feat: export fidelity and web client parity

apps/api/src/synthora/api/export.py

@@ -14,8 +14,14 @@
   body {{ font-family: Georgia, serif; max-width: 46rem; margin: 3rem auto; line-height: 1.6; color: #1c2a30; padding: 0 1.5rem; }}
   h1, h2, h3 {{ font-family: Georgia, serif; letter-spacing: -0.01em; }}
   code {{ font-family: ui-monospace, monospace; background: #f3f1ea; padding: 0.1em 0.3em; border-radius: 3px; }}
+  pre {{ background: #f3f1ea; padding: 0.75rem 1rem; overflow-x: auto; border-radius: 4px; }}
+  pre code {{ background: transparent; padding: 0; }}
   a {{ color: #0e6f6a; }}
   blockquote {{ border-left: 3px solid #0e6f6a; margin-left: 0; padding-left: 1rem; color: #3d5059; }}
+  table {{ border-collapse: collapse; width: 100%; margin: 1rem 0; }}
+  th, td {{ border: 1px solid #d8d2c4; padding: 0.35rem 0.6rem; text-align: left; }}
+  th {{ background: #f3f1ea; }}
+  hr {{ border: none; border-top: 1px solid #d8d2c4; margin: 1.5rem 0; }}
   @media print {{ body {{ margin: 0.5in; max-width: none; }} }}
 </style>
 </head>
@@ -27,38 +33,114 @@
 
 
 def markdown_to_html(markdown: str) -> str:
-    """Minimal dependency-free Markdown -> HTML for report export.
+    """Dependency-free Markdown -> HTML for report export.
 
-    Covers the subset our writers emit: headings, paragraphs, lists,
-    bold/italic, inline code, and links.
+    Covers headings, paragraphs, bullet/ordered lists, blockquotes, fenced
+    code blocks, tables, bold/italic, inline code, and links.
     """
     lines = markdown.splitlines()
     out: list[str] = []
-    in_list = False
-    for line in lines:
+    i = 0
+    list_kind: str | None = None  # "ul" | "ol"
+
+    def close_list() -> None:
+        nonlocal list_kind
+        if list_kind:
+            out.append(f"</{list_kind}>")
+            list_kind = None
+
+    while i < len(lines):
+        line = lines[i]
         stripped = line.strip()
+
+        if stripped.startswith("```"):
+            close_list()
+            fence = stripped[3:].strip()
+            i += 1
+            code_lines: list[str] = []
+            while i < len(lines) and not lines[i].strip().startswith("```"):
+                code_lines.append(lines[i])
+                i += 1
+            if i < len(lines):
+                i += 1
+            code = html_lib.escape("\n".join(code_lines))
+            lang = html_lib.escape(fence) if fence else ""
+            cls = f' class="language-{lang}"' if lang else ""
+            out.append(f"<pre><code{cls}>{code}</code></pre>")
+            continue
+
+        if stripped.startswith("|") and "|" in stripped[1:]:
+            close_list()
+            table_rows: list[str] = []
+            while i < len(lines) and lines[i].strip().startswith("|"):
+                table_rows.append(lines[i].strip())
+                i += 1
+            if len(table_rows) >= 2 and re.match(r"^\|[-: |]+\|$", table_rows[1]):
+                table_rows.pop(1)
+            if table_rows:
+                out.append("<table>")
+                for row_idx, row in enumerate(table_rows):
+                    cells = [c.strip() for c in row.strip("|").split("|")]
+                    tag = "th" if row_idx == 0 else "td"
+                    out.append("<tr>")
+                    for cell in cells:
+                        out.append(f"<{tag}>{_inline(cell)}</{tag}>")
+                    out.append("</tr>")
+                out.append("</table>")
+            continue
+
         heading = re.match(r"^(#{1,6})\s+(.*)$", stripped)
         if heading:
-            if in_list:
-                out.append("</ul>")
-                in_list = False
+            close_list()
             level = len(heading.group(1))
             out.append(f"<h{level}>{_inline(heading.group(2))}</h{level}>")
+            i += 1
+            continue
+
+        if re.match(r"^[-*]{3,}$", stripped):
+            close_list()
+            out.append("<hr />")
+            i += 1
+            continue
+
+        if stripped.startswith(">"):
+            close_list()
+            quote_lines: list[str] = []
+            while i < len(lines) and lines[i].strip().startswith(">"):
+                quote_lines.append(lines[i].strip().lstrip(">").lstrip())
+                i += 1
+            out.append(
+                f"<blockquote><p>{_inline(' '.join(quote_lines))}</p></blockquote>"
+            )
+            continue
+
+        ordered = re.match(r"^(\d+)\.\s+(.*)$", stripped)
+        if ordered:
+            if list_kind != "ol":
+                close_list()
+                out.append("<ol>")
+                list_kind = "ol"
+            out.append(f"<li>{_inline(ordered.group(2))}</li>")
+            i += 1
             continue
+
         if stripped.startswith(("- ", "* ")):
-            if not in_list:
+            if list_kind != "ul":
+                close_list()
                 out.append("<ul>")
-                in_list = True
+                list_kind = "ul"
             out.append(f"<li>{_inline(stripped[2:])}</li>")
+            i += 1
             continue
-        if in_list:
-            out.append("</ul>")
-            in_list = False
+
+        close_list()
         if not stripped:
+            i += 1
             continue
         out.append(f"<p>{_inline(stripped)}</p>")
-    if in_list:
-        out.append("</ul>")
+        i += 1
+
+    close_list()
     return "\n".join(out)
 
 
@@ -79,26 +161,17 @@ def render_html_document(markdown: str, *, title: str) -> str:
     )
 
 
-def _plain_from_markdown(markdown: str) -> str:
-    """Strip common markdown markers for PDF core fonts."""
-    text = re.sub(r"^#{1,6}\s+", "", markdown, flags=re.MULTILINE)
-    text = re.sub(r"\*\*(.+?)\*\*", r"\1", text)
-    text = re.sub(r"(?<!\*)\*([^*]+)\*(?!\*)", r"\1", text)
-    text = re.sub(r"`([^`]+)`", r"\1", text)
-    text = re.sub(r"\[([^\]]+)\]\((https?://[^)]+)\)", r"\1 (\2)", text)
-    return text
-
-
 def _pdf_safe(text: str) -> str:
     return text.encode("latin-1", "replace").decode("latin-1")
 
 
 def markdown_to_pdf_bytes(markdown: str, *, title: str = "Report") -> bytes:
-    """Render markdown to PDF bytes using fpdf2 (no network)."""
+    """Render markdown to PDF bytes using fpdf2 HTML layout (no network)."""
     from fpdf import FPDF
     from fpdf.enums import XPos, YPos
 
     pdf = FPDF()
+    pdf.set_compression(False)
     pdf.set_auto_page_break(auto=True, margin=15)
     pdf.set_margins(15, 15, 15)
     pdf.add_page()
@@ -112,17 +185,8 @@ def markdown_to_pdf_bytes(markdown: str, *, title: str = "Report") -> bytes:
     )
     pdf.ln(4)
     pdf.set_font("Helvetica", size=11)
-    body = _plain_from_markdown(markdown or "")
-    for line in body.splitlines() or [""]:
-        if not line.strip():
-            pdf.ln(6)
-            continue
-        pdf.multi_cell(
-            0,
-            6,
-            _pdf_safe(line),
-            new_x=XPos.LMARGIN,
-            new_y=YPos.NEXT,
-        )
+    body_html = markdown_to_html(markdown or "")
+    if body_html.strip():
+        pdf.write_html(_pdf_safe(body_html))
     out = pdf.output()
     return bytes(out)

apps/web/src/api.ts

@@ -456,6 +456,19 @@ export const api = {
         body: JSON.stringify({ query, max_results: maxResults }),
       },
     ),
+
+  health: () => request<{ status: string }>("/health"),
+  ready: () => request<{ status: string }>("/ready"),
+  mcpToolsList: () =>
+    request<{ tools: Array<Record<string, unknown>> }>(
+      "/api/v1/mcp/tools/list",
+      { method: "POST", body: "{}" },
+    ),
+  mcpToolsCall: (name: string, args: Record<string, unknown> = {}) =>
+    request<{ content: string }>("/api/v1/mcp/tools/call", {
+      method: "POST",
+      body: JSON.stringify({ name, arguments: args }),
+    }),
 };
 
 export function eventsSocketUrl(runId: string): string {

apps/web/src/components/History.tsx

@@ -4,14 +4,22 @@ import { api, RunSummary, SessionSummary } from "../api";
 export function History({ onOpen }: { onOpen: (runId: string) => void }) {
   const [runs, setRuns] = useState<RunSummary[]>([]);
   const [sessions, setSessions] = useState<SessionSummary[]>([]);
+  const [selectedSessionId, setSelectedSessionId] = useState<string | null>(
+    null,
+  );
   const [error, setError] = useState<string | null>(null);
   const [busy, setBusy] = useState(false);
 
-  const load = useCallback(() => {
+  const loadRuns = useCallback((sessionFilter?: string | null) => {
+    const filter =
+      sessionFilter !== undefined ? sessionFilter : selectedSessionId;
     api
-      .listRuns()
+      .listRuns(filter || undefined)
       .then(setRuns)
       .catch((e) => setError(String(e)));
+  }, [selectedSessionId]);
+
+  const loadSessions = useCallback(() => {
     api
       .listSessions()
       .then(setSessions)
@@ -21,8 +29,9 @@ export function History({ onOpen }: { onOpen: (runId: string) => void }) {
   }, []);
 
   useEffect(() => {
-    load();
-  }, [load]);
+    loadRuns();
+    loadSessions();
+  }, [loadRuns, loadSessions]);
 
   async function handleDelete(runId: string, e: MouseEvent) {
     e.stopPropagation();
@@ -46,6 +55,10 @@ export function History({ onOpen }: { onOpen: (runId: string) => void }) {
     try {
       await api.deleteSession(sessionId);
       setSessions((prev) => prev.filter((s) => s.id !== sessionId));
+      if (selectedSessionId === sessionId) {
+        setSelectedSessionId(null);
+        loadRuns(null);
+      }
     } catch (err) {
       setError(String(err));
     }
@@ -65,6 +78,22 @@ export function History({ onOpen }: { onOpen: (runId: string) => void }) {
     }
   }
 
+  async function handleSelectSession(sessionId: string) {
+    setError(null);
+    if (selectedSessionId === sessionId) {
+      setSelectedSessionId(null);
+      loadRuns(null);
+      return;
+    }
+    setSelectedSessionId(sessionId);
+    try {
+      const detail = await api.getSession(sessionId);
+      setRuns(detail.runs);
+    } catch (err) {
+      setError(String(err));
+    }
+  }
+
   function sessionTitle(id: string | null): string {
     if (!id) return "—";
     const match = sessions.find((s) => s.id === id);
@@ -75,6 +104,18 @@ export function History({ onOpen }: { onOpen: (runId: string) => void }) {
     <section className="panel">
       <div className="action-row">
         <h2 style={{ margin: 0, flex: 1 }}>Research history</h2>
+        {selectedSessionId && (
+          <button
+            type="button"
+            className="ghost"
+            onClick={() => {
+              setSelectedSessionId(null);
+              loadRuns(null);
+            }}
+          >
+            Clear session filter
+          </button>
+        )}
         {runs.length > 0 && (
           <button
             type="button"
@@ -86,6 +127,12 @@ export function History({ onOpen }: { onOpen: (runId: string) => void }) {
           </button>
         )}
       </div>
+      {selectedSessionId && (
+        <p className="muted">
+          Showing runs for session:{" "}
+          <strong>{sessionTitle(selectedSessionId)}</strong>
+        </p>
+      )}
       {error && <p className="error-text">{error}</p>}
       {runs.length === 0 && !error && <p>No research yet.</p>}
       {runs.length > 0 && (
@@ -143,7 +190,14 @@ export function History({ onOpen }: { onOpen: (runId: string) => void }) {
             {sessions.map((s) => (
               <li key={s.id} className="discourse-turn">
                 <div className="discourse-meta">
-                  <strong>{s.title}</strong>
+                  <button
+                    type="button"
+                    className="ghost"
+                    aria-pressed={selectedSessionId === s.id}
+                    onClick={() => handleSelectSession(s.id)}
+                  >
+                    <strong>{s.title}</strong>
+                  </button>
                   <span className="muted">
                     {" "}
                     · <code>{s.id.slice(0, 8)}…</code>

docs/feature-parity.md

@@ -59,7 +59,7 @@ See also [parity-audit.md](parity-audit.md).
 | Search strategy abstraction (5 strategies + aliases) | ✅ | `strategy_registry` |
 | Search engine abstraction (full catalog) | ✅ | 29 registered engines |
 | LLM provider abstraction + think-tag handling | ✅ | 11 providers |
-| Research history + export md/html/pdf | ✅ | API + web buttons |
+| Research history + export md/html/pdf | ✅ | API + web buttons; structured HTML/PDF export |
 | Docker Compose self-host | ✅ | `docker-compose.yml` |
 | Python SDK | ✅ | `packages/sdk` — full REST mirror incl. upload, export download, news GET |
 | Document library + RAG (`collection` engine) | ✅ | documents API + `document_index` |
@@ -125,6 +125,11 @@ bytes), ``get_news_subscription``, ``search_documents(max_results=...)``,
 ``health``/``ready``; web client ``getNewsSubscription`` and search
 ``max_results``; streamable MCP workspace isolation test.
 
+Closed on ``feat/export-and-web-parity``: full markdown export subset (fenced
+code, blockquotes, ordered lists, tables, rules); PDF via fpdf2 ``write_html``
+(preserves structure); web ``health``/``ready``/MCP REST wrappers; History
+session drill-down via ``getSession``.
+
 No known functional gaps remain beyond explicit non-goals below.
 
 Chat remains session-scoped ``fast_research`` with prior-report memory —