Jonas Fonseca
00c4e0bfa2
Do not attempt to read *string when string == scanner->end
...
There might be other places that needs to be reviewd for this.
2006-01-28 03:23:06 +01:00
Jonas Fonseca
d92a074e40
Fix parsing of '<a< b>' where the scanner didn't rewind to the proper place
...
Add test for this tag soup combo.
2006-01-28 03:21:27 +01:00
Jonas Fonseca
bccf5512d6
Force an incomplete token for quoted attribute values when there's no end
2006-01-28 00:56:48 +01:00
Jonas Fonseca
9d91994f3c
Propone updating the scanner->state until incompleteness has been checked
...
That way the scanner state is meaningful when resuming during incremental
parsing.
2006-01-27 07:41:42 +01:00
Jonas Fonseca
afb45aace5
Add support for scanning comment endings such as '--!>' correctly
2006-01-25 18:18:01 +01:00
Jonas Fonseca
acb1f7e74d
Refactor computation of scanner error string length to get_sgml_error_end()
2006-01-07 23:51:19 +01:00
Jonas Fonseca
534a16fff1
Improve error detection
2006-01-07 23:40:21 +01:00
Jonas Fonseca
3835bf8449
A handful of fixes related to error detection
...
- Fix assertion failure by breaking the switch if an error token is next
when previous was a processing instruction.
- Fix <!notation parsing by skipping ident chars instead of spaces.
- Improve checking of processing instruction 'target'-end and what error
string is generated.
- For now put all of the processing instruction data in the error token.
- Remove a DBG()-print.
2006-01-07 05:18:43 +01:00
Jonas Fonseca
c993a0012e
Add basic support for detection errors while scanning
...
It mostly uses the checking for incompleteness already in place. Tested
lightly so it will definately need some more work.
2006-01-07 04:26:08 +01:00
Jonas Fonseca
7ff2cb2607
Improve a comment a bit
2006-01-07 01:41:07 +01:00
Jonas Fonseca
f8d44ffe32
scan_sgml_tokens(): Drop local variable and use scanner->current
...
... so lower level scanners can change the next token to use.
2006-01-07 01:25:42 +01:00
Jonas Fonseca
f75ccffbc7
Fix SGML parsing and scanning so that all tests succeeds
...
This includes checking the return token of get_next_dom_scanner_token() and
fixing the calculated size of recovered processing instruction data tokens.
2006-01-02 21:04:51 +01:00
Jonas Fonseca
e78d43f1ac
Add mode where the SGML scanner checks for completeness
2006-01-02 17:46:09 +01:00
Jonas Fonseca
fcf7677584
Skip spaces immediately when recognising '<?ident'
2006-01-02 16:58:48 +01:00
Jonas Fonseca
58c31f44a0
Clearify the code a bit
2006-01-02 03:06:47 +01:00
Jonas Fonseca
6e9a18b444
fix a few bugs for line counting in plain text
2006-01-02 01:49:12 +01:00
Jonas Fonseca
4a766f350b
Just for fun also parse <?xml-stylesheet attributes
2005-12-31 03:13:39 +01:00
Jonas Fonseca
a578ed4667
Make the SGML scanner (optionally) keep track of line numbers
...
A new line is either \n or \f. The main logic for counting lines is in
skip_sgml{,_chars,_space}. For the general case where line numbers are not
wanted the code tries to avoid the extra checks for newlines.
This will be useful for reporting errors when loading the XBEL file.
2005-12-31 02:46:56 +01:00
Jonas Fonseca
b23beed031
Rename skip_comment() and skip_cdata_section() to conform to skip_sgml_*()
2005-12-31 02:00:09 +01:00
Jonas Fonseca
0891cda51e
Introduce skip_sgml_space() that wraps scan_sgml(..., SGML_SCAN_WHITESPACE)
2005-12-31 01:57:54 +01:00
Jonas Fonseca
7489c134f7
Make non-terminated comments and cdata sections have 'the rest' as content
2005-12-31 01:47:57 +01:00
Jonas Fonseca
8f7f6abc16
Use skip_sgml_chars() in skip_comment() and skip_cdata_section()
2005-12-31 01:40:52 +01:00
Jonas Fonseca
4e10bcf772
Drop useless code for proc. instruction scanning
2005-12-31 01:18:49 +01:00
Jonas Fonseca
e8ff8bd5f0
Fix another off-by-one error similar to the SGML comment parsing
2005-12-31 01:14:52 +01:00
Jonas Fonseca
ab7ba39d42
Introduce skip_sgml_chars() to avoid usage of memchr()
2005-12-31 00:06:12 +01:00
Jonas Fonseca
76a524ddf6
More <?xml and comment tests, fix an off-by-one error for comments skipping
2005-12-29 22:26:39 +01:00
Jonas Fonseca
bd877570d2
Test some more obscure proc. instructions and fix some assertion failures
2005-12-29 21:52:27 +01:00
Jonas Fonseca
57168e1fbc
Handle <element path=/to/%61-&\one";files/> as a self-closing tag
...
Before the '/' before '>' would be interpreted as part of the attribute
value. Hope this is sensible slurping of the markup soup.
2005-12-29 20:38:43 +01:00
Jonas Fonseca
1a177491a0
Fix SGML parsing of processing instructions (<?xml ...?>)
...
It involves adding a new scanner state which is used only to generate a new
processing instruction (PI) data token. This removes some scanner specific
code from the parser and makes handling of PIs more generic. The data of
XML PIs are still parsed as attributes and added to the PI node.
The 6th test now succeeds. Hurrah!
2005-12-29 18:31:49 +01:00
Jonas Fonseca
fb6ca9a390
Use dom_string for storing the name member of dom_scanner_string_mapping
2005-12-28 21:10:05 +01:00
Jonas Fonseca
f1015f8a6a
Make files include dom/string.h instead of util/string.h directly
2005-12-28 20:45:55 +01:00
Jonas Fonseca
6e163b186c
Make the dom_scanner_token store it's string in a dom_string struct
2005-12-28 16:23:36 +01:00
Jonas Fonseca
71533eef9a
Elute all DOM-related code and put it in src/dom
2005-12-28 14:05:14 +01:00
