From 732a122c4460b28af67d5acbfd625b89b579781d Mon Sep 17 00:00:00 2001
From: Stefan Karner <stefan.karner@onb.ac.at>
Date: Tue, 7 May 2019 15:55:44 +0200
Subject: [PATCH] Add fulltext search methods 'seed' and 'capture'

---
 webarchiv.py | 50 ++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 50 insertions(+)

diff --git a/webarchiv.py b/webarchiv.py
index ea466ba..189c7b1 100644
--- a/webarchiv.py
+++ b/webarchiv.py
@@ -149,6 +149,56 @@ class WebarchivSession:
             self._display_http_error(e)
             print('Query for "{}" not added'.format(query_string))
 
+    def fulltext_search_within_domain(self, query_string, domain, from_=None, to_=None):
+        """
+        Start a fulltext seed search query in the Webarchive.
+
+        :param query_string: String to search for
+        :param domain: Search only within this domain name
+        :param from_: Optional earliest date bound for the search
+          in the format YYYYMM.
+        :param to_: Optional latest date bound for the search
+          in the format YYYYMM.
+        :return: HTTP Response object
+        """
+        params = {'q': query_string, 'g': domain}
+        if from_:
+            params['from'] = from_
+        if to_:
+            params['to'] = to_
+
+        try:
+            response = self._get(op='/search/fulltext/seed', params=params)
+            return self.waitForResponse(response)
+
+        except HTTPError as e:
+            self._display_http_error(e)
+
+    def fulltext_search_within_url(self, query_string, url, pagesize=10, from_=None, to_=None):
+        """
+        Start a fulltext capture search query in the Webarchive.
+
+        :param query_string: String to search for
+        :param url: Search only captures starting at this exact web address
+        :param from_: Optional earliest date bound for the search
+          in the format YYYYMM.
+        :param to_: Optional latest date bound for the search
+          in the format YYYYMM.
+        :return: HTTP Response object
+        """
+        params = {'q': query_string, 'g': url, 'pagesize': pagesize}
+        if from_:
+            params['from'] = from_
+        if to_:
+            params['to'] = to_
+
+        try:
+            response = self._get(op='/search/fulltext/capture', params=params)
+            return self.waitForResponse(response)
+
+        except HTTPError as e:
+            self._display_http_error(e)
+
     def wayback_search(self, query_string, from_=None, to_=None):
         """
         Start a wayback search query in the Webarchive.
-- 
GitLab