Merge pull request #1282 from hypothesis/searchable_uri-2

Update uri mapping, use match query

h/js/displayer.coffee

@@ -42,8 +42,8 @@ class Displayer
       streamfilter
         .setPastDataNone()
         .setMatchPolicyIncludeAny()
-        .addClausesParse('references:^' + $scope.annotation.id)
-        .addClausesParse('id:=' + $scope.annotation.id)
+        .addClause('references', 'first_of', $scope.annotation.id, true)
+        .addClause('id', 'equals', $scope.annotation.id, true)
         .getFilter()
 
     $scope.change_annotation_content = (id, new_annotation) =>

h/js/streamfilter.coffee

-class ClauseParser
-  filter_fields : ['references', 'text', 'user', 'uri', 'id', 'tags', 'created', 'updated']
-  operators: [
-    '#<=', '#>=', '#<', '#>', '#=',
-    '=>', '>=', '<=', '=<', '>', '<',
-    '[', '=~', '^', '{',
-    '='
-  ]
-  operator_mapping:
-    '=': 'equals'
-    '>': 'gt'
-    '<': 'lt'
-    '=>': 'ge'
-    '>=': 'ge'
-    '=<': 'le'
-    '<=': 'le'
-    '[' : 'one_of'
-    '=~' : 'matches'
-    '^' : 'first_of'
-    '{' : 'match_of' # one_of but not exact search
-    '#=' : 'lene'
-    '#>' : 'leng'
-    '#>=' : 'lenge'
-    '#<' : 'lenl'
-    '#<=' : 'lenle'
-  insensitive_operator : 'i'
-
-  parse_clauses: (clauses) ->
-    bads = []
-    structure = []
-    unless clauses
-      return
-    clauses = clauses.split ' '
-    for clause in clauses
-      #Here comes the long and boring validation checking
-      clause = clause.trim()
-      if clause.length < 1 then continue
-
-      parts = clause.split /:(.+)/
-      unless parts.length > 1
-        bads.push [clause, 'Filter clause is not well separated']
-        continue
-
-      unless parts[0] in @filter_fields
-        bads.push [clause, 'Unknown filter field']
-        continue
-
-      field = parts[0]
-
-      if parts[1][0] is @insensitive_operator
-        sensitive = false
-        rest = parts[1][1..]
-      else
-        sensitive = true
-        rest = parts[1]
-
-      operator_found = false
-      for operator in @operators
-        if (rest.indexOf operator) is 0
-          oper = @operator_mapping[operator]
-          if operator is '[' or operator is '{'
-            value = rest[operator.length..].split ','
-          else
-            value = rest[operator.length..]
-          operator_found = true
-          if field is 'user'
-            value = 'acct:' + value + '@' + window.location.hostname
-          break
-
-      unless operator_found
-        bads.push [clause, 'Unknown operator']
-        continue
-
-      structure.push
-        'field'   : '/' + field
-        'operator': oper
-        'value'   : value
-        'case_sensitive': sensitive
-    [structure, bads]
-
-
 class StreamFilter
   strategies: ['include_any', 'include_all', 'exclude_any', 'exclude_all']
   past_modes: ['none','hits','time']
@@ -94,7 +13,6 @@ class StreamFilter
         load_past: "none"
 
   constructor: ->
-    @parser = new ClauseParser()
 
   getFilter: -> return @filter
   getPastData: -> return @filter.past_data
@@ -166,29 +84,13 @@ class StreamFilter
     @filter.clauses.push clause
     this
 
-  addClause: (field, operator, value, case_sensitive = false, es_query_string = false) ->
+  addClause: (field, operator, value, case_sensitive = false, options = {}) ->
     @filter.clauses.push
       field: field
       operator: operator
       value: value
       case_sensitive: case_sensitive
-      es_query_string: es_query_string
-    this
-
-  setClausesParse: (clauses_to_parse, error_checking = false) ->
-    res = @parser.parse_clauses clauses_to_parse
-    if res[1].length
-      console.log "Errors while parsing clause:"
-      console.log res[1]
-    if res? and (not error_checking) or (error_checking and res[1]?.length is 0)
-      @filter.clauses = res[0]
-    this
-
-  addClausesParse: (clauses_to_parse, error_checking = false) ->
-    res = @parser.parse_clauses clauses_to_parse
-    if res? and (not error_checking) or (error_checking and res[1]?.length is 0)
-      for clause in res[0]
-        @filter.clauses.push clause
+      options: options
     this
 
   resetFilter: ->
@@ -202,5 +104,4 @@ class StreamFilter
 
 
 angular.module('h.streamfilter',['bootstrap'])
-  .service('clauseparser', ClauseParser)
   .service('streamfilter', StreamFilter)
\ No newline at end of file

h/js/streamsearch.coffee

@@ -33,8 +33,14 @@ get_quote = (annotation) ->
 #      exact_match: true|false (default: true)
 #      case_sensitive: true|false (default: false)
 #      and_or: and|or for multiple values should it threat them as 'or' or 'and' (def: or)
-#      es_query_string: should the streaming backend use query_string es query for this facet
 #      operator: if given it'll use this operator regardless of other circumstances
+#
+#      options: backend specific options
+#      options.es: elasticsearch specific options
+#      options.es.query_type : can be: simple, query_string, match
+#         defaults to: simple, determines which es query type to use
+#      options.es.cutoff_frequency: if set, the query will be given a cutoff_frequency for this facet
+#      options.es.and_or: match queries can use this, defaults to and
 # }
 # The models is the direct output from visualsearch
 # The limit is the default limit
@@ -71,14 +77,13 @@ class SearchHelper
       case_sensitive = if rule.case_sensitive? then rule.case_sensitive else false
       and_or = if rule.and_or? then rule.and_or else 'or'
       mapped_field = if rule.path? then rule.path else '/'+category
-      es_query_string = if rule.es_query_string? then rule.es_query_string else false
 
       if values.length is 1
         oper_part =
           if rule.operator? then rule.operator
           else if exact_match then 'equals' else 'matches'
         value_part = if rule.formatter then rule.formatter values[0] else values[0]
-        filter.addClause mapped_field, oper_part, value_part, case_sensitive, es_query_string
+        filter.addClause mapped_field, oper_part, value_part, case_sensitive, rule.options
       else
         if and_or is 'or'
           val_list = ''
@@ -90,14 +95,14 @@ class SearchHelper
           oper_part =
             if rule.operator? then rule.operator
             else if exact_match then 'one_of' else 'match_of'
-          filter.addClause mapped_field, oper_part, val_list, case_sensitive, es_query_string
+          filter.addClause mapped_field, oper_part, val_list, case_sensitive, rule.options
         else
           oper_part =
             if rule.operator? then rule.operator
             else if exact_match then 'equals' else 'matches'
           for val in values
             value_part = if rule.formatter then rule.formatter val else val
-            filter.addClause mapped_field, oper_part, value_part, case_sensitive, es_query_string
+            filter.addClause mapped_field, oper_part, value_part, case_sensitive, rule.options
 
     if limit != 50 then categories['results'] = [limit]
     [filter.getFilter(), categories]
@@ -132,16 +137,17 @@ class StreamSearch
       and_or: 'and'
     uri:
       formatter: (uri) ->
-        uri = uri.toLowerCase()
-        if uri.match(/http:\/\//) then uri = uri.substring(7)
-        if uri.match(/https:\/\//) then uri = uri.substring(8)
-        if uri.match(/^www\./) then uri = uri.substring(4)
-        uri
+        uri.toLowerCase()
       path: '/uri'
       exact_match: false
       case_sensitive: false
-      es_query_string: true
       and_or: 'or'
+      options:
+        es:
+          query_type: 'match'
+          cutoff_frequency: 0.001
+          and_or: 'and'
+
     since:
       formatter: (past) ->
         seconds =