CVS log for parser3/src/main/untaint.C

Up to [parser3project] / parser3 / src / main

Request diff between arbitrary revisions - Display revisions graphically

Keyword substitution: kv
Default branch: MAIN

Copyright year updated, websites links changed to https://

Copyright year updated

single quote now escaped with ' in HTML tainting mode (implements feature #1241)

copyright, authors actualized

Copyright year updated

SAPI::abort was used only once, thus replaced with SAPI::die and removed

to be sure that our version of gc.h/cord.h is used, we now include it directly

as in case of <space>=?UTF-8?Q?= space is ignored, =?UTF-8?Q?= should start in case of leading space (related to bug #123)

Copyright year: 2015->2017

optimization: L_PASS_APPENDED removed as no longer required (related to feature #1091)

Copyright year updated

minor optimizations

Now char * is converted to CORD only using AS_CORD() to check for empty cord bug (fixes bug #957)

\u2028, \u2029 are now escaped in json:string[] (fixes bug #1023)

removed most of the clang compiler warnings (closes issue #1018)

json chars 0x01-0x1F now are escaped as \u00XX, minor optimizations (closes issue #896)

- do not replace ' char by _26 while exploding filespec-tainting ( new feature: #829 )

ident now works under Linux + ident displays filenames (closes issue #818)
Copyright updated

- use pa_isalpha and pa_isalnum instead of isalpha and isalnum
- bug with redundand quoting lowercased latin chars while building email body was fixed

L_HTTP_HEADER now used in http headers values, pa_http_safe_header_name added for headers names (fixes bug #195)

bugfix: first, second String::Body argument is hashcode; second, info.fragment_begin is original, not resulting length

bugfix: charset can be null (in ^string:save[] as example), thus check is added

'*' is now not urlencoded to allow header "Accept: */*" to be passed

files upload now uses binary blocks instead of L_FILE_POST tainting. (bugfix: #128)

- taint[json] was added
- escaping was slightly modified

- little hacking: for dealing with transcoding+^taint[uri] transcode_and_untaint was replaced by untaint_and_transcode

- workaround in file-spec lang for old Macs was removed => now parser can open files with Russian 'r' in filename

- bugfix: client charset whould be analized while escaping cookies but now source charset

- a bit more comments were added

- some fixes with transcode+untaint

- new language L_PARSER_CODE for auto-untaint in process was added

- some code cleanup

- optimizations and simplification of string.cstr(...)

- CORD_cat optimization

- ^taint[js] now escapes \x0D as well

- String::append optimisation

- for file post the new taint language L_FILE_POST was added

- some formating changes

- char '-' also prefixed by '\' while regex tainting

- comments removed

- cookies outputs as %uXXXX now. while decoding for backward compatibility they decoded from %XX in request:charset too

no message

- try to fix coredump on unix if print to body ^taint[sql][something] outside of connect

merged from HEAD
! mail:send << changed \r or \n to ' ' in

merged from HEAD
! mail:send << changed \r or \n to ' ' in

! mail:send << changed \r or \n to ' ' in

merged from HEAD:
+ ^taint/untaint[regex] << escapes these: \^$.[]|()?*+{}

+ ^taint/untaint[regex] << escapes these: \^$.[]|()?*+{}

2005

. 2005 ;)

! string invariant violated in passing empty strings to file::exec/cgi environment

workaround kinda bug in libxml: life requires to do xdoc::create{invalid chars}
standard disables chars less then \x20, except tab, cr, lf.

changed tainting so that those become '!'

beauty: removed outdated level 'table'

speed: there's no \n chars in output [normally], so no need to optimize them
they could be retrived from databases, though. but
1. should be replaced
2. if not optimized, no harm done

bugfix: one last char prior to "spaces<" were not made quoted-printable

2004

beauty: removed old def/ifdef

beauty: coredump @unknown untaint language now

merged: bugfix from 3.1.1 on whitespace optimization

bugfix: whitespace now optimized properly [were bug: ^taint[1 & 2]="1 &2"

*** empty log message ***

bugfix: consequences of not reading this comment: CORD_pos_chars_left /* Number of characters in cache.  <= 0 ==> none	*/

more warnings --

static const char * const IDENT

removed needless field

removed needless const

templates and anonymous unions differences on unix

string_fragments_to_cord merged to HEAD

v() functions of String::Body, String::Languages and String itself
StringBody->String::Body

$a[ok!]
$a works!

something even more works :)

something more works :)

just compiled
todo: serialize&deserialize

main idea implemented, details left

merged 3.1.0 latest changes

email addresses in forms:
  "non-ascii" <yyy>
  non-ascii <yyy>
now encoded correctly [kinda merge from 3.0.8]

email addresses in forms:
  "non-ascii" <yyy>
  non-ascii <yyy>
now encoded correctly

moved tempate_gc to HEAD

merged from HEAD:
enabled '~' letter in filenames

enabled '~' letter in filenames

started http://
  $.charset[] param
and
  http response charset detection
just compiled.
todo:test

changed transcode param converntion  along with string creating convention -- all strings are zero-terminated,
this allowed to fix one remaining String("123", 2) case

pa_vform violated String::invariant. started fixing [not compiled now]

number of string style improvements & optimizations

no message

removed redundant & in CORD_pos param passing [for it's a pointer really]

CORD_pos_advance turned out to have limit on 'n' param. worked that around

no message

unknown untaint lang now causes death

fixed clients of &get():
most to use non-ref version,
some[in tight places] to get_unchecked_ref

configured math funcs
ported pa_exec

untaint.C L_SQL

untaint.C L_MAIL_HEADER

untaint.C L_URI

string optimize bit implemented

String::ArrayFragment::append_positions fixed

attempt to do gif encoding to CORD_ec [bad]
would rewrite as realloc now

main.lib compiled

introducing StringBody [C++ CORD wrapper]

cord bug fix, but still errors
todo: clear out

more tests [bugs fixed]

more tests OK

test compiled

untaint.C 99% [except mail&sql&optimize]

untaint.C 30%,
pa_common.C [done, without http for now]

started test

started porting to gc:
PA_Object done
lots of replacements also

libgc attempt

mail: transcode fixed.
todo: still problems when transcode from charset A to A.

uchar changed to String_UL in all places [was not everywhere]
untaint test: 032

class fields of simple type not initialized with zeros :(
while class fields of class types initialized with default constructors.
learn C++

string, response compiled

removed stupid value_includes.h

*** empty log message ***

grammar compiled

ConstStringPtr died, long live StringPtr

more compiled

vrequest,vdouble,vint compiled

untaint compiled

more compiled.
struck with string::store_to needing to know source/client charsets

2002->2003

filespec russian small 'r' changed to latin 'p' bug fix

filespec russian small 'r' changed to latin 'p'

mail header closed properly

mail header ',' allowed.
still bugs in smtp [only one receiptient works, cc, bcc ignored now, and MAILED ;)]

[strncpy killed, memnchr used]

cstr(UL_UNSPECIFIED) [not _PASS_APPENDED)

mail header quoted printable changed after RFC reread

1.
file spec language changed:
now there are only few chars are untainted:
* ? ' " < > |
and, on unix, : \ ~
[russian letters and SPACES now enabled, one should use ^untaint[uri]{...} now]

2.
$list[^file:list[dir]] now returns simply tainted names in $list.name, not tainted as filespec

@russianindex[]
#dir with files with russian-lang names
$where[dir]

$dir[^file:list[$where;\.txt^$]]
^dir.menu{
	<a href=$where/^untaint[uri]{$dir.name}>$dir.name</a><br>
}

mail:send closing ?=  now closed right

quoted printable encoding stops before \s*<...>$

quoted printable ' ' now =20
and encoding stops before <...>$

$Date: now$

ident.C* removed

(bug#2) mail subject got always prepended with charset even when all letters were 7bit one

killed 8 bytes from String.head

mail subject encoding taken from .content-type.charset, not .charset

assigned exception types

removed some debug comments

removed debug code, activated commented-for debug parts

STRING_*FOREACH_ROW
changed to stop at append_row, not at link zero
now string to string appending works

//#define DEBUG_STRING_APPENDS_VS_EXPANDS

economy: 22%

space uptimized: string::link_row removed

speed uptimized: string::is_empty

strign to string append optimiziation idea and estimates

name spelling

name spelling

2002

^mail:send[
   $.charset[zzz]
addded

allowed space in filespec

uri lang now knows about client/source charsets

introducing Charset

FILE_SPEC untainting changed so that one can erase files like that, knowing their full name
introducing
	// theoretical problem with, for instance, "_2B" and "." fragments,
	// they would yield the same
	// because need_file_encode('_')=false
	// but we need to delete such files somehow, getting names from ^index

ORIGINS mode removed UHTML

ORIGINS mode was bad - forgot about optimize bit

optimized \r produced by xslt

cached some string::size() es

table.append bug fixed: forgot to switch default language for {code}, it happen to change it's language from CLEAN, so later refused to split by CLEAN \t

UL_CLEAN pieces now inherit UL_OPTIMIZED_BIT at wcontext.write

removed 1+4 bytes from String

introducing estimating String::cstr_bufsize, 0

$status:db
fixed used not initialized in db_connection & db_table

optimizing returned after debugging

memory hunging
#pragma pack(1)

16 419 844
13 169 394

String size_t replace to uint in proper places [todo:Array]

debuggging memory.
string::as_int/double now usually uses stack, not heap [200K out of 17M :)]
VString(String&) not copies reference [29M -> 17M]

switched to c++ exceptions 0

untaint lang origins table fixed.
got rid of empty_string, which caused errors - it hasnt exception but somebody[value.bark] tried to throw it on it's pool

header untainting to UL_HTTP_HEADER, UL_MAIL_HEADER made nonforced [only tainted pieces]

remove_crlf @ SAPI::log

\r to ' ' in origins mode

http_header tainting more gentle with enquoting now

added ident.awk and main.dsp splitted to main+classes+types along to .am

dom.xslt  stylesheet compiled&cached

^dom:set{<some>xml</some>} default language XML
languages+=xml

strncpy forced with zero ending in case of limit

pa_directory speeded up. moved zero 'from' check to directory constructor.
string:replace

speeded up typo-html replacements. introducing Dictionary with first-char caching

bug fix @ untaint when \r skipped \n

ORIGINS langs name abbrevations

$ORIGINS(1) output tracing mode

qp wrong name

*/
static const char *RCSId="$Id: double.C,v 1.31 2001/06/28 07:41:59 parser Exp $";

static char *RCSId="$Id$";

introducing String::cstr_bufsize, returns just size+1 for as_is target.

fixed bug in pre html untaint, wrong size used, 4* mem wasted

#include "pa_config_includes.h"
removed from most .C

^string.upper|lower[]

z

untaint without charset

mail header only once to =?

untaint - 	@test optimize whitespaces for all but 'html'

untaint - 	@test mail-header

z

fixed header "a/a"

changed urlencode here and in untaint.C to HTTP standard's " and \" mech

fixed vstring:as_vfile  length

image:measure -90

exec3. decided exec:cgi to move to file:cgi

mail:send -10 just compiled

z

sql quote. string untaint UL_SQL

mysql 0

connect&sql -1000 just compiled

string:match replace 2.  string.cstr(forced lang)

todo/tests

string::pos & cmp & piece bugs fixed.
string::split new
table:load separated from set
table:set implemented with clean \n \t searches

string: pos

string: left right mid

pool.request undone

pool.request

file:load 1. tainted

sort 0

z

z

moved default typo-table to lowlevel - it initialization could be skipped if failed in @auto. so that exception report would use some table

html-typo - moved all processing to table [except preliminary \r\n \r \n replacements to "\n"]

html-typo - decided to move all the processing to table

returned table originating. useful for reporting typo table problems origin. think would be useful somewhere else

typo  \r \r\n \n properly handled

typo & typo-default

html-typo 2

html-typo 0
html-typo 1. noticed double default content-type prob. run cgi.cmd tomorrow first
html-typo sample. detected << problem: they become ltlt too fast

cgi: reinvent document_root and request_uri under iis. autoconf updated. introducing ap_config_includes

shifted apache branch on main trunc

String::UL_

z

file 1

z

auto @brief

doxygen include/, Value

cookie class

taint[uri

proper @exceptions params tainting

think constructing flag as is  is bad

z