 Michaël Van Canneyt
|
54af8218fd
* PChar -> PAnsiChar
|
2 years ago |
 sergei
|
8afc5b8ebf
+ Basic implementation of TXPathNSResolver class, Mantis #25183.
|
12 years ago |
 florian
|
39aef35bdc
* pass not only exception address but also frame to raise:
|
13 years ago |
|
sergei
|
1876fb1c74
XPath: now when predicate filtering is no longer dependent on axis, it becomes possible to use the same code for TStep and TXPathFilterNode. Inherited TStep from TXPathFilterNode, removed duplicating code.
|
15 years ago |
|
sergei
|
ae7aef0861
XPath: reworked step processing:
|
15 years ago |
|
sergei
|
c902b96673
* XPath, change parsing of function call so that function arguments are parsed before creating the function node. This way function nodes can validate their arguments at creation time.
|
15 years ago |
|
sergei
|
5352b78c4a
* XPath, fixed parsing of 'prefix:*' node tests:
|
15 years ago |
|
sergei
|
0f5795baaf
XPath, use a perfect hash to recognize all possible keywords.
|
15 years ago |
|
sergei
|
83711ff972
XPath improvements:
|
15 years ago |
|
sergei
|
d44a1603db
* TXPathScanner.ParseStep split into two functions in order to reduce complexity and improve readability.
|
15 years ago |
|
sergei
|
3bb9bfd2b8
* XPath, fixed parent axis of attribute nodes: it must consist of the owner element of an attribute.
|
15 years ago |
|
sergei
|
1ed130e2e1
* Fixed XPath functions name() and local-name(), which should behave different from DOM properties of the same names. name() is empty for text, comment and document nodes. local-name() is the same as name() for non-prefixed attributes and processing instructions.
|
15 years ago |
|
sergei
|
d44c6876cd
xpath.pp:
|
16 years ago |
|
sergei
|
a283bc4f46
xpath.pp:
|
16 years ago |
|
sergei
|
7e731c78b6
xpath.pp: cleaned up and simplified TXPathScanner.GetToken()
|
16 years ago |
|
sergei
|
3f5ce0d228
xpath.pp: Fixed runtime range and overflow checks, at least those which could be detected with existing test suite.
|
16 years ago |
|
sergei
|
2a924d0335
* For some reason, any assignment of NaN cannot be compiled with overflow checking, issue #14748.
|
16 years ago |
 Jonas Maebe
|
2db72437e8
* fixed range check errors
|
16 years ago |
|
sergei
|
f748a01baf
* Once the #14143 is fixed, I may (and actually must) remove the related workarounds.
|
16 years ago |
|
sergei
|
a9d9841eee
* XPath: number-to-string conversion is now compliant to the specs (no scientific notation, decimal separator is a hardcoded period, correct output values for INF/NAN) + tests.
|
16 years ago |
|
Jonas Maebe
|
1f6ec379de
* moved field definitions before method/property definitions (see mantis
|
16 years ago |
|
sergei
|
106a605e05
xpath.pp:
|
16 years ago |
|
sergei
|
a45049155e
XPath parser improvements:
|
16 years ago |
|
sergei
|
538f82091a
XPath improvements:
|
16 years ago |
|
sergei
|
26767125f0
* Refactored TXPathLocationPathNode.Evaluate(), split that awful 3-level nested procedure into
|
16 years ago |
|
sergei
|
e66e35ac2c
XPath: reducing memory load:
|
16 years ago |
|
sergei
|
448b621ff4
xpath.pp: A slightly more efficient implementation of TXPathCompareNode.Evaluate().
|
16 years ago |
|
sergei
|
7a478e64c2
xpath.pp: Honor axis direction when filtering step result with predicates. For reverse axes, ContextPosition is now relative to the last node of the set. The third argument of TXPathExprNode.EvalPredicate() is redundant and has been removed.
|
16 years ago |
|
sergei
|
4cabfad02a
xpath.pp: Fixed lexical analysis in ParseStep(), now conformant with paragraph 3.7 of the specs:
|
16 years ago |
|
sergei
|
f3e64cf8ce
xpath.pp: replaced TXPathScanner.SaveState/RestoreState by PeekToken().
|
16 years ago |
