From f4525880ed5f965ac4c241638933842a27a2acf7 Mon Sep 17 00:00:00 2001 From: potato Date: Tue, 6 Sep 2016 11:47:27 +0200 Subject: [PATCH] [enh] dictionary engine added --- searx/engines/dictionary.py | 70 +++++++++++++++++++++++++++++++++++++ searx/settings.yml | 4 +++ 2 files changed, 74 insertions(+) create mode 100644 searx/engines/dictionary.py diff --git a/searx/engines/dictionary.py b/searx/engines/dictionary.py new file mode 100644 index 000000000..1849322f5 --- /dev/null +++ b/searx/engines/dictionary.py @@ -0,0 +1,70 @@ +import re +from lxml import html +from searx.engines.xpath import extract_text +from searx.languages import language_codes + +categories = [] +url = 'http://dictzone.com/{from_lang}-{to_lang}-dictionary/{query}' +weight = 100 + +parser_re = re.compile(u'.*?([a-z]+)-([a-z]+) (.+)', re.I) +results_xpath = './/table[@id="r"]/tr' + + +def request(query, params): + m = parser_re.match(unicode(query, 'utf8')) + if not m: + return params + + from_lang, to_lang, query = m.groups() + + if len(from_lang) == 2: + lan = filter(lambda x: x[0][:2] == from_lang, language_codes) + if lan: + from_lang = lan[0][1].lower() + else: + return params + + if len(to_lang) == 2: + lan = filter(lambda x: x[0][:2] == to_lang, language_codes) + if lan: + to_lang = lan[0][1].lower() + else: + return params + + params['url'] = url.format(from_lang=from_lang, to_lang=to_lang,query=query) + params['from_lang'] = from_lang + params['to_lang'] = to_lang + params['query'] = query + + return params + +def response(resp): + results = [] + answers = [] + + dom = html.fromstring(resp.text) + + for result in dom.xpath(results_xpath)[1:]: + try: + from_result, to_results_raw = result.xpath('./td') + except: + continue + + to_results = [] + for to_result in to_results_raw.xpath('./p/a'): + t = to_result.text_content() + if t.strip(): + to_results.append(to_result.text_content()) + + results.append({ + 'answer': u'{0} - {1}'.format( + from_result.text_content(), + '; '.join(to_results) + ), + 'url': url + }) + + return results + + diff --git a/searx/settings.yml b/searx/settings.yml index 2c034a863..b998e0c0d 100644 --- a/searx/settings.yml +++ b/searx/settings.yml @@ -495,6 +495,10 @@ engines: timeout: 6.0 categories : science + - name : dictionary + engine : dictionary + shortcut : dc + #The blekko technology and team have joined IBM Watson! -> https://blekko.com/ # - name : blekko images # engine : blekko_images