Find file
Fetching contributors…
Cannot retrieve contributors at this time
378 lines (354 sloc) 15.7 KB
-- Copyright 2016 Roland Bouman,
-- Licensed under the Apache License, Version 2.0 (the "License");
-- you may not use this file except in compliance with the License.
-- You may obtain a copy of the License at
-- Unless required by applicable law or agreed to in writing, software
-- distributed under the License is distributed on an "AS IS" BASIS,
-- See the License for the specific language governing permissions and
-- limitations under the License.
drop PROCEDURE p_parse_xml
-- Parses xml and returns a dom like parse tree as a table variable.
-- Primary use case of this procedure is parsing model XML stored in the _SYS_REPO.ACTIVE_OBJECT table.
-- XML definitions of analytical, attribute and calculation views is stored here.
-- This procedure offers at least some kind of way to inspect and query the contents of these views.
-- (Since SAP HANA system views and tools like HANA studio offer only limited facilities for tracking and managing dependecies of these types of views)
-- With the primary use case in mind, please note that this is not a fully featured xml parser.
-- A non-exhaustive list of limitations follows below:
-- - It does not do any actual error checking - not even a well-formedness check, let alone validation.
-- - There is no support for namespaces (that is, element and attributes are returned as is without checking or parsing namespace uri or prefix.
-- - There is no resolving of external entities. Only basic entity replacement is performed using p_decode_xml_entities.
-- - It is quite slow.
-- However, it does keep track of the tree structure, provided a well-formed xml document is passed.
create PROCEDURE p_parse_xml (
p_xml nclob
, out p_dom table (
node_id int -- unique id of the node
, parent_node_id int -- id of the parent node
, node_type tinyint -- dom node type constant: 1=element, 2=attribute, 3=text, 4=cdata, 5=entityref, 6=entity, 7=processing instruction, 8=comment, 9=document, 10=document type, 11=document fragment, 12=notation
, node_name nvarchar(64) -- dom node name: tagname for element, attribute name for attribute, target for processing instruction, document type name for document type, "#text" for text and cdata, "#comment" for comment, "#document" for document, "#document-fragment" for document fragment.
, node_value nclob -- dom node value: text for text, comment, and cdata nodes, data for processing instruction node, null otherwise.
, pos int -- character position of token
, len int -- lenght of token.
, out p_error table (
error_code int
, error_message nvarchar(255)
, position int
, node_name nvarchar(64)
, p_strip_empty_text tinyint default 1
reads sql data
-- default regexp flag: s: . includes newline; m: ^ and $ match start/end of input (not of line)
declare RX_FLAG CONSTANT nchar(2) default 'sm';
-- less than
declare RX_LT CONSTANT nchar(1) default '<';
declare RX_LT_LEN CONSTANT tinyint default length(RX_LT);
-- greater than
declare RX_GT CONSTANT nchar(1) default '>';
-- single quote
declare RX_APOS CONSTANT nchar(8) default '''';
-- double quote
declare RX_QUOT CONSTANT nchar(8) default '"';
-- name start char
declare RX_NSCHAR CONSTANT nchar(27) default ':_A-Za-z\xC0-\xD6\xD8-\xF6';
-- name char
declare RX_NCHAR CONSTANT nchar(38) default '\-\.'||RX_NSCHAR||'0-9\xB7';
-- name: one or more word characters. (TODO: xml probably allows more chars than \w)
declare RX_NAME CONSTANT nchar(70) default '['||RX_NSCHAR||']['||RX_NCHAR||']*';
-- qualified name: name with optional prefix
declare RX_QNAME CONSTANT nchar(146) default '(('||RX_NAME||':)?'||RX_NAME||')';
-- whitespace
declare RX_WS CONSTANT nchar(3) default '\s+';
-- optional whitespace
declare RX_OPTWS CONSTANT nchar(3) default '\s*';
-- single quoted string: single quote, followed by anything byt a single quote or a left angle parenthesis, followed by single quote
declare RX_SQSTR CONSTANT nchar(8) default RX_APOS||'[^'||RX_APOS||RX_LT||']*'||RX_APOS;
-- double quoted string: double quote, followed by anything but a double quote or a left angle parenthesis, followed by double quote
declare RX_DQSTR CONSTANT nchar(8) default RX_QUOT||'[^'||RX_QUOT||RX_LT||']*'||RX_QUOT;
-- quoted string: either a double or a single quoted string
declare RX_QSTR CONSTANT nchar(19) default '('||RX_SQSTR||'|'||RX_DQSTR||')';
-- attribute: whitespace, qname, optional whitespace, equals sign, optional whitespace, quoted string.
declare RX_ATT CONSTANT nchar(178) default '('||RX_WS||RX_QNAME||RX_OPTWS||'='||RX_OPTWS||RX_QSTR||')';
-- literal question mark (used in pi)
declare RX_Q CONSTANT nchar(2) default '\?';
-- start pi
declare RX_SPI CONSTANT nchar(3) default RX_LT||RX_Q;
-- end pi
declare RX_EPI CONSTANT nchar(3) default RX_Q||RX_GT;
-- processing instruction: name, mandatory whitespace, followed by anything that is not a end pi delimiter
declare RX_PI CONSTANT nchar(193) default RX_SPI||'('||RX_NAME||')('||RX_WS||'.*(?<!'||RX_EPI||'))'||RX_EPI;
-- dash
declare RX_DASH CONSTANT nchar(1) default '-';
-- dashdash
declare RX_DASHDASH CONSTANT nchar(2) default RX_DASH||RX_DASH;
-- start comment
declare RX_SCOMM CONSTANT nchar(4) default RX_LT||'!'||RX_DASHDASH;
declare RX_SCOMM_LEN CONSTANT tinyint default length(RX_SCOMM);
-- end comment
declare RX_ECOMM CONSTANT nchar(3) default RX_DASHDASH||RX_GT;
declare RX_ECOMM_LEN CONSTANT tinyint default length(RX_ECOMM);
-- no dash
declare RX_NODASH CONSTANT nchar(4) default '[^-]';
-- comment:
declare RX_COMMENT CONSTANT nchar(20) default RX_SCOMM||'('||RX_NODASH||'|'||RX_DASH||RX_NODASH||')*'||RX_ECOMM;
-- start cdata:
declare RX_SCDATA CONSTANT nchar(11) default RX_LT||'!\[CDATA\[';
declare RX_SCDATA_LEN CONSTANT tinyint default length(RX_SCDATA);
-- end cdata:
declare RX_ECDATA CONSTANT nchar(5) default '\]\]'||RX_GT;
declare RX_ECDATA_LEN CONSTANT tinyint default length(RX_ECDATA);
-- cdata
declare RX_CDATA CONSTANT nchar(28) default RX_SCDATA||'.*(?<!'||RX_ECDATA||')'||RX_ECDATA;
-- external id
declare RX_EXTID CONSTANT nchar(61) default '((SYSTEM|PUBLIC'||RX_WS||RX_QSTR||')'||RX_WS||RX_QSTR||')';
-- doctype start
declare RX_SDOCTYPE CONSTANT nchar(9) default RX_LT||'!DOCTYPE';
-- doctype
declare RX_DOCTYPE CONSTANT nchar(155) default RX_SDOCTYPE||RX_WS||'('||RX_NAME||')('||RX_WS||RX_EXTID||')?'||RX_OPTWS||RX_GT;
declare RX_DOCTYPE_LEN CONSTANT tinyint default length(RX_SDOCTYPE);
-- opening tag
declare RX_STAG CONSTANT nchar(334) default RX_LT||RX_QNAME||'('||RX_ATT||'*)'||RX_OPTWS||'/?'||RX_GT;
-- opening tag
declare RX_CTAG CONSTANT nchar(149) default RX_LT||'/'||RX_QNAME||RX_GT;
-- text: any content between > and <
declare RX_TEXT CONSTANT nchar(16) default '(?<='||RX_GT||')[^'||RX_LT||']+(?='||RX_LT||')';
declare REGXP CONSTANT nchar(901) default RX_PI
declare CLOSE_ELEMENT CONSTANT tinyint default 0; -- pseudo-nodetype used to signal closing element
-- DOM node types.
declare ELEMENT_NODE CONSTANT tinyint default 1;
declare ATTRIBUTE_NODE CONSTANT tinyint default 2;
declare TEXT_NODE CONSTANT tinyint default 3;
declare CDATA_SECTION_NODE CONSTANT tinyint default 4;
declare ENTITY_REFERENCE_NODE CONSTANT tinyint default 5;
declare ENTITY_NODE CONSTANT tinyint default 6;
declare COMMENT_NODE CONSTANT tinyint default 8;
declare DOCUMENT_NODE CONSTANT tinyint default 9;
declare DOCUMENT_TYPE_NODE CONSTANT tinyint default 10;
declare DOCUMENT_FRAGMENT_NODE CONSTANT tinyint default 11;
declare NOTATION_NODE CONSTANT tinyint default 12;
declare v_node_id integer default 0;
declare v_element_stack integer array;
declare v_parent_node_id integer default 0;
declare v_node_type tinyint;
declare v_node_name nvarchar(64);
declare v_node_value nclob;
declare v_chars char(12);
declare v_index integer default 1;
declare v_length integer;
declare v_end integer default length(:p_xml);
declare v_token nclob;
declare v_atts nclob;
declare v_atts_index integer;
declare v_atts_length integer;
declare v_atts_end integer;
declare v_att nclob;
declare v_att_name nvarchar(64);
declare v_att_value nclob;
declare v_row_num integer default 1;
declare v_node_ids integer array;
declare v_parent_node_ids integer array;
declare v_node_types tinyint array;
declare v_node_names nvarchar(64) array;
declare v_node_values nclob array;
declare v_positions integer array;
declare v_lengths integer array;
declare exit handler for sqlexception
p_error = select
::SQL_ERROR_CODE error_code
, ::SQL_ERROR_MESSAGE error_message
, v_index position
, v_node_name node_name
from dummy;
p_dom = unnest(
, :v_parent_node_ids
, :v_node_types
, :v_node_names
, :v_node_values
, :v_positions
, :v_lengths
) as dom (
, parent_node_id
, node_type
, node_name
, node_value
, pos
, len
v_element_stack[v_row_num] = v_node_id;
v_node_ids[v_row_num] = v_node_id;
v_parent_node_ids[v_row_num] = null;
v_node_types[v_row_num] = DOCUMENT_NODE;
v_node_names[v_row_num] = '#document';
v_node_values[v_row_num] = null;
v_positions[v_row_num] = 1;
v_lengths[v_row_num] = v_end;
while v_index < v_end do
select substr_regexpr(REGXP flag RX_FLAG in :p_xml from v_index)
into v_token
from dummy;
v_length = length(v_token);
v_node_id = v_node_id + 1;
v_parent_node_id = :v_element_stack[CARDINALITY(:v_element_stack)];
v_node_name = null;
v_node_type = null;
v_node_value = null;
v_atts = null;
if v_token is null then
-- check for whitespace trailing the document
select substr_regexpr('\s+$' flag RX_FLAG in :p_xml from v_index)
into v_token
from dummy;
if length(v_token) > 0 then
v_index = v_index + length(v_token);
v_node_type = 0;
signal sql_error_code 10000
set message_text = 'No token found at '||cast(v_index as varchar(12));
end if;
elseif left(v_token, RX_LT_LEN) = RX_LT then
if substr(v_token, 2, 1) = '?' then
select substr_regexpr(RX_PI flag RX_FLAG in v_token group 1)
, substr_regexpr(RX_PI flag RX_FLAG in v_token group 2)
into v_node_name
, v_node_value
from dummy;
v_atts = v_node_value;
elseif left(v_token, RX_SCOMM_LEN) = RX_SCOMM then
v_node_type = COMMENT_NODE;
v_node_name = '#comment';
v_node_value = substr(v_token, RX_SCOMM_LEN + 1, v_length - RX_SCOMM_LEN - RX_ECOMM_LEN);
elseif left(v_token, RX_SCDATA_LEN) = RX_SCDATA then
v_node_type = CDATA_SECTION_NODE;
v_node_name = '#text';
v_node_value = substr(v_token, RX_SCDATA_LEN + 1, v_length - RX_SCDATA_LEN - RX_ECDATA_LEN);
elseif left(v_token, RX_DOCTYPE_LEN) = RX_SDOCTYPE then
v_node_type = DOCUMENT_TYPE_NODE;
select substr_regexpr(RX_DOCTYPE flag RX_FLAG in v_token group 1)
into v_node_name
from dummy;
elseif substr(v_token, 2, 1) = '/' then
v_node_type = CLOSE_ELEMENT;
v_node_id = v_node_id - 1;
v_element_stack = trim_array(:v_element_stack, 1);
v_node_type = ELEMENT_NODE;
select substr_regexpr(RX_STAG in v_token group 1)
, substr_regexpr(RX_STAG in v_token group 3)
into v_node_name
, v_atts
from dummy;
v_chars = substr(v_token, v_length - 1, 1);
if v_chars != '/' then
v_element_stack[CARDINALITY(:v_element_stack) + 1] = v_node_id;
end if;
end if;
v_node_type = TEXT_NODE;
v_node_name = '#text';
call p_decode_xml_entities(
, v_node_value
end if;
-- lose non-significant whitespace.
if p_strip_empty_text != 0 and v_node_type = TEXT_NODE then
select case count(*) when 1 then 0 else v_node_type end
, v_node_id - count(*)
into v_node_type, v_node_id
from dummy
where replace_regexpr ('^\s+$' flag RX_FLAG in v_token with '') = '';
end if;
if v_node_type > 0 then
v_row_num = v_row_num + 1;
v_node_ids[v_row_num] = v_node_id;
v_parent_node_ids[v_row_num] = v_parent_node_id;
v_node_types[v_row_num] = v_node_type;
v_node_names[v_row_num] = v_node_name;
v_node_values[v_row_num] = v_node_value;
v_positions[v_row_num] = v_index;
v_lengths[v_row_num] = v_length;
if not v_atts is null then
v_parent_node_id = v_node_id;
v_atts_index = 1;
v_atts_end = length(v_atts);
while v_atts_index < v_atts_end do
select substr_regexpr(RX_ATT flag RX_FLAG in v_atts from v_atts_index group 1)
, substr_regexpr(RX_ATT flag RX_FLAG in v_atts from v_atts_index group 2)
, substr_regexpr(RX_ATT flag RX_FLAG in v_atts from v_atts_index group 4)
into v_att, v_att_name, v_att_value
from dummy;
v_atts_length = length(v_att);
if v_att_name is null then
signal sql_error_code 10000
set message_text = 'No attribute found in '||v_atts||' at index '||cast(v_atts_index as varchar(12));
v_node_id = v_node_id + 1;
call p_decode_xml_entities(
substr(v_att_value, 2, length(v_att_value) - 2)
, v_att_value
v_row_num = v_row_num + 1;
v_node_ids[v_row_num] = v_node_id;
v_parent_node_ids[v_row_num] = v_parent_node_id;
v_node_types[v_row_num] = ATTRIBUTE_NODE;
v_node_names[v_row_num] = v_att_name;
v_node_values[v_row_num] = v_att_value;
v_positions[v_row_num] = v_index + v_atts_index;
v_lengths[v_row_num] = v_atts_length;
end if;
v_atts_index = v_atts_index + v_atts_length;
end while;
end if;
end if;
v_index = v_index + v_length;
end while;
p_dom = unnest(
, :v_parent_node_ids
, :v_node_types
, :v_node_names
, :v_node_values
, :v_positions
, :v_lengths
) as dom (
, parent_node_id
, node_type
, node_name
, node_value
, pos
, len