From d923b2a420b77a3a29375516ed9929beeda0924a Mon Sep 17 00:00:00 2001 From: Thomas Pointhuber Date: Mon, 1 Jun 2015 09:15:50 +0200 Subject: [enh] implement swisscows engine, FIX #347 TODO: implement image engine --- searx/engines/swisscows.py | 86 ++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 86 insertions(+) create mode 100644 searx/engines/swisscows.py (limited to 'searx/engines') diff --git a/searx/engines/swisscows.py b/searx/engines/swisscows.py new file mode 100644 index 000000000..fe7a6ce0b --- /dev/null +++ b/searx/engines/swisscows.py @@ -0,0 +1,86 @@ +""" + Swisscows (Web) + + @website https://swisscows.ch + @provide-api no + + @using-api no + @results HTML (using search portal) + @stable no (HTML can change) + @parse url, title, content +""" + +from json import loads +from urllib import urlencode, unquote +import re + +# engine dependent config +categories = ['general'] +paging = True +language_support = True + +# search-url +base_url = 'https://swisscows.ch/' +search_string = '?{query}&page={page}' + +# regex +regex_json = re.compile('initialData: {"Request":(.|\n)*}\]},\s*environment') +regex_json_remove_start = re.compile('^initialData:\s*') +regex_json_remove_end = re.compile(',\s*environment$') +regex_img_url_remove_start = re.compile('^https?://i\.swisscows\.ch/\?link=') + + +# do search-request +def request(query, params): + if params['language'] == 'all': + ui_language = 'browser' + region = 'browser' + else: + region = params['language'].replace('_', '-') + ui_language = params['language'].split('_')[0] + + search_path = search_string.format( + query=urlencode({'query': query, + 'uiLanguage': ui_language, + 'region': region}), + page=params['pageno']) + + params['url'] = base_url + search_path + + return params + + +# get response from search-request +def response(resp): + results = [] + + json_regex = regex_json.search(resp.content) + + # check if results are returned + if not json_regex: + return [] + + json_raw = regex_json_remove_end.sub('', regex_json_remove_start.sub('', json_regex.group())) + json = loads(json_raw) + + # parse normal results + for result in json['Results'].get('items', []): + # append result + results.append({'url': result['Url'].replace(u'\uE000', '').replace(u'\uE001', ''), + 'title': result['Title'].replace(u'\uE000', '').replace(u'\uE001', ''), + 'content': result['Description'].replace(u'\uE000', '').replace(u'\uE001', '')}) + + # parse images + for result in json.get('Images', []): + # decode image url + img_url = unquote(regex_img_url_remove_start.sub('', result['Url'])) + + # append result + results.append({'url': result['SourceUrl'], + 'title': result['Title'], + 'content': '', + 'img_src': img_url, + 'template': 'images.html'}) + + # return results + return results -- cgit v1.2.3 From 2d81079384706aab4c39efcc8a2c4267af4397a5 Mon Sep 17 00:00:00 2001 From: Thomas Pointhuber Date: Mon, 1 Jun 2015 12:30:07 +0200 Subject: [enh] implement image support for swisscows engine --- searx/engines/swisscows.py | 38 ++++++++++++++++++++++++++++++-------- 1 file changed, 30 insertions(+), 8 deletions(-) (limited to 'searx/engines') diff --git a/searx/engines/swisscows.py b/searx/engines/swisscows.py index fe7a6ce0b..2d31264ca 100644 --- a/searx/engines/swisscows.py +++ b/searx/engines/swisscows.py @@ -1,5 +1,5 @@ """ - Swisscows (Web) + Swisscows (Web, Images) @website https://swisscows.ch @provide-api no @@ -15,7 +15,7 @@ from urllib import urlencode, unquote import re # engine dependent config -categories = ['general'] +categories = ['general', 'images'] paging = True language_support = True @@ -24,7 +24,7 @@ base_url = 'https://swisscows.ch/' search_string = '?{query}&page={page}' # regex -regex_json = re.compile('initialData: {"Request":(.|\n)*}\]},\s*environment') +regex_json = re.compile('initialData: {"Request":(.|\n)*},\s*environment') regex_json_remove_start = re.compile('^initialData:\s*') regex_json_remove_end = re.compile(',\s*environment$') regex_img_url_remove_start = re.compile('^https?://i\.swisscows\.ch/\?link=') @@ -45,6 +45,10 @@ def request(query, params): 'region': region}), page=params['pageno']) + # image search query is something like 'image?{query}&page={page}' + if params['category'] == 'images': + search_path = 'image' + search_path + params['url'] = base_url + search_path return params @@ -63,12 +67,30 @@ def response(resp): json_raw = regex_json_remove_end.sub('', regex_json_remove_start.sub('', json_regex.group())) json = loads(json_raw) - # parse normal results + # parse results for result in json['Results'].get('items', []): - # append result - results.append({'url': result['Url'].replace(u'\uE000', '').replace(u'\uE001', ''), - 'title': result['Title'].replace(u'\uE000', '').replace(u'\uE001', ''), - 'content': result['Description'].replace(u'\uE000', '').replace(u'\uE001', '')}) + result_title = result['Title'].replace(u'\uE000', '').replace(u'\uE001', '') + + # parse image results + if result.get('ContentType', '').startswith('image'): + img_url = unquote(regex_img_url_remove_start.sub('', result['Url'])) + + # append result + results.append({'url': result['SourceUrl'], + 'title': result['Title'], + 'content': '', + 'img_src': img_url, + 'template': 'images.html'}) + + # parse general results + else: + result_url = result['Url'].replace(u'\uE000', '').replace(u'\uE001', '') + result_content = result['Description'].replace(u'\uE000', '').replace(u'\uE001', '') + + # append result + results.append({'url': result_url, + 'title': result_title, + 'content': result_content}) # parse images for result in json.get('Images', []): -- cgit v1.2.3