def genParseObject(self,cLocale):
sym = DateFormatSymbols(cLocale)
val = sym.getEras()
frag = Or(CaselessLiteral(val[0].encode('utf-8')))
for item in val[1:]:
frag.append(CaselessLiteral(item.encode('utf-8')))
return frag.setResultsName('era')
ddlWord("KEY"),
ddlWord("INDEX"),
ddlWord("UNIQUE"),
ddlWord("CHECK"),
])
ddlColumn = Group(Optional(ddlConstraint).setResultsName(
"isConstraint") + OneOrMore(MatchFirst([
ddlUnsigned, ddlNotNull, ddlAutoValue, ddlDefaultValue,
ddlFunctionWord("NOW"), ddlTerm, ddlNum, ddlColumnComment, ddlString,
ddlArguments, ddlMathCombined, ddlInlineComment
])))
ddlIfNotExists = Optional(Group(ddlWord("IF") + ddlWord("NOT") + ddlWord(
"EXISTS")).setResultsName("ifNotExists"))
createTable = Group(ddlWord("CREATE") + ddlWord("TABLE") + ddlIfNotExists +
ddlName.setResultsName("tableName") + "(" + Group(
delimitedList(ddlColumn)).setResultsName(
"columns") + ")").setResultsName("create")
#ddlString.setDebug(True) #uncomment to debug pyparsing
ddl = ZeroOrMore(Suppress(SkipTo(createTable, False)) + createTable)
ddlComment = oneOf(["--", "#"]) + restOfLine
ddl.ignore(ddlComment)
# MAP SQL TYPES
types = {
'tinyint': 'tinyint',
'smallint': 'smallint',
'smallserial': 'smallint', # PostgreSQL
'integer': 'integer',
# S -> WHERE C EOF
stringLit = Group('"' + w + '"')
intLit = Word(nums)
C = Forward()
T = ((stringLit | intLit | w) + oneOf(_ops) +
(stringLit | intLit | w)) | (Suppress("(") + C + Suppress(")"))
C << ((T + combinators + T) | T)
bvalues = Group(Suppress("(") + values + Suppress(")"))
svalues = (bvalues | '*').setResultsName("columns")
where_clause = CaselessLiteral("WHERE") + C
where = Optional(where_clause, "").setResultsName("where")
op = operators.setResultsName("op")
sgram = op + svalues + where + StringEnd()
insert = CaselessLiteral("INSERT").setResultsName(
"op") + bvalues.setResultsName('ivalues') + StringEnd()
delete = CaselessLiteral("DELETE").setResultsName("op") + where + StringEnd()
stmt = sgram | insert | delete
def parse_sql(s):
d = {'stmt': s}
try:
r = stmt.parseString(s)
except pyPE, e:
raise ParseException(e)
o = r.op.lower()
if o == "insert":
qualified_identifier = Word(alphas + "_.", alphas + alphanums + "_.")
BASIC_TYPES = ["bool", "i8", "u8", "i16", "u16", "i32", "u32", "i64", "u64"]
basic_type = Or([Keyword(t) for t in BASIC_TYPES])
bit_width = Word(nums)
dec_literal = Word(nums)
hex_literal = Combine("0x" + Word(hexnums))
comment = cppStyleComment
field = Group(
Optional(comment).setResultsName("doc") +
identifier.setResultsName("name") + ':' +
basic_type.setResultsName("type") +
Optional(':' + bit_width.setResultsName("width")) +
';'
)
struct = originalTextFor(
Group(
Optional(comment).setResultsName("doc") +
Keyword("struct") +
identifier.setResultsName("name") + "{" +
OneOrMore(field.setResultsName("fields", listAllMatches=True)) + "}"
), asString=False
)
vector = Group(
Keyword("vector") + "<" + qualified_identifier.setResultsName("type") + ">"
bom = Regex(BOM)
msg_utf8 = bom + utf_8_string
msg_any = utf_8_string
msg = Combine(Or([msg_utf8, msg_any])).setResultsName("MSG")
sd_name = CharsNotIn('= ]"', 1, 32)
param_name = sd_name.setResultsName("SD_PARAM_NAME")
param_value = QuotedString(quoteChar='"', escChar="\\", multiline=True)
param_value = param_value.setResultsName("SD_PARAM_VALUE")
sd_id = sd_name.setResultsName("SD_ID")
sd_param = Group(param_name + Regex("=") + param_value)
sd_params = Group(ZeroOrMore(Group(sp + sd_param.setResultsName("SD_PARAM"))))
sd_element = Group("[" + sd_id + sd_params.setResultsName("SD_PARAMS") + "]")
sd_element = sd_element.setResultsName("SD_ELEMENT")
sd_elements = Group(OneOrMore(sd_element))
structured_data = Or([nilvalue, sd_elements.setResultsName("SD_ELEMENTS")])
structured_data = structured_data.setResultsName("STRUCTURED_DATA")
time_hour = Regex("0[0-9]|1[0-9]|2[0-3]")
time_minute = Regex("[0-5][0-9]")
time_second = time_minute
time_secfrac = Regex("\.[0-9]{1,6}")
time_numoffset = Or([Regex("\+"), Regex("-")]) + time_hour + ":" + time_minute
time_offset = Or([Regex("Z"), time_numoffset])
partial_time = time_hour + ":" + time_minute + ":" + time_second + Optional(time_secfrac)
full_time = partial_time + time_offset
date_mday = Regex("[0-9]{2}")
date_month = Regex("0[1-9]|1[0-2]")
date_fullyear = Regex("[0-9]{4}")
full_date = date_fullyear + "-" + date_month + "-" + date_mday
timestamp = Combine(Or([nilvalue, full_date + "T" + full_time]))
timestamp = timestamp.setResultsName("TIMESTAMP")
msgid = Or([nilvalue, CharsNotIn('= ]"', 1, 32)])
bom = Regex(BOM)
msg_utf8 = bom + utf_8_string
msg_any = utf_8_string
msg = Combine(Or([msg_utf8, msg_any])).setResultsName('MSG')
sd_name = CharsNotIn('= ]"', 1, 32)
param_name = sd_name.setResultsName('SD_PARAM_NAME')
param_value = QuotedString(quoteChar='"', escChar='\\', multiline=True)
param_value = param_value.setResultsName('SD_PARAM_VALUE')
sd_id = sd_name.setResultsName('SD_ID')
sd_param = Group(param_name + Regex('=') + param_value)
sd_params = Group(ZeroOrMore(Group(sp + sd_param.setResultsName('SD_PARAM'))))
sd_element = Group('[' + sd_id + sd_params.setResultsName('SD_PARAMS') + ']')
sd_element = sd_element.setResultsName('SD_ELEMENT')
sd_elements = Group(OneOrMore(sd_element))
structured_data = Or([nilvalue, sd_elements.setResultsName('SD_ELEMENTS')])
structured_data = structured_data.setResultsName('STRUCTURED_DATA')
time_hour = Regex('0[0-9]|1[0-9]|2[0-3]')
time_minute = Regex('[0-5][0-9]')
time_second = time_minute
time_secfrac = Regex('\.[0-9]{1,6}')
time_numoffset = Or([Regex('\+'), Regex('-')]) + time_hour + ':' + time_minute
time_offset = Or([Regex('Z'), time_numoffset])
partial_time = time_hour + ':' + time_minute + ':' + time_second + \
Optional(time_secfrac)
full_time = partial_time + time_offset
date_mday = Regex('[0-9]{2}')
date_month = Regex('0[1-9]|1[0-2]')
date_fullyear = Regex('[0-9]{4}')
full_date = date_fullyear + '-' + date_month + '-' + date_mday
timestamp = Combine(Or([nilvalue, full_date + 'T' + full_time]))
timestamp = timestamp.setResultsName('TIMESTAMP')
# 13.1
# DefinitiveOID = "{" + DefinitiveObjIdComponentList + "}"
# 13.1
DefinitiveIdentification = Or([
# DefinitiveOID,
# DefinitiveOIDandIRI,
empty
])
# 13.1
ModuleIdentifier = And([
modulereference,
# DefinitiveIdentification
])
# 13.1
ModuleDefinition = And([
ModuleIdentifier.setResultsName("ModuleIdentifier"),
DEFINITIONS,
EncodingReferenceDefault.setResultsName("EncodingReferenceDefault"),
TagDefault.setResultsName("TagDefault"),
ExtensionDefault.setResultsName("ExtensionDefault"),
assignment,
BEGIN,
Group(ModuleBody).setResultsName("ModuleBody"),
# EncodingControlSection + \
END
]).setParseAction(parse_ModuleDefinition)
bom = Regex(BOM)
msg_utf8 = bom + utf_8_string
msg_any = utf_8_string
msg = Combine(Or([msg_utf8, msg_any])).setResultsName('MSG')
sd_name = CharsNotIn('= ]"', 1, 32)
param_name = sd_name.setResultsName('SD_PARAM_NAME')
param_value = QuotedString(quoteChar='"', escChar='\\', multiline=True)
param_value = param_value.setResultsName('SD_PARAM_VALUE')
sd_id = sd_name.setResultsName('SD_ID')
sd_param = Group(param_name + Regex('=') + param_value)
sd_params = Group(ZeroOrMore(Group(sp+sd_param.setResultsName('SD_PARAM'))))
sd_element = Group('['+sd_id+sd_params.setResultsName('SD_PARAMS')+']')
sd_element = sd_element.setResultsName('SD_ELEMENT')
sd_elements = Group(OneOrMore(sd_element))
structured_data = Or([nilvalue, sd_elements.setResultsName('SD_ELEMENTS')])
structured_data = structured_data.setResultsName('STRUCTURED_DATA')
time_hour = Regex('0[0-9]|1[0-9]|2[0-3]')
time_minute = Regex('[0-5][0-9]')
time_second = time_minute
time_secfrac = Regex('\.[0-9]{1,6}')
time_numoffset = Or([Regex('\+'), Regex('-')]) + \
time_hour + ':' + time_minute
time_offset = Or([Regex('Z'), time_numoffset])
partial_time = time_hour + ':' + time_minute + ':' + time_second + \
Optional(time_secfrac)
full_time = partial_time + time_offset
date_mday = Regex('[0-9]{2}')
date_month = Regex('0[1-9]|1[0-2]')
date_fullyear = Regex('[0-9]{4}')
full_date = date_fullyear + '-' + date_month + '-' + date_mday
timestamp = Combine(Or([nilvalue, full_date + 'T' + full_time]))
dec_literal = Word(nums)
hex_literal = Combine("0x" + Word(hexnums))
signed_literal = Combine(Optional('-') + (dec_literal ^ hex_literal))
comment = cppStyleComment
enumValue = Group(
Optional(comment).setResultsName("doc") +
identifier.setResultsName("name") +
Optional('=' + signed_literal.setResultsName("constant")))
enum = originalTextFor(Group(
Optional(comment).setResultsName("doc") + Keyword("enum") +
identifier.setResultsName("name") + ':' +
basic_type.setResultsName("type") + '{' + delimitedList(
enumValue.setResultsName("enum_values", listAllMatches=True), ",") +
Optional(',') + '}'),
asString=False)
field = Group(
Optional(comment).setResultsName("doc") +
identifier.setResultsName("name") + ':' +
qualified_identifier.setResultsName("type") +
Optional(':' + bit_width.setResultsName("width")) + ';')
struct = originalTextFor(Group(
Optional(comment).setResultsName("doc") + Keyword("struct") +
identifier.setResultsName("name") + "{" +
OneOrMore(field.setResultsName("fields", listAllMatches=True)) + "}"),
asString=False)
ddlWord("SMALLSERIAL"),
ddlWord("SERIAL"),
ddlWord("BIGSERIAL"),
]).setResultsName("hasAutoValue")
ddlColumnComment = Group(ddlWord("COMMENT") + ddlString).setResultsName("comment")
ddlConstraint = Or([
ddlWord("CONSTRAINT"),
ddlWord("PRIMARY"),
ddlWord("FOREIGN"),
ddlWord("KEY"),
ddlWord("INDEX"),
ddlWord("UNIQUE"),
])
ddlColumn = Group(Optional(ddlConstraint).setResultsName("isConstraint") + OneOrMore(MatchFirst([ddlNotNull, ddlAutoValue, ddlDefaultValue, ddlFunctionWord("NOW"), ddlTerm, ddlNum, ddlColumnComment, ddlString, ddlArguments])))
ddlIfNotExists = Optional(Group(ddlWord("IF") + ddlWord("NOT") + ddlWord("EXISTS")).setResultsName("ifNotExists"))
createTable = Group(ddlWord("CREATE") + ddlWord("TABLE") + ddlIfNotExists + ddlName.setResultsName("tableName") + "(" + Group(delimitedList(ddlColumn)).setResultsName("columns") + ")").setResultsName("create")
#ddlString.setDebug(True) #uncomment to debug pyparsing
ddl = ZeroOrMore(Suppress(SkipTo(createTable, False)) + createTable)
ddlComment = oneOf(["--", "#"]) + restOfLine
ddl.ignore(ddlComment)
# MAP SQL TYPES
types = {
'tinyint': 'tinyint',
'smallint': 'smallint',
'smallserial': 'smallint', # PostgreSQL
'integer': 'integer',
'int': 'integer',
'serial': 'integer', # PostgreSQL