This document gives a functional overview of the Natural statements for internet and XML access, specifies the general prerequisites for using these statements in a mainframe environment, informs about restrictions that apply and contains a list of further references. To take full advantage of these statements, a thorough knowledge of the underlying communication standards is required.
The following topics are covered:
The following Natural statements are available for access to the internet and to XML documents:
This statement enables you to use the Hypertext Transfer Protocol (HTTP) and - under z/OS only - the Hypertext Transfer Protocol Secure (HTTPS) in order to access documents on the web with a given Uniform Resource Identifier (URI) or Uniform Resource Locator (URL), that is, the internet or intranet address of a web site.
REQUEST DOCUMENT implements an HTTP client at
Natural statement level, which allows applications to access any HTTP server on
either the intranet or the internet. The statement has a set of operands, which
allows it to formulate HTTP requests according to the needs of the user
application. For example, using outbound operands it is possible to send
user-defined HTTP headers, form data, or entire documents to an HTTP server.
The inbound operands can be used to retrieve a document from the server, to
view the entire HTTP header block returned from the server, or to return the
values of dedicated headers, etc. Via binary format operands, binary objects
such as gif files can be exchanged with the HTTP server as well. For basic
authorization purposes, user ID and password operands can be specified. The
content of this operand is sent with base64 encoding over the line, according
to HTTP standards.
Natural supports the following REQUEST-METHODs:
GET - retrieve a document and HTTP headers,
HEAD - retrieve HTTP headers only,
POST - transfer form data to an HTTP server,
and
PUT - upload a file to an HTTP server.
The REQUEST-METHOD is normally evaluated
automatically, based on the operands coded for the executed REQUEST
DOCUMENT statement. However, the predetermined
REQUEST-METHOD can be overwritten by an explicit user
specification of a REQUEST-METHOD header.
In addition to the standard
REQUEST-METHODs mentioned above, the following methods can be
specified in a REQUEST-METHOD header:
DELETE - delete a document from an HTTP
server,
PATCH - modify a document on an HTTP server,
OPTIONS - retrieve the
REQUEST-METHODs supported by an HTTP server, and
TRACE - retrieve the message received by an
HTTP server.
Data transfer with the REQUEST DOCUMENT
statement normally does not involve any code page conversion. If you want to
have the outgoing and/or incoming data encoded in a specific code page, you can
use the DATA ALL clause and/or the RETURN PAGE clause
of the REQUEST DOCUMENT statement to specify this.
In order to simplify data exchange from EBCDIC-based
mainframes with HTTP servers, which in most cases work with UTF-8 or ISO-8859-1
encoded data, the statement provides ENCODED clauses to allow
implicit or automatic conversion of outbound and inbound document data.
The following is an example of how the REQUEST
DOCUMENT statement can be used to access an externally-located document:
REQUEST DOCUMENT FROM "http://bolsap1:5555/invoke/sap.demo/handle_RFC_XML_POST" WITH USER #User PASSWORD #Password DATA NAME 'XMLData' VALUE #Queryxml NAME 'repServerName' VALUE 'NT2' RETURN PAGE #Resultxml RESPONSE #rc
The implementation of the REQUEST DOCUMENT
statement mainly consists of two layers:
an independent runtime layer, where the entire HTTP processing, URL analysis, data conversion, etc., is executed; and
a layer where an environment-dependent routine processes the TCP/IP communication between Natural and the HTTP server. This layer is implemented based on IBM LE (Language Environment) sockets for z/OS, VSE; SMARTS sockets for Com-plete and Natural Development Server; and CRTE sockets for BS2000. For CICS, the appropriate socket library is included into the build process.
Natural for Mainframes supports the HTTP protocol version 1.0 only, meaning that no persistent connection to the server is maintained. Since virtually every corporate network processes access to the internet via a proxy server from the client, Natural can be configured with the adequate settings for the proxy server and the port on which the proxy server runs. Moreover, it is possible to specify local domain name suffixes (intranet sites), which shall be accessed directly instead via the proxy server. See also Overview of Applicable Natural Parameters.
The proxy server, which is located between client (user) and internet, serves the following purposes: It receives the request from the clients, forwards it to the target server, caches the returned document and forwards it to the client. A proxy server is advantageous because of its improved performance, which is due to the caching, and because it helps to avoid security issues (most proxy servers are working as firewall as well).
The syntax of the REQUEST DOCUMENT statement and
detailed application hints are to be found in the
Statements documentation.
The REQUEST DOCUMENT statement is supported on the
following mainframe platforms:
z/OS: Batch, TSO, CICS, Com-plete, IMS TM
z/VSE: Batch, Com-plete, CICS
BS2000: Batch, TIAM, openUTM *
* See also Preconditions for the Support of XML-Related Statements under openUTM below.
Moreover this statement is available on all OpenSystems platforms that are supported by Natural.
The contingent of Internet Protocol Version 4 (IPv4) addresses (approximately 4 billion) is almost used up. New internet addresses are supplied with the Internet Protocol Version 6 (IPv6) providing an address space for about 3.4 x 1038 of IPv6 addresses (a single IPv6 address is 128 bits).
Natural supports both internet protocols to retain the
functionality of the REQUEST DOCUMENT statement in an internet
with a growing number of IPv6-based HTTP servers. Since IPv6 is not compatible
with IPv4, Natural provides additional logic and configuration parameters to
support the new protocol version.
- Prerequisites for IPv6 Support
For the prerequisites required to support
REQUEST DOCUMENTstatements with IPv6, refer to Prerequisites in Installation for REQUEST DOCUMENT and PARSE XML Statements in the Natural Installation documentation for z/OS, z/VSE and BS2000.- IPv6 Verification at Natural Session Start
IPv6 support is enabled with the
RDIPV6keyword subparameter of theXMLprofile parameter (see the Parameter Reference documentation). Natural checks whether an IPv6 TCP/IP stack is available on the local host at session start. If no IPv6 stack is available on the local host, Natural issues a corresponding warning message and disables IPv6 for the execution of theREQUEST DOCUMENTstatement.
- IPv4 and IPv6 Address Notations
The most common way to address an HTTP server is specifying the symbolic name substituted for the IP address like www.mycompany.com. However, you can also directly address an HTTP server by using the IP address.
The common notation for an IPv4 address is a group of four decimal encoded octets, separated by periods like 192.168.0.1.
The following rules apply to the notation of IPv6 addresses (according to RFC 4291, IP Version 6 Addressing Architecture):
Write an IPv6 address as eight groups of hexadecimal digits and separate them by colons. Each group consists of four hexadecimal digits. Example:
2031:AF04:87C2:0000:0412:988:7F2C:1C1BYou can omit leading zeros within a group.
You can omit a single group or multiple (consecutive) groups that consist of zeros only (
0000) by entering double colons (::).You cannot apply Rule 3 more than once.
For embedded IPv4 addresses, you can use the old decimal notation for the last 4 bytes. Example:
0000::FFFF:192.203.55.07Note:
Embedded IPv4 addresses are not supported by Natural.If you want to specify an IPv6 address in a
REQUEST DOCUMENTstatement, enclose the address in square brackets ([ ]). Example:http://[2BFC:5022:4081:0000::C:41]:8080If you want an IPv6 address to be recognized as a local address, the notation must exactly map to an IPv6 no-proxy specification, that is, the address in the example above would not map to the following no-proxy specification:
2BFC:5022:4081:0::C:41. You can, however, use the following prefix notation for a local site:2BFC:5022:4081:0000::.
The PARSE
XML statement allows you to parse XML documents from within a
Natural program.
The PARSE XML statement integrates a full XML
parser into Natural, thus allowing Natural applications to parse XML documents
in order to easily process their content. The PARSE XML statement
opens a processing loop and returns, whenever one of a list of events occurs
during the parse process, the respective path through the document, name and
value of parsed elements together with some parser status system variables.
For parsing XML documents the following parsing strategies or models are most common:
DOM (Document Object Model), an object oriented approach
SAX (Simple Access to XML), a stream-oriented parsing method
The implementation of the PARSE XML statement in
Natural for Mainframes is based on the SAX method, using a mainframe port of
the SAX parser Expat.
Parsing is processed internally on a UTF-16 encoded image of the document to parse, that is, if the document is not delivered in this encoding, an internal conversion to UTF-16 is performed before the parsing starts. This has to be considered at Natural installation time, for example, when the thread size for the TP environment is evaluated.
The encoding of the document to parse is checked automatically.
A check for a BOM (Byte Order Mark), which marks the document's encoding, is done.
If no BOM is found, a check for ASCII, EBCDIC, or UTF-16 (BE or LE: big endian or little endian) is done.
If an EBCDIC or ASCII encoding is identified, a search for an encoding processing instruction is performed.
If no encoding can be identified, an adequate error message is issued and the parse process is terminated. Internally, the parser works with UTF-16BE, so the document to parse is always converted to this encoding before it is passed to the Expat parser.
If an encoding PI (Processing Instruction) is found, the following defaults apply:
for ASCII, UTF-8 is assumed as encoding
for EBCDIC, the Natural default code page (see system
variable *CODEPAGE)
is assumed as encoding
The parse process itself consists of two phases.
In the first phase, the parser is called repeatedly to announce a well-defined set of callback entries. Those entries are entered by the parser whenever a corresponding element is encountered in the current parsed document. The occurrence of a start tag is, for instance, such an event which triggers a callback to the corresponding entry. The callback entries expose the Natural runtime logic for the execution of the parse process.
The second phase is the actual parse process. The parser is called with the document to parse as input operand. Now, each element is parsed, and for each element type its corresponding callback routine is called. The Natural runtime then processes the returned element, updates the return operands, and enters the parse loop for processing those operands. Then the parser is restarted to continue the parse process. The parse process is finished either if the document is completely parsed or if an XML syntax error occurs in the current document, meaning the document is not well formed.
Note:
For technical reasons, nested parse loops are not supported
in Natural for Mainframes.
The syntax of the PARSE
XML statement and detailed application hints are to be found
in the Statements documentation.
The PARSE XML statement is supported on the
following mainframe platforms:
z/OS: Batch, TSO, CICS, Com-plete, IMS TM *
z/VSE: Batch, Com-plete, CICS
BS2000: Batch, TIAM, openUTM **
* See Restriction Concerning IMS TM.
* * See also Preconditions for the Support of XML-Related Statements under openUTM.
Moreover this statement is available on all Open Systems platforms that are supported by Natural.
This section describes the general prerequisites that apply if you
wish to use the Natural statements REQUEST DOCUMENT and
PARSE XML.
To enable the use of the Natural statements REQUEST
DOCUMENT and PARSE XML, the installation steps described in
the Installation documentation must be performed; see:
Installation for REQUEST DOCUMENT and PARSE XML Statements on z/OS
Installation for REQUEST DOCUMENT and PARSE XML Statements on z/VSE
Installation for REQUEST DOCUMENT and PARSE XML Statements on BS2000
Since REQUEST DOCUMENT as well as PARSE
XML, at least internally, always have to convert data from one encoding
to another, Natural has to be driven with active ICU support. Therefore, the
ICU
library must be installed.
If REQUEST DOCUMENT or PARSE XML is to
be executed, the following prerequisites must be fulfilled:
a TCP/IP stack must be available and enabled for the execution environment,
a DNS (Domain Name System) server or DNS services must be
available in the execution environment to resolve internet address resolution
requests (gehthostbyname function),
a Natural driver must be installed LE enabled (in IBM environments) or CRTE enabled (in BS2000 environments),
support of HTTPS under Com-plete requires APS Version 2.7.2 Patch Level 16.
The following is an overview of Natural profile and/or session
parameters that enable/disable or influence the support of the statements
REQUEST DOCUMENT and/or PARSE XML:
| Parameter | Purpose |
|---|---|
XML |
This Natural profile parameter and/or
the corresponding parameter macro
NTXML
in conjunction with their keyword subparameters are used to activate/deactivate
the statements REQUEST DOCUMENT and PARSE XML.
In addition, there are various options that can be set
with the keyword subparameters of As a prerequisite for using the
|
CFICU |
This Natural profile parameter and/or
the corresponding parameter macro
NTCFICU
in conjunction with their keyword subparameters are used to enable Unicode and
code page support.
|
CP |
This Natural profile parameter defines the default code page for Natural data and Natural sources. |
CPCVERR |
This Natural profile and session parameter specifies whether a conversion error that occurs when converting results in a Natural error or not. |
To activate the support of the statements REQUEST
DOCUMENT and PARSE XML for the current
session
To activate both statements together, set the
Natural profile parameter XML (or the
corresponding parameter macro NTXML) and also its
keyword subparameters RDOC and
PARSE to
ON.
Or:
To activate the support individually, set only the
corresponding XML/NTXML keyword
subparameter to ON:
RDOC to
enable support of REQUEST DOCUMENT
PARSE to
enable support of PARSE XML
If the installation platform operates behind an internet
firewall or if the internet traffic is routed via a proxy server, the
XML/NTXML
keyword
subparameters for proxy and proxyport have to be specified
accordingly.
To activate the support of the statements REQUEST
DOCUMENT and PARSE XML for all sessions
Ask your system administrator to add the parameters and/or macros listed in the Overview of Applicable Natural Parameters to the Natural parameter module and to set the values correspondingly.
To deactivate the support of the statements REQUEST
DOCUMENT and PARSE XML
To activate both statements together, set the
Natural profile parameter XML or the parameter macro
NTXML to OFF.
Or:
To deactivate the support individually, set only
the corresponding XML/NTXML
keyword subparameter to OFF:
RDOC to
disable support of REQUEST DOCUMENT
PARSE to
disable support of PARSE XML
For information, see XML - Activate PARSE XML and REQUEST DOCUMENT Statements in the Parameter Reference documentation
To enable Unicode support
Set the profile parameter CFICU must
be set to ON.
For information on the various options that can be set with the
keyword subparameters of profile parameter CFICU, see
CFICU - Unicode
Support in the Parameter Reference
documentation.
See also the paragraphs relating to the statements
PARSE XML and
REQUEST DOCUMENT in the section
Statements, which is part of the
section Unicode and Code Page Support in the Natural Programming
Language in the Unicode and Code Page Support
documentation.
HTTPS, short for Hypertext Transfer Protocol Secure, is an additional security layer between HTTP and the TCP/IP protocol stack:
| Layer | Protocol |
|---|---|
| Application layer | HTTP(S) |
| Security layer | TLS/SSL |
| Transport layer | TCP |
| Network layer | IP |
It was introduced to enable encryption and communication partner authentication for a secure data communication over the internet.
The HTTPS URI scheme is used to indicate, that the HTTP communication is secured. For the encryption of the data the SSL (Secure Socket Layer) protocol or its successor TLS (Transport Layer Security) is used. Authentication is hereby provided by the exchange of certificates, which guarantee the identity of the communication partners.
In most cases of HTTPS communication, however, only the server identifies itself with a certificate against the client. Client authentication with a client certificate occurs quite seldom.
SSL communication is established in several steps:
It starts with identification and authentication of the communication partners over the so-called SSL handshake protocol (Client Hello, Server Hello).
The handshake is followed by the exchange of a symmetric session key via asymmetric encryption (private – public key proceeding). The public key, which is hereby used by the client, is an essential part of the server certificate.
After the handshake and key exchange have been performed, the encrypted payload request messages are communicated. The symmetric session key, which was negotiated in the preceding steps, is used for the encryption/decryption of those messages.
The HTTPS protocol uses port numbers which differ from the standard HTTP port numbers. While HTTP normally uses port 80, the default port number for HTTPS is 443.
HTTP access to the internet from a client, which is connected to a LAN (Local Area Network), is normally processed via special HTTP servers, so-called proxies. Proxies are gateways to the internet from the LAN, which perform security policies, provide caches and validation routines or filter functions and act as firewalls. HTTPS secured internet access is most often performed over a proxy server of its own, which maintains the connections to the remote servers. This proxy is also known as "SSL proxy".
Certificates are binary documents, which contain, among other information items, information about the owner and the issuer of the certificate, the public key for the encryption of the session key data, an expiration date and a digital signature. The certificates, which are presented by HTTPS servers, are normally the lowest link of an entire chain of certificates. Such a certificate chain is called a Public Key Infrastructure (PKI). The certificate on top of such a chain is called a root certificate. Root certificates are generally issued by special organizations, named Certificate Authorities (CA). Root certificates, which are issued and signed by a CA are also called CA (root) certificates. For further information, see HTTP Developers Manual and other sources in the internet.
HTTPS support for the Natural REQUEST DOCUMENT
statement is based on the z/OS Communication Server component AT-TLS
(Application Transparent-Transport Layer Security).
AT-TLS provides TLS/SSL encryption as a configurable service for sockets applications. It is realized as an additional layer on top of the TCP/IP protocol stack, which exploits the SSL functionality in nearly or even fully transparent mode to sockets applications. AT-TLS offers three modes of operation. See z/OS Communications Server, IP Programmer’s Guide and Reference, Version 1, Release 9, Chapter 15, IBM manual SC31-8787-09).
These modes are:
The sockets application runs without modification in transparent mode, unaware of performing encrypted communication via AT-TLS. Thus legacy applications can run in secured mode without source code modification.
The application is aware of running in secured mode and is able to query TLS status information.
The sockets application is aware of AT-TLS and controls the use of AT-TLS encryption services itself. This means, the application is able to switch between secured and non secured communication.
Natural for Mainframes uses Controlling mode to switch on secured mode for HTTPS requests only, while HTTP requests remain unencrypted.
Certificates, which are to be used with AT-TLS, can be maintained in two ways under z/OS. They are stored in RACF key rings or in key databases, which are located in the z/OS UNIX file system. Which of these proceedings actually applies is defined in the AT-TLS Policy Agent Configuration file for the z/OS TCP/IP stack, which is used by the Natural HTTPS client.
IBM delivers a set of commonly used CA root certificates with each z/OS system delivery. If key rings are going to be used to hold server certificates, those root certificates must be manually imported into the key rings by the system administrator. If IBM delivers newer replacements for expired root certificates, all affected key rings have to be updated accordingly.
Unlike key rings, key databases contain the current set of root certificates automatically after they have been newly created. However, the need for maintaining always the latest set of root certificates applies to the key database alternative as well.
Certificates to be used by the Natural HTTPS client, must be flagged as trusted. If they are part of a Public Key Infrastructure, the corresponding CA root certificate has to be flagged as trusted.
In RACF, digital certificates are stored in so-called key rings.
The RACF command RACDCERT is used to create and
maintain key rings and certificates, which are contained in those key
rings.
See z/OS Security Server RACF Security Administrator’s Guide, IBM manual SA22-7683-11, and z/OS Security Server RACF Command Language Reference, IBM manual SA22-7687-11.
Alternatively to RACF, certificates can be kept in key databases,
which reside in the z/OS UNIX services file system. For the creation and
maintenance of key databases, the GSKKYMAN utility has to be
used.
See z/OS Cryptographic Services PKI Services Guide and Reference, IBM manual SA22-7693-10.
The following restriction applies if you wish to use the Natural
statements REQUEST DOCUMENT and PARSE XML in an IMS
TM environment:
The PARSE XML statement can be executed under the
TP monitor IMS TM with the restriction that no I/O statement is allowed within
an active PARSE loop. If an I/O occurs within a PARSE
loop, error NAT0967 is issued.
For further restrictions, see the corresponding notes in the statement descriptions.
During an active parse loop with I/O operations, the
openUTM function call PGWT must be used. This means:
The openUTM application must be started with not less than 2 tasks, otherwise an openUTM error K319 with subsequent dump will occur.
PGWT conditions must be defined for the
KDCDEF.
Define the maximum wait time (in seconds) for input messages
during a PGWT call.
Example:
MAX PGWTTIME=60
Define the maximum number of openUTM tasks for
PGWT calls.
Example:
MAX TASKS-IN-PGWT=1
PGWT can be controlled using either the
TAC-PRIORITIES instruction or the TACCLASS
concept:
Control of PGWT using the
TAC-PRIORITIES instruction:
Example:
DEFAULT TAC TYPE=D,PROGRAM=NATUTM,. . . . . . . TAC NAT,ADMIN=NO,TIME=(0,0),PGWT=YES,TACCLASS=1 TAC-PRIORITIES DIAL-PRIO=EQ
Control of PGWT using the
TACCLASS concept:
Example:
DEFAULT TAC TYPE=D,PROGRAM=NATUTM,. . . . . . . TAC NAT,ADMIN=NO,TIME=(0,0),TACCLASS=1 TAC NAT1,ADMIN=NO,TIME=(0,0),TACCLASS=2 TACCLASS 1,TASKS=2 TACCLASS 2,TASKS=1,PGWT=YES
The keyword subparameter
ILCS of parameter macro
NURENT must be set to
ILCS=CRTE.
The following sample program shows the usage of the REQUEST
DOCUMENT and the PARSE XML statement.
Further sample programs are provided at the end of the description
of each statement and in the Natural library SYSEXV.
DEFINE DATA
LOCAL
1 #FROM (A) DYNAMIC
1 #HEADER (A) DYNAMIC
1 #PAGE (A) DYNAMIC
1 #RC (I4)
1 #COL (N8)
1 #COL1 (I4)
1 #COL2 (I4)
1 #COL3 (I4)
1 #LOC (A30)
1 #CP (A) DYNAMIC
1 #PATH (A) DYNAMIC
1 #NAME (A) DYNAMIC
1 #VALUE (A) DYNAMIC
1 #RTERR (I4)
END-DEFINE
*
ASSIGN #FROM = 'HTTP://SI15.HQ.SAG/autos6.xml'
**
REQUEST DOCUMENT FROM #FROM
RETURN
HEADER ALL #HEADER
PAGE #PAGE ENCODED FOR TYPES 'TEXT/XML'
CODEPAGE ' '
RESPONSE #RC
GIVING #RTERR
**
IF #RC NE 200 /* TEST FOR HTTP RESPONSE 200 = 'OK'
WRITE 'HTTP RESPONSE' #RC 'RECEIVED'
ESCAPE ROUTINE
END-IF
EJECT
PRINT #HEADER
/ '_'(79)
PRINT #PAGE
/ '_'(79)
/ '_'(79)
ASSIGN #CP = *CODEPAGE
EXAMINE #PAGE FOR 'encoding' GIVING POSITION #COL1
IF #COL1 GT 0
EXAMINE #PAGE FOR '?>' GIVING POSITION #COL3
IF #COL3 GT #COL1
EXAMINE #PAGE FOR 'ISO-8859-1' GIVING POSITION #COL2
END-IF
IF #COL2 GT #COL1 AND #COL2 LT #COL3
EXAMINE #PAGE FOR 'ISO-8859-1' REPLACE #CP
END-IF
END-IF
PRINT #PAGE
/ '_'(79)
EJECT
PARSE XML #PAGE INTO PATH #PATH NAME #NAME VALUE #VALUE
PRINT #PATH / 'NAME=' #NAME / 'VALUE=' #VALUE / '_'(79)
END-PARSE
END
Note:
The URL accessed in the above program addresses an intranet site
and cannot be accessed from the internet.
Output of the sample program:
HTTP/1.1 200 OK?Date: Thu, 10 Aug 2006 16:26:22 GMT?Server: Apache/1.3.19 ( BS2000)?Last-Modified: Thu, 27 Jul 2006 16:44:42 GMT?ETag: "2602c-111-44c8ed7a" ?Accept-Ranges: bytes?Content-Length: 273?Connection: close?Content-Type: text/ xml?? _______________________________________________________________________________ <?xml version="1.0" encoding="ISO-8859-1" ?><autos>?<make></make>?<make>Ford</ make>?<model>Thunderbird</model>?<make>Merceds-Benz</make><model>S400</model>< make>BWM</make><model version="latest">330I</model>?<make><label><company> Mercedes</company></label></make>?</autos>? _______________________________________________________________________________ <?xml version="1.0" encoding="IBM01140" ?><autos>?<make></make>?<make>Ford</ make>?<model>Thunderbird</model>?<make>Merceds-Benz</make><model>S400</model>< make>BWM</make><model version="latest">330I</model>?<make><label><company> Mercedes</company></label></make>?</autos>? _______________________________________________________________________________
MORE
autos Name= autos Value= _______________________________________________________________________________ autos/$ Name= Value= ? _______________________________________________________________________________ autos/make Name= make Value= _______________________________________________________________________________ autos/make// Name= make Value= _______________________________________________________________________________ autos/$ Name= Value= ? _______________________________________________________________________________ autos/make Name= make Value= VVVV Name= autos Value= _______________________________________________________________________________ autos/$ Name= Value= ? _______________________________________________________________________________ autos/make Name= make Value= _______________________________________________________________________________
How to use the XML keyword subparameters (e.g. RDP and RDNOP)
How to determine proxy server, port number and HTTP server at a site?
How to decide if a problem is a TCP/IP or HTTP issue or if it is a Natural issue?
How can I check if I can reach a website from my mainframe without using Natural?
Which is the preferable method for maintaining certificates?
Documentation for Natural on mainframe states that "The Natural ICU handler must be linked to the Natural nucleus".
The codepage support is needed, as on mainframe platforms, the
document to be parsed is always internally converted to UTF-16 (if the document
is not already encoded in UTF-16). In most cases the document is not in UTF-16
and a conversion will take place. For more detailed information, see the
PARSE XML statement documentation and PARSE XML in
the Unicode and Code Page Support documentation.
The ICU library is needed to interpret incoming HTTP headers and
convert outgoing HTTP headers. Typically the incoming headers are ISO 8859-1
encoded and on mainframe always have to be converted to the Natural default
codepage (see also system variable *CODEPAGE)
- on PC a conversion is not always necessary.
On the PC, the REQUEST DOCUMENT statement executes
the Internet Explorer and uses the settings as defined there.
On the mainframe, the URL of the (intranet) proxy server through
which all requests have to be routed has to be specified with the
NTXML/XML
keyword subparameter RDP. With the
keyword subparameter RDNOP,
local domain(s) which are to be addressed directly, not via the proxy server
can be defined.
Information about proxy server, port number and HTTP server at a site has to be provided by the network administrator.
You can also look into your browser which proxy server is defined for your site.
For example, in the Internet Explorer under: Tools > Internet Options > LAN settings > Advanced
You can also search the web for tools which provide such information. For example (untested): http://www.sharewareconnection.com/titles/proxy-settings.htm
HTTP response is returned via the
RESPONSE
specification of the REQUEST DOCUMENT statement. See also
HTTP/HTTPS Responses
Redirected and Denied in the
Statements documentation.
The range for these errors is 8300 ff.
Especially error NAT8304 gives more detailed information about a failing HTTP request.
As the TCP/IP error number may be different depending on the installed environment, the text returned by NAT8304 is the best reference.
Additional information:
See buffer RDOCWA at Offset 480
Quite often these errors are ICU errors: Recommendation is to
set profile or session parameter CPCVERR to
OFF.
To determine if a problem is related to the Natural installation
or if there is a more general problem, you can do a
PING from within TSO.
For example, in the TSO command shell enter:
TSO PING www.google.com
This returns:
CS V1R9: Pinging host WWW.GOOGLE.COM (66.249.91.99) Ping #1 response took 0.018 seconds.
From within the Natural session you then can test the access to this website with the following small program.
For example, start Natural with:
NATvr CFICU=ON XML=(ON,RDOC=ON,PARSE=ON,RDP='HTTPPROX.HQ.SAG',RDPPORT=8080,RDNOP='*.EUR.AD.SAG; *.HQ.SAG;*.SOFTWAREAG.COM')
where vr represents the relevant product version.
These values from an internal environment and a profile were used
to store it. You have to get your settings for the keyword subparameters
RDP, RDPPORT and
RDNOP from your network administrator, or try the values
as defined in your browser (Internet Explorer).
Execute:
DEFINE DATA LOCAL 1 #RESULTXML (A) DYNAMIC 1 #RC (I4) END-DEFINE REQUEST DOCUMENT FROM "HTTP://WWW.GOOGLE.DE" RETURN HEADER ALL #HEADER RESPONSE #RC WRITE #RC WRITE #HEADER (AL=79) END
This is a frequent user error: An XML document is converted
implicitly or explicitly from one code page to another, for example, from
ISO-8859-1 to the code page found in system variable
*CODEPAGE.
The document's encoding PI encoding="ISO-8859-1", however, has not
been not adapted to the changed encoding. In this case, the parser terminates
with an error already on the first character of the document to parse.
Set session parameter CPCVERR to
OFF.
Self-signed certificates can be used on an intranet server for test purposes, using the open ssl sdk. After these certificates have been imported into a key database or a RACF key ring, they must be flagged as trusted.
The necessary effort for the RACF key ring approach seems to be much higher than for using key databases. Key rings must be created for every user who wants to access an HTTPS server, whereas key databases can be shared by multiple users.
Proceed as follows:
In the TCP/IP configuration file, set the option
TTLS in the TCPCONFIG statement.
Configure and start the AT-TLS Policy Agent. This agent is called by TCP/IP on each new TCP connection to check if the connection is SSL.
Create the Policy Agent file containing the AT-TLS rules. The Policy Agent file contains the rules to stipulate which connection is SSL.
See also z/OS Communications Server: IP Configuration Guide, Chapter 18 Application Transparent Transport Layer Security (AT-TLS) data protection.
The Sample Policy Agent file defines all outgoing connections as
application controlled TLS. This should not affect any other TCP/IP application
except the Natural REQUEST DOCUMENT support, because the rule is
defined as application controlled. That means the application is allowed to set
the connection status as SSL. As long as the application does not set this
status, it is not affected. However, the Policy Agent file allows also to
restrict the application controlled SSL connections to particular ports, users
or address spaces. The sample expects the certificate database on the HFS file
/ u/admin/CERT.kdb.
TTLSRule ConnRule01~1
{
LocalAddrSetRef addr1
RemoteAddrSetRef addr1
LocalPortRangeRef portR1
Direction Outbound
Priority 255
TTLSGroupActionRef gAct1~AllUsersAsClient
TTLSEnvironmentActionRef eAct1~AllUsersAsClient
TTLSConnectionActionRef cAct1~AllUsersAsClient
}
TTLSGroupAction gAct1~AllUsersAsClient
{
TTLSEnabled On
Trace 6
}
TTLSEnvironmentAction eAct1~AllUsersAsClient
{
HandshakeRole Client
EnvironmentUserInstance 0
TTLSKeyringParmsRef keyR1
}
TTLSConnectionAction cAct1~AllUsersAsClient
{
HandshakeRole Client
TTLSCipherParmsRef cipher1~AT-TLS__Silver
TTLSConnectionAdvancedParmsRef cAdv1~AllUsersAsClient
Trace 0
}
TTLSConnectionAdvancedParms cAdv1~AllUsersAsClient
{
ApplicationControlled On
}
TTLSKeyringParms keyR1
{
Keyring /u/admin/CERT.kdb
KeyringStashFile /u/admin/CERT.sth
}
TTLSCipherParms cipher1~AT-TLS__Silver
{
V3CipherSuites TLS_RSA_WITH_DES_CBC_SHA
V3CipherSuites TLS_RSA_WITH_3DES_EDE_CBC_SHA
V3CipherSuites TLS_RSA_WITH_AES_128_CBC_SHA
}
IpAddrSet addr1
{
Prefix 0.0.0.0/0
}
PortRange portR1
{
Port 1024-65535
}
Check Policy-Agent job output JESMSGLG for:
EZZ8771I PAGENT CONFIG POLICY PROCESSING COMPLETE FOR <your TCP/IP address space>: TTLS
This message indicates a successful initialization.
Check Policy-Agent job output JESMSGLG for:
EZZ8438I PAGENT POLICY DEFINITIONS CONTAIN ERRORS FOR <your TCP/IP address space>: TTLS
This message indicates errors in the configuration file. Check the
syslog.log file for further information.
Does the configuration rule cover the client?
Check syslog.log for:
EZD1281I TTLS Map CONNID: 00002909 LOCAL: 10.20.91.61..1751 REMOTE: 10.20.91.117..443 JOBNAME: KSP USERID: KSP TYPE: OutBound STATUS: Appl Control RULE: ConnRule01 ACTIONS: gAct1 eAct1 AllUsersAsClient
The above entry indicates that the connection to Port 443 by user
KSP is application controlled.
See also z/OS V1R8.0 Comm Svr: IP Diagnosis Guide: 3.23, Chapter 29 Diagnosing Application Transparent Transport Layer Security (AT-TLS)
See Comm Svr: IP Configuration Reference, Chapter 20 Syslog deamon and Comm Svr: IP Configuration Guide, Chapter 1.5.1 Configuring the syslog daemon (syslogd)
Find return code RC and corresponding
GSK_ function name in P-agent trace.
See System SSL Programming and locate the
RC in Chapter 12.1 SSL Function Return
Codes.
Sample trace with trace=255:
EZD1281I TTLS Map CONNID: 00002909 LOCAL: 10.20.91.61..1751 REMOTE: 10.20.91.117..443 JOBNAME: KSP USERID: KSP TYPE: OutBound STATUS: A EZD1283I TTLS Event GRPID: 00000003 ENVID: 00000000 CONNID: 00002909 RC: 0 Connection Init EZD1282I TTLS Start GRPID: 00000003 ENVID: 00000002 CONNID: 00002909 Initial Handshake ACTIONS: gAct1 eAct1 AllUsersAsClient HS-Client EZD1284I TTLS Flow GRPID: 00000003 ENVID: 00000002 CONNID: 00002909 RC: 0 Call GSK_SECURE_SOCKET_OPEN - 7EE4F718 EZD1284I TTLS Flow GRPID: 00000003 ENVID: 00000002 CONNID: 00002909 RC: 0 Set GSK_SESSION_TYPE - CLIENT EZD1284I TTLS Flow GRPID: 00000003 ENVID: 00000002 CONNID: 00002909 RC: 0 Set GSK_V3_CIPHER_SPECS - 090A2F EZD1284I TTLS Flow GRPID: 00000003 ENVID: 00000002 CONNID: 00002909 RC: 0 Set GSK_FD - 00002909 EZD1284I TTLS Flow GRPID: 00000003 ENVID: 00000002 CONNID: 00002909 RC: 0 Set GSK_USER_DATA - 7EEE9B50 EZD1284I TTLS Flow GRPID: 00000003 ENVID: 00000002 CONNID: 00002909 RC: 435 Call GSK_SECURE_SOCKET_INIT - 7EE4F718 EZD1283I TTLS Event GRPID: 00000003 ENVID: 00000002 CONNID: 00002909 RC: 435 Initial Handshake 00000000 7EEE8118 EZD1286I TTLS Error GRPID: 00000003 ENVID: 00000002 CONNID: 00002909 JOBNAME: KSP USERID: KSP RULE: ConnRule01 RC: 435 Initial Handshake EZD1283I TTLS Event GRPID: 00000003 ENVID: 00000002 CONNID: 00002909 RC: 0 Connection Close 00000000 7EEE8118
Below is a list of resources that you may find useful.
Below is a collection of links that may be of interest.
World Wide Web Consortium (W3C): http://www.w3.org/
Extensible Markup Language (XML): http://www.w3.org/XML/
HyperText Markup Language (HTML) Home Page: http://www.w3.org/MarkUp/
W3 Schools: https://www.w3schools.com/