refactor: reasoning parser

This commit is contained in:
cogwheel0
2025-09-05 22:44:04 +05:30
parent 2580bf961e
commit f8650cf809
3 changed files with 454 additions and 173 deletions

View File

@@ -0,0 +1,24 @@
import 'tool_calls_parser.dart';
import 'reasoning_parser.dart';
/// Unified segment representing ordered pieces of a message:
/// - `text`: plain text/markdown to render
/// - `toolCall`: a parsed tool call entry to render as a tile
/// - `reasoning`: a parsed reasoning entry to render as a tile
class MessageSegment {
final String? text;
final ToolCallEntry? toolCall;
final ReasoningEntry? reasoning;
const MessageSegment._({this.text, this.toolCall, this.reasoning});
factory MessageSegment.text(String text) => MessageSegment._(text: text);
factory MessageSegment.tool(ToolCallEntry tool) =>
MessageSegment._(toolCall: tool);
factory MessageSegment.reason(ReasoningEntry entry) =>
MessageSegment._(reasoning: entry);
bool get isText => text != null;
bool get isTool => toolCall != null;
bool get isReasoning => reasoning != null;
}

View File

@@ -100,6 +100,203 @@ class ReasoningParser {
return null;
}
/// Splits content into ordered segments of plain text and reasoning entries
/// (in the order they appear). Supports multiple reasoning blocks.
/// - Handles `<details type="reasoning" ...>` with optional summary/duration/done
/// - Handles raw tag pairs like `<think>...</think>` and `<reasoning>...</reasoning>`
/// - Handles incomplete/streaming cases by emitting a partial reasoning entry
static List<ReasoningSegment>? segments(
String content, {
List<String>? customTagPair,
bool detectDefaultTags = true,
}) {
if (content.isEmpty) return null;
// Build raw tag pairs to check
final tagPairs = <List<String>>[];
if (customTagPair != null && customTagPair.length == 2) {
tagPairs.add(customTagPair);
}
if (detectDefaultTags) {
tagPairs.addAll(defaultReasoningTagPairs);
}
final segs = <ReasoningSegment>[];
int index = 0;
while (index < content.length) {
final nextDetails = content.indexOf('<details', index);
// Find earliest raw tag start among known pairs
int nextRawStart = -1;
List<String>? rawPair; // [start, end]
for (final pair in tagPairs) {
final s = content.indexOf(pair[0], index);
if (s != -1 && (nextRawStart == -1 || s < nextRawStart)) {
nextRawStart = s;
rawPair = pair;
}
}
// Determine which comes first: reasoning <details> or raw tag
int nextIdx;
String kind; // 'details' or 'raw' or 'none'
if (nextDetails == -1 && nextRawStart == -1) {
nextIdx = -1;
kind = 'none';
} else if (nextDetails != -1 && (nextRawStart == -1 || nextDetails < nextRawStart)) {
nextIdx = nextDetails;
kind = 'details';
} else {
nextIdx = nextRawStart;
kind = 'raw';
}
if (kind == 'none') {
if (index < content.length) {
segs.add(ReasoningSegment.text(content.substring(index)));
}
break;
}
// Add text before the next block
if (nextIdx > index) {
segs.add(ReasoningSegment.text(content.substring(index, nextIdx)));
}
if (kind == 'details') {
// Try to parse the opening <details ...>
final openEnd = content.indexOf('>', nextIdx);
if (openEnd == -1) {
// Malformed tag; treat rest as text and stop
segs.add(ReasoningSegment.text(content.substring(nextIdx)));
break;
}
final openTag = content.substring(nextIdx, openEnd + 1);
// Parse attributes
final attrs = <String, String>{};
final attrRegex = RegExp(r'(\w+)="(.*?)"');
for (final m in attrRegex.allMatches(openTag)) {
attrs[m.group(1)!] = m.group(2) ?? '';
}
final isReasoning = (attrs['type'] ?? '') == 'reasoning';
// Find matching closing tag with nesting awareness
int depth = 1;
int i = openEnd + 1;
while (i < content.length && depth > 0) {
final nextOpen = content.indexOf('<details', i);
final nextClose = content.indexOf('</details>', i);
if (nextClose == -1 && nextOpen == -1) break;
if (nextOpen != -1 && (nextClose == -1 || nextOpen < nextClose)) {
depth++;
i = nextOpen + 8; // '<details'
} else {
depth--;
i = (nextClose != -1) ? nextClose + 10 : content.length;
}
}
if (!isReasoning) {
// Not a reasoning block; keep entire block as text if closed,
// else append remainder and stop (streaming/malformed)
if (depth != 0) {
segs.add(ReasoningSegment.text(content.substring(nextIdx)));
break;
} else {
final full = content.substring(nextIdx, i);
segs.add(ReasoningSegment.text(full));
index = i;
continue;
}
}
// Reasoning block
final done = (attrs['done'] ?? 'true') == 'true';
final duration = int.tryParse(attrs['duration'] ?? '0') ?? 0;
if (depth != 0) {
// Unclosed; treat as streaming partial
final after = content.substring(openEnd + 1);
final summaryMatch = RegExp(r'<summary>([^<]*)<\/summary>').firstMatch(after);
final summary = (summaryMatch?.group(1) ?? '').trim();
final reasoning = after
.replaceAll(RegExp(r'^\s*<summary>[\s\S]*?<\/summary>'), '')
.trim();
segs.add(
ReasoningSegment.entry(
ReasoningEntry(
reasoning: reasoning,
summary: summary,
duration: duration,
isDone: false,
),
),
);
// No more content after partial block
break;
} else {
// Closed block: extract inner content
final inner = content.substring(openEnd + 1, i - 10); // without </details>
final sumMatch = RegExp(r'<summary>([^<]*)<\/summary>').firstMatch(inner);
final summary = (sumMatch?.group(1) ?? '').trim();
final reasoning = inner
.replaceAll(RegExp(r'<summary>[\s\S]*?<\/summary>'), '')
.trim();
segs.add(
ReasoningSegment.entry(
ReasoningEntry(
reasoning: reasoning,
summary: summary,
duration: duration,
isDone: done,
),
),
);
index = i;
continue;
}
} else if (kind == 'raw' && rawPair != null) {
final startTag = rawPair[0];
final endTag = rawPair[1];
final start = nextIdx;
final end = content.indexOf(endTag, start + startTag.length);
if (end == -1) {
// Unclosed raw tag => streaming partial
final inner = content.substring(start + startTag.length);
segs.add(
ReasoningSegment.entry(
ReasoningEntry(
reasoning: inner.trim(),
summary: '',
duration: 0,
isDone: false,
),
),
);
break;
} else {
final inner = content.substring(start + startTag.length, end);
segs.add(
ReasoningSegment.entry(
ReasoningEntry(
reasoning: inner.trim(),
summary: '',
duration: 0,
isDone: true,
),
),
);
index = end + endTag.length;
continue;
}
}
}
return segs.isEmpty ? null : segs;
}
/// Checks if a message contains reasoning content
static bool hasReasoningContent(String content) {
if (content.contains('<details type="reasoning"')) return true;
@@ -176,3 +373,41 @@ class ReasoningContent {
.trim();
}
}
/// Lightweight reasoning block for segmented rendering
class ReasoningEntry {
final String reasoning;
final String summary;
final int duration;
final bool isDone;
const ReasoningEntry({
required this.reasoning,
required this.summary,
required this.duration,
required this.isDone,
});
String get formattedDuration => ReasoningParser.formatDuration(duration);
String get cleanedReasoning {
return reasoning
.split('\n')
.map((line) => line.startsWith('>') ? line.substring(1).trim() : line)
.join('\n')
.trim();
}
}
/// Ordered segment that is either plain text or a reasoning entry
class ReasoningSegment {
final String? text;
final ReasoningEntry? entry;
const ReasoningSegment._({this.text, this.entry});
factory ReasoningSegment.text(String text) => ReasoningSegment._(text: text);
factory ReasoningSegment.entry(ReasoningEntry entry) =>
ReasoningSegment._(entry: entry);
bool get isReasoning => entry != null;
}