class InstantsearchMainWindowExtension(WindowExtension):
uimanager_xml = '''
<ui>
<menubar name='menubar'>
<menu action='tools_menu'>
<placeholder name='plugin_items'>
<menuitem action='instantsearch'/>
</placeholder>
</menu>
</menubar>
</ui>
'''
gui = "";
@action(_('_Instantsearch'), accelerator='<ctrl>e') # T: menu item
def instantsearch(self):
#init
self.cached_titles = []
#self.menu = defaultdict(_MenuItem)
self.lastQuery = "" # previous user input
self.queryO = None
self.caret = {'pos':0, 'altPos':0, 'text':""} # cursor position
self.originalPage = self.window.ui.page.name # we return here after escape
self.selection = None
if not self.plugin.preferences['isCached']:
# reset last search results
State.reset()
self.menuPage = None
self.isClosed = False
self.lastPage = None
# preferences
self.title_match_char = self.plugin.preferences['title_match_char']
self.start_search_length = self.plugin.preferences['start_search_length']
self.keystroke_delay = self.plugin.preferences['keystroke_delay']
self.open_when_unique = self.plugin.preferences['open_when_unique']
# building quick title cache
def build(start = ""):
if hasattr(self.window.ui.notebook, 'pages'):
o = self.window.ui.notebook.pages
else: # for Zim 0.66-
o = self.window.ui.notebook.index
for s in o.list_pages(Path(start or ":")):
start2 = (start + ":" if start else "") + s.basename
self.cached_titles.append((start2, start2.lower()))
build(start2)
build()
# Gtk
self.gui = Dialog(self.window.ui, _('Search'), buttons=None, defaultwindowsize=(300, -1))
self.gui.resize(300, 100) # reset size
self.inputEntry = InputEntry()
self.inputEntry.connect('key_press_event', self.move)
self.inputEntry.connect('changed', self.change) # self.change is needed by GObject or something
self.gui.vbox.pack_start(self.inputEntry, False)
self.labelObject = gtk.Label((''))
self.labelObject.set_usize(300, -1)
self.gui.vbox.pack_start(self.labelObject, False)
#gui geometry
px, py = self.window.get_position()
pw, ph = self.window.get_size()
x, y = self.gui.get_position()
if self.plugin.preferences['position'] == InstantsearchPlugin.POSITION_RIGHT:
self.gui.move((pw-300), 0)
elif self.plugin.preferences['position'] == InstantsearchPlugin.POSITION_CENTER:
self.gui.resize(300, 100)
self.gui.move(px + (pw / 2) - 150, py + (ph / 2) - 250)
else:
raise AttributeError("Instant search: Wrong position preference.")
self.gui.show_all()
self.labelVar = ""
self.timeout = ""
self.timeoutOpenPage = None
#lastPage = ""
#pageTitleOnly = False
menu = []
#queryTime = 0
def change(self, _): #widget, event,text
if self.timeout:
gobject.source_remove(self.timeout)
q = self.inputEntry.get_text()
#print("Change. {} {}".format(input, self.lastQuery))
if q == self.lastQuery: return
if q == self.title_match_char: return
if q and q[-1] == "∀": # easter egg: debug option for zim --standalone
q = q[:-1]
import ipdb; ipdb.set_trace()
self.state = State.setCurrent(q)
if not self.state.isFinished:
self.isSubset = True if self.lastQuery and q.startswith(self.lastQuery) else False
self.state.checkTitleSearch(self.title_match_char)
self.startSearch()
else: # search completed before
#print("Search already cached.")
self.startSearch() # update the results in a page has been modified meanwhile (not if something got deleted in the notebook #16 )
self.checkLast()
self.soutMenu()
self.lastQuery = q
def startSearch(self):
""" Search string has certainly changed. We search in indexed titles and/or we start zim search.
Normally, zim gives 11 points bonus if the search-string appears in the titles.
If we are ignoring subpages, the search "foo" will match only page "journal:foo",
but not "journal:foo:subpage" (and score of the parent page will get slightly higher by 1.)
However, if there are occurences of the string in the fulltext of the subpage,
subpage remains in the result, but gets bonus only 2 points (not 11).
"""
query = self.state.query
menu = self.state.menu
isInQuery = re.compile(r"(^|:|\s|\()" + query).search # 'te' matches this page titles: 'test' or 'Journal:test' or 'foo test' or 'foo (test)'
if self.isSubset and len(query) < self.start_search_length:
# letter(s) was/were added and full search has not yet been activated
for path in _MenuItem.titles:
if path in self.state.menu and not isInQuery(path.lower()): # 'te' didnt match 'test' etc
del menu[path] # we pop out the result
else:
menu[path].sure = True
else: # perform new search in cached_titles
_MenuItem.titles = set()
found = 0
if self.state.firstSeen:
for path, pathLow in self.cached_titles: # quick search in titles
if isInQuery(pathLow): # 'te' matches 'test' or 'Journal:test' etc
_MenuItem.titles.add(path)
if query in path.lower() and query not in path.lower().split(":")[-1]: # "raz" in "raz:dva", but not in "dva"
self.state.menu[":".join(path.split(":")[:-1])].bonus += 1 # 1 point for subpage
menu[path].bonus = -11
menu[path].score += 10 # 10 points for title (zim default) (so that it gets displayed before search finishes)
menu[path].intitle = True
menu[path].path = path
found += 1
if found >= 10: # we dont want more than 10 results; we would easily match all of the pages
break
else:
menu[path].intitle = False
self.processMenu() # show for now results of title search
if len(query) >= self.start_search_length:
self.timeout = gobject.timeout_add(self.keystroke_delay, self.startZimSearch) # ideal delay between keystrokes
def startZimSearch(self):
""" Starts search for the input. """
self.timeout = ""
self.caret['altPos'] = 0 # possible position of caret - beginning
s = '"*{}*"'.format(self.state.query) if self.plugin.preferences['isWildcarded'] else self.state.query
self.queryO = Query(unicode(s)) # beware when searching for unicode character. Update the row when going to Python3.
lastSel = self.selection if self.isSubset and self.state.previous.isFinished else None # it should be quicker to find the string, if we provide this subset from last time (in the case we just added a letter, so that the subset gets smaller)
self.selection = SearchSelection(self.window.ui.notebook)
state = self.state # this is thread, so that self.state would can before search finishes
self.selection.search(self.queryO, selection=lastSel, callback=self._search_callback(self.state.rawQuery))
state.isFinished = True
for item in list(state.menu): # remove all the items that we didnt encounter during the search
if not state.menu[item].sure:
del state.menu[item]
if state == self.state:
self.checkLast()
self.processMenu(state=state)
def checkLast(self):
""" opens the page if there is only one option in the menu """
if self.open_when_unique and len(self.state.menu) == 1:
self._open_page(Path(self.state.menu.keys()[0]), excludeFromHistory=False)
self.close()
def _search_callback(self, query):
def _search_callback(results, path):
if results is not None:
self._update_results(results, State.get(query)) # we finish the search even if another search is running. If we returned False, the search would be cancelled-
while gtk.events_pending():
gtk.main_iteration(block=False)
return True
return _search_callback
def _update_results(self, results, state):
"""
This method may run many times, due to the _update_results, which are updated many times.
I may set that _update_results would run only once, but this is nice - the results are appearing one by one.
"""
changed = False
state.lastResults = results
for option in results.scores:
if state.pageTitleOnly and state.query not in option.name: # hledame jen v nazvu stranky
continue
if option.name not in state.menu: # new item found
if state == self.state and option.name == self.caret['text']: # this is current search
self.caret['altPos'] = len(state.menu)-1 #karet byl na tehle pozici, pokud se zuzil vyber, budeme vedet, kam karet opravne umistit
if option.name not in state.menu or (state.menu[option.name].bonus < 0 and state.menu[option.name].score == 0):
changed = True
if not state.menu[option.name].sure:
state.menu[option.name].sure = True
changed = True
state.menu[option.name].score = results.scores[option] #zaradit mezi moznosti
if changed: # we added a page
self.processMenu(state=state, sort=False)
else:
pass
def processMenu(self, state=None, sort=True):
""" Sort menu and generate items and sout menu. """
if state is None:
state = self.state
if sort:
state.items = sorted(state.menu, reverse=True, key=lambda item: (state.menu[item].intitle, state.menu[item].score + state.menu[item].bonus, -item.count(":"), item))
else: # when search results are being updated, it's good when the order doesnt change all the time. So that the first result does not become for a while 10th and then become first back.
state.items = sorted(state.menu, reverse=True, key=lambda item: (state.menu[item].intitle, -state.menu[item].lastOrder))
state.items = [item for item in state.items if (state.menu[item].score + state.menu[item].bonus) > 0] # i dont know why there are items with 0 score
if state == self.state:
self.soutMenu()
def soutMenu(self, displayImmediately=False):
""" Displays menu and handles caret position. """
if self.timeoutOpenPage:
gobject.source_remove(self.timeoutOpenPage)
self.gui.resize(300, 100) # reset size
# treat possible caret deflection
if self.caret['pos'] < 0 or self.caret['pos'] > len(self.state.items)-1: # place the caret to the beginning or the end of list
self.caret['pos'] = self.caret['altPos']
text = ""
i = 0
for item in self.state.items:
score = self.state.menu[item].score + self.state.menu[item].bonus
self.state.menu[item].lastOrder = i
if i == self.caret['pos']:
self.caret['text'] = item # caret is at this position
text += '→ {} ({}) {}\n'.format(item, score, "" if self.state.menu[item].sure else "?")
else:
try:
text += '{} ({}) {}\n'.format(item, score, "" if self.state.menu[item].sure else "?")
except:
text += "CHYBA\n"
text += item[0:-1] + "\n"
i += 1
self.labelObject.set_text(text)
self.menuPage = Path(self.caret['text'] if len(self.state.items) else self.originalPage)
if not displayImmediately:
self.timeoutOpenPage = gobject.timeout_add(self.keystroke_delay, self._open_page, self.menuPage) # ideal delay between keystrokes
else:
self._open_page(self.menuPage)
def move(self, widget, event):
""" Move caret up and down. Enter to confirm, Esc closes search."""
keyname = gtk.gdk.keyval_name(event.keyval)
if keyname == "Up" or keyname == "ISO_Left_Tab":
self.caret['pos'] -= 1
self.soutMenu(displayImmediately=False)
if keyname == "Down" or keyname == "Tab":
self.caret['pos'] += 1
self.soutMenu(displayImmediately=False)
if keyname == "KP_Enter" or keyname == "Return":
self._open_page(self.menuPage, excludeFromHistory=False)
self.close()
if keyname == "Escape":
self._open_original()
# GTK closes the windows itself on Escape, no self.close() needed
return
## Safely closes
# Xwhen closing directly, Python gave allocation error
def close(self):
if not self.isClosed:
self.isClosed = True
self.gui.emit("close")
def _open_original(self):
self._open_page(Path(self.originalPage))
def _open_page(self, page, excludeFromHistory=True):
""" Open page and highlight matches """
self.timeoutOpenPage = None # no delayed page will be open
if self.isClosed == True:
return
if page and page.name and page.name != self.lastPage:
self.lastPage = page.name
#print("*** HISTORY BEF", self.window.ui.history._history[-3:])
self.window.ui.open_page(page)
if excludeFromHistory:
# there is no public API, so lets use protected _history instead
self.window.ui.history._history.pop()
self.window.ui.history._current = len(self.window.ui.history._history) - 1
if not excludeFromHistory and self.window.ui.history.get_current().name is not page.name:
# we insert the page to the history because it was likely to be just visited and excluded
self.window.ui.history.append(page)
# Popup find dialog with same query
if self.queryO:# and self.queryO.simple_match:
string = self.state.query
string = string.strip('*') # support partial matches
if self.plugin.preferences['highlight_search']:
self._get_mainwindow().pageview.show_find(string, highlight=True)
def _get_mainwindow(self): # #18
try:
return self.window.ui._mainwindow
except AttributeError:
return self.window.ui.mainwindow
class InstantSearchMainWindowExtension(MainWindowExtension):
gui: "Dialog"
state: "State"
def __init__(self, plugin, window):
super().__init__(plugin, window)
self.timeout = None
self.timeout_open_page = None # will open page after keystroke delay
self.timeout_open_page_preview = None # will open page after keystroke delay
self.cached_titles = None
self.last_query = None
self.query_o = None
self.caret = None
self.original_page = None
self.original_history = None
self.selection = None
self.menu_page = None
self.is_closed = None
self.last_page = self.last_page_preview = None
self.label_object = None
self.input_entry = None
self.label_preview = None
self.preview_pane = None
self._last_update = 0
# preferences
State.title_match_char = self.plugin.preferences['title_match_char']
State.start_search_length = self.plugin.preferences[
'start_search_length']
self.keystroke_delay_open = self.plugin.preferences[
'keystroke_delay_open']
self.keystroke_delay = self.plugin.preferences['keystroke_delay']
# noinspection PyArgumentList,PyUnresolvedReferences
@action(_('_Instant search'), accelerator='<ctrl>e') # T: menu item
def instant_search(self):
# init
self.cached_titles = []
self.last_query = "" # previous user input
self.query_o = None
self.caret = SimpleNamespace(pos=0, text="",
stick=False) # cursor position
self.original_page = self.window.page.name # we return here after escape
self.original_history = list(self.window.history.uistate["list"])
self.selection = None
if not self.plugin.preferences['is_cached']:
# reset last search results
State.reset()
self.menu_page = None
self.is_closed = False
self.last_page = None
# building quick title cache
def build(start=""):
o = self.window.notebook.pages
for s in o.list_pages(Path(start or ":")):
start2 = (start + ":" if start else "") + s.basename
self.cached_titles.append((start2, start2.lower()))
build(start2)
build()
# Gtk
self.gui = Dialog(self.window,
_('Search'),
buttons=None,
defaultwindowsize=(300, -1))
self.gui.resize(300, 100) # reset size
self.input_entry = InputEntry()
self.input_entry.connect('key_press_event', self.move)
self.input_entry.connect(
'changed',
self.change) # self.change is needed by GObject or something
self.gui.vbox.pack_start(self.input_entry,
expand=False,
fill=True,
padding=0)
# noinspection PyArgumentList
self.label_object = Gtk.Label(label='')
self.label_object.set_size_request(300, -1)
self.gui.vbox.pack_start(self.label_object,
expand=False,
fill=True,
padding=0)
# preview pane
self.label_preview = Gtk.Label(label='...loading...')
# not sure if this has effect, longer lines without spaces still make window inflate
self.label_preview.set_line_wrap(True)
self.label_preview.set_xalign(0) # align to the left
self.preview_pane = Gtk.VBox()
self.preview_pane.pack_start(self.label_preview, False, False, 5)
self.window.pageview.pack_start(self.preview_pane, False, False, 5)
# gui geometry
self.geometry(init=True)
self.gui.show_all()
def geometry(self, init=False, repeat=True, force=False):
if repeat and not init:
# I do not know how to catch callback when result list's width is final, so we align several times
[
GObject.timeout_add(
x, lambda: self.geometry(repeat=False, force=force))
for x in (30, 50, 70, 400)
]
# it is not worthy we continue now because often the Gtk redraw is delayed which would mean
# the Dialog dimensions change twice in a row
return
px, py = self.window.get_position()
pw, ph = self.window.get_size()
if init:
x, y = None, None
w, h = 300, 100
else:
x, y = self.gui.get_position()
w, h = self.gui.get_allocated_width(
), self.gui.get_allocated_height()
if self.plugin.preferences[
'position'] == InstantSearchPlugin.POSITION_RIGHT:
x2, y2 = pw - w, 0
elif self.plugin.preferences[
'position'] == InstantSearchPlugin.POSITION_CENTER:
x2, y2 = px + (pw / 2) - w / 2, py + (ph / 2) - 250
else:
raise AttributeError("Instant search: Wrong position preference.")
if init or x != x2 or force:
self.gui.resize(300, 100)
self.gui.move(x2, y2)
def title(self, title=""):
self.gui.set_title("Search " + title)
def change(self, _): # widget, event,text
if self.timeout:
GObject.source_remove(self.timeout)
self.timeout = None
q = self.input_entry.get_text()
if q == self.last_query:
return
if q == State.title_match_char:
return
if q and q[-1] == "∀": # easter egg: debug option for zim --standalone
q = q[:-1]
import ipdb
ipdb.set_trace()
self.state = State.set_current(q)
if not self.state.is_finished:
if self.start_search():
self.process_menu()
else: # search completed before
# update the results if a page has been modified meanwhile
# (not if something got deleted in the notebook #16 )
self.start_search()
self.check_last()
self.sout_menu()
self.last_query = q
def start_search(self):
""" Search string has certainly changed. We search in indexed titles and/or we start fulltext search.
:rtype: True if no other search is needed and we may output the menu immediately.
"""
query = self.state.query
menu = self.state.menu
if not query:
return self.process_menu() # show for now results of title search
# 'te' matches these page titles: 'test' or 'Journal:test' or 'foo test' or 'foo (test)'
sub_queries = [
re.compile(r"(^|:|\s|\()?" + q, re.IGNORECASE)
for q in query.split(" ")
]
def in_query(txt):
""" False if any part of the query does not match.
If the query is longer >3 characters:
* +10 for every query part that matches a title part beginning
Ex: query 'te' -> +3 for these page titles:
'test' or 'Journal:test' or 'foo test' or 'foo (test)'
* +1 for every query part
Ex: query 'st' -> +1 for those page titles
If the query is shorter <=3 characters:
+10 for every query part that matches a title part beginning 'te' for 'test'
False otherwise ('st' for 'test') so that you do not end up messed
with page titles, after writing a single letter.
"""
arr = (q.search(txt) for q in sub_queries)
try:
if len(query) <= 3:
# raises if subquery m does not match or is not at a page chunk beginning
return sum(10 if m.group(1) is not None else None
for m in arr)
else:
# raises if subquery m does not match
return sum(10 if m.group(1) is not None else 1
for m in arr)
except (AttributeError, TypeError
): # one of the sub_queries is not part of the page title
return False
# we loop either all cached page titles or menu that should be built from previous superset-query menu
it = ((x, x.lower())
for x in list(menu)) if menu else self.cached_titles
for path, path_low in it: # quick search in titles
score = in_query(path_low)
if score: # 'te' matches 'test' or 'Journal:test' etc
# "foo" in "foo:bar", but not in "bar"
# when looping "foo:bar", page "foo" receives +1 for having a subpage
if all(q in path.lower() for q in query) \
and any(q not in path.lower().split(":")[-1] for q in query):
menu[":".join(path.split(
":")[:-1])].bonus += 1 # 1 point for having a subpage
# Normally, zim search gives 11 points bonus if the search-string appears in the titles.
# If we are ignoring sub-pages, the search "foo" will match only page "journal:foo",
# but not "journal:foo:subpage"
# (and score of the parent page will get slightly higher by 1.)
# However, if there are occurrences of the string in the fulltext of the subpage,
# subpage remains in the result, but gets bonus only 2 points (not 11).
# But internal zim search is now disabled.
# menu[path].bonus = -11
else:
# 10 points for title (zim default) (so that it gets displayed before search finishes)
menu[
path].bonus += score # will be added to score (score will be reset)
# if score > 9, it means this might be priority match, not fulltext header search
# ex "te" for "test" is priority, whereas "st" is just fulltext
menu[path].in_title = True if score > 9 else False
menu[path].path = path
# menu[path].sure = True
if self.state.page_title_only:
return True
else:
if not self.state.previous or len(
query) == State.start_search_length:
# quickly show page title search results before longer fulltext search is ready
# Either there is no previous state – query might have been copied into input
# or the query is finally long enough to start fulltext search.
# It is handy to show out filtered page names before because
# it is often use case to jump to queries matched in page names.
self.process_menu(ignore_geometry=True)
self.title("..")
self.timeout = GObject.timeout_add(
self.keystroke_delay,
self.start_zim_search) # ideal delay between keystrokes
def start_zim_search(self):
""" Starts search for the input. """
self.title("...")
if self.timeout:
GObject.source_remove(self.timeout)
self.timeout = None
self.query_o = Query(self.state.query)
# it should be quicker to find the string, if we provide this subset from last time
# (in the case we just added a letter, so that the subset gets smaller)
# last_sel = self.selection if self.is_subset and self.state.previous and self.state.previous.is_finished
# else None
selection = self.selection = SearchSelection(self.window.notebook)
state = self.state # this is a thread, so that self.state might change before search finishes
# internal search disabled - it was way too slower
# selection.search(self.query_o, selection=last_sel, callback=self._search_callback(state))
# self._update_results(selection, state, force=True)
# self.title("....")
# fulltext external search
# Loop either all .txt files in the notebook or narrow the search with a previous state
if state.previous and state.previous.is_finished and state.previous.matching_files is not None:
paths_set = state.previous.matching_files
# see below paths_cached_set = (p for p in files_set if p in InstantSearchPlugin.file_cache)
else:
paths_set = (f for f in pathlib.Path(
str(self.window.notebook.folder)).rglob("*.txt")
if f.is_file())
# see below paths_cached_set = (p for p in InstantSearchPlugin.file_cache)
state.matching_files = []
# This cached search takes about 60 ms, so I let it commented.
# However on HDD disks this may boost performance.
# We may do an option: "empty cache immediately after close (default)",
# "search cache first and then do the fresh search (HDD)"
# "use cache always (empties cache after Zim restart)"
# "empty cache after 5 minutes"
# and then prevent to clear the cache in .close().
# Or rather we may read file mtime and re-read if only it has been changed since last search.
# if not InstantSearchPlugin.file_cache_fresh:
# # Cache might not be fresh but since it is quick, perform quick non-fresh-cached search
# # and then do a fresh search. If we are lucky enough, results will not change.
# # using temporary selection so that files will not received double points for both cached and fresh loop
# selection_temp = SearchSelection(self.window.notebook)
# self.start_external_search(selection_temp, state, paths_cached_set)
# InstantSearchPlugin.file_cache_fresh = True
# InstantSearchPlugin.file_cache.clear()
self.start_external_search(selection, state, paths_set)
state.is_finished = True
# for item in list(state.menu): # remove all the items that we didnt encounter during the search
# if not state.menu[item].sure:
# del state.menu[item]
if state == self.state:
self.check_last()
self.process_menu(state=state)
self.title()
def start_external_search(self, selection, state: "State", paths):
""" Zim internal search is not able to find out text with markup.
Ex:
'economical' is not recognized as 'economi**cal**' (however highlighting works great),
as 'economi[[inserted link]]cal'
as 'any text with [[http://economical.example.com|link]]'
This fulltext search loops all .txt files in the notebook directory
and tries to recognize the patterns.
"""
# divide query to independent words "foo economical" -> "foo", "economical", page has to contain both
# strip markup: **bold**, //italic//, __underline__, ''verbatim'', ~~strike through~~
# matches query "economi**cal**"
def letter_split(q):
""" Every letter is divided by a any-formatting-match-group and escaped.
'foo.' -> 'f[*/'_~]o[*/'_~]o[*/'_~]\\.'
"""
return r"[*/'_~]*".join((re.escape(c) for c in list(q)))
sub_queries = state.query.split(" ")
# regex to identify in all sub_queries present in the text
queries = [
re.compile(letter_split(q), re.IGNORECASE) for q in sub_queries
]
# regex to identify the very query is present
exact_query = re.compile(letter_split(
state.query), re.IGNORECASE) if len(sub_queries) > 1 else None
# regex to count the number of the sub_queries present and to optionally add information about header used
header_queries = [
re.compile("(\n=+ .*)?" + letter_split(q), re.IGNORECASE)
for q in sub_queries
]
# regex to identify inner link contents
link = re.compile(
r"\[\[(.*?)\]\]",
re.IGNORECASE) # matches all links "economi[[inserted link]]cal"
for p in paths:
if p not in InstantSearchPlugin.file_cache:
s = p.read_text() # strip header
if s.startswith('Content-Type: text/x-zim-wiki'):
# XX will that work on Win?
# I should use more general separator IMHO in the whole file rather than '\n'.
s = s[s.find("\n\n"):]
InstantSearchPlugin.file_cache[p] = s
else:
s = InstantSearchPlugin.file_cache[p]
matched_links = []
def matched_link(match):
matched_links.append(match.group(1))
return ""
# pull out links "economi[[inserted link]]cal" -> "economical" + "inserted link"
txt_body = link.sub(matched_link, s)
txt_links = "".join(matched_links)
if all(
query.search(txt_body) or query.search(txt_links)
for query in queries):
path = self.window.notebook.layout.map_file(File(str(p)))[0]
# score = header order * 3 + body match count * 1
# if there are '=' equal chars before the query, it is header. The bigger number, the bigger header.
# Header 5 corresponds to 3 points, Header 1 to 7 points.
score = sum([
len(m.group(1)) * 3 if m.group(1) else 1
for q in header_queries for m in q.finditer(txt_body)
])
if exact_query: # there are sub-queries, we favourize full-match
score += 50 * len(exact_query.findall(txt_body))
# noinspection PyProtectedMember
# score might be zero because we are not re-checking against txt_links matches
selection._count_score(path, score or 1)
state.matching_files.append(p)
self._update_results(selection, state, force=True)
def check_last(self):
""" opens the page if there is only one option in the menu """
if len(self.state.menu
) == 1 and self.plugin.preferences['open_when_unique']:
self._open_page(Path(list(self.state.menu)[0]),
exclude_from_history=False)
self.close()
elif not len(self.state.menu):
self._open_original()
def _search_callback(self, state):
def _(results, _path):
if results is not None:
# we finish the search even if another search is running.
# If returned False, the search would be cancelled
self._update_results(results, state)
while Gtk.events_pending():
Gtk.main_iteration()
return True
return _
def _update_results(self, results, state: "State", force=False):
"""
This method may run many times, due to the _update_results, which are updated many times,
the results are appearing one by one. However, if called earlier than 0.2 s, ignored.
Measures:
If every callback would be counted, it takes 3500 ms to build a result set.
If callbacks earlier than 0.6 s -> 2300 ms, 0.3 -> 2600 ms, 0.1 -> 2800 ms.
"""
if not force and time(
) < self._last_update + 0.2: # if update callback called earlier than 200 ms, ignore
return
self._last_update = time()
changed = False
for option in results.scores:
if option.name not in state.menu or (
state.menu[option.name].bonus < 0
and state.menu[option.name].score == 0):
changed = True
o: _MenuItem = state.menu[option.name]
# if not o.sure:
# o.sure = True
# changed = True
o.score = results.scores[option] # includes into options
if changed: # we added a page
self.process_menu(state=state, sort=False)
else:
pass
def process_menu(self, state=None, sort=True, ignore_geometry=False):
""" Sort menu and generate items and sout menu. """
if state is None:
state = self.state
if sort:
state.items = sorted(
state.menu,
reverse=True,
key=lambda item:
(state.menu[item].in_title, state.menu[item].score + state.
menu[item].bonus, -item.count(":"), item))
else:
# when search results are being updated, it's good when the order doesnt change all the time.
# So that the first result does not become for a while 10th and then become first back.
state.items = sorted(
state.menu,
reverse=True,
key=lambda item:
(state.menu[item].in_title, -state.menu[item].last_order))
# I do not know why there are items with score 0 if internal Zim search used
state.items = [
item for item in state.items
if (state.menu[item].score + state.menu[item].bonus) > 0
]
if state == self.state:
self.sout_menu(ignore_geometry=ignore_geometry)
def sout_menu(self,
display_immediately=False,
caret_move=None,
ignore_geometry=False):
""" Displays menu and handles caret position. """
if self.timeout_open_page:
GObject.source_remove(self.timeout_open_page)
self.timeout_open_page = None
if self.timeout_open_page_preview:
GObject.source_remove(self.timeout_open_page_preview)
self.timeout_open_page_preview = None
# caret:
# by default stays at position 0
# If moved to a page, it keeps the page.
# If moved back to position 0, stays there.
if caret_move is not None:
if caret_move == 0:
self.caret.pos = 0
else:
self.caret.pos += caret_move
self.caret.stick = self.caret.pos != 0
elif self.state.items and self.caret.stick:
# identify current caret position, depending on the text
self.caret.pos = next((i for i, item in enumerate(self.state.items)
if item == self.caret.text), 0)
# treat possible caret deflection
if self.caret.pos < 0:
# place the caret to the beginning or the end of list
self.caret.pos = 0
elif self.caret.pos >= len(self.state.items):
self.caret.pos = 0 if caret_move == 1 else len(
self.state.items) - 1
text = []
i = 0
for item in self.state.items:
score = self.state.menu[item].score + self.state.menu[item].bonus
self.state.menu[item].last_order = i
pieces = item.split(":")
pieces[-1] = f"<b>{pieces[-1]}</b>"
s = ":".join(pieces)
if i == self.caret.pos:
self.caret.text = item # caret is at this position
# text += f'→ {s} ({score}) {"" if self.state.menu[item].sure else "?"}\n'
text.append(f'→ {s} ({score})')
else:
# text += f'{s} ({score}) {"" if self.state.menu[item].sure else "?"}\n'
text.append(f'{s} ({score})')
i += 1
text = "No result" if not text and self.state.is_finished else "\n".join(
text)
self.label_object.set_markup(text)
self.menu_page = Path(
self.caret.text if len(self.state.items) else self.original_page)
if not display_immediately:
if self.plugin.preferences[
'preview_mode'] != InstantSearchPlugin.PREVIEW_ONLY:
self.timeout_open_page = GObject.timeout_add(
self.keystroke_delay_open, self._open_page,
self.menu_page) # ideal delay between keystrokes
if self.plugin.preferences[
'preview_mode'] != InstantSearchPlugin.FULL_ONLY:
self.timeout_open_page_preview = GObject.timeout_add(
self.keystroke_delay, self._open_page_preview,
self.menu_page) # ideal delay between keystrokes
else:
self._open_page(self.menu_page)
# we force here geometry to redraw because often we end up with "No result" page that is very tall
# because of a many records just hidden
if not ignore_geometry:
self.geometry(force=True)
def move(self, widget, event):
""" Move caret up and down. Enter to confirm, Esc closes search."""
key_name = Gdk.keyval_name(event.keyval)
# handle basic caret movement
moves = {
"Up": -1,
"ISO_Left_Tab": -1,
"Down": 1,
"Tab": 1,
"Page_Up": -10,
"Page_Down": 10
}
if key_name in moves:
self.sout_menu(display_immediately=False,
caret_move=moves[key_name])
elif key_name in ("Home", "End"):
if event.state & Gdk.ModifierType.CONTROL_MASK or event.state & Gdk.ModifierType.SHIFT_MASK:
# Ctrl/Shift+Home jumps to the query input text start
return
if key_name == "Home": # Home jumps at the result list start
self.sout_menu(display_immediately=False, caret_move=0)
widget.emit_stop_by_name("key-press-event")
else:
self.sout_menu(display_immediately=False,
caret_move=float("inf"))
widget.emit_stop_by_name("key-press-event")
# confirm or cancel
elif key_name == "KP_Enter" or key_name == "Return":
self._open_page(self.menu_page, exclude_from_history=False)
self.close()
elif key_name == "Escape":
self._open_original()
self.is_closed = True # few more timeouts are on the way probably
self.close()
return
def close(self):
""" Safely (closes gets called when hit Enter) """
if not self.is_closed: # if hit Esc, GTK has already emitted close itself
self.is_closed = True
self.gui.emit("close")
# remove preview pane and show current text editor
self._hide_preview()
self.preview_pane.destroy()
InstantSearchPlugin.file_cache.clear(
) # until next search, pages might change
InstantSearchPlugin.file_cache_fresh = False
def _open_original(self):
self._open_page(Path(self.original_page))
# we already have HistoryPath objects in the self.original_history, we cannot add them in the constructor
# XX I do not know what is that good for
hl = HistoryList([])
hl.extend(self.original_history)
self.window.history.uistate["list"] = hl
# noinspection PyProtectedMember
def _open_page(self, page, exclude_from_history=True):
""" Open page and highlight matches """
self._hide_preview()
if self.timeout_open_page: # no delayed page will be open
GObject.source_remove(self.timeout_open_page)
self.timeout_open_page = None
if self.timeout_open_page_preview: # no delayed preview page will be open
GObject.source_remove(self.timeout_open_page_preview)
self.timeout_open_page_preview = None
# open page
if page and page.name and page.name != self.last_page:
self.last_page = page.name
self.window.navigation.open_page(page)
if exclude_from_history and list(
self.window.history._history
)[-1:][0].name != self.original_page:
# there is no public API, so lets use protected _history instead
self.window.history._history.pop()
self.window.history._current = len(
self.window.history._history) - 1
if not exclude_from_history and self.window.history.get_current(
).name is not page.name:
# we insert the page to the history because it was likely to be just visited and excluded
self.window.history.append(page)
# Popup find dialog with same query
if self.query_o: # and self.query_o.simple_match:
string = self.state.query
string = string.strip('*') # support partial matches
if self.plugin.preferences['highlight_search']:
# unfortunately, we can highlight single word only
self.window.pageview.show_find(string.split(" ")[0],
highlight=True)
def _hide_preview(self):
self.preview_pane.hide()
# noinspection PyProtectedMember
self.window.pageview._hack_hbox.show()
def _open_page_preview(self, page):
""" Open preview which is far faster then loading and
building big parse trees into text editor buffer when opening page. """
# note: if the dialog is already closed, we do not want a preview to open, but page still can be open
# (ex: after hitting Enter the dialog can close before opening the page)
if self.timeout_open_page_preview:
# no delayed preview page will be open, however self.timeout_open_page might be still running
GObject.source_remove(self.timeout_open_page_preview)
self.timeout_open_page_preview = None
# it does not pose a problem if we re-load preview on the same page;
# the query text might got another letter to highlight
if page and not self.is_closed:
# show preview pane and hide current text editor
self.last_page_preview = page.name
local_file = self.window.notebook.layout.map_page(page)[0]
path = pathlib.Path(str(local_file))
if path in InstantSearchPlugin.file_cache:
s = InstantSearchPlugin.file_cache[path]
else:
try:
s = InstantSearchPlugin.file_cache[path] = local_file.read(
)
except newfs.base.FileNotFoundError:
s = f"page {page} has no content" # page has not been created yet
lines = s.splitlines()
# the file length is very small, prefer to not use preview here
if self.plugin.preferences[
'preview_mode'] != InstantSearchPlugin.PREVIEW_ONLY and len(
lines) < 50:
return self._open_page(page, exclude_from_history=True)
self.label_preview.set_markup(
self._get_preview_text(lines, self.state.query))
# shows GUI (hidden in self._hide_preview()
self.preview_pane.show_all()
# noinspection PyProtectedMember
self.window.pageview._hack_hbox.hide()
def _get_preview_text(self, lines, query):
max_lines = 200
# check if the file is a Zim markup file and if so, skip header
if lines[0] == 'Content-Type: text/x-zim-wiki':
for i, line in enumerate(lines):
if line == "":
lines = lines[i + 1:]
break
if query.strip() == "":
return "\n".join(line for line in lines[:max_lines])
# searching for "a" cannot match "&a", since markup_escape_text("&") -> "'"
# Ignoring q == "b", it would interfere with multiple queries:
# Ex: query "f b", text "foo", matched with "f" -> "<b>f</b>oo", matched with "b" -> "<<b>b</b>>f</<b>b</b>>"
query_match = (re.compile("(" + re.escape(q) + ")", re.IGNORECASE)
for q in query.split(" ") if q != "b")
# too long lines caused strange Gtk behaviour – monitor brightness set to maximum, without any logged warning
# so that I decided to put just extract of such long lines in preview
# This regex matches query chunk in the line, prepends characters before and after.
# When there should be the same query chunk after the first, it stops.
# Otherwise, the second chunk might be halved and thus not highlighted.
# Ex: query "test", text: "lorem ipsum text dolor text text sit amet consectetur" ->
# ["ipsum text dolor ", "text ", "text sit amet"] (words "lorem" and "consectetur" are strip)
line_extract = [
re.compile(
"(.{0,80}" + re.escape(q) + "(?:(?!" + re.escape(q) +
").){0,80})", re.IGNORECASE) for q in query.split(" ")
if q != "b"
]
# grep some lines
keep_all = not self.plugin.preferences["preview_short"] and len(
lines) < max_lines
lines_iter = iter(lines)
chosen = [
next(lines_iter)
] # always include header as the first line, even if it does not contain the query
for line in lines_iter:
if len(
chosen
) > max_lines: # file is too long which would result the preview to not be smooth
break
elif keep_all or any(q in line.lower() for q in query.split(" ")):
# keep this line since it contains a query chunk
if len(line) > 100:
# however, this line is too long to display, try to extract query and its neighbourhood
s = "...".join("...".join(q.findall(line))
for q in line_extract).strip(".")
if not s: # no query chunk was find on this line, the keep_all is True for sure
chosen.append(line[:100] + "...")
else:
chosen.append("..." + s + "...")
else:
chosen.append(line)
if not keep_all or len(chosen) > max_lines:
# note that query might not been found, ex: query "foo" would not find line with a bold 'o': "f**o**o"
chosen.append("...")
txt = markup_escape_text("\n".join(line for line in chosen))
# bold query chunks in the text
for q in query_match:
txt = q.sub(r"<b>\g<1></b>", txt)
# preserve markup_escape_text entities
# correct ex: '&a<b>m</b>p;' -> '&' if searching for 'm'
bold_tag = re.compile("</?b>")
broken_entity = re.compile("&[a-z]*<b[^;]*;")
txt = broken_entity.sub(lambda m: bold_tag.sub("", m.group(0)), txt)
return txt
class InstantsearchMainWindowExtension(MainWindowExtension):
uimanager_xml = '''
<ui>
<menubar name='menubar'>
<menu action='tools_menu'>
<placeholder name='plugin_items'>
<menuitem action='instantsearch'/>
</placeholder>
</menu>
</menubar>
</ui>
'''
gui = ""
@action(_('_Instantsearch'), accelerator='<ctrl>e') # T: menu item
def instantsearch(self):
# init
self.cached_titles = []
# self.menu = defaultdict(_MenuItem)
self.lastQuery = "" # previous user input
self.queryO = None
self.caret = {'pos': 0, 'altPos': 0, 'text': ""} # cursor position
self.originalPage = self.window.page.name # we return here after escape
self.selection = None
if not self.plugin.preferences['isCached']:
# reset last search results
State.reset()
self.menuPage = None
self.isClosed = False
self.lastPage = None
# preferences
self.title_match_char = self.plugin.preferences['title_match_char']
self.start_search_length = self.plugin.preferences[
'start_search_length']
self.keystroke_delay = self.plugin.preferences['keystroke_delay']
self.open_when_unique = self.plugin.preferences['open_when_unique']
# building quick title cache
def build(start=""):
if hasattr(self.window.notebook, 'pages'):
o = self.window.notebook.pages
else: # for Zim 0.66-
o = self.window.notebook.index
for s in o.list_pages(Path(start or ":")):
start2 = (start + ":" if start else "") + s.basename
self.cached_titles.append((start2, start2.lower()))
build(start2)
build()
# Gtk
self.gui = Dialog(self.window,
_('Search'),
buttons=None,
defaultwindowsize=(300, -1))
self.gui.resize(300, 100) # reset size
self.inputEntry = InputEntry()
self.inputEntry.connect('key_press_event', self.move)
self.inputEntry.connect(
'changed',
self.change) # self.change is needed by GObject or something
self.gui.vbox.pack_start(self.inputEntry,
expand=False,
fill=True,
padding=0)
self.labelObject = Gtk.Label(label=(''))
self.labelObject.set_size_request(300, -1)
self.gui.vbox.pack_start(self.labelObject,
expand=False,
fill=True,
padding=0)
# gui geometry
px, py = self.window.get_position()
pw, ph = self.window.get_size()
x, y = self.gui.get_position()
if self.plugin.preferences[
'position'] == InstantsearchPlugin.POSITION_RIGHT:
self.gui.move((pw - 300), 0)
elif self.plugin.preferences[
'position'] == InstantsearchPlugin.POSITION_CENTER:
self.gui.resize(300, 100)
self.gui.move(px + (pw / 2) - 150, py + (ph / 2) - 250)
else:
raise AttributeError("Instant search: Wrong position preference.")
self.gui.show_all()
self.labelVar = ""
self.timeout = ""
self.timeoutOpenPage = None
# lastPage = ""
# pageTitleOnly = False
menu = []
# queryTime = 0
def change(self, _): # widget, event,text
if self.timeout:
GObject.source_remove(self.timeout)
q = self.inputEntry.get_text()
# print("Change. {} {}".format(input, self.lastQuery))
if q == self.lastQuery: return
if q == self.title_match_char: return
if q and q[-1] == "∀": # easter egg: debug option for zim --standalone
q = q[:-1]
import ipdb
ipdb.set_trace()
self.state = State.setCurrent(q)
if not self.state.isFinished:
self.isSubset = True if self.lastQuery and q.startswith(
self.lastQuery) else False
self.state.checkTitleSearch(self.title_match_char)
self.startSearch()
else: # search completed before
# print("Search already cached.")
self.startSearch(
) # update the results in a page has been modified meanwhile (not if something got deleted in the notebook #16 )
self.checkLast()
self.soutMenu()
self.lastQuery = q
def startSearch(self):
""" Search string has certainly changed. We search in indexed titles and/or we start zim search.
Normally, zim gives 11 points bonus if the search-string appears in the titles.
If we are ignoring subpages, the search "foo" will match only page "journal:foo",
but not "journal:foo:subpage" (and score of the parent page will get slightly higher by 1.)
However, if there are occurences of the string in the fulltext of the subpage,
subpage remains in the result, but gets bonus only 2 points (not 11).
"""
query = self.state.query
menu = self.state.menu
isInQuery = re.compile(
r"(^|:|\s|\()" + query
).search # 'te' matches this page titles: 'test' or 'Journal:test' or 'foo test' or 'foo (test)'
if self.isSubset and len(query) < self.start_search_length:
# letter(s) was/were added and full search has not yet been activated
for path in _MenuItem.titles:
if path in self.state.menu and not isInQuery(
path.lower()): # 'te' didnt match 'test' etc
del menu[path] # we pop out the result
else:
menu[path].sure = True
else: # perform new search in cached_titles
_MenuItem.titles = set()
found = 0
if self.state.firstSeen:
for path, pathLow in self.cached_titles: # quick search in titles
if isInQuery(
pathLow
): # 'te' matches 'test' or 'Journal:test' etc
_MenuItem.titles.add(path)
if query in path.lower() and query not in path.lower(
).split(":"
)[-1]: # "raz" in "raz:dva", but not in "dva"
self.state.menu[":".join(path.split(
":")[:-1])].bonus += 1 # 1 point for subpage
menu[path].bonus = -11
menu[
path].score += 10 # 10 points for title (zim default) (so that it gets displayed before search finishes)
menu[path].intitle = True
menu[path].path = path
found += 1
if found >= 10: # we dont want more than 10 results; we would easily match all of the pages
break
else:
menu[path].intitle = False
self.processMenu() # show for now results of title search
if len(query) >= self.start_search_length:
self.timeout = GObject.timeout_add(
self.keystroke_delay,
self.startZimSearch) # ideal delay between keystrokes
def startZimSearch(self):
""" Starts search for the input. """
self.timeout = ""
self.caret['altPos'] = 0 # possible position of caret - beginning
s = '"*{}*"'.format(
self.state.query
) if self.plugin.preferences['isWildcarded'] else self.state.query
self.queryO = Query(
s
) # Xunicode(s) beware when searching for unicode character. Update the row when going to Python3.
lastSel = self.selection if self.isSubset and self.state.previous.isFinished else None # it should be quicker to find the string, if we provide this subset from last time (in the case we just added a letter, so that the subset gets smaller)
self.selection = SearchSelection(self.window.notebook)
state = self.state # this is thread, so that self.state would can before search finishes
self.selection.search(self.queryO,
selection=lastSel,
callback=self._search_callback(
self.state.rawQuery))
state.isFinished = True
for item in list(
state.menu
): # remove all the items that we didnt encounter during the search
if not state.menu[item].sure:
del state.menu[item]
if state == self.state:
self.checkLast()
self.processMenu(state=state)
def checkLast(self):
""" opens the page if there is only one option in the menu """
if self.open_when_unique and len(self.state.menu) == 1:
self._open_page(Path(self.state.menu.keys()[0]),
excludeFromHistory=False)
self.close()
def _search_callback(self, query):
def _search_callback(results, path):
if results is not None:
self._update_results(
results, State.get(query)
) # we finish the search even if another search is running. If we returned False, the search would be cancelled-
while Gtk.events_pending():
Gtk.main_iteration_do(blocking=False)
return True
return _search_callback
def _update_results(self, results, state):
"""
This method may run many times, due to the _update_results, which are updated many times.
I may set that _update_results would run only once, but this is nice - the results are appearing one by one.
"""
changed = False
state.lastResults = results
for option in results.scores:
if state.pageTitleOnly and state.query not in option.name: # hledame jen v nazvu stranky
continue
if option.name not in state.menu: # new item found
if state == self.state and option.name == self.caret[
'text']: # this is current search
self.caret['altPos'] = len(
state.menu
) - 1 # karet byl na tehle pozici, pokud se zuzil vyber, budeme vedet, kam karet opravne umistit
if option.name not in state.menu or (
state.menu[option.name].bonus < 0
and state.menu[option.name].score == 0):
changed = True
if not state.menu[option.name].sure:
state.menu[option.name].sure = True
changed = True
state.menu[option.name].score = results.scores[
option] # zaradit mezi moznosti
if changed: # we added a page
self.processMenu(state=state, sort=False)
else:
pass
def processMenu(self, state=None, sort=True):
""" Sort menu and generate items and sout menu. """
if state is None:
state = self.state
if sort:
state.items = sorted(
state.menu,
reverse=True,
key=lambda item:
(state.menu[item].intitle, state.menu[item].score + state.menu[
item].bonus, -item.count(":"), item))
else: # when search results are being updated, it's good when the order doesnt change all the time. So that the first result does not become for a while 10th and then become first back.
state.items = sorted(
state.menu,
reverse=True,
key=lambda item:
(state.menu[item].intitle, -state.menu[item].lastOrder))
state.items = [
item for item in state.items
if (state.menu[item].score + state.menu[item].bonus) > 0
] # i dont know why there are items with 0 score
if state == self.state:
self.soutMenu()
def soutMenu(self, displayImmediately=False):
""" Displays menu and handles caret position. """
if self.timeoutOpenPage:
GObject.source_remove(self.timeoutOpenPage)
self.gui.resize(300, 100) # reset size
# treat possible caret deflection
if self.caret['pos'] < 0 or self.caret['pos'] > len(
self.state.items
) - 1: # place the caret to the beginning or the end of list
self.caret['pos'] = self.caret['altPos']
text = ""
i = 0
for item in self.state.items:
score = self.state.menu[item].score + self.state.menu[item].bonus
self.state.menu[item].lastOrder = i
if i == self.caret['pos']:
self.caret['text'] = item # caret is at this position
text += '→ {} ({}) {}\n'.format(
item, score, "" if self.state.menu[item].sure else "?")
else:
try:
text += '{} ({}) {}\n'.format(
item, score, "" if self.state.menu[item].sure else "?")
except:
text += "CHYBA\n"
text += item[0:-1] + "\n"
i += 1
self.labelObject.set_text(text)
self.menuPage = Path(
self.caret['text'] if len(self.state.items) else self.originalPage)
if not displayImmediately:
self.timeoutOpenPage = GObject.timeout_add(
self.keystroke_delay, self._open_page,
self.menuPage) # ideal delay between keystrokes
else:
self._open_page(self.menuPage)
def move(self, widget, event):
""" Move caret up and down. Enter to confirm, Esc closes search."""
keyname = Gdk.keyval_name(event.keyval)
if keyname == "Up" or keyname == "ISO_Left_Tab":
self.caret['pos'] -= 1
self.soutMenu(displayImmediately=False)
if keyname == "Down" or keyname == "Tab":
self.caret['pos'] += 1
self.soutMenu(displayImmediately=False)
if keyname == "KP_Enter" or keyname == "Return":
self._open_page(self.menuPage, excludeFromHistory=False)
self.close()
if keyname == "Escape":
self._open_original()
# GTK closes the windows itself on Escape, no self.close() needed
return
## Safely closes
# Xwhen closing directly, Python gave allocation error
def close(self):
if not self.isClosed:
self.isClosed = True
self.gui.emit("close")
def _open_original(self):
self._open_page(Path(self.originalPage))
def _open_page(self, page, excludeFromHistory=True):
""" Open page and highlight matches """
self.timeoutOpenPage = None # no delayed page will be open
if self.isClosed == True:
return
if page and page.name and page.name != self.lastPage:
self.lastPage = page.name
# print("*** HISTORY BEF", self.window.ui.history._history[-3:])
self.window.open_page(page)
if excludeFromHistory:
# there is no public API, so lets use protected _history instead
self.window.history._history.pop()
self.window.history._current = len(
self.window.history._history) - 1
if not excludeFromHistory and self.window.history.get_current(
).name is not page.name:
# we insert the page to the history because it was likely to be just visited and excluded
self.window.history.append(page)
# Popup find dialog with same query
if self.queryO: # and self.queryO.simple_match:
string = self.state.query
string = string.strip('*') # support partial matches
if self.plugin.preferences['highlight_search']:
self.window.pageview.show_find(string, highlight=True)
