search.py

# ##### BEGIN GPL LICENSE BLOCK #####
#
#  This program is free software; you can redistribute it and/or
#  modify it under the terms of the GNU General Public License
#  as published by the Free Software Foundation; either version 2
#  of the License, or (at your option) any later version.
#
#  This program is distributed in the hope that it will be useful,
#  but WITHOUT ANY WARRANTY; without even the implied warranty of
#  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
#  GNU General Public License for more details.
#
#  You should have received a copy of the GNU General Public License
#  along with this program; if not, write to the Free Software Foundation,
#  Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
#
# ##### END GPL LICENSE BLOCK #####

if "bpy" in locals():
    import imp

    imp.reload(paths)
    imp.reload(utils)
else:
    from blenderkit import paths, utils, categories, ui, version_checker
import blenderkit
from bpy.app.handlers import persistent

from bpy.props import (  # TODO only keep the ones actually used when cleaning
    IntProperty,
    FloatProperty,
    FloatVectorProperty,
    StringProperty,
    EnumProperty,
    BoolProperty,
    PointerProperty,
)
from bpy.types import (
    Operator,
    Panel,
    AddonPreferences,
    PropertyGroup,
    UIList
)

import requests, os, random
import time
import threading
import tempfile
import json
import bpy

search_start_time = 0
prev_time = 0


def check_errors(rdata):
    if rdata.get('statusCode') == 401:
        if rdata.get('detail') == 'Invalid token.':
            return False, 'Missing or wrong api_key in addon preferences'
    return True, ''


search_threads = []
thumb_sml_download_threads = {}
thumb_full_download_threads = {}
reports = ''


@persistent
def scene_load(context):
    wm = bpy.context.window_manager
    fetch_server_data()
    # following doesn't necessarilly happen if version isn't checked yet or similar, first run.
    wm['bkit_update'] = version_checker.compare_versions(blenderkit)
    utils.load_categories()


def fetch_server_data():
    ''' download categories and addon version'''
    user_preferences = bpy.context.preferences.addons['blenderkit'].preferences
    url = paths.BLENDERKIT_ADDON_URL
    api_key = user_preferences.api_key
    version_checker.check_version_thread(url, api_key, blenderkit)
    categories.fetch_categories_thread(api_key)


@bpy.app.handlers.persistent
def timer_update():  # TODO might get moved to handle all blenderkit stuff.

    global search_threads
    # don't do anything while dragging - this could switch asset type during drag, and make results list lenght different,
    # causing a lot of throuble literally.
    if len(search_threads) == 0 or bpy.context.scene.blenderkitUI.dragging:
        return 1
    for thread in search_threads:  # TODO this doesn't check all processess when removal... mostly 1 process will be running however.

        if not thread[0].is_alive():
            search_threads.remove(thread)  #
            icons_dir = thread[1]
            scene = bpy.context.scene
            # these 2 lines should update the previews enum and set the first result as active.
            s = bpy.context.scene
            asset_type = thread[2]
            if asset_type == 'model':
                props = scene.blenderkit_models
                json_filepath = os.path.join(icons_dir, 'model_searchresult.json')
                search_name = 'bkit model search'
            if asset_type == 'scene':
                props = scene.blenderkit_scene
                json_filepath = os.path.join(icons_dir, 'scene_searchresult.json')
                search_name = 'bkit scene search'
            if asset_type == 'material':
                props = scene.blenderkit_mat
                json_filepath = os.path.join(icons_dir, 'material_searchresult.json')
                search_name = 'bkit material search'
            if asset_type == 'brush':
                props = scene.blenderkit_brush
                json_filepath = os.path.join(icons_dir, 'brush_searchresult.json')
                search_name = 'bkit brush search'

            s[search_name] = []

            global reports
            if reports != '':
                props.report = str(reports)
                return .2
            with open(json_filepath, 'r') as data_file:
                rdata = json.load(data_file)

            result_field = []
            ok, error = check_errors(rdata)
            if ok:

                for r in rdata['results']:

                    if r['assetType'] == asset_type:
                        # print(r)
                        if len(r['files']) > 0:
                            furl = None
                            tname = None
                            allthumbs = []
                            durl, tname = None, None
                            for f in r['files']:
                                if f['fileType'] == 'thumbnail':
                                    tname = paths.extract_filename_from_url(f['fileThumbnailLarge'])
                                    small_tname = paths.extract_filename_from_url(f['fileThumbnail'])
                                    allthumbs.append(tname)  # TODO just first thumb is used now.

                                tdict = {}
                                for i, t in enumerate(allthumbs):
                                    tdict['thumbnail_%i'] = t
                                if f['fileType'] == 'blend':
                                    durl = f['downloadUrl'].split('?')[0]
                                    # fname = paths.extract_filename_from_url(f['filePath'])
                            if durl and tname:

                                tooltip = generate_tooltip(r)
                                # utils.pprint(print(r))
                                asset_data = {'thumbnail': tname,
                                              'thumbnail_small': small_tname,
                                              # 'thumbnails':allthumbs,
                                              'download_url': durl,
                                              'id': r['id'],
                                              'asset_base_id': r['assetBaseId'],
                                              'name': r['name'],
                                              'asset_type': r['assetType'],
                                              'tooltip': tooltip,
                                              'tags': r['tags'],
                                              'can_download': r.get('canDownload', True),
                                              'verification_status': r['verificationStatus']
                                              # 'description': r['description'],
                                              # 'author': r['description'],
                                              }
                                asset_data['downloaded'] = 0

                                # parse extra params needed for blender here
                                params = params_to_dict(r['parameters'])

                                if asset_type == 'model':
                                    if params.get('boundBoxMinX') != None:
                                        bbox = {
                                            'bbox_min': (
                                                float(params['boundBoxMinX']),
                                                float(params['boundBoxMinY']),
                                                float(params['boundBoxMinZ'])),
                                            'bbox_max': (
                                                float(params['boundBoxMaxX']),
                                                float(params['boundBoxMaxY']),
                                                float(params['boundBoxMaxZ']))
                                        }

                                    else:
                                        bbox = {
                                            'bbox_min': (-.5, -.5, 0),
                                            'bbox_max': (.5, .5, 1)
                                        }
                                    asset_data.update(bbox)
                                if asset_type == 'material':
                                    asset_data['texture_size_meters'] = params.get('textureSizeMeters', 1.0)

                                asset_data.update(tdict)
                                if r['assetBaseId'] in scene.get('assets used', {}).keys():
                                    asset_data['downloaded'] = 100

                                result_field.append(asset_data)

                                # results = rdata['results']
                s[search_name] = result_field
                s['search results'] = result_field
                s['search results orig'] = rdata
                load_previews()
                ui_props = bpy.context.scene.blenderkitUI
                if len(result_field) < ui_props.scrolloffset:
                    ui_props.scrolloffset = 0
                props.is_searching = False
                props.search_error = False
                props.report = 'Open assetbar to see %i results. ' % len(s['search results'])
                if len(s['search results']) == 0:
                    props.report = 'No matching results found.'

            # (rdata['next'])
            # if rdata['next'] != None:
            # search(False, get_next = True)
            else:
                print('error', error)
                props.report = error
                props.search_error = True

            print('finished search thread')
            mt('preview loading finished')
    return .2


def load_previews():
    mappingdict = {
        'MODEL': 'model',
        'SCENE': 'scene',
        'MATERIAL': 'material',
        'TEXTURE': 'texture',
        'BRUSH': 'brush'
    }
    scene = bpy.context.scene
    # FIRST START SEARCH
    props = scene.blenderkitUI

    directory = paths.get_temp_dir('%s_search' % mappingdict[props.asset_type])
    s = bpy.context.scene
    results = s.get('search results')
    #
    if results is not None:
        inames = []
        tpaths = []

        i = 0
        for r in results:

            tpath = os.path.join(directory, r['thumbnail_small'])

            iname = utils.previmg_name(i)

            if os.path.exists(tpath):  # sometimes we are unlucky...
                img = bpy.data.images.get(iname)
                if img is None:
                    img = bpy.data.images.load(tpath)
                    img.name = iname
                elif img.filepath != tpath:
                    # had to add this check for autopacking files...
                    if img.packed_file is not None:
                        img.unpack(method='USE_ORIGINAL')
                    img.filepath = tpath
                    img.reload()

            i += 1
    print('previews loaded')


#  line splitting for longer texts...
def split_subs(text):
    if text == '':
        return []
    threshold = 40
    text = text.rstrip()
    lines = []
    while len(text) > threshold:
        i = text.rfind(' ', 0, threshold)
        i1 = text.rfind(',', 0, threshold)
        i = max(i, i1)
        if i == -1:
            i = threshold
        lines.append(text[:i])
        text = text[i:]
    lines.append(text)
    return lines


def list_to_str(input):
    output = ''
    for i, text in enumerate(input):
        output += text
        if i < len(input) - 1:
            output += ', '
    return output


def writeblock(t, input):  # for longer texts
    dlines = split_subs(input)
    for i, l in enumerate(dlines):
        t += '%s\n' % l
    return t


def writeblockm(tooltip, mdata, key='', pretext=None):  # for longer texts
    if mdata.get(key) == None:
        return tooltip
    else:
        intext = mdata[key]
        if type(intext) == list:
            intext = list_to_str(intext)
        intext = str(intext)
        if intext.rstrip() == '':
            return tooltip
        if pretext == None:
            pretext = key
        if pretext != '':
            pretext = pretext + ': '
        text = pretext + intext
        dlines = split_subs(text)
        for i, l in enumerate(dlines):
            tooltip += '%s\n' % l

    return tooltip


def fmt_length(prop):
    prop = str(round(prop, 2)) + 'm'
    return prop


def has(mdata, prop):
    if mdata.get(prop) is not None and mdata[prop] is not None and mdata[prop] is not False:
        return True
    else:
        return False


def params_to_dict(params):
    params_dict = {}
    for p in params:
        params_dict[p['parameterType']] = p['value']
    return params_dict


def generate_tooltip(mdata):
    if type(mdata['parameters']) == list:
        mparams = params_to_dict(mdata['parameters'])
    else:
        mparams = mdata['parameters']
    t = ''
    t = writeblock(t, mdata['name'])

    t = writeblockm(t, mdata, key='description', pretext='')
    if mdata['description'] != '':
        t += '\n'

    bools = (('rig', None), ('animated', None), ('manifold', 'non-manifold'), ('scene', None), ('simulation', None),
             ('uv', None))
    for b in bools:
        if mparams.get(b[0]):
            mdata['tags'].append(b[0])
        elif b[1] != None:
            mdata['tags'].append(b[1])

    bools_data = ('adult',)
    for b in bools_data:
        if mdata.get(b) and mdata[b]:
            mdata['tags'].append(b)
    t = writeblockm(t, mparams, key='designer', pretext='designer')
    t = writeblockm(t, mparams, key='manufacturer', pretext='manufacturer')
    t = writeblockm(t, mparams, key='designCollection', pretext='design collection')

    # t = writeblockm(t, mparams, key='engines', pretext='engine')
    # t = writeblockm(t, mparams, key='model_style', pretext='style')
    # t = writeblockm(t, mparams, key='material_style', pretext='style')
    # t = writeblockm(t, mdata, key='tags')
    # t = writeblockm(t, mparams, key='condition', pretext='condition')
    # t = writeblockm(t, mparams, key='productionLevel', pretext='production level')
    if has(mdata, 'purePbr'):
        t = writeblockm(t, mparams, key='pbrType', pretext='pbr')

    t = writeblockm(t, mparams, key='designYear', pretext='design year')

    if has(mparams, 'dimensionX'):
        t += 'size: %s, %s, %s\n' % (fmt_length(mparams['dimensionX']),
                                     fmt_length(mparams['dimensionY']),
                                     fmt_length(mparams['dimensionZ']))
    if has(mparams, 'faceCount'):
        t += 'face count: %s, render: %s\n' % (mparams['faceCount'], mparams['faceCountRender'])

    # t = writeblockm(t, mparams, key='meshPolyType', pretext='mesh type')
    # t = writeblockm(t, mparams, key='objectCount', pretext='nubmber of objects')

    # t = writeblockm(t, mparams, key='materials')
    # t = writeblockm(t, mparams, key='modifiers')
    # t = writeblockm(t, mparams, key='shaders')

    if has(mparams, 'textureSizeMeters'):
        t = writeblockm(t, mparams, key='textureSizeMeters', pretext='texture size in meters')

    if has(mparams, 'textureResolutionMax') and mparams['textureResolutionMax'] > 0:
        if mparams['textureResolutionMin'] == mparams['textureResolutionMax']:
            t = writeblockm(t, mparams, key='textureResolutionMin', pretext='texture resolution')
        else:
            t += 'tex resolution: %i - %i\n' % (mparams['textureResolutionMin'], mparams['textureResolutionMax'])

    if has(mparams, 'thumbnailScale'):
        t = writeblockm(t, mparams, key='thumbnailScale', pretext='preview scale')

    # t += 'uv: %s\n' % mdata['uv']
    t += '\n'
    # t = writeblockm(t, mdata, key='license')

    # generator is for both upload preview and search, this is only after search
    if mdata.get('versionNumber'):
        # t = writeblockm(t, mdata, key='versionNumber', pretext='version')

        t += 'author: %s %s\n' % (mdata['author']['firstName'], mdata['author']['lastName'])
        # t += '\n'

    at = mdata['assetType']
    t += '\n'
    if at == 'brush' or at == 'texture':
        t += 'click to link %s' % mdata['assetType']
    if at == 'model' or at == 'material':
        t += 'click or drag in scene to link/append %s' % mdata['assetType']

    return t


def get_items_models(self, context):
    global search_items_models
    return search_items_models


def get_items_brushes(self, context):
    global search_items_brushes
    return search_items_brushes


def get_items_materials(self, context):
    global search_items_materials
    return search_items_materials


def get_items_textures(self, context):
    global search_items_textures
    return search_items_textures


class ThumbDownloader(threading.Thread):
    query = None

    def __init__(self, url, path):
        super(ThumbDownloader, self).__init__()
        self.url = url
        self.path = path
        self._stop_event = threading.Event()

    def stop(self):
        self._stop_event.set()

    def stopped(self):
        return self._stop_event.is_set()

    def run(self):
        r = requests.get(self.url, stream=False)
        if r.status_code == 200:
            with open(self.path, 'wb') as f:
                f.write(r.content)
            # ORIGINALLY WE DOWNLOADED THUMBNAILS AS STREAM, BUT THIS WAS TOO SLOW.
            # with open(path, 'wb') as f:
            #     for chunk in r.iter_content(1048576*4):
            #         f.write(chunk)


class Searcher(threading.Thread):
    query = None

    def __init__(self, query, params):
        super(Searcher, self).__init__()
        self.query = query
        self.params = params
        self._stop_event = threading.Event()

    def stop(self):
        self._stop_event.set()

    def stopped(self):
        return self._stop_event.is_set()

    def run(self):
        maxthreads = 300
        maximages = 50
        query = self.query
        params = self.params
        global reports

        t = time.time()
        mt('search thread started')
        tempdir = paths.get_temp_dir('%s_search' % query['asset_type'])
        json_filepath = os.path.join(tempdir, '%s_searchresult.json' % query['asset_type'])

        if query['token'] != '':
            headers = {
                "accept": "application/json",
                "Authorization": "Bearer %s" % query['token'],
                # "Content-Type": "application/json",
            }
        else:
            headers = {
                "accept": "application/json",
                # "Content-Type": "application/json",
            }
        rdata = {}
        rdata['results'] = []

        if params['get_next']:
            with open(json_filepath, 'r') as infile:
                try:
                    origdata = json.load(infile)
                    urlquery = origdata['next']
                    if urlquery == None:
                        return;
                except:
                    # in case no search results found on drive we don't do next page loading.
                    params['get_next'] = False
        if not params['get_next']:
            # build a new request
            url = paths.get_bkit_url() + 'search/'

            nquery = {
                # 'tags': query['keywords'],
                'asset_type': query['asset_type'],
            }
            if query.get('category'):
                nquery['category_subtree'] = query['category']

            # build request manually
            # TODO use real queries
            requeststring = '?query=' + query['keywords'].lower() + '+'
            #
            for i, q in enumerate(nquery):
                requeststring += q + ':' + str(nquery[q])
                if i < len(nquery) - 1:
                    requeststring += '+'

            requeststring += '&addon_version=%s' % params['addon_version']
            if params.get('scene_uuid') is not None:
                requeststring += '&scene_uuid=%s' % params['scene_uuid']

            urlquery = url + requeststring

        try:
            # print(urlquery)
            r = requests.get(urlquery, headers=headers)
            reports = ''
            # print(r.text)
        except requests.exceptions.RequestException as e:
            print(e)
            reports = e
            # props.report = e
            return
        mt('response is back ')
        try:
            rdata = r.json()
        except Exception as inst:
            reports = r.text
            print(inst)

        mt('data parsed ')

        # filter results here:
        # todo remove this in future
        nresults = []
        for d in rdata.get('results', []):
            # TODO this code is for filtering brush types, should vanish after we implement filter in Elastic
            mode = None
            if query['asset_type'] == 'brush':
                for p in d['parameters']:
                    if p['parameterType'] == 'mode':
                        mode = p['value']
            if query['asset_type'] != 'brush' or (
                    query.get('brushType') != None and query['brushType']) == mode:
                nresults.append(d)
        rdata['results'] = nresults

        print('number of results: ', len(rdata.get('results', [])))
        if self.stopped():
            print('stopping search : ' + query['keywords'])
            return

        mt('search finished')
        i = 0

        thumb_small_urls = []
        thumb_small_filepaths = []
        thumb_full_urls = []
        thumb_full_filepaths = []
        # END OF PARSING
        for d in rdata.get('results', []):
            for f in d['files']:
                # TODO move validation of published assets to server, too manmy checks here.
                if f['fileType'] == 'thumbnail' and f['fileThumbnail'] != None and f['fileThumbnailLarge'] != None:
                    if f['fileThumbnail'] == None:
                        f['fileThumbnail'] = 'NONE'
                    if f['fileThumbnailLarge'] == None:
                        f['fileThumbnailLarge'] = 'NONE'

                    thumb_small_urls.append(f['fileThumbnail'])
                    thumb_full_urls.append(f['fileThumbnailLarge'])

                    imgname = paths.extract_filename_from_url(f['fileThumbnail'])
                    imgpath = os.path.join(tempdir, imgname)
                    thumb_small_filepaths.append(imgpath)

                    imgname = paths.extract_filename_from_url(f['fileThumbnailLarge'])
                    imgpath = os.path.join(tempdir, imgname)
                    thumb_full_filepaths.append(imgpath)

        sml_thbs = zip(thumb_small_filepaths, thumb_small_urls)
        full_thbs = zip(thumb_full_filepaths, thumb_full_urls)

        # we save here because a missing thumbnail check is in the previous loop
        # we can also prepend previous results. These have allready thumbnails downloaded...
        if params['get_next']:
            rdata['results'][0:0] = origdata['results']

        with open(json_filepath, 'w') as outfile:
            json.dump(rdata, outfile)

        killthreads_sml = []
        for k in thumb_sml_download_threads.keys():
            if k not in thumb_small_filepaths:
                killthreads_sml.append(k)  # do actual killing here?

        killthreads_full = []
        for k in thumb_full_download_threads.keys():
            if k not in thumb_full_filepaths:
                killthreads_full.append(k)  # do actual killing here?
        # TODO do the killing/ stopping here! remember threads might have finished inbetween!

        if self.stopped():
            print('stopping search : ' + query['keywords'])
            return

        # this loop handles downloading of small thumbnails
        for imgpath, url in sml_thbs:
            if imgpath not in thumb_sml_download_threads and not os.path.exists(imgpath):
                thread = ThumbDownloader(url, imgpath)
                # thread = threading.Thread(target=download_thumbnail, args=([url, imgpath]),
                #                           daemon=True)
                thread.start()
                thumb_sml_download_threads[imgpath] = thread
                # threads.append(thread)

                if len(thumb_sml_download_threads) > maxthreads:
                    while len(thumb_sml_download_threads) > maxthreads:
                        threads_copy = thumb_sml_download_threads.copy()  # because for loop can erase some of the items.
                        for tk, thread in threads_copy.items():
                            if not thread.is_alive():
                                thread.join()
                                # print(x)
                                del (thumb_sml_download_threads[tk])
                                print('fetched thumbnail ', i)
                                i += 1
        if self.stopped():
            print('stopping search : ' + query['keywords'])
            return
        idx = 0
        while len(thumb_sml_download_threads) > 0:
            threads_copy = thumb_sml_download_threads.copy()  # because for loop can erase some of the items.
            for tk, thread in threads_copy.items():
                if not thread.is_alive():
                    thread.join()
                    del (thumb_sml_download_threads[tk])
                    i += 1

        if self.stopped():
            print('stopping search : ' + query['keywords'])
            return

        # start downloading full thumbs in the end
        for imgpath, url in full_thbs:
            if imgpath not in thumb_full_download_threads and not os.path.exists(imgpath):
                thread = ThumbDownloader(url, imgpath)
                # thread = threading.Thread(target=download_thumbnail, args=([url, imgpath]),
                #                           daemon=True)
                thread.start()
                thumb_full_download_threads[imgpath] = thread
        mt('thumbnails finished')


def build_query_common(query, props):
    user_preferences = bpy.context.preferences.addons['blenderkit'].preferences
    query_common = {
        "token": user_preferences.api_key,
        "keywords": props.search_keywords
    }
    query.update(query_common)


# def query_add_range(query, name, rmin, rmax):

def build_query_model():
    '''use all search input to request results from server'''

    props = bpy.context.scene.blenderkit_models
    query = {
        "asset_type": 'model',
        "engine": props.search_engine,
        "adult": props.search_adult,
    }
    if props.search_style != 'ANY':
        if props.search_style != 'OTHER':
            query["style"] = props.search_style
        else:
            query["style"] = props.search_style_other
    if props.search_advanced:
        if props.search_condition != 'UNSPECIFIED':
            query["condition"] = props.search_condition
        if props.search_design_year:
            query["designYearMin"] = props.search_design_year_min
            query["designYearMax"] = props.search_design_year_max
        if props.search_polycount:
            query["polyCountMin"] = props.search_polycount_min
            query["polyCountMax"] = props.search_polycount_max
        if props.search_texture_resolution:
            query["textureResolutionMin"] = props.search_texture_resolution_min
            query["textureResolutionMax"] = props.search_texture_resolution_max

    build_query_common(query, props)
    # query = {
    #     "query": {
    #         "exists": {"field": "faceCount"},
    #
    #         "range": {
    #
    #             "faceCount": {
    #                 "gte": query["polyCountMin"],
    #                 "lte": query["polyCountMax"],
    #                 "boost": 2.0
    #             }
    #
    #         },
    #
    #          "match": {
    #              "asset_type": 'model',
    #          }
    #
    #     }
    # }

    return query


def build_query_scene():
    '''use all search input to request results from server'''

    props = bpy.context.scene.blenderkit_scene
    query = {
        "asset_type": 'scene',
        "engine": props.search_engine,
        # "adult": props.search_adult,
    }
    build_query_common(query, props)
    return query


def build_query_material():
    props = bpy.context.scene.blenderkit_mat
    query = {
        "asset_type": 'material',

    }
    if props.search_engine == 'NONE':
        query["engine"] = ''
    if props.search_engine != 'OTHER':
        query["engine"] = props.search_engine
    else:
        query["engine"] = props.search_engine_other
    if props.search_style != 'ANY':
        if props.search_style != 'OTHER':
            query["style"] = props.search_style
        else:
            query["style"] = props.search_style_other
    build_query_common(query, props)

    return query


def build_query_texture():
    props = bpy.context.scene.blenderkit_tex
    query = {
        "asset_type": 'texture',

    }

    if props.search_style != 'ANY':
        if props.search_style != 'OTHER':
            query["search_style"] = props.search_style
        else:
            query["search_style"] = props.search_style_other

    build_query_common(query, props)

    return query


def build_query_brush():
    props = bpy.context.scene.blenderkit_brush

    brush_type = ''
    if bpy.context.sculpt_object is not None:
        brush_type = 'sculpt'

    elif bpy.context.image_paint_object:  # could be just else, but for future p
        brush_type = 'texture_paint'

    query = {
        "asset_type": 'brush',

        "brushType": brush_type
    }

    build_query_common(query, props)

    return query


def mt(text):
    global search_start_time, prev_time
    alltime = time.time() - search_start_time
    since_last = time.time() - prev_time
    prev_time = time.time()
    print(text, alltime, since_last)


def add_search_process(query, params):
    global search_threads

    while (len(search_threads) > 0):
        old_thread = search_threads.pop(0)
        old_thread[0].stop()
        # TODO CARE HERE FOR ALSO KILLING THE THREADS...AT LEAST NOW SEARCH DONE FIRST WON'T REWRITE AN OLDER ONE

    tempdir = paths.get_temp_dir('%s_search' % query['asset_type'])
    thread = Searcher(query, params)
    # thread = threading.Thread(target=Searcher, args=([query]), daemon=True)
    thread.start()

    search_threads.append([thread, tempdir, query['asset_type']])

    mt('thread started')


def search(own=False, category='', get_next=False, free_only=False):
    ''' initialize searching'''
    global search_start_time

    search_start_time = time.time()
    mt('start')
    scene = bpy.context.scene
    uiprops = scene.blenderkitUI

    if uiprops.asset_type == 'MODEL':
        if not hasattr(scene, 'blenderkit'):
            return;
        props = scene.blenderkit_models
        query = build_query_model()

    if uiprops.asset_type == 'SCENE':
        if not hasattr(scene, 'blenderkit_scene'):
            return;
        props = scene.blenderkit_scene
        query = build_query_scene()

    if uiprops.asset_type == 'MATERIAL':
        if not hasattr(scene, 'blenderkit_mat'):
            return;
        props = scene.blenderkit_mat
        query = build_query_material()

    if uiprops.asset_type == 'TEXTURE':
        if not hasattr(scene, 'blenderkit_tex'):
            return;
        # props = scene.blenderkit_tex
        # query = build_query_texture()

    if uiprops.asset_type == 'BRUSH':
        if not hasattr(scene, 'blenderkit_brush'):
            return;
        props = scene.blenderkit_brush
        query = build_query_brush()

    if props.is_searching and get_next == True:
        return;
    query['own'] = own
    if category != '':
        query['category'] = category

    print('searching')
    props.is_searching = True

    params = {
        'scene_uuid': bpy.context.scene.get('uuid', None),
        'addon_version': version_checker.get_addon_version(),
        'get_next': get_next
    }

    if free_only:
        query['keywords'] += '+is_free:true'

    add_search_process(query, params)
    props.report = 'BlenderKit searching....'


def search_update(self, context):
    if self.search_keywords != '':
        search()


class SearchOperator(Operator):
    """Tooltip"""
    bl_idname = "view3d.blenderkit_search"
    bl_label = "BlenderKit asset search"

    own: BoolProperty(name="own assets only",
                      description="Find all own assets",
                      default=False)
    category: StringProperty(
        name="category",
        description="search only subtree of this category",
        default="")

    get_next: BoolProperty(name="next page",
                           description="get next page from previous search",
                           default=False)

    @classmethod
    def poll(cls, context):
        return True

    def execute(self, context):
        search(own=self.own, category=self.category, get_next=self.get_next)
        bpy.ops.view3d.blenderkit_asset_bar()

        return {'FINISHED'}


classes = [
    SearchOperator
]


def register_search():
    bpy.app.handlers.load_post.append(scene_load)

    for c in classes:
        bpy.utils.register_class(c)

    # bpy.app.timers.register(timer_update, persistent = True)

    utils.load_categories()


def unregister_search():
    bpy.app.handlers.load_post.remove(scene_load)

    for c in classes:
        bpy.utils.unregister_class(c)

    # bpy.app.timers.unregister(timer_update)


'''
search - 
build query
START THREAD
send query (bg allready)
get result - metadata, small thumbnails, big thumbnails paths (now genereate this?)
write metadata, possibly to 
download small thumbnails first
start big thumbnails download. these don't have to be there on updates, if they aren't the Image in image editor doesn't get updated.
parse metadata, save it in json in the temp dir which gets read on each update of the search.
END THREAD
when download is triggered, get also this metadata from json. E
pass this single - item metadata in the download functions, threads.
'''