CVS log for parser3/src/targets/cgi/parser3.C

Up to [parser3project] / parser3 / src / targets / cgi

Request diff between arbitrary revisions - Display revisions graphically

Keyword substitution: kv
Default branch: MAIN

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

avoid parser version requests logging in parser3.log

$status:log-filename added (related to feature #1262)

$status:mode (cgi/console/mail/httpd/apache/isapi) added (related to feature #1262)

more correct arguments processing to block -fl config.file log.file logic (related to feature #1262)

now only one log location is selected and tried. -l - supported for stderr specification (related to feature #1262)

-l /path/to/log options added (implements feature #1262)

Exceptions in @unhandled_exception are now processed in request::core for correct logging and reporting (feature #1256)

unhandled exception now called if exception occures during $response:download processing (related to bug #1255)

sync CGI mode behavior with httpd mode and avoid exit status 1 on die

headers buffering to avoid extra headers in errors, custom send_error in cgi/httpd mode to avoid memory allocations (fixes bug #1255)

avoid GPF in very low memory case, when static initialization failed even before main() start

Copyright year updated

*** empty log message ***

under Windows back_slashes_to_slashes should be used on argv[0] as well to match document root.

parser3 t.html now behaves identically to /usr/bin/parser3 t.html in terms of /etc/parser3/auto.p load (fixes bug #1244)

avoid "parser allready configured" exception when ./parser3.cgi <file> is used;
rsplit(..., '/') || rsplit(..., '\\') is not correct when / and \ are mixed in path

full_disk_path now uses pa_strcat()

copyright, authors actualized

more clear error messages

only SAPI_Info_CGI can read POST, SAPI_Info will return 0 in read_post and post_size != content_length will be thrown

argv[0] -> parser3_filespec

MSVC SOCKET warnings war

default sock_ready interval changed to 0.5 sec

pa_strncpy added ans used to avoid strncpy drawbacks

waitpid in file::exec returned "No child processes" when SIGCHLD was SIG_IGN (fixes bug #1215)

^httpd-main[] -> ^httpd:main[]

request_info now on stack to avoid yet another THREAD_LOCAL memory deallocation issue (now with cookies), explained in #1203

added by PAF in rev 201 to "write to error_log uri of currently processed document", but never
used and conflicts with how libgc stops threads on FreeBSD (#define SIG_SUSPEND SIGUSR1), thus removed.

+HAVE_TLS check

clear_response_headers added

do not send error via network if it was network write error.

*** empty log message ***

log -> pa_log

*** empty log message ***

pa_get_thread_id() used for logging

we need to ignore "void" connections from browsers
(browsers open connections in advance and they will be empty if user does not request more pages)

avoid previous uri in logs

global ::request and RequestController not needed in httpd mode

Copyright year updated

msvs support fixes

$main:HTTPD.mode implemented

*** empty log message ***

initial support for MULTITHREADED and PARALLEL httpd server modes

msvc warning war

msvc warnings war

SYSTEM_CONFIG_FILE used

locate_config call returned to original place as it requires pa_thread_request() under Windows

*** empty log message ***

renaming

thread request check added

httpd config_handler added, stdout flush added.

filespec_4log added, pa_strcat used.

pa_strcat implemented and used

*** empty log message ***

*** empty log message ***

spelling

renaming

args_skip removed, locate_config relocated

execution_canceled was for sigpipe before request processing, which is weired

filespec_to_process now variable, can't be empty but can be null.
httpd-main is now main method name for httpd mode

if filename to process is not specified, auto.p should be present.

clearing config_filespec when config was not found, thus flag not required and removed.

+ $env:name[value] and $form:name[value] features

pa_atoi / pa_atoui / pa_atoul usage checked. 10 is now default base, no hex autodetect by default

connection socket closing in destructor + accept exception handling

warning war

*** empty log message ***

*** empty log message ***

global try exception handling unified

excaption in unhandled exception code optimization

request_info now global variable

[host:]port syntax support in httpd mode

populate_env added

initial httpd mode implementation

options together

SAPI::die used in exception in unhandled exception handler to report 500 instead of 200 it was before.

*** empty log message ***

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

whitespace

locate_config() extracted, whitespace.

duplicate info in signal logging removed, whitespace
was: SIGPIPE received while executing code. uri=/_tmp.html, method=GET, cl=0 [uri=/_tmp.html, method=GET, cl=0]

avoid empty ?

PATH_INFO check in reinventing DOCUMENT_ROOT (related to feature #1164)

IIS5 support removed to simplify fcgiwrap setup (implements feature #1164)

real_parser_handler logic simplified, but no issues found (related to feature #1164)

actualized

GC_dont_gc=1 moved to pa_globals_init

pa_socks_init/pa_socks_done moved inside pa_globals_init (optimization)

warning war

warning war: gcc 6.x issues fixed

Copyright year: 2015->2017

Request::SKIP_INTERRUPTED, Request::SKIP_RETURN added and implemented,
Request::SKIP_* now checked after each possible process/execute call (fixes bug #1077)

VS compatibility fix

HTTP response code used as exit status if < 100 (implements feature #1099)

whitespace

whitespace

remove_absolute_uri -> strip_absolute_uri

request_info.remove_absolute_uri added to remove possible http://site.name/ from $request:uri (implements feature #1112)

Copyright year updated

year updated

$request:headers, $request:method added (implements feature #1007)

minor header usage optimization

PA_DEBUG_DISABLE_GC compilation fix

headers cleanup, garbage cleanup

includes cleanup

2012->2013

gcc compilation warnings fixed

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

option to hide $status used in apache module removed

vsnprintf now not called twice in die_or_abort (backport from 3.4.1)

vsnprintf now not called twice in die_or_abort (bugfix: #106)

- don't run as cgi if env:PARSER_VERSION was set for preventing infinite loot ( bugfix: #12 )

- core in sigpipe was fixed

- pass pre-capitalized headers
- don't use format function because it calls malloc

- back to storing response http headers in lowercase
- capitalize them during output

- http headers were changed: content-type => Content-type and so on

- year was updated :)

- new classes implementation

- type size_t was changed to int in main (better compatibility)

- size_t -> int for calming down compirer

make g++ happy with the type of argc in main()

- some constants changed their names

- new year in copyright %-)

- buffer size for parser3.log increased

- $request:argv

- little syntax changes

- annoying 'SIGPIPE' messages in parser3.log switched off by default. If someone really still need it: use $SIGPIPE(1)

- bugxif. failed when request cgi

- don't print headers if $console:line[data] was used during cgi execution.

- 2007 in help ;)

+ ^break[] ^continue[], in ^for, ^while, ^menu, ^hash/hashfile.foreach

! switched to VS2005 (projects and object<info explicitly typed>.for_each(, info))
+ $response:field[] setting void means removing
+ grammar $name` name stops now, for mysql `$field`
+ started ^break[] ^continue[], in ^for. TODO: to other iterators (while, menu, foreach)
+ all log messages += [uri=, method=, cl=]

. style

2005

. 2005 ;)

forgotten makefiles.
couple warnings in parser3.C

now, when version (compiled on HOST), we need more hspace

no message

debug: PA_DEBUG_CGI_ENTRY_EXIT
if on, writes basics to c:\parser3.log

bugfix: win32: system&parser exceptions in release mode reported properly
parser exception in exception handler WERE mistakenly reported as system exception, without details
[in apache & isapi]
+some beauty in exception text

bugfix: win32: system&parser exceptions in release mode reported properly
parser exception in exception handler WERE mistakenly reported as system exception, without details

bugfix: isspace((unsigned char)c) everywhere. failed on russian letters

patched:  ^file:send by Victor Fedoseev <vvf_ru@mail.ru>
todo: turn it to $response:download[
$.filename[filename]
$.option[]
,,
]

bugfix: iis specific
http://parser3/_bug.html?404;http://hpsv/test/
now
$request:uri
/_bug.html?404;http://hpsv/test/
--------------------------------------------------------------------------------
$request:query
IIS-STATUS=404&IIS-DOCUMENT=http://hpsv/test/
--------------------------------------------------------------------------------
IIS-STATUS="404"
IIS-DOCUMENT="http://hpsv/test/"

new: iis specific
http://parser3/_bug.html?404;http://server/_bug.html?f=v
now
$request:query
IIS-STATUS=404&IIS-DOCUMENT=http://server/_bug.html&f=v
$form:fields
IIS-STATUS="404"
IIS-DOCUMENT="http://server/_bug.html"
f="v"

todo: isapi too

new: CGI_PARSER_LOG env variable allows to specify where to put parser log file

bugfix: sql connections now are disconnected

debug helper added

migrated to latest libxml[2.6.5] and libgdome[1.1.2]
simplified lib build processes

static const char * const IDENT

resurrected: pool idea. now only for destructing objects at request processing end

hashfile implemented get/put [raw]

beauty: SIGUSRX removed qs= printing [already included into uri=xxx]

moved tempate_gc to HEAD

no message

initialized request_info properly

no message

attributed_meaning_to_string added L_UNSPECIFIED piece
which was sortof OK, but violated string invariant [assertion barked on that]
changed to L_PASS_APPEND, made that param obligatory

number of string style improvements & optimizations

unknown untaint lang now causes death

more understandable error message in case of errors in @unhandled_exception

gif encoder bug fix [gdGrowingBuf]

array::get/put inlined

converted debug hacks to ANSI assertions
started configure.in --enable-assertions

memory bug debugged down: gdome uses glib memory, and stores last pointer to xmlDoc there, gc misses that and collects valid memory
fixed by remembering xmlDoc from dom object in xdoc.
todo: do something with premature free of xdoc with xnodes/node values referring into it

xml memory debugging showed no errors. todo: debug parser memory

more build configurations

started xml lib memory debugging, simple checks failed:
it seems that library does realloc(bad ptr), and recording those ptrs in heap [for debugging] was bad:
heap situation changes = everything works fine

forgot to merge XML memory funcs replacement from gc branch

GC_java_finalization turned off [was 'not recommened' in gc.h, on win32 noticed no difference]

convention changed: all resulting strings are zero-terminated

disabled gc, become even slower

Array::put(index, T>>&<< removed
after Stack::pop wiping removed [moved to separate func]
other Array & removed

pre-evaluated .count() in some places

a number of bugfixes [while testing first real site (aval)]

fixed dom language [values are now considered tainted.
it was a bug in parser. updated 107 test result

all compiled.
todo:link

introducing StringBody [C++ CORD wrapper]

*** empty log message ***

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

started test

started porting to gc:
PA_Object done
lots of replacements also

no message

libgc attempt

lt_dlexit called

set_base, set_derived simplified [counter leaks fixed]

no message

1. more detailed log on signals.
2. sigpipe before request constructor now causes death

SIGPIPE now can be intercepted
and does not cause exception in exception handler

no message

todo: move to latest xml version on win32 and run memleak tests again

charsets: see some strange things with old xml lib-- it's internal memory handling has faults

fixed: mem leak from copy/paste bug

test:
97 file::load http://  [bugs fixed]

removed -H command line key, now testing using .sh file with SERVER_SOFTWARE=xxx

SAPI::abort << abort.  die now just exits

methoded_array now contains all Mxxx classes
SAPI::get_env now returns info on pool
more tests

$cookie:name[$.expires[date << can be now]]

written makefile with regression tests [raw]

MAKE_TEST must be used inside of 'make tests' only
[it's not forcing CGI mode now]
now use -H to output CGI header when parser used in command line [useful for tests also]

MAKE_TEST environment variable switches on CGI mode, and is used in regression tests

env fixed

no message

release mode now compiles OK
test to show benefits of free

form values passed OK now

no message

found&kill all remaning leaks resulted from pooled::malloc calls
[excluding in gd -- planning complete rewrite extremely ugly code]

few bugs in debug output fixed

fixed more leaks resulted from pooled::malloc calls

couple mem leaks rusulted from old pooled::malloc usage.
todo:find more like those

all compiled, only 14 link errors :)

classes.lib - 0 error(s), 0 warning(s)
parser.C compiled

pa_request compiled

grammar compiled

2002->2003

removed last \n appending in non-win32 non-cgi [script] runs

merged die&iis changes

1. on win32 in die: abort() reverted to exit(1)
2. more flexible iilegal call check

::die now tries to write core dump

::die now tries to write core dump

::die now tries to write core dump

more checks on http://domain/parser.cgi start [maybe some getenv returns "", checked that now

more checks on http://domain/parser.cgi start [maybe some getenv returns "", checked that now

error logging made unbuffered [so that out-of-mem errors reached log]

mail:send MIME-Version default

no message

logging @signal += query_string

comment

SIGNALS overriden @ main top

no message

SIGUSR1 >> writes to error_log uri of currently processed document
SIGPIPE >> interrupts request processing [exception = DB rollback]

^file:exec/cgi [pa_exec] pipe read errors now checked

removed #ifdef WIN32 around check of CGI: Illegal call

compiling to system classes disabled

when auto.p beside binary [cgi, isapi] not accessible [due to bad rights or whatever] it's error now

$mail.received.file.value.content-type fixed

removed fixopt stupidity. on stupid linux use
cd /document/root
../cgi/parser3 script

fixopt now preprocesses command line params, splitting them by space, excluding argv[0], argv[argc-1]

-f config file

fixed .log file dir

no message

removed #define DEBUG*

$Date: now$

ident.C* removed

usage to stdout now

removed pa_config_paths.h

/cgi-bin/parser empty filename checked

#ifdef WITH_MAILRECEIVE

uue to separate file
introducting $message:received
^mail:send rewritten with backward comp

marged latest HEAD updates, mainly file:justname&co & @auto/conf[filespec

parser3.conf renamed to auto.p

removed root conf define creation

parser3.conf now one and only

file::exec/cgi to pass HTTP_ vars
introducing SAPI::environment

fixed .dsp-s along with reorganized cvs modules dirs structure

info.uri now ""

removed 100*40 bytes per request memory leak [nonpool malloc in globals]

removed 100*40 bytes per request memory leak [nonpool malloc in globals]

request.uri now never 0

merged from 0001

setenv in .htaccess when cgi is not under that dir got REDIRECT_ prefix before HTTP_PARSER_x_CONFIG, now that took into account

setenv in .htaccess when cgi is not under that dir got REDIRECT_ prefix before HTTP_PARSER_x_CONFIG, now that took into account

MSVC profile targets

cheched date:create(days) param for validity [later were assumed valid and crashed on invalid onces]

OP_CALL -> OP_CALL__WRITE and used that for removing VString wrapper

ctime sometimes can be just "", checked that

switch in hash constructor fixed [were problems with using of stacked wwrapper after it's death]

removed
{
  char *a;
  {
     /*sub local*/char b[...];
     a=b;
situations

int/double get_string now not pool.malloc(MAX_NUMBER) but really neaded

document root in standalone version = current dir

document root in standalone version = current dir

assigned exception types

http_site_config_filespec

isapi site config beside .dll

//#define DEBUG_STRING_APPENDS_VS_EXPANDS

speed uptimized: string::is_empty

strign to string append optimiziation idea and estimates

PARSER_ROOT_CONFIG

*** empty log message ***

removed libtool usage from build mech

baltic charset file generated, in generation script ispunct check added

name spelling

name spelling

2002

lowered indent.awk back to src/

moved etc&www out of src, moved configure&co out of src
added windows-1250.cfg
straightened installation scripts procedure
made targets/cgi/parser3.C be dependant on ./configure --sysconfdir change

gnome libs just compiled in.
no refcounting
no xpath
no xslt
  yet

going away from xalan&xerces, started the process.
globals [initialization
pool [charset update
charset [transcodings
vxnode vxdoc [DOM calls

msvc heap debugging flag

removed 'expires' from header outputs

removed debug info

fatal error reporting fixed.
in IIS5 there were a prob with illegal call check

fatal error reporting fixed.
in IIS5 there were a prob with illegal call check

optimized \r produced by xslt

removed debug cgi||1

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

#if _MSC_VER _set_new_handler

set_new_handler in apache&isapi

pool debug @ die

introducing estimating String::cstr_bufsize, 0

die in pool::fail

die in pool::fail

exception with dynamic buffer re-back
now with counter

std::

set_new_handler started

apache 'status' class enabler

<Location /parser-status.html>
ParserStatusAllowed
</Location>

read can return 0, that means sort of sig_pipe on freebsd [moko's experiments].
 checked that in cgi, on apache already done :)

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

removed debug defines

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

cgi += pool cleanups

hashfile 0

stack backtrace

switched to c++ exceptions 0

isapi&cgi couple non-pooled mallocs in main handler

xalan&xerces multithread bug fixed. initialization&free on each thread!

xml transform win32 errors intercepted [xalan transformer patched and incorporated]

xalan patches starting

remove_crlf @ SAPI::log

z

z

#ifdef SMTP

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

merged3 with before_xml
[merge2 actually didn't happen - i've created tags on wrong branch. so this merge is in fact re-merge]

merged latest bugfixes from before_xml branch
^if(def $hash) now true only when ^hash:_count[]!=0
moved
.string .save .file from dom to dnode

expressions unary+
^hash.count[]
bugfix: in expression compound($aa.zz.xx) names now can have minus '-' in name after '.'

ParserRootConfig .../parser3.conf
ParserSiteConfig .../parser3.conf

cgi&isapi looks for {configure|c:\windows}/parser3.conf

z

#define STRINGIZE(name) #name
does not macro expantion on param, simply qoutes whatever passed, undone

#define STRINGIZE(name) #name
invented

full_file_spec bug

getenvcheck

zz

no message

zzz

no message

configure --sysconfdir=sysadmin-controlled auto.p location for targets/cgi/parser3, default[/usr/local/etc]

/configure cgi SYSCONFDIR

z

compiled under cygwin

strncpy forced with zero ending in case of limit

parser3 test.html [auto.p from current dir loading]

not cgi extra \n 3

not cgi extra \n 2

not cgi extra \n

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

static char *RCSId="$Id$";

^log ^exp

// no _ conversions in @exception[params]

z

argv can be just "parser3". made site_auto_path "." in that case

z

z

pcre_tables=pcre_default_tables;

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

z

removed array debug. before vstring rebasing

array debugged; adjusted

think that all must grow lineary, not exponentialy

string fixed bug with fullchunk cmps

main loss here: 5673321/     70041=        81

detected huge mem allocation: size/times malloc 27809390/368771, calloc 3232/83. would test now

renamed pa_methoded back to classes/classes.h

classes/classes renamet to include|main/pa_methoded

z

Methoded reorganized. todo: methoded-configure

removed ^a.menu{$name} ability. now
$a{^menu{$name}}
or
^a.menu{$a.name}

started beautifying

illegal call check a bit improved, but still under iis no mapping of dir with parser allowed!

redo failed
/// @test noticed series in isapi, check how initialize_random_class is called! [must be called only once]

/// @test with commandline start "parser3 a.html" so that ^load[a.cfg] worked! [now doesnt]

cgi cmdline ver

cgi /// @test disable /cgi-bin/parser3/auto.p

$MAIN:LOCALE

fixed header "a/a"

z

exec @jav

compile-1

forced UL_FILE_NAME of posted file name

all targets exec

sapi exec dead end. badly parsed args in apache:util.script

exec win32 env

exec class just compiled. moved post read to request core

problems with ^exec:cgi post data. they are already read by vform

mail:send 1

todo/tests

content-disposition

main:post-process

vfile in response:body

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

$request:browser [.type .version]; unknown.get_double now = 0.

file:load 1. tainted

continue on sort

z

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: sapi::log

nt error log experiments failed - cgi under iis got no access to log. registereventsource simply failed with 5(illegal call) code, while worked OK as standalone executable.
isapi: used HSE_APPEND_LOG_PARAMETER feature

auto.p monkey every target

auto.p monkey

isapi: think it would be better to rewrite as filter then to deal with pool

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

cgi: reinvent document_root under iis

cgi: only one pool now

sapi made object

all: sapi beauty

isapi minor env beauty

isapi document_root 0

isapi 1

isapi 0

started isapi

config of auto.p path: parser_root_auto_path parser_site_auto_path

z

in cookie & env. todo:post

shifted apache branch on main trunc

starting mod_parser3

automake compiled and works. todo:make win32 ifdefs work and add ifdef unistd

z

String::UL_