Handle search expressions with boundary conditions

This commit is contained in:
Kovid Goyal 2020-01-22 15:57:23 +05:30
parent b0701cc4a6
commit 7cdd0c10ef
No known key found for this signature in database
GPG Key ID: 06BC317B515ACE7C
3 changed files with 49 additions and 3 deletions

View File

@ -117,7 +117,10 @@ class SearchResult(object):
@property @property
def for_js(self): def for_js(self):
return {'file_name': self.file_name, 'spine_idx': self.spine_idx, 'index': self.index, 'text': self.text} return {
'file_name': self.file_name, 'spine_idx': self.spine_idx, 'index': self.index, 'text': self.text,
'before': self.before, 'after': self.after, 'mode': self.search_query.mode
}
def is_or_is_after(self, result_from_js): def is_or_is_after(self, result_from_js):
return result_from_js['spine_idx'] == self.spine_idx and self.index >= result_from_js['index'] and result_from_js['text'] == self.text return result_from_js['spine_idx'] == self.spine_idx and self.index >= result_from_js['index'] and result_from_js['text'] == self.text
@ -148,7 +151,10 @@ def searchable_text_for_name(name):
stack.append(tail) stack.append(tail)
if children: if children:
stack.extend(reversed(children)) stack.extend(reversed(children))
return ''.join(ans) # Normalize whitespace to a single space, this will cause failures
# when searching over spaces in pre nodes, but that is a lesser evil
# since the DOM converts \n, \t etc to a single space
return regex.sub(r'\s+', ' ', ''.join(ans))
def search_in_name(name, search_query, ctx_size=50): def search_in_name(name, search_query, ctx_size=50):

View File

@ -46,7 +46,9 @@ from read_book.shortcuts import (
from read_book.toc import update_visible_toc_anchors from read_book.toc import update_visible_toc_anchors
from read_book.touch import create_handlers as create_touch_handlers from read_book.touch import create_handlers as create_touch_handlers
from read_book.viewport import scroll_viewport from read_book.viewport import scroll_viewport
from utils import debounce, html_escape, is_ios from utils import (
apply_cloned_selection, clone_selection, debounce, html_escape, is_ios
)
FORCE_FLOW_MODE = False FORCE_FLOW_MODE = False
CALIBRE_VERSION = '__CALIBRE_VERSION__' CALIBRE_VERSION = '__CALIBRE_VERSION__'
@ -537,6 +539,29 @@ class IframeBoss:
if not window.find(sr.text, True, False, False, False, False): if not window.find(sr.text, True, False, False, False, False):
self.send_message('search_result_not_found', search_result=sr) self.send_message('search_result_not_found', search_result=sr)
break break
if sr.mode is not 'normal':
# verify we have the correct match since regexes can have
# boundary conditions
sel = window.getSelection()
ranges = clone_selection(sel)
r = ranges[0]
if sr.before:
p = r.cloneRange()
p.collapse(True)
sel = apply_cloned_selection(v'[p]')
sel.modify('extend', 'left', 'character')
if sel.toString() is not sr.before[-1]:
apply_cloned_selection(ranges)
continue
if sr.after:
p = r.cloneRange()
p.collapse(False)
sel = apply_cloned_selection(v'[p]')
sel.modify('extend', 'right', 'character')
if sel.toString() is not sr.after[0]:
apply_cloned_selection(ranges)
continue
apply_cloned_selection(ranges)
idx += 1 idx += 1
if idx > -1 and current_layout_mode() is not 'flow': if idx > -1 and current_layout_mode() is not 'flow':
snap_to_selection() snap_to_selection()

View File

@ -252,6 +252,21 @@ def sandboxed_html(html, style, sandbox):
return ans return ans
def clone_selection(sel):
ans = v'[]'
for i in range(sel.rangeCount):
ans.push(sel.getRangeAt(i).cloneRange())
return ans
def apply_cloned_selection(ranges):
sel = window.getSelection()
sel.removeAllRanges()
for r in ranges:
sel.addRange(r)
return sel
if __name__ is '__main__': if __name__ is '__main__':
from pythonize import strings from pythonize import strings
strings() strings()