added search

Author: asifpy

pysmartprice/price_parser.py renamed to pysmartprice/abstract.py

@@ -2,15 +2,16 @@
 from bs4 import BeautifulSoup
 
 from pysmartprice.results import SmartPriceResult
-from pysmartprice.helpers import scrape
+from pysmartprice.helpers import scrape, scrape_helper
 from pysmartprice import constants
 
 
-class PriceListParser(object):
+class BaseParser(object):
     def __init__(self, mapper, **kwargs):
         self.mapper = mapper
+        self.params = kwargs
         self.url = constants.URL_MAPPER[self.mapper]
-        self.response = scrape(self._make_url(self.url))
+        self.response = scrape(self._make_url(self.url), **kwargs)
         self.soup = BeautifulSoup(self.response, 'lxml')
         self.result = [
             SmartPriceResult(self.get_product_attrs(item))
@@ -21,17 +22,14 @@ def _make_url(self, target):
         return '{}{}'.format(constants.SMARTPRICE_WEB_URL, target)
 
     @property
-    def get_page_range(self):
-        page_range = self.soup.findAll(
-            'span', attrs={'class': 'pgntn__rslt-page'})
+    def price_results(self):
+        if self.get_page_range:
+            return self.process_multiple_pages()
 
-        if not page_range:
-            return None
+        return self.result
 
-        first_page = int(page_range[0].text)
-        last_page = int(page_range[1].text)
-        return first_page, last_page
 
+class ParserMixin(object):
     def get_product_attrs(self, item):
         return dict(
             img=item.find('img').get('src'),
@@ -48,12 +46,6 @@ def products_html(self):
         html = self.soup.findAll('div', attrs={'class': 'prdct-item'})
         return html
 
-    @property
-    def get_paged_url(self):
-        i = self.url.find(self.mapper)
-        paged_url = '{}pages/{}'.format(self.url[:i], self.url[i:])
-        return paged_url
-
     def process_multiple_pages(self):
         results = self.result
         first_page, last_page = self.get_page_range
@@ -62,12 +54,15 @@ def process_multiple_pages(self):
 
         for page in range(first_page+1, last_page+1):
             url = paged_url.replace('.html', '-{}.html'.format(page))
-            page_urls.append(self._make_url(url))
+            params = self.params.copy()
+            if self.params.get('page', None):
+                params.update({'page': page})
+            page_urls.append((self._make_url(url), params))
 
         # Scrape pages in parallel
         pool = multiprocessing.Pool(processes=multiprocessing.cpu_count()*2)
 
-        for page in pool.map(scrape, page_urls):
+        for page in pool.map(scrape_helper, page_urls):
             self.soup = BeautifulSoup(page, 'lxml')
 
             results += [
@@ -77,8 +72,13 @@ def process_multiple_pages(self):
         return results
 
     @property
-    def price_results(self):
-        if self.get_page_range:
-            return self.process_multiple_pages()
+    def get_page_range(self):
+        page_range = self.soup.findAll(
+            'span', attrs={'class': 'pgntn__rslt-page'})
 
-        return self.result
+        if not page_range:
+            return None
+
+        first_page = int(page_range[0].text)
+        last_page = int(page_range[1].text)
+        return first_page, last_page

pysmartprice/base.py

@@ -1,11 +1,14 @@
-from pysmartprice.price_parser import PriceListParser
+from pysmartprice.smartparser import(
+    PriceListParser,
+    SearchParser
+)
 from pysmartprice.constants import SMARTPRICE_ATTRS
 
 
 class SmartPrice(object):
 
-    def parser_results(self, product):
-        parser = PriceListParser(product)
+    def parser_results(self, product, **kwargs):
+        parser = PriceListParser(product, **kwargs)
         return parser.price_results
 
     def __getattr__(self, attr):
@@ -15,3 +18,8 @@ def __getattr__(self, attr):
 
         setattr(self, attr, self.parser_results(SMARTPRICE_ATTRS[attr]))
         return getattr(self, attr)
+
+    def search(self, search_key):
+        params = dict(s=search_key, page=1)
+        parser = SearchParser('search', **params)
+        return parser.price_results

pysmartprice/constants.py

@@ -32,7 +32,10 @@
     'samsung-tv': 'electronics/pricelist/samsung-tv-price-list-in-india.html',
     'sony-tv': 'electronics/pricelist/sony-tv-price-list-in-india.html',
     'lg-tv': 'electronics/pricelist/lg-tv-price-list-in-india.html',
+    'panasonic-tv': 'electronics/pricelist/panasonic-tv-price-list-in-india.html',
+    'sharp-tv': 'electronics/pricelist/sharp-tv-price-list-in-india.html',
 
+    'search': 'msp/search/msp_search_new.php',
 }
 
 

pysmartprice/helpers.py

@@ -1,6 +1,10 @@
 import requests
 
 
-def scrape(url):
-    resp = requests.get(url)
+def scrape(url, **kwargs):
+    resp = requests.get(url, params=kwargs)
     return resp.text
+
+
+def scrape_helper(args):
+    return scrape(args[0], **args[1])

pysmartprice/smartparser.py

@@ -0,0 +1,18 @@
+from pysmartprice.abstract import(
+    BaseParser,
+    ParserMixin
+)
+
+
+class PriceListParser(BaseParser, ParserMixin):
+    @property
+    def get_paged_url(self):
+        i = self.url.find(self.mapper)
+        paged_url = '{}pages/{}'.format(self.url[:i], self.url[i:])
+        return paged_url
+
+
+class SearchParser(BaseParser, ParserMixin):
+    @property
+    def get_paged_url(self):
+        return self.url