FabledCurator/backend/app/utils/prosemirror.py

"""Render Patreon's post body to HTML.

Patreon deprecated the flat `content` HTML field (it returns `null` now, on the
feed AND the detail endpoint, for every post type) and moved the real body to
`content_json_string` — a ProseMirror/TipTap document (a JSON string of typed
nodes + marks), returned only under the DEFAULT post fieldset. This module
converts that doc to the HTML the rest of the pipeline already expects (it then
flows through the existing nh3 sanitizer + inline-image localization). #842.

Built from the observed Patreon doc shape: doc → [paragraph|heading|image|list…];
paragraph content → [text(+marks: bold/italic/underline/strike/code/link),
hardBreak, image]. Unknown node types degrade to rendering their children so we
never silently drop text.
"""

from __future__ import annotations

import json
from html import escape

# ProseMirror text-mark type → wrapping HTML tag (link handled separately — it
# needs an href attribute).
_MARK_TAGS = {
    "bold": "strong",
    "strong": "strong",
    "italic": "em",
    "em": "em",
    "underline": "u",
    "strike": "s",
    "strikethrough": "s",
    "code": "code",
}


def post_body_html(attrs: dict | None) -> str | None:
    """Resolve a Patreon post's body HTML from its `attributes`.

    Prefers the legacy flat `content` HTML (older posts still carry it); falls
    back to converting the `content_json_string` ProseMirror doc (current posts).
    Returns None when neither holds a body. The ONE place body-source precedence
    lives, so the feed-parse path and the detail-fetch path can't diverge.
    """
    if not isinstance(attrs, dict):
        return None
    content = attrs.get("content")
    if isinstance(content, str) and content.strip():
        return content
    return prosemirror_to_html(attrs.get("content_json_string"))


def prosemirror_to_html(doc_json: str | None) -> str | None:
    """Render a ProseMirror/TipTap doc (a JSON string) to HTML, or None for an
    empty / blank / unparseable doc."""
    if not isinstance(doc_json, str) or not doc_json.strip():
        return None
    try:
        doc = json.loads(doc_json)
    except (ValueError, TypeError):
        return None
    if not isinstance(doc, dict):
        return None
    return _render_nodes(doc.get("content")) or None


def _render_nodes(nodes: object) -> str:
    if not isinstance(nodes, list):
        return ""
    return "".join(_render_node(n) for n in nodes if isinstance(n, dict))


def _render_node(node: dict) -> str:
    ntype = node.get("type")
    if ntype == "text":
        return _render_text(node)
    if ntype == "hardBreak":
        return "<br>"
    if ntype == "image":
        return _render_image(node)
    if ntype == "horizontalRule":
        return "<hr>"

    inner = _render_nodes(node.get("content"))
    if ntype == "paragraph":
        return f"<p>{inner}</p>"
    if ntype == "heading":
        level = (node.get("attrs") or {}).get("level", 1)
        if not isinstance(level, int) or not 1 <= level <= 6:
            level = 1
        return f"<h{level}>{inner}</h{level}>"
    if ntype == "blockquote":
        return f"<blockquote>{inner}</blockquote>"
    if ntype == "bulletList":
        return f"<ul>{inner}</ul>"
    if ntype == "orderedList":
        return f"<ol>{inner}</ol>"
    if ntype == "listItem":
        return f"<li>{inner}</li>"
    if ntype == "codeBlock":
        return f"<pre><code>{inner}</code></pre>"
    # Unknown / unhandled container — emit children so text is never dropped.
    return inner


def _render_text(node: dict) -> str:
    text = node.get("text")
    if not isinstance(text, str) or not text:
        return ""
    out = escape(text)
    for mark in node.get("marks") or []:
        if not isinstance(mark, dict):
            continue
        mtype = mark.get("type")
        if mtype == "link":
            href = (mark.get("attrs") or {}).get("href") or ""
            out = f'<a href="{escape(str(href), quote=True)}">{out}</a>'
        else:
            tag = _MARK_TAGS.get(mtype)
            if tag:
                out = f"<{tag}>{out}</{tag}>"
    return out


def _render_image(node: dict) -> str:
    attrs = node.get("attrs") or {}
    src = attrs.get("src")
    if not isinstance(src, str) or not src:
        return ""
    alt = attrs.get("alt") or ""
    return f'<img src="{escape(src, quote=True)}" alt="{escape(str(alt), quote=True)}">'