v0.5.2

Что то сделал

word2md/emit.py

@@ -0,0 +1,249 @@
+"""Emit md2gost dialect Markdown from IR blocks."""
+
+from __future__ import annotations
+
+from .blocks import (
+    Block,
+    CaptionBlock,
+    EquationBlock,
+    HeadingBlock,
+    ImageBlock,
+    InlineSpan,
+    ListBlock,
+    ListingBlock,
+    PageBreakBlock,
+    ParagraphBlock,
+    TableBlock,
+    TocPlaceholder,
+)
+
+# Characters that must stay literal after DOCX→MD (Word stores bare '*', not '\*').
+# Do not escape '_': md2gost labels use it (@Рисунок:fig1_1).
+_MD_SPECIAL = str.maketrans(
+    {
+        "\\": "\\\\",
+        "*": "\\*",
+        "`": "\\`",
+    }
+)
+
+
+def escape_md_inline(text: str) -> str:
+    """Escape Markdown inline metacharacters in plain text from DOCX."""
+    if not text:
+        return ""
+    return text.translate(_MD_SPECIAL)
+
+
+def _escape_md_cell(text: str) -> str:
+    return escape_md_inline(text or "").replace("|", "\\|").replace("\n", " ")
+
+
+def _render_spans(spans: list[InlineSpan]) -> str:
+    parts: list[str] = []
+    for s in spans:
+        if s.math:
+            parts.append(f"${s.math}$")
+            continue
+        text = escape_md_inline(s.text or "")
+        if s.href:
+            parts.append(f"[{text}]({s.href})")
+            continue
+        if s.bold and s.italic:
+            text = f"***{text}***"
+        elif s.bold:
+            text = f"**{text}**"
+        elif s.italic:
+            text = f"*{text}*"
+        if s.strike:
+            text = f"~~{text}~~"
+        parts.append(text)
+    return "".join(parts)
+
+
+def _emit_caption_line(unique_id: str, text: str | None = None, *, landscape: bool = False) -> str:
+    title = escape_md_inline((text or "").strip())
+    flags = " +landscape" if landscape else ""
+    if title:
+        return f"%{unique_id} {title}{flags}".rstrip()
+    return f"%{unique_id}{flags}".rstrip()
+
+
+def _last_nonempty(chunks: list[str]) -> str:
+    for prev in reversed(chunks):
+        if prev.strip():
+            return prev
+    return ""
+
+
+def _emit_table(tbl: TableBlock) -> str:
+    if not tbl.rows:
+        return ""
+    lines: list[str] = []
+    for r, row in enumerate(tbl.rows):
+        cells = []
+        for c, cell in enumerate(row):
+            if tbl.merges and r < len(tbl.merges) and c < len(tbl.merges[r]):
+                v, h = tbl.merges[r][c]
+                if h == "continue":
+                    cells.append(">")
+                    continue
+                if v == "continue":
+                    cells.append("^")
+                    continue
+            cells.append(_escape_md_cell(cell if cell not in ("^", ">") else cell))
+        lines.append("| " + " | ".join(cells) + " |")
+        if r == 0:
+            lines.append("| " + " | ".join("---" for _ in cells) + " |")
+    return "\n".join(lines)
+
+
+def emit_markdown(blocks: list[Block], *, emit_page_breaks: bool = False) -> str:
+    chunks: list[str] = []
+    i = 0
+    while i < len(blocks):
+        b = blocks[i]
+
+        if isinstance(b, HeadingBlock):
+            hashes = "#" * max(1, min(b.level, 6))
+            title = escape_md_inline(b.text or "")
+            if not b.numbered:
+                # Dialect flag '*' after '# ' is structural — not escaped.
+                chunks.append(f"{hashes} *{title}")
+            else:
+                chunks.append(f"{hashes} {title}")
+            chunks.append("")
+            i += 1
+            continue
+
+        if isinstance(b, TocPlaceholder):
+            chunks.append("[TOC]")
+            chunks.append("")
+            i += 1
+            continue
+
+        if isinstance(b, CaptionBlock):
+            if b.is_continuation:
+                i += 1
+                continue
+            nxt = blocks[i + 1] if i + 1 < len(blocks) else None
+            if isinstance(nxt, ImageBlock) and b.kind == "figure":
+                nxt.caption_id = nxt.caption_id or b.unique_id
+                nxt.caption_text = nxt.caption_text or b.text
+                i += 1
+                continue
+            if isinstance(nxt, TableBlock) and b.kind == "table":
+                nxt.caption_id = nxt.caption_id or b.unique_id
+                nxt.caption_text = nxt.caption_text or b.text
+                if b.landscape:
+                    nxt.landscape = True
+                chunks.append(_emit_caption_line(
+                    nxt.caption_id or b.unique_id,
+                    nxt.caption_text or b.text,
+                    landscape=nxt.landscape,
+                ))
+                chunks.append("")
+                i += 1
+                continue
+            if isinstance(nxt, ListingBlock) and b.kind == "listing":
+                nxt.caption_id = nxt.caption_id or b.unique_id
+                nxt.caption_text = nxt.caption_text or b.text
+                chunks.append(_emit_caption_line(nxt.caption_id or b.unique_id, nxt.caption_text))
+                chunks.append("")
+                i += 1
+                continue
+            chunks.append(_emit_caption_line(b.unique_id, b.text, landscape=b.landscape))
+            chunks.append("")
+            i += 1
+            continue
+
+        if isinstance(b, ImageBlock):
+            # Safety: caption still after image in IR
+            nxt = blocks[i + 1] if i + 1 < len(blocks) else None
+            if (
+                isinstance(nxt, CaptionBlock)
+                and nxt.kind == "figure"
+                and not nxt.is_continuation
+            ):
+                b.caption_id = b.caption_id or nxt.unique_id
+                b.caption_text = b.caption_text or nxt.text
+                # skip the caption block
+                i += 1
+            title = ""
+            if b.caption_id:
+                title_body = escape_md_inline((b.caption_text or "").strip())
+                title = f" \"%{b.caption_id}" + (f" {title_body}" if title_body else "") + "\""
+            alt = escape_md_inline((b.alt or "рисунок").replace('"', "'"))
+            chunks.append(f"![{alt}]({b.rel_path}{title})")
+            chunks.append("")
+            i += 1
+            continue
+
+        if isinstance(b, TableBlock):
+            if b.caption_id:
+                prev = _last_nonempty(chunks)
+                if not prev.startswith(f"%{b.caption_id}"):
+                    chunks.append(_emit_caption_line(b.caption_id, b.caption_text, landscape=b.landscape))
+                    chunks.append("")
+            chunks.append(_emit_table(b))
+            chunks.append("")
+            i += 1
+            continue
+
+        if isinstance(b, ListingBlock):
+            if b.caption_id:
+                prev = _last_nonempty(chunks)
+                if not prev.startswith(f"%{b.caption_id}"):
+                    chunks.append(_emit_caption_line(b.caption_id, b.caption_text))
+                    chunks.append("")
+            lang = b.language or ""
+            body = "\n".join(line.rstrip("\n") for line in b.lines)
+            chunks.append(f"```{lang}")
+            chunks.append(body)
+            chunks.append("```")
+            chunks.append("")
+            i += 1
+            continue
+
+        if isinstance(b, EquationBlock):
+            if b.unique_id:
+                chunks.append(f"%{b.unique_id}")
+            chunks.append("$$")
+            chunks.append(b.latex.strip())
+            chunks.append("$$")
+            chunks.append("")
+            i += 1
+            continue
+
+        if isinstance(b, ListBlock):
+            for idx, item in enumerate(b.items, start=1):
+                body = escape_md_inline(item)
+                if b.ordered:
+                    chunks.append(f"{idx}. {body}")
+                else:
+                    chunks.append(f"- {body}")
+            chunks.append("")
+            i += 1
+            continue
+
+        if isinstance(b, ParagraphBlock):
+            line = _render_spans(b.spans).strip()
+            if line:
+                chunks.append(line)
+                chunks.append("")
+            i += 1
+            continue
+
+        if isinstance(b, PageBreakBlock):
+            if emit_page_breaks:
+                chunks.append("---")
+                chunks.append("")
+            i += 1
+            continue
+
+        i += 1
+
+    text = "\n".join(chunks)
+    while "\n\n\n" in text:
+        text = text.replace("\n\n\n", "\n\n")
+    return text.strip() + "\n"