omarsol commited on
Commit
3c4adeb
·
1 Parent(s): b155332

Enhance citation handling and UI components

Browse files

- Updated citation handling in chat messages to include citation numbers, improving clarity and usability.
- Introduced citation chips in Markdown rendering for better visual representation of citations.
- Refactored CitationRow and TextBlock components to support new citation number functionality, ensuring consistent display across the application.
- Enhanced citation normalization and numbering logic to ensure accurate citation tracking in messages.

app/prompts.py CHANGED
@@ -140,8 +140,10 @@ ANSWERING_RULES = """## Answering rules
140
  introduces or explains the code instead.
141
  - Do not put all citations only in a final "Sources" section. A short sources
142
  recap is okay only if the answer already has inline citations.
143
- - Do not cite wiki pages unless the wiki page itself is the source of the
144
- claim. Prefer raw docs and course lessons.
 
 
145
  - If retrieval results are weak or missing, say the topic is not well
146
  covered by the current knowledge base rather than guessing.
147
  - Synthesize retrieved material into a clear teaching explanation. Do not
 
140
  introduces or explains the code instead.
141
  - Do not put all citations only in a final "Sources" section. A short sources
142
  recap is okay only if the answer already has inline citations.
143
+ - Never cite `wiki/` or `generated/` KB paths: they are navigation indexes,
144
+ not sources, and they do not resolve to citable source cards. When a wiki
145
+ page led you to a claim, cite the underlying `raw/` page it references
146
+ (open it with `run_kb_command` if you have not already).
147
  - If retrieval results are weak or missing, say the topic is not well
148
  covered by the current knowledge base rather than guessing.
149
  - Synthesize retrieved material into a clear teaching explanation. Do not
frontend/app/globals.css CHANGED
@@ -386,6 +386,28 @@ textarea {
386
  text-underline-offset: 0.14em;
387
  }
388
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
389
  .markdown-block hr {
390
  border: 0;
391
  border-top: 1px solid var(--hr-color);
 
386
  text-underline-offset: 0.14em;
387
  }
388
 
389
+ .markdown-block a.citation-chip {
390
+ display: inline-flex;
391
+ align-items: center;
392
+ justify-content: center;
393
+ min-width: 1.05rem;
394
+ height: 1.05rem;
395
+ margin-left: 0.18rem;
396
+ padding: 0 0.28rem;
397
+ border-radius: 9999px;
398
+ background: var(--accent-faint);
399
+ color: var(--accent);
400
+ font-size: 10px;
401
+ font-weight: 650;
402
+ line-height: 1;
403
+ text-decoration: none;
404
+ vertical-align: 0.32em;
405
+ }
406
+
407
+ .markdown-block a.citation-chip:hover {
408
+ background: var(--surface-strong);
409
+ }
410
+
411
  .markdown-block hr {
412
  border: 0;
413
  border-top: 1px solid var(--hr-color);
frontend/components/chat-message.tsx CHANGED
@@ -37,6 +37,8 @@ import type {
37
  } from "@/lib/chat-ui";
38
  import {
39
  buildActivityItems,
 
 
40
  getMessageCitations,
41
  getOrderedMessageBlocks,
42
  prettifyToolName,
@@ -77,6 +79,7 @@ export function ChatMessage({
77
  const isAssistant = message.role === "assistant";
78
  const contentBlocks = getOrderedMessageBlocks(message);
79
  const citations = isAssistant && !isStreaming ? getMessageCitations(message) : [];
 
80
  const editTextareaRef = useRef<HTMLTextAreaElement>(null);
81
 
82
  useEffect(() => {
@@ -145,10 +148,16 @@ export function ChatMessage({
145
  key={block.key}
146
  block={block}
147
  isActive={isActive}
 
148
  />
149
  );
150
  })}
151
- {citations.length > 0 ? <CitationRow citations={citations} /> : null}
 
 
 
 
 
152
  </div>
153
  )}
154
  </article>
@@ -280,14 +289,16 @@ function MessageActionButton({
280
  function ContentBlock({
281
  block,
282
  isActive = false,
 
283
  }: {
284
  block: TutorMessageBlock;
285
  isActive?: boolean;
 
286
  }) {
287
  if (block.kind === "activity") {
288
  return <ActivityPanel parts={block.parts} isActive={isActive} />;
289
  }
290
- return <TextBlock parts={block.parts} />;
291
  }
292
 
293
  function ActivityPanel({
@@ -601,13 +612,20 @@ function formatToolResultSummary({
601
  return "";
602
  }
603
 
604
- function TextBlock({ parts }: { parts: TutorMessagePart[] }) {
 
 
 
 
 
 
605
  return (
606
  <div className="space-y-3 text-[15px] leading-[1.72] tracking-[-0.012em] text-[var(--ink)]">
607
  {parts.map((part, index) => (
608
  <MarkdownBlock
609
  key={`text-${index}`}
610
  className="text-[15px] leading-[1.72] tracking-[-0.012em] text-[var(--ink)]"
 
611
  >
612
  {part.text ?? ""}
613
  </MarkdownBlock>
@@ -651,7 +669,13 @@ const CITATION_KIND_META: Record<
651
  doc: { icon: BookOpen, label: "Docs" },
652
  };
653
 
654
- function CitationRow({ citations }: { citations: MessageCitation[] }) {
 
 
 
 
 
 
655
  return (
656
  <section className="mt-2 space-y-2 border-t border-[var(--line)] pt-3">
657
  <div className="inline-flex items-center gap-1.5 text-[10px] font-semibold uppercase tracking-[0.12em] text-[var(--muted)]">
@@ -665,6 +689,7 @@ function CitationRow({ citations }: { citations: MessageCitation[] }) {
665
  {citations.map((citation, index) => {
666
  const meta = CITATION_KIND_META[citation.kind];
667
  const Icon = meta.icon;
 
668
  return (
669
  <a
670
  key={`${citation.kind}-${citation.url}-${index}`}
@@ -674,6 +699,11 @@ function CitationRow({ citations }: { citations: MessageCitation[] }) {
674
  title={`${meta.label}${citation.sublabel ? ` · ${citation.sublabel}` : ""}`}
675
  className="group inline-flex max-w-full items-center gap-1.5 rounded-full border border-[var(--line-strong)] bg-[var(--accent-faint)] px-2.5 py-1.5 text-xs font-medium text-[var(--accent)] transition hover:border-[var(--accent)] hover:bg-[var(--surface-strong)]"
676
  >
 
 
 
 
 
677
  <Icon className="h-3.5 w-3.5 shrink-0" />
678
  <span className="truncate">{citation.label}</span>
679
  <ExternalLink className="h-3 w-3 shrink-0 opacity-60 transition group-hover:opacity-100" />
 
37
  } from "@/lib/chat-ui";
38
  import {
39
  buildActivityItems,
40
+ buildCitationNumbers,
41
+ citationNumberFor,
42
  getMessageCitations,
43
  getOrderedMessageBlocks,
44
  prettifyToolName,
 
79
  const isAssistant = message.role === "assistant";
80
  const contentBlocks = getOrderedMessageBlocks(message);
81
  const citations = isAssistant && !isStreaming ? getMessageCitations(message) : [];
82
+ const citationNumbers = isAssistant ? buildCitationNumbers(message) : undefined;
83
  const editTextareaRef = useRef<HTMLTextAreaElement>(null);
84
 
85
  useEffect(() => {
 
148
  key={block.key}
149
  block={block}
150
  isActive={isActive}
151
+ citationNumbers={citationNumbers}
152
  />
153
  );
154
  })}
155
+ {citations.length > 0 ? (
156
+ <CitationRow
157
+ citations={citations}
158
+ citationNumbers={citationNumbers}
159
+ />
160
+ ) : null}
161
  </div>
162
  )}
163
  </article>
 
289
  function ContentBlock({
290
  block,
291
  isActive = false,
292
+ citationNumbers,
293
  }: {
294
  block: TutorMessageBlock;
295
  isActive?: boolean;
296
+ citationNumbers?: Map<string, number>;
297
  }) {
298
  if (block.kind === "activity") {
299
  return <ActivityPanel parts={block.parts} isActive={isActive} />;
300
  }
301
+ return <TextBlock parts={block.parts} citationNumbers={citationNumbers} />;
302
  }
303
 
304
  function ActivityPanel({
 
612
  return "";
613
  }
614
 
615
+ function TextBlock({
616
+ parts,
617
+ citationNumbers,
618
+ }: {
619
+ parts: TutorMessagePart[];
620
+ citationNumbers?: Map<string, number>;
621
+ }) {
622
  return (
623
  <div className="space-y-3 text-[15px] leading-[1.72] tracking-[-0.012em] text-[var(--ink)]">
624
  {parts.map((part, index) => (
625
  <MarkdownBlock
626
  key={`text-${index}`}
627
  className="text-[15px] leading-[1.72] tracking-[-0.012em] text-[var(--ink)]"
628
+ citationNumbers={citationNumbers}
629
  >
630
  {part.text ?? ""}
631
  </MarkdownBlock>
 
669
  doc: { icon: BookOpen, label: "Docs" },
670
  };
671
 
672
+ function CitationRow({
673
+ citations,
674
+ citationNumbers,
675
+ }: {
676
+ citations: MessageCitation[];
677
+ citationNumbers?: Map<string, number>;
678
+ }) {
679
  return (
680
  <section className="mt-2 space-y-2 border-t border-[var(--line)] pt-3">
681
  <div className="inline-flex items-center gap-1.5 text-[10px] font-semibold uppercase tracking-[0.12em] text-[var(--muted)]">
 
689
  {citations.map((citation, index) => {
690
  const meta = CITATION_KIND_META[citation.kind];
691
  const Icon = meta.icon;
692
+ const number = citationNumberFor(citationNumbers, citation.url);
693
  return (
694
  <a
695
  key={`${citation.kind}-${citation.url}-${index}`}
 
699
  title={`${meta.label}${citation.sublabel ? ` · ${citation.sublabel}` : ""}`}
700
  className="group inline-flex max-w-full items-center gap-1.5 rounded-full border border-[var(--line-strong)] bg-[var(--accent-faint)] px-2.5 py-1.5 text-xs font-medium text-[var(--accent)] transition hover:border-[var(--accent)] hover:bg-[var(--surface-strong)]"
701
  >
702
+ {number !== undefined ? (
703
+ <span className="shrink-0 rounded-full bg-[var(--surface)] px-1.5 text-[10px] font-semibold text-[var(--accent)]">
704
+ {number}
705
+ </span>
706
+ ) : null}
707
  <Icon className="h-3.5 w-3.5 shrink-0" />
708
  <span className="truncate">{citation.label}</span>
709
  <ExternalLink className="h-3 w-3 shrink-0 opacity-60 transition group-hover:opacity-100" />
frontend/components/markdown-block.tsx CHANGED
@@ -1,15 +1,38 @@
 
1
  import ReactMarkdown from "react-markdown";
2
  import rehypeHighlight from "rehype-highlight";
3
  import remarkGfm from "remark-gfm";
 
4
 
5
  type MarkdownBlockProps = {
6
  children: string;
7
  className?: string;
 
8
  };
9
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
10
  export function MarkdownBlock({
11
  children,
12
  className = "",
 
13
  }: MarkdownBlockProps) {
14
  return (
15
  <div className={`markdown-block ${className}`.trim()}>
@@ -17,13 +40,32 @@ export function MarkdownBlock({
17
  remarkPlugins={[remarkGfm]}
18
  rehypePlugins={[[rehypeHighlight, { detect: true, ignoreMissing: true }]]}
19
  components={{
20
- a: ({ ...props }) => (
21
- <a
22
- {...props}
23
- target="_blank"
24
- rel="noreferrer"
25
- />
26
- ),
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
27
  }}
28
  >
29
  {children}
 
1
+ import type { ReactNode } from "react";
2
  import ReactMarkdown from "react-markdown";
3
  import rehypeHighlight from "rehype-highlight";
4
  import remarkGfm from "remark-gfm";
5
+ import { citationNumberFor } from "@/lib/chat-ui";
6
 
7
  type MarkdownBlockProps = {
8
  children: string;
9
  className?: string;
10
+ citationNumbers?: Map<string, number>;
11
  };
12
 
13
+ // Above this length the link text is a claim the model wrapped in a link, not
14
+ // a source title; keep the text as prose and append the chip after it.
15
+ const CITATION_TEXT_KEEP_THRESHOLD = 48;
16
+
17
+ function nodeToText(node: ReactNode): string {
18
+ if (typeof node === "string" || typeof node === "number") {
19
+ return String(node);
20
+ }
21
+ if (Array.isArray(node)) {
22
+ return node.map(nodeToText).join("");
23
+ }
24
+ if (node && typeof node === "object" && "props" in node) {
25
+ return nodeToText(
26
+ (node as { props: { children?: ReactNode } }).props.children,
27
+ );
28
+ }
29
+ return "";
30
+ }
31
+
32
  export function MarkdownBlock({
33
  children,
34
  className = "",
35
+ citationNumbers,
36
  }: MarkdownBlockProps) {
37
  return (
38
  <div className={`markdown-block ${className}`.trim()}>
 
40
  remarkPlugins={[remarkGfm]}
41
  rehypePlugins={[[rehypeHighlight, { detect: true, ignoreMissing: true }]]}
42
  components={{
43
+ a: ({ children: linkChildren, href, ...props }) => {
44
+ const number = citationNumberFor(citationNumbers, href);
45
+ if (number !== undefined) {
46
+ const label = nodeToText(linkChildren).trim();
47
+ const keepText = label.length > CITATION_TEXT_KEEP_THRESHOLD;
48
+ return (
49
+ <>
50
+ {keepText ? <span>{label}</span> : null}
51
+ <a
52
+ href={href}
53
+ target="_blank"
54
+ rel="noreferrer"
55
+ title={label}
56
+ className="citation-chip"
57
+ >
58
+ {number}
59
+ </a>
60
+ </>
61
+ );
62
+ }
63
+ return (
64
+ <a {...props} href={href} target="_blank" rel="noreferrer">
65
+ {linkChildren}
66
+ </a>
67
+ );
68
+ },
69
  }}
70
  >
71
  {children}
frontend/lib/chat-ui.ts CHANGED
@@ -180,6 +180,47 @@ export function getMessageCitations(message: TutorMessage): MessageCitation[] {
180
  return citations;
181
  }
182
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
183
  function cleanTitle(raw: string | undefined) {
184
  const trimmed = raw?.trim();
185
  if (!trimmed) {
 
180
  return citations;
181
  }
182
 
183
+ const MARKDOWN_LINK_PATTERN = /(!?)\[([^\]]*)\]\(\s*<?([^)\s>]+)>?(?:\s+"[^"]*")?\s*\)/g;
184
+
185
+ function normalizeCitationUrl(url: string) {
186
+ // Mirror the server's normalize_url so inline links and data-source parts
187
+ // resolve to the same key.
188
+ return url.replace(/#.*$/, "").replace(/\/+$/, "");
189
+ }
190
+
191
+ // Number the citation links in a message's answer text by order of first
192
+ // appearance, deduped by normalized URL. The same map drives the inline
193
+ // citation chips and the numbers on the sources row.
194
+ export function buildCitationNumbers(message: TutorMessage): Map<string, number> {
195
+ const numbers = new Map<string, number>();
196
+ for (const part of message.parts) {
197
+ if (!("type" in part) || part.type !== "text") {
198
+ continue;
199
+ }
200
+ const text = (part as TutorMessagePart).text ?? "";
201
+ for (const match of text.matchAll(MARKDOWN_LINK_PATTERN)) {
202
+ if (match[1] === "!") {
203
+ continue;
204
+ }
205
+ const key = normalizeCitationUrl(match[3] ?? "");
206
+ if (key && !numbers.has(key)) {
207
+ numbers.set(key, numbers.size + 1);
208
+ }
209
+ }
210
+ }
211
+ return numbers;
212
+ }
213
+
214
+ export function citationNumberFor(
215
+ numbers: Map<string, number> | undefined,
216
+ url: string | undefined,
217
+ ) {
218
+ if (!numbers || !url) {
219
+ return undefined;
220
+ }
221
+ return numbers.get(normalizeCitationUrl(url));
222
+ }
223
+
224
  function cleanTitle(raw: string | undefined) {
225
  const trimmed = raw?.trim();
226
  if (!trimmed) {