Skip to content

Commit

Permalink
Improved performance of smart completions
Browse files Browse the repository at this point in the history 
On lengthy SQL statements with lots of identifiers and aliases
the performance was poor, this commit improves performance of
smart completions for most widely used case - when prefix
does not contain any dots (no parent object specified).
  • Loading branch information
@tkopets
tkopets committed May 26, 2017
1 parent a101c8e commit 0e529f6
Showing 1 changed file with 97 additions and 88 deletions.
185 changes: 97 additions & 88 deletions SQLToolsAPI/Completion.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -19,29 +19,31 @@
def _stripQuotes(ident):
return ident.strip('"\'`')


# used for formatting output
def _stripQuotesOnDemand(ident, doStrip=True):
if doStrip:
return _stripQuotes(ident)
return ident


def _startsWithQuote(ident):
# str.startswith can be matched against a tuple
quotes = ('`', '"')
return ident.startswith(quotes)


def _stripPrefix(text, prefix):
if text.startswith(prefix):
return text[len(prefix):]
return text


class CompletionItem(namedtuple('CompletionItem', ['type', 'ident', 'score'])):
class CompletionItem(namedtuple('CompletionItem', ['type', 'ident'])):
"""
Represents a potential or actual completion item.
* type - Type of item e.g. (Table, Function, Column)
* ident - identifier e.g. ("tablename.column", "database.table", "alias")
* score - the lower score, the better is match for completion item
"""
__slots__ = ()

Expand Down Expand Up @@ -100,9 +102,10 @@ def prefixMatchScore(self, search, exactly=False):
targetList = target.split('.')
targetObject = _stripQuotes(targetList.pop())
targetParent = _stripQuotes(targetList.pop())
if (searchParent == targetParent and
self._stringMatched(targetObject, searchObject, exactly)):
return 1 # highest score
if (searchParent == targetParent):
if self._stringMatched(targetObject, searchObject, exactly):
return 1 # highest score
return 0

# second part matches ?
if '.' in target:
Expand All @@ -119,6 +122,13 @@ def prefixMatchScore(self, search, exactly=False):
return 0
return 0

def prefixMatchListScore(self, searchList, exactly=False):
for item in searchList:
score = self.prefixMatchScore(item, exactly)
if score:
return score
return 0

# format completion item according to sublime text completions format
def format(self, stripQuotes=False):
typeDisplay = ''
Expand Down Expand Up @@ -147,9 +157,9 @@ def format(self, stripQuotes=False):

class Completion:
def __init__(self, uppercaseKeywords, allTables, allColumns, allFunctions):
self.allTables = [CompletionItem('Table', table, 0) for table in allTables]
self.allColumns = [CompletionItem('Column', column, 0) for column in allColumns]
self.allFunctions = [CompletionItem('Function', func, 0) for func in allFunctions]
self.allTables = [CompletionItem('Table', table) for table in allTables]
self.allColumns = [CompletionItem('Column', column) for column in allColumns]
self.allFunctions = [CompletionItem('Function', func) for func in allFunctions]

self.allKeywords = []
for keyword in keywords_list:
Expand All @@ -158,7 +168,7 @@ def __init__(self, uppercaseKeywords, allTables, allColumns, allFunctions):
else:
keyword = keyword.lower()

self.allKeywords.append(CompletionItem('Keyword', keyword, 0))
self.allKeywords.append(CompletionItem('Keyword', keyword))

def getBasicAutoCompleteList(self, prefix):
prefix = prefix.lower()
Expand All @@ -168,17 +178,17 @@ def getBasicAutoCompleteList(self, prefix):
for item in self.allColumns:
score = item.prefixMatchScore(prefix)
if score:
autocompleteList.append(CompletionItem(item.type, item.ident, score))
autocompleteList.append(item)

for item in self.allTables:
score = item.prefixMatchScore(prefix)
if score:
autocompleteList.append(CompletionItem(item.type, item.ident, score))
autocompleteList.append(item)

for item in self.allFunctions:
score = item.prefixMatchScore(prefix)
if score:
autocompleteList.append(CompletionItem(item.type, item.ident, score))
autocompleteList.append(item)

if len(autocompleteList) == 0:
return None
Expand Down Expand Up @@ -263,94 +273,93 @@ def _noDotsCompletions(self, prefix, identifiers, joinAlias=None):
Order: statement aliases -> statement cols -> statement tables -> statement functions,
then: other cols -> other tables -> other functions that match the prefix in their names
"""
# get join conditions
joinConditions = []
if joinAlias:
joinConditions = self._joinConditionCompletions(identifiers, joinAlias)

# use set, as we are interested only in unique identifiers
sqlAliases = set()
sqlTables = set()
sqlColumns = set()
sqlFunctions = set()
sqlTables = []
sqlColumns = []
sqlFunctions = []
otherTables = []
otherColumns = []
otherFunctions = []
otherKeywords = []
otherJoinConditions = []

# utilitary temp lists
identTables = set()
identColumns = set()
identFunctions = set()

for ident in identifiers:
if ident.has_alias():
sqlAliases.add(CompletionItem('Alias', ident.alias, 0))
aliasItem = CompletionItem('Alias', ident.alias)
score = aliasItem.prefixMatchScore(prefix)
if score and aliasItem.ident != prefix:
sqlAliases.add(aliasItem)

if ident.is_function:
functions = [
fun
for fun in self.allFunctions
if fun.prefixMatchScore(ident.full_name, exactly=True) > 0
]
sqlFunctions.update(functions)
else:
tables = [
table
for table in self.allTables
if table.prefixMatchScore(ident.full_name, exactly=True) > 0
]
sqlTables.update(tables)
prefixForColumnMatch = ident.name + '.'
columns = [
col
for col in self.allColumns
if col.prefixMatchScore(prefixForColumnMatch, exactly=True) > 0
]
sqlColumns.update(columns)

autocompleteList = []

for condition in joinConditions:
if condition.ident.lower().startswith(prefix):
autocompleteList.append(CompletionItem(condition.type, condition.ident, 1))

# first of all list aliases and identifiers related to currently parsed statement
for item in sqlAliases:
score = item.prefixMatchScore(prefix)
if score and item.ident != prefix:
autocompleteList.append(CompletionItem(item.type, item.ident, score))
identFunctions.add(ident.full_name)
elif ident.is_table_alias:
identTables.add(ident.full_name)
identColumns.add(ident.name + '.' + prefix)

for item in sqlColumns:
score = item.prefixMatchScore(prefix)
for table in self.allTables:
score = table.prefixMatchScore(prefix, exactly=False)
if score:
autocompleteList.append(CompletionItem(item.type, item.ident, score))
if table.prefixMatchListScore(identTables, exactly=True) > 0:
sqlTables.append(table)
else:
otherTables.append(table)

for item in sqlTables:
score = item.prefixMatchScore(prefix)
for col in self.allColumns:
score = col.prefixMatchScore(prefix, exactly=False)
if score:
autocompleteList.append(CompletionItem(item.type, item.ident, score))
if col.prefixMatchListScore(identColumns, exactly=False) > 0:
sqlColumns.append(col)
else:
otherColumns.append(col)

for item in sqlFunctions:
score = item.prefixMatchScore(prefix)
for fun in self.allFunctions:
score = fun.prefixMatchScore(prefix, exactly=False)
if score:
autocompleteList.append(CompletionItem(item.type, item.ident, score))
if fun.prefixMatchListScore(identFunctions, exactly=True) > 0:
sqlColumns.append(fun)
else:
otherColumns.append(fun)

# add keywords to auto-complete results
# keywords
for item in self.allKeywords:
score = item.prefixMatchScore(prefix)
if score:
autocompleteList.append(CompletionItem(item.type, item.ident, score))
otherKeywords.append(item)

# add the rest of the columns, tables and functions that also match the prefix
for item in self.allColumns:
score = item.prefixMatchScore(prefix)
if score:
if item not in sqlColumns:
autocompleteList.append(CompletionItem(item.type, item.ident, score))
# join conditions
if joinAlias:
joinConditions = self._joinConditionCompletions(identifiers, joinAlias)

for item in self.allTables:
score = item.prefixMatchScore(prefix)
if score:
if item not in sqlTables:
autocompleteList.append(CompletionItem(item.type, item.ident, score))
for condition in joinConditions:
if condition.ident.lower().startswith(prefix):
otherJoinConditions.append(condition)

for item in self.allFunctions:
score = item.prefixMatchScore(prefix)
if score:
if item not in sqlFunctions:
autocompleteList.append(CompletionItem(item.type, item.ident, score))
# collect the results in prefered order
autocompleteList = []

# first of all list join conditions (if applicable)
autocompleteList.extend(otherJoinConditions)

# then aliases and identifiers related to currently parsed statement
autocompleteList.extend(sqlAliases)

# then cols, tables, functions related to current statement
autocompleteList.extend(sqlColumns)
autocompleteList.extend(sqlTables)
autocompleteList.extend(sqlFunctions)

# then other matching cols, tables, functions
autocompleteList.extend(otherKeywords)
autocompleteList.extend(otherColumns)
autocompleteList.extend(otherTables)
autocompleteList.extend(otherFunctions)

return autocompleteList, False

Expand Down Expand Up @@ -390,7 +399,7 @@ def _singleDotCompletions(self, prefix, identifiers, joinAlias=None):
aliasPrefix = prefixParent + '.'
if condition.ident.lower().startswith(aliasPrefix):
autocompleteList.append(CompletionItem(condition.type,
_stripPrefix(condition.ident, aliasPrefix), 1))
_stripPrefix(condition.ident, aliasPrefix)))

# first of all expand table aliases to real table names and try
# to match their columns with prefix of these expanded identifiers
Expand All @@ -400,23 +409,23 @@ def _singleDotCompletions(self, prefix, identifiers, joinAlias=None):
for item in self.allColumns:
score = item.prefixMatchScore(prefix_to_match)
if score:
autocompleteList.append(CompletionItem(item.type, item.ident, score))
autocompleteList.append(item)

# try to match all our other objects (tables, columns, functions) with prefix
for item in self.allColumns:
score = item.prefixMatchScore(prefix)
if score:
autocompleteList.append(CompletionItem(item.type, item.ident, score))
autocompleteList.append(item)

for item in self.allTables:
score = item.prefixMatchScore(prefix)
if score:
autocompleteList.append(CompletionItem(item.type, item.ident, score))
autocompleteList.append(item)

for item in self.allFunctions:
score = item.prefixMatchScore(prefix)
if score:
autocompleteList.append(CompletionItem(item.type, item.ident, score))
autocompleteList.append(item)

inhibit = len(autocompleteList) > 0
# in case prefix parent is a query alias we simply don't know what those
Expand All @@ -432,7 +441,7 @@ def _multiDotCompletions(self, prefix, identifiers):
for item in self.allColumns:
score = item.prefixMatchScore(prefix)
if score:
autocompleteList.append(CompletionItem(item.type, item.ident, score))
autocompleteList.append(item)

if len(autocompleteList) > 0:
return autocompleteList, True
Expand All @@ -450,7 +459,7 @@ def _joinConditionCompletions(self, identifiers, joinAlias=None):

for ident in identifiers:
if ident.has_alias() and not ident.is_function:
sqlTableAliases.add(CompletionItem('Alias', ident.alias, 0))
sqlTableAliases.add(CompletionItem('Alias', ident.alias))

prefixForColumnMatch = ident.name + '.'
columns = [
Expand Down Expand Up @@ -486,7 +495,7 @@ def _joinConditionCompletions(self, identifiers, joinAlias=None):
sideA = joinAlias + '.' + joinColumn.name
sideB = otherAlias + '.' + otherColumn.name

joinCandidatesCompletions.append(CompletionItem('Condition', sideA + ' = ' + sideB, 0))
joinCandidatesCompletions.append(CompletionItem('Condition', sideB + ' = ' + sideA, 0))
joinCandidatesCompletions.append(CompletionItem('Condition', sideA + ' = ' + sideB))
joinCandidatesCompletions.append(CompletionItem('Condition', sideB + ' = ' + sideA))

return joinCandidatesCompletions

0 comments on commit 0e529f6

Please sign in to comment.