| 1 |
# vim:fileencoding=utf-8 |
| 2 |
# License: GPL v3 Copyright: 2021, Kovid Goyal <kovid at kovidgoyal.net> |
| 3 |
from __python__ import bound_methods, hash_literals |
| 4 |
|
| 5 |
from traceback import format_exception |
| 6 |
|
| 7 |
from read_book.db import DB |
| 8 |
from read_book.resources import text_from_serialized_html |
| 9 |
|
| 10 |
GET_SPINE_FAILED = 1 |
| 11 |
CONNECT_FAILED = 2 |
| 12 |
UNHANDLED_ERROR = 3 |
| 13 |
DB_ERROR = 4 |
| 14 |
|
| 15 |
_RE_ESCAPE = /[\/\\^$*+?.()|[\]{}]/g |
| 16 |
quote_map = {'"': '"“”', "'": "'‘’"} |
| 17 |
qpat = /(['"])/g |
| 18 |
spat = /(\s+)/g |
| 19 |
invisible_chars = '(?:[\u00ad\u200c\u200d]{0,1})' |
| 20 |
# Matches each "atom" in an escaped string: an escape sequence (\X) or a |
| 21 |
# single Unicode code point. The u flag ensures surrogate pairs are kept |
| 22 |
# together as one code point. |
| 23 |
_ESCAPED_ATOM = /\\[\s\S]|[\s\S]/gu |
| 24 |
|
| 25 |
|
| 26 |
def escape(string): |
| 27 |
return string.replace(_RE_ESCAPE, '\\$&') |
| 28 |
|
| 29 |
|
| 30 |
def split_string(pat, string): |
| 31 |
pat.lastIndex = 0 |
| 32 |
return string.split(pat) |
| 33 |
|
| 34 |
|
| 35 |
def text_to_regex(text): |
| 36 |
if text and not text.strip(): |
| 37 |
return r'\s+' |
| 38 |
has_leading = text.lstrip() is not text |
| 39 |
has_trailing = text.rstrip() is not text |
| 40 |
ans = [] |
| 41 |
if has_leading: |
| 42 |
ans.push(r'\s+') |
| 43 |
for wpart in split_string(spat, text.strip()): |
| 44 |
if not wpart.strip(): |
| 45 |
ans.push(r'\s+') |
| 46 |
else: |
| 47 |
for part in split_string(qpat, wpart): |
| 48 |
r = quote_map[part] |
| 49 |
if r: |
| 50 |
ans.push('[' + r + ']') |
| 51 |
elif part: |
| 52 |
atoms = escape(part).match(_ESCAPED_ATOM) |
| 53 |
if atoms: |
| 54 |
ans.push(atoms.join(invisible_chars)) |
| 55 |
if has_trailing: |
| 56 |
ans.push(r'\s+') |
| 57 |
return ans.join('') |
| 58 |
|
| 59 |
|
| 60 |
class ToCOffsetMap: |
| 61 |
def __init__(self, toc_nodes, offset_map, previous_toc_node, parent_map): |
| 62 |
self.toc_nodes = toc_nodes or [] |
| 63 |
self.offset_map = offset_map or {} |
| 64 |
self.previous_toc_node = previous_toc_node or None |
| 65 |
self.parent_map = parent_map or {} |
| 66 |
|
| 67 |
def toc_nodes_for_offset(self, offset): |
| 68 |
matches = [] |
| 69 |
for node in self.toc_nodes: |
| 70 |
q = self.offset_map[node.id] |
| 71 |
if q?: |
| 72 |
if q > offset: |
| 73 |
break |
| 74 |
matches.push(node) |
| 75 |
if not matches and self.previous_toc_node: |
| 76 |
matches.push(self.previous_toc_node) |
| 77 |
ans = [] |
| 78 |
if matches.length: |
| 79 |
ancestors = [] |
| 80 |
node = matches[-1] |
| 81 |
parent = self.parent_map[node.id] |
| 82 |
while parent?: |
| 83 |
ancestors.push(parent) |
| 84 |
parent = self.parent_map[parent.id] |
| 85 |
if ancestors.length > 1: |
| 86 |
# last item in ancestors is the root node |
| 87 |
for v'var i = ancestors.length - 1; i-- > 0;': |
| 88 |
ans.push(ancestors[i].id) |
| 89 |
ans.push(node.id) |
| 90 |
return ans |
| 91 |
|
| 92 |
|
| 93 |
def toc_offset_map_for_name(name): |
| 94 |
if wc.toc_offset_map_cache[name]: |
| 95 |
return wc.toc_offset_map_cache[name] |
| 96 |
anchor_map = wc.text_cache[name][1] |
| 97 |
toc_data = wc.current_book.toc_data |
| 98 |
idx = toc_data.spine_idx_map[name] |
| 99 |
toc_nodes = toc_data.spine_toc_map[name] |
| 100 |
if not idx? or idx < 0: |
| 101 |
wc.toc_offset_map_cache[name] = ToCOffsetMap() |
| 102 |
return wc.toc_offset_map_cache[name] |
| 103 |
offset_map = {} |
| 104 |
for node in toc_nodes: |
| 105 |
node_id = node.id |
| 106 |
if node_id?: |
| 107 |
aid = node.frag |
| 108 |
offset = anchor_map[aid] or 0 |
| 109 |
offset_map[node_id] = offset |
| 110 |
prev_toc_node = None |
| 111 |
for v'var i = idx; i-- > 0;': |
| 112 |
spine_name = wc.current_book.spine[i] |
| 113 |
ptn = toc_data.spine_toc_map[spine_name] |
| 114 |
if ptn?: |
| 115 |
prev_toc_node = ptn[-1] |
| 116 |
break |
| 117 |
wc.toc_offset_map_cache[name] = ToCOffsetMap(toc_nodes, offset_map, prev_toc_node, toc_data.parent_map) |
| 118 |
return wc.toc_offset_map_cache[name] |
| 119 |
|
| 120 |
|
| 121 |
def toc_nodes_for_search_result(sr): |
| 122 |
sidx = sr.spine_idx |
| 123 |
name = wc.current_book.spine[sidx] |
| 124 |
if not name: |
| 125 |
return [] |
| 126 |
tmap = toc_offset_map_for_name(name) |
| 127 |
return tmap.toc_nodes_for_offset(sr.offset) |
| 128 |
|
| 129 |
|
| 130 |
class Worker: |
| 131 |
def __init__(self): |
| 132 |
self.db = None |
| 133 |
self.connected_to_db = False |
| 134 |
self.pending_search = None |
| 135 |
self.searching = False |
| 136 |
self.current_query = None |
| 137 |
self.current_query_id = None |
| 138 |
self.current_book = None |
| 139 |
self.regex = None |
| 140 |
self.result_num = 0 |
| 141 |
self.clear_caches() |
| 142 |
|
| 143 |
@property |
| 144 |
def initialize_error_msg(self): |
| 145 |
return self.db?.initialize_error_msg |
| 146 |
|
| 147 |
def clear_caches(self): |
| 148 |
self.text_cache = {} |
| 149 |
self.toc_offset_map_cache = {} |
| 150 |
|
| 151 |
|
| 152 |
wc = Worker() |
| 153 |
|
| 154 |
|
| 155 |
def send_search_complete(): |
| 156 |
self.postMessage({'type': 'search_complete', 'id': wc.current_query_id}) |
| 157 |
wc.current_query = wc.current_query_id = None |
| 158 |
|
| 159 |
|
| 160 |
def search_in_text_of(name): |
| 161 |
ctx_size = 75 |
| 162 |
r = wc.regex |
| 163 |
r.lastIndex = 0 |
| 164 |
haystack = wc.text_cache[name][0] or '' |
| 165 |
match_counts = {} |
| 166 |
spine_idx = wc.current_book.spine.indexOf(name) |
| 167 |
while True: |
| 168 |
m = r.exec(haystack) |
| 169 |
if not m: |
| 170 |
break |
| 171 |
text = m[0] |
| 172 |
start, end = m.index, r.lastIndex |
| 173 |
before = haystack[Math.max(0, start - ctx_size):start] |
| 174 |
after = haystack[end:end + ctx_size] |
| 175 |
q = (before or '')[-15:] + text + (after or '')[:15] |
| 176 |
match_counts[q] = match_counts[q] or 0 |
| 177 |
wc.result_num += 1 |
| 178 |
result = { |
| 179 |
'file_name': name, 'spine_idx': spine_idx, 'index': match_counts[q], 'offset': start, |
| 180 |
'text': text, 'before': before, 'after': after, 'mode': wc.current_query.query.mode, |
| 181 |
'q': q, 'result_num': wc.result_num, 'on_discovery': wc.current_query_id, |
| 182 |
} |
| 183 |
result.toc_nodes = toc_nodes_for_search_result(result) |
| 184 |
self.postMessage({'type': 'search_result', 'id': wc.current_query_id, 'result': result}) |
| 185 |
if wc.current_query.query.only_first_match: |
| 186 |
break |
| 187 |
match_counts[q] += 1 |
| 188 |
|
| 189 |
|
| 190 |
def queue_next_spine_item(spine_idx, allow_current_name): |
| 191 |
spine_idx = spine_idx % wc.current_book.spine.length |
| 192 |
name = wc.current_book.spine[spine_idx] |
| 193 |
if wc.current_query.query.only_first_match and wc.result_num > 0: |
| 194 |
send_search_complete() |
| 195 |
return |
| 196 |
if not name or (not allow_current_name and name is wc.current_query.current_name): |
| 197 |
send_search_complete() |
| 198 |
return |
| 199 |
if wc.text_cache[name]: |
| 200 |
search_in_text_of(name) |
| 201 |
setTimeout(queue_next_spine_item.bind(None, spine_idx + 1), 0) |
| 202 |
return |
| 203 |
query = wc.current_query |
| 204 |
wc.db.get_book_file(wc.current_book.book_hash, wc.current_book.stored_files, name, got_spine_item.bind(None, query.id, spine_idx)) |
| 205 |
|
| 206 |
|
| 207 |
def got_spine_item(query_id, spine_idx, result): |
| 208 |
if query_id is not wc.current_query_id: |
| 209 |
return |
| 210 |
if result.ok: |
| 211 |
name = wc.current_book.spine[spine_idx] |
| 212 |
data = result.result |
| 213 |
if v'data instanceof Blob': # happens for titlepage which is stored as html for some reason |
| 214 |
queue_next_spine_item(spine_idx + 1) |
| 215 |
return |
| 216 |
wc.text_cache[name] = text_from_serialized_html(data, True) |
| 217 |
search_in_text_of(name) |
| 218 |
setTimeout(queue_next_spine_item.bind(None, spine_idx + 1), 0) |
| 219 |
else: |
| 220 |
if result.details is 'ENOENT': |
| 221 |
queue_next_spine_item(spine_idx + 1) |
| 222 |
else: |
| 223 |
send_error(GET_SPINE_FAILED, result.details) |
| 224 |
wc.current_query = wc.current_query_id = None |
| 225 |
|
| 226 |
|
| 227 |
def regex_for_query(query): |
| 228 |
expr = query.text |
| 229 |
flags = 'umg' |
| 230 |
if not query.case_sensitive: |
| 231 |
flags += 'i' |
| 232 |
if query.mode is not 'regex': |
| 233 |
if query.mode is 'word': |
| 234 |
words = [] |
| 235 |
for part in expr.split(' '): |
| 236 |
words.push(r'\b' + text_to_regex(part) + r'\b') |
| 237 |
expr = words.join(r'\s+') |
| 238 |
else: |
| 239 |
expr = text_to_regex(expr) |
| 240 |
return new RegExp(expr, flags) |
| 241 |
|
| 242 |
|
| 243 |
def perform_search(query): |
| 244 |
wc.current_query = query |
| 245 |
wc.current_query_id = query.id |
| 246 |
wc.result_num = 0 |
| 247 |
if not wc.current_book or not wc.current_book.spine?.length or not query.query.text: |
| 248 |
send_search_complete() |
| 249 |
return |
| 250 |
wc.regex = regex_for_query(query.query) |
| 251 |
|
| 252 |
idx = wc.current_book.spine.indexOf(query.current_name) |
| 253 |
if idx < 0: |
| 254 |
idx = 0 |
| 255 |
# Ensure current_name is always a valid spine item name so that the search |
| 256 |
# terminates after going through all spine items exactly once (the search |
| 257 |
# loop ends when it wraps back around to current_name). |
| 258 |
wc.current_query.current_name = wc.current_book.spine[idx] |
| 259 |
queue_next_spine_item(idx, True) |
| 260 |
|
| 261 |
|
| 262 |
def worker_connection_done(): |
| 263 |
wc.connected_to_db = True |
| 264 |
if not wc.initialize_error_msg: |
| 265 |
if wc.pending_search: |
| 266 |
s = wc.pending_search |
| 267 |
wc.pending_search = None |
| 268 |
perform_search(s) |
| 269 |
|
| 270 |
|
| 271 |
def send_error(code, msg, error): |
| 272 |
self.postMessage({'type': 'error', 'code': code, 'msg': msg, 'id': wc.current_query_id, 'error': error}) |
| 273 |
|
| 274 |
|
| 275 |
def on_worker_db_error(title, msg, details): |
| 276 |
send_error(DB_ERROR, msg, {'title': title, 'msg': msg, 'details': details}) |
| 277 |
|
| 278 |
|
| 279 |
def worker_main(): |
| 280 |
wc.db = DB(worker_connection_done, on_worker_db_error) |
| 281 |
|
| 282 |
self.onmessage = def (e): |
| 283 |
if e.data.type is 'search': |
| 284 |
wc.current_query_id = e.data.id |
| 285 |
if wc.connected_to_db: |
| 286 |
if wc.initialize_error_msg: |
| 287 |
send_error(CONNECT_FAILED, wc.initialize_error_msg) |
| 288 |
else: |
| 289 |
perform_search(e.data) |
| 290 |
else: |
| 291 |
wc.pending_search = e.data |
| 292 |
elif e.data.type is 'clear_caches': |
| 293 |
wc.clear_caches() |
| 294 |
wc.current_book = e.data.book |
| 295 |
|
| 296 |
self.addEventListener('error', def (e): |
| 297 |
send_error(UNHANDLED_ERROR, f'{e.lineno}:{e.message}' + '\n\n' + format_exception(e.error).join('')) |
| 298 |
) |