Improve data security by mandating token search

app/local_settings.example.py

@@ -16,6 +16,45 @@ OPENSEARCH_MAIN_WILDCARD_ANON = False
 OPENSEARCH_MAIN_SOURCES = ["irc", "dis", "all"]
 DRILLDOWN_RESULTS_PER_PAGE = 15
 
+# Encryption
+ENCRYPTION = False
+ENCRYPTION_KEY = b""
+
+# Hashing
+HASHING = True
+HASHING_KEY = "xxx"
+
+# Common to encryption and hashing
+WHITELIST_FIELDS = [
+    "ts",
+    "date",
+    "time",
+    "sentiment",
+    "version_sentiment",
+    "tokens",
+    "num_chans",
+    "num_users",
+    "tokens",
+    "src",
+    "exemption",
+    "hidden",
+]
+
+# Don't obfuscate these parameters, or lookup hashes in them
+NO_OBFUSCATE_PARAMS = [
+    "query",
+    "query_full",
+    "size",
+    "source",
+    "sorting",
+    "tags",
+    "index",
+    "dedup",
+    "check_sentiment",
+    "sentiment_method",
+    "dates",
+]
+
 OPENSEARCH_BLACKLISTED = {}
 
 # URLs

core/lib/opensearch.py

@@ -5,7 +5,16 @@ from opensearchpy import OpenSearch
 from opensearchpy.exceptions import NotFoundError, RequestError
 
 from core.lib.threshold import annotate_num_chans, annotate_num_users, annotate_online
-from core.views.helpers import dedup_list, encrypt_list, hash_list, hash_lookup
+from core.views.helpers import (
+    SearchDenied,
+    dedup_list,
+    encrypt_list,
+    hash_list,
+    hash_lookup,
+)
+
+# from json import dumps
+# pp = lambda x: print(dumps(x, indent=2))
 
 
 def initialise_opensearch():
@@ -141,18 +150,17 @@ def filter_blacklisted(user, response):
     response["hits"]["hits"] = [hit for hit in response["hits"]["hits"] if hit]
 
 
-def construct_query(query, size):
+def construct_query(query, size, use_query_string=True, tokens=False):
     """
     Accept some query parameters and construct an OpenSearch query.
     """
     if not size:
         size = 5
-    query = {
+    query_base = {
         "size": size,
-        "query": {
+        "query": {"bool": {"must": []}},
-            "bool": {
+    }
-                "must": [
+    query_string = {
-                    {
         "query_string": {
             "query": query,
             # "fields": fields,
@@ -177,11 +185,31 @@ def construct_query(query, size):
             "auto_generate_synonyms_phrase_query": True,
         }
     }
-                ]
+    query_tokens = {
+        "simple_query_string": {
+            # "tokens": query,
+            "query": query,
+            "fields": ["tokens"],
+            "flags": "ALL",
+            "fuzzy_transpositions": True,
+            "fuzzy_max_expansions": 50,
+            "fuzzy_prefix_length": 0,
+            "default_operator": "and",
+            "analyzer": "standard",
+            "lenient": True,
+            "boost": 1,
+            "quote_field_suffix": "",
+            "analyze_wildcard": False,
+            "auto_generate_synonyms_phrase_query": False,
         }
-        },
     }
-    return query
+    if tokens:
+        query_base["query"]["bool"]["must"].append(query_tokens)
+        # query["query"]["bool"]["must"].append(query_string)
+        # query["query"]["bool"]["must"][0]["query_string"]["fields"] = ["tokens"]
+    elif use_query_string:
+        query_base["query"]["bool"]["must"].append(query_string)
+    return query_base
 
 
 def run_main_query(client, user, query, custom_query=False, index=None, size=None):
@@ -261,6 +289,7 @@ def query_results(
     dedup=False,
     dedup_fields=None,
     lookup_hashes=True,
+    tags=None,
 ):
     """
     API helper to alter the OpenSearch return format into something
@@ -276,12 +305,15 @@ def query_results(
     add_top = []
     add_top_negative = []
     sort = None
+    query_created = False
 
     # Lookup the hash values but don't disclose them to the user
     if lookup_hashes:
         if settings.HASHING:
             query_params = deepcopy(query_params)
-            hash_lookup(query_params)
+            hash_lookup(request.user, query_params)
+            if tags:
+                hash_lookup(request.user, tags)
 
     if request.user.is_anonymous:
         sizes = settings.OPENSEARCH_MAIN_SIZES_ANON
@@ -366,15 +398,53 @@ def query_results(
             range_query_precise["match"]["sentiment"] = 0
             add_top_negative.append(range_query_precise)
 
+    # Only one of query or query_full can be active at once
+    # We prefer query because it's simpler
     if "query" in query_params:
         query = query_params["query"]
-        search_query = construct_query(query, size)
+        search_query = construct_query(query, size, tokens=True)
+        query_created = True
+    elif "query_full" in query_params:
+        query_full = query_params["query_full"]
+        if request.user.has_perm("query_search"):
+            search_query = construct_query(query_full, size)
+            query_created = True
+        else:
+            message = "You cannot search by query string"
+            message_class = "danger"
+            return {"message": message, "class": message_class}
     else:
         if custom_query:
             search_query = custom_query
+
+    if tags:
+        # Get a blank search query
+        if not query_created:
+            search_query = construct_query(None, size, use_query_string=False)
+            query_created = True
+        for tagname, tagvalue in tags.items():
+            add_bool.append({tagname: tagvalue})
+
+    required_any = ["query_full", "query", "tags"]
+    if not any([field in query_params.keys() for field in required_any]):
+        if not custom_query:
+            message = "Empty query!"
+            message_class = "warning"
+            return {"message": message, "class": message_class}
+
     if add_bool:
+        # if "bool" not in search_query["query"]:
+        #     search_query["query"]["bool"] = {}
+        # if "must" not in search_query["query"]["bool"]:
+        #     search_query["query"]["bool"] = {"must": []}
+
         for item in add_bool:
-            search_query["query"]["bool"]["must"].append({"match": item})
+            k, v = list(item.items())[0]
+            if isinstance(v, SearchDenied):
+                message = f"Access denied: search by protected field {k}: {v.value}"
+                message_class = "danger"
+                return {"message": message, "class": message_class}
+            search_query["query"]["bool"]["must"].append({"match_phrase": item})
     if add_top:
         for item in add_top:
             search_query["query"]["bool"]["must"].append(item)
@@ -398,7 +468,6 @@ def query_results(
                 return {
                     "message": message,
                     "class": message_class,
-                    "params": query_params,
                 }
             if index == "meta":
                 index = settings.OPENSEARCH_INDEX_META
@@ -410,7 +479,6 @@ def query_results(
                 return {
                     "message": message,
                     "class": message_class,
-                    "params": query_params,
                 }
 
     else:
@@ -461,7 +529,6 @@ def query_results(
     if not request.user.has_perm("view_plain"):
         if settings.HASHING:
             hash_list(request.user, results_parsed)
-
     # process_list(reqults)
 
     # IMPORTANT! - DO NOT PASS query_params to the user!

core/templates/base.html

@@ -15,6 +15,7 @@
     <link rel="stylesheet" href="{% static 'css/bulma-slider.min.css' %}">
     <link rel="stylesheet" href="{% static 'css/bulma-calendar.min.css' %}">
     <link rel="stylesheet" href="{% static 'css/bulma-tagsinput.min.css' %}">
+    <link rel="stylesheet" href="{% static 'css/bulma-switch.min.css' %}">
     <script src="{% static 'js/bulma-calendar.min.js' %}" integrity="sha384-DThNif0xGXbopX7+PE+UabkuClfI/zELNhaVqoGLutaWB76dyMw0vIQBGmUxSfVQ" crossorigin="anonymous"></script>
     <script src="{% static 'js/bulma-slider.min.js' %}" integrity="sha384-wbyps8iLG8QzJE02viYc/27BtT5HSa11+b5V7QPR1/huVuA8f4LRTNGc82qAIeIZ" crossorigin="anonymous"></script>
     <script defer src="{% static 'js/htmx.min.js' %}" integrity="sha384-cZuAZ+ZbwkNRnrKi05G/fjBX+azI9DNOkNYysZ0I/X5ZFgsmMiBXgDZof30F5ofc" crossorigin="anonymous"></script>

core/templates/ui/drilldown/drilldown.html

@@ -39,60 +39,23 @@
         } catch {
           var value = spl[1];
         }
-        populateSearch(field, value);
         return `${field}: ${value}`;
       });
       inputTags.BulmaTagsInput().on('after.remove', function(item) {
         var spl = item.split(": ");
         var field = spl[0];
         var value = spl[1].trim();
-        populateSearch(field, value);
       });
     }
     function populateSearch(field, value) {
-      var queryElement = document.getElementById('query');
+      var inputTags = document.getElementById('tags');
-
+      inputTags.BulmaTagsInput().add(field+": "+value);
-      var present = true;
-      if (present == true) {
-        var combinations = [`${field}: "${value}"`,
-                            `${field}: "${value}"`,
-                            `${field}: ${value}`,
-                            `${field}:${value}`,
-                            `${field}:"${value}"`];
-        var toAppend = ` AND ${field}: "${value}"`;
-      } else {
-        var combinations = [`NOT ${field}: "${value}"`,
-                            `NOT ${field}: "${value}"`,
-                            `NOT ${field}: ${value}`,
-                            `NOT ${field}:${value}`,
-                            `NOT ${field}:"${value}"`];
-      }
-      var contains = combinations.some(elem => queryElement.value.includes(elem));
-      if (!contains) {
-          queryElement.value+=toAppend;
-      } else {
-        for (var index in combinations) {
-          combination = combinations[index];
-          queryElement.value = queryElement.value.replaceAll("AND "+combination, "");
-          queryElement.value = queryElement.value.replaceAll(combination, "");
-        }
-      }
-
-      if (field == "src") {
-        document.getElementById("source").selectedIndex = 2;
-      }
-      if (queryElement.value.startsWith(" AND ")) {
-        queryElement.value = queryElement.value.replace(" AND ", "");
-      }
-      if (queryElement.value.startsWith("AND ")) {
-        queryElement.value = queryElement.value.replace("AND ", "");
-      }
       htmx.trigger("#search", "click");
     }
   </script>
   <div>
     {% include 'partials/notify.html' %}
-    <form method="POST" hx-post="{% url 'search' %}"
+    <form class="skipEmptyFields" method="POST" hx-post="{% url 'search' %}"
           hx-trigger="change"
           hx-target="#results"
           hx-swap="innerHTML"
@@ -102,12 +65,17 @@
       <div class="columns">
         <div class="column">
           <div class="field has-addons">
-            <div class="control is-expanded has-icons-left">
+            <div id="query" class="control is-expanded has-icons-left">
               <input
                 hx-post="{% url 'search' %}"
                 hx-trigger="keyup changed delay:200ms"
                 hx-target="#results"
-                hx-swap="innerHTML" id="query" name="query" value="{{ params.query }}" class="input" type="text" placeholder="msg: science AND nick: BillNye AND channel: #science">
+                hx-swap="innerHTML"
+                name="query"
+                value="{{ params.query }}"
+                class="input"
+                type="text"
+                placeholder="(science | tech | art) + (interest) -hello">
               <span class="icon is-small is-left">
                 <i class="fas fa-magnifying-glass"></i>
               </span>
@@ -166,6 +134,19 @@
                 </a>
               </p>
             </div>
+            <div class="control">
+              <div class="field">
+                <input
+                  id="full_query"
+                  type="checkbox"
+                  class="switch"
+                  {% if params.query_full is not None %}checked="checked"{% else %}none{% endif %}
+                  data-script="on click toggle .is-hidden on #query_full">
+                <label
+                  class="{% if not perms.core.query_search %}is-disabled{% endif %}"
+                  for="full_query">Full query </label>
+              </div>
+            </div>
           </div>
           <div class="column is-narrow">
             <div class="field has-addons block">
@@ -411,24 +392,45 @@
           </div>
         </div>
       </div>
+      <div id="query_full" class="block {% if params.query_full is None %}is-hidden{% endif %}">
+        <div class="control is-expanded has-icons-left">
+          <input
+            hx-post="{% url 'search' %}"
+            hx-trigger="keyup changed delay:200ms"
+            hx-target="#results"
+            hx-swap="innerHTML"
+            name="query_full"
+            value="{{ params.query_full }}"
+            class="input"
+            type="text"
+            placeholder="msg: science AND nick: BillNye AND channel: #science">
+          <span class="icon is-small is-left">
+            <i class="fas fa-magnifying-glass"></i>
+          </span>
+        </div>
+      </div>
+      <div class="block">
+        <input
+          hx-trigger="change"
+          hx-post="{% url 'search' %}"
+          hx-target="#results"
+          hx-swap="innerHTML"
+          id="tags"
+          class="input"
+          type="tags"
+          name="tags"
+          placeholder="Add tags"
+          value="{{ params.tags }}">
+      </div>
       <div class="is-hidden"></div>
     </form>
   </div>
-  <div class="block">
-    <input id="tags" class="input" type="tags" placeholder="Add query" value="{{ tags|joinsep:',' }}">
-  </div>
   <div class="block">
     <div id="results">
-      <!-- {% if results %}
-        {% include 'ui/drilldown/results.html' %}
-      {% endif %} -->
-      {% if table %}
       {% include 'ui/drilldown/table_results.html' %}
-      {% else %}
       <script>
         setupTags();
       </script>
-      {% endif %}
     </div>
   </div>
   <div id="modals-here">

core/templates/ui/drilldown/table_results.html

@@ -37,23 +37,3 @@
   {% endif %}
   {% include 'ui/drilldown/table_results_partial.html' %}
 {% endif %}
-{# Update the tags in case the user changed the query #}
-{# Check for focus and refocus #}
-<script>
-  var inputTags = document.getElementsByClassName('tags-input');
-  var inputBox = document.querySelector("[placeholder='Add query']");
-  var isFocused = (document.activeElement === inputBox);
-  for (index = 0; index < inputTags.length; index++) {
-    if (index == 0) {
-      inputTags[0].outerHTML = '<input id="tags" class="input" type="tags" placeholder="Add query" value="{{ tags|joinsep:',' }}">';
-    } else {
-      inputTags[index].remove();
-    }
-  }
-  // inputTags[0].outerHTML = '<input id="tags" class="input" type="tags" placeholder="Add query" value="{{ tags|joinsep:',' }}">';
-  setupTags();
-  var inputBox = document.querySelector("[placeholder='Add query']");
-  if (isFocused) {
-    inputBox.focus();
-  }
-</script>

core/templates/ui/drilldown/table_results_partial.html

@@ -143,7 +143,7 @@
                                 </span>
                               </td>
                             {% elif column.name == 'tokens' %}
-                              <td class="{{ column.name }}">
+                              <td class="{{ column.name }} wrap" style="max-width: 10em">
                                 {{ cell|joinsep:',' }}
                               </td>
                             {% elif column.name == 'src' %}

core/views/helpers.py

@@ -10,6 +10,11 @@ from sortedcontainers import SortedSet
 from core import r
 
 
+class SearchDenied:
+    def __init__(self, value):
+        self.value = value
+
+
 def dedup_list(data, check_keys):
     """
     Remove duplicate dictionaries from list.
@@ -90,7 +95,10 @@ def hash_list(user, data, hash_keys=False):
     for index, item in enumerate(data_copy):
         if isinstance(item, dict):
             for key, value in list(item.items()):
-                if key not in settings.WHITELIST_FIELDS:
+                if (
+                    key not in settings.WHITELIST_FIELDS
+                    and key not in settings.NO_OBFUSCATE_PARAMS
+                ):
                     if isinstance(value, int):
                         value = str(value)
                     if isinstance(value, bool):
@@ -122,16 +130,33 @@ def hash_list(user, data, hash_keys=False):
         r.hmset(cache, hash_table)
 
 
-def hash_lookup(data_dict):
+def hash_lookup(user, data_dict):
     cache = "cache.hash"
     hash_list = SortedSet()
-    for key, value in data_dict.items():
+    for key, value in list(data_dict.items()):
+        if (
+            key not in settings.WHITELIST_FIELDS
+            and key not in settings.NO_OBFUSCATE_PARAMS
+        ):
             if not value:
                 continue
             # hashes = re.findall("\|([^\|]*)\|", value)  # noqa
+            if isinstance(value, str):
                 hashes = re.findall("[A-Z0-9]{12,13}", value)
-        if not hashes:
+            elif isinstance(value, dict):
+                hashes = []
+                for key, value in value.items():
+                    if not value:
                         continue
+                    hashes_iter = re.findall("[A-Z0-9]{12,13}", value)
+                    for h in hashes_iter:
+                        hashes.append(h)
+            if not hashes:
+                # Otherwise the user could inject plaintext search queries
+                if not user.has_perm("bypass_hashing"):
+                    data_dict[key] = SearchDenied(value=data_dict[key])
+                    # del data_dict[key]
+
             for hash in hashes:
                 hash_list.add(hash)
 
@@ -147,8 +172,17 @@ def hash_lookup(data_dict):
         for key in data_dict.keys():
             for hash in total:
                 if data_dict[key]:
+                    if isinstance(data_dict[key], str):
                         if hash in data_dict[key]:
-                        data_dict[key] = data_dict[key].replace(f"{hash}", total[hash])
+                            print("Replacing", data_dict[key], "with", total[hash])
+                            data_dict[key] = data_dict[key].replace(
+                                f"{hash}", total[hash]
+                            )
+                    elif isinstance(data_dict[key], dict):
+                        for k2, v2 in data_dict[key].items():
+                            if hash in v2:
+                                print("Replacing", v2, "with", total[hash])
+                                data_dict[key][k2] = v2.replace(f"{hash}", total[hash])
 
 
 def encrypt_list(user, data, secret):

core/views/ui/drilldown.py

@@ -55,6 +55,21 @@ def create_tags(query):
     return tags
 
 
+def parse_tags(tags_pre):
+    """
+    Parse the tags from the variable tags_pre.
+    """
+    tags = {}
+    tags_spl = tags_pre.split(",")
+    if tags_spl:
+        for tag in tags_spl:
+            tag = tag.split(": ")
+            if len(tag) == 2:
+                key, val = tag
+                tags[key] = val
+    return tags
+
+
 def make_table(context):
     table = DrilldownTable(context["object_list"])
     context["table"] = table
@@ -78,6 +93,8 @@ def make_graph(results):
 
 
 def drilldown_search(request, return_context=False, template=None):
+    extra_params = {}
+
     if not template:
         template_name = "ui/drilldown/table_results.html"
     else:
@@ -105,6 +122,10 @@ def drilldown_search(request, return_context=False, template=None):
     query_params.update(tmp_post)
     query_params.update(tmp_get)
 
+    # URI we're passing to the template for linking
+    if "csrfmiddlewaretoken" in query_params:
+        del query_params["csrfmiddlewaretoken"]
+
     # Parse the dates
     if "dates" in query_params:
         dates = parse_dates(query_params["dates"])
@@ -118,24 +139,34 @@ def drilldown_search(request, return_context=False, template=None):
             query_params["to_time"] = dates["to_time"]
 
     if "query" in query_params:
-        context = query_results(request, query_params)
+        # Remove null values
-
+        if query_params["query"] == "":
+            del query_params["query"]
             # Turn the query into tags for populating the taglist
-        tags = create_tags(query_params["query"])
+            # tags = create_tags(query_params["query"])
-        context["tags"] = tags
+            # context["tags"] = tags
+    # else:
+    #     context = {"object_list": []}
+
+    # Remove null values
+    if "query_full" in query_params:
+        if query_params["query_full"] == "":
+            del query_params["query_full"]
+
+    if "tags" in query_params:
+        if query_params["tags"] == "":
+            del query_params["tags"]
         else:
-        context = {"object_list": []}
+            tags = parse_tags(query_params["tags"])
+            extra_params["tags"] = tags
+
+    context = query_results(request, query_params, **extra_params)
 
     # Valid sizes
     context["sizes"] = sizes
 
-    # URI we're passing to the template for linking
-    if "csrfmiddlewaretoken" in query_params:
-        del query_params["csrfmiddlewaretoken"]
-
     url_params = urllib.parse.urlencode(query_params)
     context["client_uri"] = url_params
-
     context["params"] = query_params
     if "message" in context:
         response = render(request, template_name, context)
@@ -158,6 +189,17 @@ def drilldown_search(request, return_context=False, template=None):
     clean_url_params = urllib.parse.urlencode(clean_params)
     context["uri"] = clean_url_params
 
+    # Warn users trying to use query string that the simple query supersedes it
+    if all([x in query_params for x in ["query", "query_full"]]):
+        context["message"] = (
+            "You are searching with both query types. "
+            "The simple query will be used. "
+            "The full query will be ignored. "
+            "Remove the text from the simple query if you wish "
+            "to use the full query."
+        )
+        context["class"] = "warning"
+
     response = render(request, template_name, context)
     if request.GET:
         if request.htmx:
@@ -260,7 +302,7 @@ class DrilldownContextModal(APIView):
         # Lookup the hash values but don't disclose them to the user
         if settings.HASHING:
             SAFE_PARAMS = deepcopy(query_params)
-            hash_lookup(SAFE_PARAMS)
+            hash_lookup(request.user, SAFE_PARAMS)
         else:
             SAFE_PARAMS = query_params
 
@@ -383,7 +425,7 @@ class ThresholdInfoModal(APIView):
         # Lookup the hash values but don't disclose them to the user
         if settings.HASHING:
             SAFE_PARAMS = request.data.dict()
-            hash_lookup(SAFE_PARAMS)
+            hash_lookup(request.user, SAFE_PARAMS)
         safe_net = SAFE_PARAMS["net"]
         safe_nick = SAFE_PARAMS["nick"]
         safe_channel = SAFE_PARAMS["channel"]