basexdb / basex Goto Github PK

Example:

delete node //*[position() <= 10000000]

might create 10 mio temporary entries in the pending update list, the mem. consumption of which should be further reduced.

Implement WebDAV binding for BaseX: http://www.webdav.org/specs/rfc4918.html

eXist's solution: http://exist.sourceforge.net/webdav.html

Fix remaining issues; see e.g.

https://github.com/BaseXdb/basex/blob/3e46b6f1af952fe4acac20a1129d1758f4375ffb/src/test/java/org/basex/test/query/FNFormatIntTest.java

• add "encoding" parser option to CSV and TEXT importers.
• add option to GUI dialogs.

Check if result of this function is correct:

file:write-binary("file.out",
  xs:base64Binary(xs:hexBinary("484950")))

Skip single malformed documents if collections are parsed, and output names of erroneous docs.

Support context item declarations in query prolog:
http://www.w3.org/TR/xquery-30/#id-context-item-declarations

Start from directory of main query to resolve module paths.
Otherwise, exceptions are thrown, à la...

[XQST0059] Module not found: ".../query.xqm".

Query:

declare function local:rec($x) {
  if($x)
  then
    for $a in 1 to $x
    order by $a
    return (local:rec($x - 1), $a)
  else
    0
};
local:rec(2)

Problem: Wrong result; may as well apply to GFLWOR queries.
Reason:: Same Order.vl instance is used (and overwritten) by recursive calls.

Some XPath suggestions are shown in the input field, even if typed in path will yield no more results.

Document: factbook.xml

Input: /m/

Suggestions: continent, country, ...

BaseX 6.5.1 on Windows XP SP3 and classical Czech keyboard as seen here http://freepages.genealogy.rootsweb.ancestry.com/~elainetmaddox/czech_keyboard.jpg

• type some text into xquery edit box
• place cursor in the middle of the line
• try to use Shift-Home od Shift-End to select text from the cursor to the line start or end

The cursor does not move at all

Steps to reproduce:

public static void main(final String[] args) throws BaseXException {
    final Context ctx = new Context();
    new CreateDB("test",
            "<?xml version=\"1.0\"?><root feed=\"http://www.example.com/index.rss\" />")
            .execute(ctx);
    System.out.println(new XQuery("/root").execute(ctx));
}

and then open the db in the GUI -- it will show the following db:

info table yields (in the GUI):

PRE DIS SIZ ATS NS KIND CONTENT

0 1 3 1 0 DOC
1 1 2 2 0 ELEM root
2 1 1 1 0 ATTR feed=""

The following query..

for $x in //buch
where $x//autor//text() contains text 'Krieger'
return $x

..is rewritten to..

FTIndexAccess("ams", "Krieger")/ancestor::*:autor/
  parent::*:lieferinfo/parent::*:abschnitt/
  parent::*:buch/parent::*:buch
    [ancestor::document-node()]

parent::*:buch should only occur once.

Example:
let $numbers := (1234.567, 789, 1234567.765)
for $i in $numbers return format-number($i, "#,###.##")

BaseX output:
1,234.57 789 1234,567.76

Should output:
1,234.57 789 1,234,567.76

see http://www.w3.org/TR/xquery-30/#id-decimal-format-decl

• remove Commands.hidden() method
• add documentation for CHECK command
• remove QUIT command

Entries are missing in the log, so these steps can't be reviewed.

Explore GUI and/or Web version.

See mailing list:
https://mailman.uni-konstanz.de/pipermail/basex-talk/2011-January/001081.html

…are not detected in the XQuery editor.

Example:

for $a in
  for $a in <a>1</a>
  return $a/text()
return <x>{ $a }</x>

Currently, inner variable $a shadows outer variables – which is wrong. Instead, inner variable declaration should be reset after each evaluation of RETURN.

The following glob filter for creating files...

SET CREATEFILTER *.xml

...won't accept files with multiple dots, such as...

a.b.xml

This is because a trailing dot...

SET CREATEFILTER *.

...is rewritten to a pattern that is supposed to find files without any dots, such as

a

Support for notification triggers for clients; to be added in 6.7

Settings in configuration file for SERVERPORT and JAXRXPORT might clash with default settings in Java classes command line settings:

...WARN::failed [email protected]:7998: java.net.BindException: Address already in use
...WARN::failed Server@1e0bc08: java.net.BindException: Address already in use

GUI Language cannot be changed in 6.6 (still working in 6.5.1, and on command line)

zip:entries("*.docx") doesnt show the real paths for the entries

dbnotify: Like inotify/FSEvents on FS side.
Relates to GUI external events update?

Example:

1. start BaseXServer
2. run python AddExample.py
3. Path 'world//World.xml' added in 32.19 ms. --> see the double "/"

…sent in via the mailing list:

1. Double click on the blue bar to get a full view (All other views are send to the background. When double clicking the other views appear again)
2. Put a close button on the blue bar
3. Folder view; add entry in the right-click menu to collapse in or out all nodes
4. Add support for intelligent xpath autocompletion in the search bar
5. Possibility in the right click menu to duplicate a node
6. Possibility to place a node in comment /uncomment
7. In text mode highlight a node when the cursor is standing on an opening or ending tag, so that the corresponding pair will be remark
8. Make "Enter" equal to double clicking when navigation in a xml
9. Option to show line numbers in text mode
10. Launch a xml in a browser, xml editor or IDE from the menu

Long run:

1. A XML Grid View (like xmlSpy or Stylus Studio)
2. XSD support for validating and generating schemas
3. XSLT support for transforming a xml db into another xml db and alternative query posibillities
4. Visualizing differences between two databases
5. Export xml to text, csv, xls, json etc
6. XQuery Documentation Generator (XQDoc Support)
7. Use colors to show structure of an xml in the visualization views (this option can be turned on and off)
8. Support for forms (Create forms as a front end for basex and distribute this together with the xml db as a jar file)
9. XML editing mode and read only mode (xacml)
10. Focus on support for adding thousands of xml files into one db

Currently, the query compiler makes no difference between the current context item and the data instance(s), which is/are valid in the current context. This complicates optimizations, e.g. index access on collections, as a collection, which is bound as current context, may point to several db's.

Proposed solution: Add current data instance in analogy with the context item; update reference whenever necessary/appropriate (e.g. in mixed location steps).

Missing:

- fn:function-name(...)
- fn:function-arity(...)
- fn:map(...)
- fn:filter(...)
- fn:fold-left(...)
- fn:fold-right(...)
- fn:map-pairs(...)

Partially Supported:

fn:format-integer() 
fn:format-number() 
fn:format-dateTime() 
fn:format-date() 
fn:format-time()

A command is needed to mirror (i.e., create copies of existing) databases. Such a command can be applied to perform updates on the database copy and keep the main instance available for read-only requests.

Syntax:

COPY DB [name] [newname]

Howto reproduce:
Start BaseX
Open the Query Editor
Enter some text,  + {A,X,C,V} is ignored

…are now rewritten to always use the index, no matter if the index terms have been stemmed, etc. As a consequence, the following query

doc("input")//li[ text() contains text "Exercise" ]

returns no results anymore. Solution: index match options have to be applied to query terms.

Include version information in command-line help texts, à la

 BaseX 6.5.1 [Standalone]
 Usage: BaseX [-bdiosuvVwz] [-cq] [file]
 [file]     Execute XQuery file
 ...

Should not be allowed (XPTY0004):

xs:date("2000-07-01") + xs:duration("P1Y")

Wrong result ("0000-07-01", expected: "-0001-07-01"):

xs:date("2000-07-01") - xs:yearMonthDuration("P2000Y")

Runtime exception:

xs:date("2000-07-01") - xs:yearMonthDuration("P2000Y") - xs:yearMonthDuration("P2000Y")

If the path index can evaluate how many results a location path will return, it could be wrapped with a positional filter predicate:

Old: /a/b/c

New: (/a/b/c)[position() <= max]

Improve placement of fn:trace() output.

Index will not be used for queries such as the following...

//*[ text() contains text 'term' ]

...if a language has been chosen. Index will be applied to the following query instead:

declare ft-option using language 'English';
//*[text() contains text 'specific']

This is kinda counter intuitive and should be handled by...

• using/recognizing English as default setting
• giving some more user (yet concise) feedback in the GUI Properties dialog

An even more far reaching approach would be to use the index options as default for evaluating full-text requests. This would be the most intuitive solution, but it could conflict with the specs, test cases, etc. Another problem could be that the database to be touched is not always known at compile time.

the db title in the GUI randomly changes its content (map visualization)

it takes random text snippets out of the db as db title

sample db can be found here:
http://students.fachschaft.inf.uni-konstanz.de/~krause/spiegel.zip

for $ n in //product//edition_no
where $n/data() > 100
return insert node text {'- greater 100'} into $n

on ****otek.xml

java.lang.ArrayIndexOutOfBoundsException: 66
  org.basex.util.Compress.pull(Compress.java:143)
  org.basex.util.Compress.unpack(Compress.java:109)
  org.basex.data.DiskData.txt(DiskData.java:207)
  org.basex.data.DiskData.text(DiskData.java:174)
  org.basex.data.Serializer.node(Serializer.java:298)
  org.basex.data.Nodes.serialize(Nodes.java:230)
  org.basex.data.Nodes.serialize(Nodes.java:224)
  org.basex.gui.view.text.TextView.setText(TextView.java:155)
  org.basex.gui.view.text.TextView.refreshContext(TextView.java:116)
  org.basex.gui.view.text.TextView.refreshUpdate(TextView.java:127)
  org.basex.gui.view.ViewNotifier.update(ViewNotifier.java:220)
  org.basex.gui.GUI.exec(GUI.java:441)
  org.basex.gui.GUI$5.run(GUI.java:351)

Example

replace value of node db:open-pre('probit', 511) with 'x'
delete node db:open-pre('probit', 511)

add "Hello World!"
"null" (Line 1): The element type "x" must be terminated by the matching end-tag "".

Check if w3 test suite issues remain to be fixed.

Conceptual: XQuery functions vs. importers/exporters?

Invalid XML with 2 Namespaces in an element.

as proposed by lukas

The following snippet creates invalid XML:

element test {
    attribute {QName('foo', 'foo')} { 'bar' },
    attribute {QName('bar', 'bar')} { 'baz' }
}

BaseX 6.5.1 on Windows XP SP3 and classical Czech keyboard as seen here http://freepages.genealogy.rootsweb.ancestry.com/~elainetmaddox/czech_keyboard.jpg

type some text into xquery edit box and then use Shift-§ key (key beside enter key) to write exclamation mark - all content gets deleted and only the exclamation mark remains.

In past I reported similar problem with problem to write $ by means of AltGr-ů (second key left to enter - see https://mailman.uni-konstanz.de/pipermail/basex-talk/2011-February/001184.html)

Especially helpful for the development of XQuery modules. Some conceptual questions have to be answered as well, e.g.:

• which query is to be evaluated if several tabs are opened?
• which query is to be evaluated if the current tab contains a module?