add exact result support (#80)

PaulMcInnis · Oct 6, 2020 · 616d351 · 616d351
1 parent a6c60cd
commit 616d351
Show file tree

Hide file tree

Showing 7 changed files with 25 additions and 1 deletion.
diff --git a/jobfunnel/backend/scrapers/base.py b/jobfunnel/backend/scrapers/base.py
@@ -54,6 +54,9 @@ def __init__(self, session: Session, config: 'JobFunnelConfigManager',
         self.session = session
         self.config = config
         self.query = ' '.join(config.search_config.keywords)
+        # if we match exact result, we add quotes.
+        if self.config.search_config.exact_result: 
+            self.query = f'"{self.query}"'
         if self.headers:
             self.session.headers.update(self.headers)
 

diff --git a/jobfunnel/config/cli.py b/jobfunnel/config/cli.py
@@ -226,6 +226,13 @@ def parse_cli(args: List[str]) -> Dict[str, Any]:
              '(NOTE: this is only available for Indeed provider).',
     )
 
+    search_group.add_argument(
+        '--exact-result',
+        dest='search.exact_result',
+        action='store_true',
+        help='Match exact search query',
+    )
+
     # Proxy stuff. TODO: way to tell argparse if proxy is seen all are req'd?
     proxy_group = cli_parser.add_argument_group('proxy')
     proxy_group.add_argument(
@@ -354,6 +361,7 @@ def get_config_manager(config: Dict[str, Any]) -> JobFunnelConfigManager:
         city=config['search']['city'],
         distance_radius=config['search']['radius'],
         return_similar_results=config['search']['similar_results'],
+        exact_result=config['search']['exact_result'],
         max_listing_days=config['search']['max_listing_days'],
         blocked_company_names=config['search']['company_block_list'],
         locale=Locale[config['search']['locale']],

diff --git a/jobfunnel/config/manager.py b/jobfunnel/config/manager.py
@@ -29,6 +29,7 @@ def __init__(self,
                  no_scrape: Optional[bool] = False,
                  bs4_parser: Optional[str] = BS4_PARSER,
                  return_similar_results: Optional[bool] = False,
+                 exact_result: Optional[bool] = False,
                  delay_config: Optional[DelayConfig] = None,
                  proxy_config: Optional[ProxyConfig] = None) -> None:
         """Init a config that determines how we will scrape jobs from Scrapers
@@ -69,7 +70,6 @@ def __init__(self,
         self.log_level = log_level
         self.no_scrape = no_scrape
         self.bs4_parser = bs4_parser  # NOTE: this is not currently configurable
-        self.return_similar_results = return_similar_results
         if not delay_config:
             # We will always use a delay config to be respectful
             self.delay_config = DelayConfig()

diff --git a/jobfunnel/config/search.py b/jobfunnel/config/search.py
@@ -22,6 +22,7 @@ def __init__(self,
                  city: Optional[str] = None,
                  distance_radius: Optional[int] = None,
                  return_similar_results: bool = False,
+                 exact_result: bool = False,
                  max_listing_days: Optional[int] = None,
                  blocked_company_names: Optional[List[str]] = None,
                  domain: Optional[str] = None,
@@ -38,6 +39,7 @@ def __init__(self,
                 DEFAULT_SEARCH_RADIUS.
             return_similar_results (Optional[bool], optional): return similar.
                 results (indeed), Defaults to False.
+            exact_result: search for exact query.
             max_listing_days (Optional[int], optional): oldest listing to show.
                 Defaults to DEFAULT_MAX_LISTING_DAYS.
             blocked_company_names (Optional[List[str]]): list of names of
@@ -54,6 +56,7 @@ def __init__(self,
         self.providers = providers
         self.keywords = keywords
         self.return_similar_results = return_similar_results  # Indeed.X thing
+        self.exact_result = exact_result
         self.max_listing_days = max_listing_days or DEFAULT_MAX_LISTING_DAYS
         self.blocked_company_names = blocked_company_names
         self.remoteness = remoteness

diff --git a/jobfunnel/config/settings.py b/jobfunnel/config/settings.py
@@ -66,6 +66,11 @@
                 'type': 'boolean',
                 'default': DEFAULT_RETURN_SIMILAR_RESULTS,
             },
+            'exact_result': {
+                'required': False,
+                'type': 'boolean',
+                'default': DEFAULT_EXACT_RESULT,
+            },
             'keywords': {
                 'required': True,
                 'type': 'list',

diff --git a/jobfunnel/resources/defaults.py b/jobfunnel/resources/defaults.py
@@ -21,6 +21,7 @@
 DEFAULT_PROVIDERS = [Provider.MONSTER, Provider.INDEED] #, Provider.GLASSDOOR]
 DEFAULT_PROVIDER_NAMES = [p.name for p in DEFAULT_PROVIDERS]
 DEFAULT_RETURN_SIMILAR_RESULTS = False
+DEFAULT_EXACT_RESULT = False
 DEFAULT_RANDOM_DELAY = False
 DEFAULT_RANDOM_CONVERGING_DELAY = False
 DEFAULT_REMOTENESS = Remoteness.ANY

diff --git a/tests/config/test_cli.py b/tests/config/test_cli.py
@@ -78,6 +78,10 @@ def test_parse_cli_build_config_dict(argv, exp_exception):
             assert cfg['search']['similar_results']
         else:
             assert not cfg['search']['similar_results']
+        if '--exact-result' in argv:
+            assert cfg['search']['exact_result']
+        else:
+            assert not cfg['search']['exact_result']
 
         assert cfg['delay']['algorithm'] == 'LINEAR'
         assert cfg['delay']['max_duration'] == 8