Spaces:
Running
Running
Commit ·
23b4d13
1
Parent(s): b1f23a6
feat: enhance footnote extraction and display at bottom of page
Browse files
webapp/tipitaka-api/app/services/page_service.py
CHANGED
|
@@ -1,4 +1,5 @@
|
|
| 1 |
import re
|
|
|
|
| 2 |
from app.database.sqlite_db import SQLiteDB
|
| 3 |
from app.services.search_service import SearchService
|
| 4 |
|
|
@@ -273,14 +274,44 @@ class PageService:
|
|
| 273 |
cleaned = self.search_service.clean_text(row["content_text"])
|
| 274 |
is_blank = self._is_blank_page(cleaned)
|
| 275 |
|
| 276 |
-
# Extract end markers from raw content_html
|
| 277 |
raw_html = row["content_html"] or ""
|
| 278 |
end_markers = []
|
|
|
|
|
|
|
| 279 |
if not is_blank and raw_html:
|
|
|
|
| 280 |
for m in re.finditer(r'<B>(.*?จบ.*?)</B>', raw_html, re.IGNORECASE):
|
| 281 |
tm = m.group(1).strip()
|
| 282 |
if tm and tm not in end_markers:
|
| 283 |
end_markers.append(tm)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 284 |
|
| 285 |
return {
|
| 286 |
"id": row["id"],
|
|
@@ -294,6 +325,7 @@ class PageService:
|
|
| 294 |
"title": row["title"],
|
| 295 |
"sections": sections,
|
| 296 |
"end_markers": end_markers,
|
|
|
|
| 297 |
"is_blank": is_blank,
|
| 298 |
}
|
| 299 |
|
|
@@ -366,14 +398,17 @@ class PageService:
|
|
| 366 |
|
| 367 |
for line in lines:
|
| 368 |
line_strip = line.strip()
|
|
|
|
|
|
|
|
|
|
| 369 |
# Definition start pattern: matches @ followed by optional bracket/paren and the id
|
| 370 |
# e.g. @๑, @(๑), @[๑], @๑-
|
| 371 |
start_pattern = rf'^@\s*[(\[]?{re.escape(clean_id)}[)\]-]?.*'
|
| 372 |
|
| 373 |
-
if re.match(start_pattern,
|
| 374 |
capturing = True
|
| 375 |
-
# Strip the @ and marker part
|
| 376 |
-
content = re.sub(rf'^@\s*[(\[]?{re.escape(clean_id)}[)\]-]?\s*', '',
|
| 377 |
if content:
|
| 378 |
found_lines.append(content)
|
| 379 |
elif capturing:
|
|
|
|
| 1 |
import re
|
| 2 |
+
from typing import Optional
|
| 3 |
from app.database.sqlite_db import SQLiteDB
|
| 4 |
from app.services.search_service import SearchService
|
| 5 |
|
|
|
|
| 274 |
cleaned = self.search_service.clean_text(row["content_text"])
|
| 275 |
is_blank = self._is_blank_page(cleaned)
|
| 276 |
|
| 277 |
+
# Extract end markers and footnotes from raw content_html
|
| 278 |
raw_html = row["content_html"] or ""
|
| 279 |
end_markers = []
|
| 280 |
+
footnotes = []
|
| 281 |
+
|
| 282 |
if not is_blank and raw_html:
|
| 283 |
+
# End markers
|
| 284 |
for m in re.finditer(r'<B>(.*?จบ.*?)</B>', raw_html, re.IGNORECASE):
|
| 285 |
tm = m.group(1).strip()
|
| 286 |
if tm and tm not in end_markers:
|
| 287 |
end_markers.append(tm)
|
| 288 |
+
|
| 289 |
+
# All Footnotes
|
| 290 |
+
# Strip line numbers first as they might precede the @ on the same line
|
| 291 |
+
fn_html = re.sub(r'<span\s+class="LineNumber">.*?</span>', '', raw_html)
|
| 292 |
+
lines = fn_html.split('\n')
|
| 293 |
+
current_fn = None
|
| 294 |
+
for line in lines:
|
| 295 |
+
line_strip = line.strip()
|
| 296 |
+
if not line_strip.startswith('@'):
|
| 297 |
+
continue
|
| 298 |
+
|
| 299 |
+
line_plain = re.sub(r'<[^>]+>', '', line_strip)
|
| 300 |
+
# Match @ marker content
|
| 301 |
+
# e.g. @๑ เนื้อหา... or @(๑) เนื้อหา...
|
| 302 |
+
m = re.match(r'^@\s*([(\[]?[\u0E50-\u0E59\d]+[)\]-]?)\s*(.*)', line_plain)
|
| 303 |
+
if m:
|
| 304 |
+
marker = m.group(1).strip()
|
| 305 |
+
text = m.group(2).strip()
|
| 306 |
+
# Clean marker for lookup: [๑] -> ๑
|
| 307 |
+
clean_marker = re.sub(r'[()\[\]-]', '', marker).strip()
|
| 308 |
+
|
| 309 |
+
if clean_marker and not re.search(r'เชิงอรรถ', marker):
|
| 310 |
+
current_fn = {"id": marker, "content": text}
|
| 311 |
+
footnotes.append(current_fn)
|
| 312 |
+
elif current_fn:
|
| 313 |
+
# Continuation line starting with @
|
| 314 |
+
current_fn["content"] += " " + line_plain.lstrip('@').strip()
|
| 315 |
|
| 316 |
return {
|
| 317 |
"id": row["id"],
|
|
|
|
| 325 |
"title": row["title"],
|
| 326 |
"sections": sections,
|
| 327 |
"end_markers": end_markers,
|
| 328 |
+
"footnotes": footnotes,
|
| 329 |
"is_blank": is_blank,
|
| 330 |
}
|
| 331 |
|
|
|
|
| 398 |
|
| 399 |
for line in lines:
|
| 400 |
line_strip = line.strip()
|
| 401 |
+
# Remove tags for matching to handle @<I>๑</I>
|
| 402 |
+
line_plain = re.sub(r'<[^>]+>', '', line_strip)
|
| 403 |
+
|
| 404 |
# Definition start pattern: matches @ followed by optional bracket/paren and the id
|
| 405 |
# e.g. @๑, @(๑), @[๑], @๑-
|
| 406 |
start_pattern = rf'^@\s*[(\[]?{re.escape(clean_id)}[)\]-]?.*'
|
| 407 |
|
| 408 |
+
if re.match(start_pattern, line_plain):
|
| 409 |
capturing = True
|
| 410 |
+
# Strip the @ and marker part from the plain text to get content
|
| 411 |
+
content = re.sub(rf'^@\s*[(\[]?{re.escape(clean_id)}[)\]-]?\s*', '', line_plain)
|
| 412 |
if content:
|
| 413 |
found_lines.append(content)
|
| 414 |
elif capturing:
|
webapp/tipitaka-web/src/components/layout/ReaderPanel.tsx
CHANGED
|
@@ -152,7 +152,9 @@ const ReaderPanel: React.FC = () => {
|
|
| 152 |
x: rect.left + rect.width / 2,
|
| 153 |
y: rect.bottom,
|
| 154 |
id: target.innerText,
|
| 155 |
-
type: isAbbrev ? 'abbrev' : 'footnote'
|
|
|
|
|
|
|
| 156 |
});
|
| 157 |
}
|
| 158 |
}
|
|
@@ -210,6 +212,26 @@ const ReaderPanel: React.FC = () => {
|
|
| 210 |
</>
|
| 211 |
)}
|
| 212 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 213 |
<footer className="mt-12 pt-6 border-t border-[#888]/10 text-center">
|
| 214 |
<span className="text-xs text-[#888]">
|
| 215 |
— เล่ม {currentVolume} หน้า {currentPage} —
|
|
|
|
| 152 |
x: rect.left + rect.width / 2,
|
| 153 |
y: rect.bottom,
|
| 154 |
id: target.innerText,
|
| 155 |
+
type: isAbbrev ? 'abbrev' : 'footnote',
|
| 156 |
+
vol: currentVolume,
|
| 157 |
+
page: currentPage
|
| 158 |
});
|
| 159 |
}
|
| 160 |
}
|
|
|
|
| 212 |
</>
|
| 213 |
)}
|
| 214 |
|
| 215 |
+
{/* Footnotes Section */}
|
| 216 |
+
{content.footnotes && content.footnotes.length > 0 && (
|
| 217 |
+
<div className="mt-16 pt-8 border-t border-[#c8860a]/20">
|
| 218 |
+
<h5 className="text-[#c8860a] text-xs font-bold tracking-widest uppercase mb-6 flex items-center gap-2">
|
| 219 |
+
<svg className="w-3.5 h-3.5" fill="none" viewBox="0 0 24 24" stroke="currentColor">
|
| 220 |
+
<path strokeLinecap="round" strokeLinejoin="round" strokeWidth={2} d="M16 4v12l-4-2-4 2V4M6 20h12a2 2 0 002-2V6a2 2 0 00-2-2H6a2 2 0 00-2 2v12a2 2 0 002 2z" />
|
| 221 |
+
</svg>
|
| 222 |
+
เชิงอรรถ (Footnotes)
|
| 223 |
+
</h5>
|
| 224 |
+
<div className="space-y-4">
|
| 225 |
+
{content.footnotes.map((fn: any, idx: number) => (
|
| 226 |
+
<div key={idx} className="flex gap-3 text-[13px] leading-relaxed text-white/50 hover:text-white/80 transition-colors">
|
| 227 |
+
<span className="text-[#c8860a] font-bold min-w-[24px] text-right shrink-0">{fn.id}</span>
|
| 228 |
+
<span className="font-light">{fn.content}</span>
|
| 229 |
+
</div>
|
| 230 |
+
))}
|
| 231 |
+
</div>
|
| 232 |
+
</div>
|
| 233 |
+
)}
|
| 234 |
+
|
| 235 |
<footer className="mt-12 pt-6 border-t border-[#888]/10 text-center">
|
| 236 |
<span className="text-xs text-[#888]">
|
| 237 |
— เล่ม {currentVolume} หน้า {currentPage} —
|
webapp/tipitaka-web/src/components/reader/ReferencePopup.tsx
CHANGED
|
@@ -6,6 +6,8 @@ export interface RefPopupPos {
|
|
| 6 |
y: number;
|
| 7 |
id: string;
|
| 8 |
type: 'abbrev' | 'footnote';
|
|
|
|
|
|
|
| 9 |
}
|
| 10 |
|
| 11 |
interface Props {
|
|
@@ -79,36 +81,62 @@ const ReferencePopup: React.FC<Props> = ({ pos, onClose }) => {
|
|
| 79 |
position: 'fixed',
|
| 80 |
left: pos.x,
|
| 81 |
top: pos.y,
|
| 82 |
-
transform: 'translate(-50%, 8px)',
|
| 83 |
zIndex: 9998,
|
| 84 |
}}
|
| 85 |
-
className="w-
|
| 86 |
>
|
| 87 |
-
|
| 88 |
-
|
| 89 |
-
<span className="text-xs text-[#888]">
|
| 90 |
-
{pos.type === 'abbrev' ? 'ขยายความส่วนที่ย่อ' : 'เชิงอรรถ'}
|
| 91 |
-
</span>
|
| 92 |
-
{isLoading && <span className="ml-auto animate-pulse text-[10px]">กำลังโหลด...</span>}
|
| 93 |
-
</div>
|
| 94 |
|
| 95 |
-
<div className="
|
| 96 |
-
|
| 97 |
-
<div className="
|
| 98 |
-
<
|
| 99 |
-
|
| 100 |
-
<
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 101 |
</div>
|
| 102 |
-
) : (
|
| 103 |
-
<p className="text-[#e8e4da]/90">
|
| 104 |
-
{content || "ไม่พบข้อมูลอ้างอิง"}
|
| 105 |
-
</p>
|
| 106 |
)}
|
| 107 |
</div>
|
| 108 |
-
|
|
|
|
| 109 |
<div
|
| 110 |
-
className="pointer-events-none absolute left-1/2 -translate-x-1/2 -top-[
|
| 111 |
-
w-
|
| 112 |
/>
|
| 113 |
</motion.div>
|
| 114 |
)}
|
|
|
|
| 6 |
y: number;
|
| 7 |
id: string;
|
| 8 |
type: 'abbrev' | 'footnote';
|
| 9 |
+
vol: number;
|
| 10 |
+
page: number;
|
| 11 |
}
|
| 12 |
|
| 13 |
interface Props {
|
|
|
|
| 81 |
position: 'fixed',
|
| 82 |
left: pos.x,
|
| 83 |
top: pos.y,
|
| 84 |
+
transform: 'translate(-50%, 8px)',
|
| 85 |
zIndex: 9998,
|
| 86 |
}}
|
| 87 |
+
className="w-80 rounded-2xl shadow-[0_20px_50px_rgba(0,0,0,0.5)] bg-[#1a1a2e]/90 backdrop-blur-xl border border-white/10 text-[#e8e4da] overflow-hidden"
|
| 88 |
>
|
| 89 |
+
{/* Gradient top bar */}
|
| 90 |
+
<div className="h-1 bg-gradient-to-r from-[#c8860a]/0 via-[#c8860a] to-[#c8860a]/0" />
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 91 |
|
| 92 |
+
<div className="p-4">
|
| 93 |
+
<div className="flex items-center justify-between mb-3 pb-2 border-b border-white/5">
|
| 94 |
+
<div className="flex items-center gap-2">
|
| 95 |
+
<span className="bg-[#c8860a] text-[#1a1a2e] text-xs font-bold px-1.5 py-0.5 rounded">
|
| 96 |
+
{pos.id}
|
| 97 |
+
</span>
|
| 98 |
+
<span className="text-xs font-medium text-white/50 tracking-wide uppercase">
|
| 99 |
+
{pos.type === 'abbrev' ? 'Abbreviation' : 'Footnote'}
|
| 100 |
+
</span>
|
| 101 |
+
</div>
|
| 102 |
+
<button
|
| 103 |
+
onClick={onClose}
|
| 104 |
+
className="text-white/30 hover:text-white/70 transition-colors"
|
| 105 |
+
>
|
| 106 |
+
<svg className="w-4 h-4" fill="none" viewBox="0 0 24 24" stroke="currentColor">
|
| 107 |
+
<path strokeLinecap="round" strokeLinejoin="round" strokeWidth={2} d="M6 18L18 6M6 6l12 12" />
|
| 108 |
+
</svg>
|
| 109 |
+
</button>
|
| 110 |
+
</div>
|
| 111 |
+
|
| 112 |
+
<div className="max-h-60 overflow-y-auto custom-scrollbar pr-1">
|
| 113 |
+
{isLoading ? (
|
| 114 |
+
<div className="space-y-3 py-1">
|
| 115 |
+
<div className="h-3 bg-white/10 rounded-full w-full animate-pulse" />
|
| 116 |
+
<div className="h-3 bg-white/10 rounded-full w-5/6 animate-pulse" />
|
| 117 |
+
<div className="h-3 bg-white/10 rounded-full w-4/6 animate-pulse" />
|
| 118 |
+
</div>
|
| 119 |
+
) : (
|
| 120 |
+
<div className="text-[15px] leading-relaxed font-light text-white/90">
|
| 121 |
+
{content || (
|
| 122 |
+
<span className="text-white/40 italic">ไม่พบข้อมูลอ้างอิง</span>
|
| 123 |
+
)}
|
| 124 |
+
</div>
|
| 125 |
+
)}
|
| 126 |
+
</div>
|
| 127 |
+
|
| 128 |
+
{pos.type === 'abbrev' && !isLoading && (
|
| 129 |
+
<div className="mt-3 pt-2 border-t border-white/5 flex items-center gap-1.5">
|
| 130 |
+
<div className="w-1.5 h-1.5 rounded-full bg-[#c8860a] animate-pulse" />
|
| 131 |
+
<span className="text-[10px] text-white/30 uppercase tracking-tighter">AI Expanded</span>
|
| 132 |
</div>
|
|
|
|
|
|
|
|
|
|
|
|
|
| 133 |
)}
|
| 134 |
</div>
|
| 135 |
+
|
| 136 |
+
{/* Tail arrow */}
|
| 137 |
<div
|
| 138 |
+
className="pointer-events-none absolute left-1/2 -translate-x-1/2 -top-[5px]
|
| 139 |
+
w-2.5 h-2.5 bg-[#1a1a2e] border-l border-t border-white/10 rotate-45"
|
| 140 |
/>
|
| 141 |
</motion.div>
|
| 142 |
)}
|