software.
-
-Except as contained in this notice, the names of individuals
-credited with contribution to this software shall not be used in
-advertising or otherwise to promote the sale, use or other
-dealings in this Software without prior written authorization
-from the individuals in question.
-
-Any stylesheet derived from this Software that is publically
-distributed will be identified with a different name and the
-version strings in any derived Software will be changed so that
-no possibility of confusion between the derived package and this
-Software will exist.
-
-## Warranty
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
-EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES
-OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
-NONINFRINGEMENT.  IN NO EVENT SHALL NORMAN WALSH OR ANY OTHER
-CONTRIBUTOR BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
-WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
-OTHER DEALINGS IN THE SOFTWARE.