shaarli / netscape-bookmark-parser Goto Github PK

TODO:

• add PHPUnit configuration
• add a basic test
• add basic Travis support

There is no rule on how tags should be separated (and for anything for that matter), and we have a few unit tests where tags are actually separated by a whitespace.

However, it seems that the comma is a common separator for tags.
Regarding shaarli/Shaarli#594, I suggest that if tags contains a comma, we use the comma as a separator, otherwise we use the whitespace.

• Add test coverage for parsing bookmark exports from popular browsers:
  • #21 Firefox
  • #22 Chromium / Chrome
  • #23 IE

X-issue from shaarli/Shaarli#1726 - quoting @cy7yz2rj:

Shaarli is case insensitive for tags but should be case preserving when importing.

Steps for reproducing:

export bookmarks to HTML file from Tools > Export Database
delete all bookmarks
import the file from step 1
all tags are now lowercase despite having correct case information in the HTML file

Firefox correctly preserves case when importing the same bookmark file.

The release zip files and tarballs (such as https://github.com/shaarli/netscape-bookmark-parser/archive/v1.0.1.tar.gz) don't include composer.json.

I think it should be included in releases, so that composer install will work. Also Debian has tools for packaging PHP libraries (pkg-php-tools) that use Composer.

use composer require shaarli/netscape-bookmark-parser
no rule to match 'icon'

download zip
There are matching rules about ‘icon‘’

I exported my database from Shaarli v0.6.4, so before you refactored the bookmark exporting.

In this version, the first link is on the same line as the title:

<!DOCTYPE NETSCAPE-Bookmark-file-1>
<!-- This is an automatically generated file.
     It will be read and overwritten.
     DO NOT EDIT! -->
<!-- Shaarli all bookmarks export on 2016/08/08 10:04:26 -->
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=UTF-8">
<TITLE>Bookmarks</TITLE>
<H1>Bookmarks</H1><DT><A HREF="https://gist.github.com/rodrigomuniz/8408734" ADD_DATE="1470640652" PRIVATE="1" TAGS="css">A Pen by Rodrigo Muniz.</A>
<DD>simple on/off button 
<DT><A HREF="https://www.addedbytes.com/articles/for-beginners/url-rewriting-for-beginners/" ADD_DATE="1469950052" PRIVATE="0" TAGS="apache">URL Rewriting for Beginners - Web Development in Brighton - Added Bytes</A>

When I try to import back this file (using your PR with this parser), the first link is not imported. We would expect it to be able to import files from Shaarli older version.

Goal

• provide a Debian package to enable packaging Shaarli > 0.8

TODO

• this repo:
  • #27 keep Composer metadata
  • add a CHANGELOG.md
• create/register a Debian package/project
  • add Debian-specific metadata
  • automate metadata generation (copyright, changelog, etc.)

Resources

Meta:

• http://keepachangelog.com/en/0.3.0/

Shaarli package:

• https://packages.debian.org/sid/shaarli
• https://wiki.debian.org/FreedomBox/Manual

Debian packaging:

• https://www.debian.org/doc/devel-manuals#packaging-tutorial
• https://www.debian.org/doc/debian-policy/
• https://www.debian.org/doc/manuals/developers-reference/
• https://wiki.debian.org/PackagingWithGit
• https://wiki.debian.org/HowToPackageForDebian
• https://feeding.cloud.geek.nz/posts/whats-in-a-debian-directory/

Debian tools for PHP packaging:

• https://packages.debian.org/sid/pkg-php-tools
• https://sources.debian.net/src/pkg-php-tools/1.34/debian/README.Composer/
• https://wiki.debian.org/ReleaseGoals/pkg-php-tools

Debian CI:

• https://ci.debian.net/
• https://ci.debian.net/doc/file.MAINTAINERS.html
• https://ci.debian.net/doc/file.TUTORIAL.html

It would be more elegant and maintainable to describe the Netscape variants as a grammar, and generate/write a corresponding tokenizer/lexer/parser to process data.

Inspired by:

• https://tomassetti.me/parsing-in-java/
• https://tomassetti.me/parsing-in-python/

See:

• https://en.wikipedia.org/wiki/Comparison_of_parser_generators
• https://stackoverflow.com/questions/8002617/php-lexer-and-parser-generator
• https://stackoverflow.com/questions/2093228/lex-and-yacc-in-php
• https://stackoverflow.com/questions/2245962/is-there-an-alternative-for-flex-bison-that-is-usable-on-8-bit-embedded-systems
• https://stackoverflow.com/questions/25049751/constructing-an-abstract-syntax-tree-with-a-list-of-tokens/25106688#25106688

TODO:

• add Composer metadata
  • authors
  • supported PHP versions
  • version
• kafene#6 - tag and release a new version under kafene/netscape-bookmark-parser
• create a Packagist account for Shaarli community packages
• #26 - update Composer metadata to use the shaarli/netscape-bookmark-parser namespace
• tag and release a new version
• create the actual package entry
• generate the package archive
• setup a GitHub -> Packagist webhook

See:

• https://getcomposer.org/doc/04-schema.md
• https://packagist.org/

Hello! I'm trying to use this in one of my Service classes in a Laravel project, but I can't seem to get it to work.

I've run

composer require shaarli/netscape-bookmark-parser

I've added to composer.json:

"autoload": {
    "psr-4": {
        "App\\": "app/",
        "Database\\Factories\\": "database/factories/",
        "Database\\Seeders\\": "database/seeders/",
        "Shaarli\\NetscapeBookmarkParser\\":"vendor/shaarli/netscape-bookmark-parser/"
    }
},

I've run

composer dump-autoload

The following line of code yields the proceeding Symfony \ Component \ ErrorHandler \ Error \ FatalError error:

$parser = new \Shaarli\NetscapeBookmarkParser\NetscapeBookmarkParser();

Cannot declare class NetscapeBookmarkParser, because the name is already in use

I've tried some other approaches with no success. If I remove (or change) the autoload line from composer, then the error is that the class NetscapeBookmarkParser couldn't be found.

Anyone know what I'm doing wrong here? Thanks!

When trying to import an empty bookmarks file, the process fails with a PHP error:

ksort() expects parameter 1 to be array, null given at /app/vendor/shaarli/netscape-bookmark-parser/NetscapeBookmarkParser.php:203

Instead of this behavior, the parser should either throw a custom exception, or simply return an empty array. The former would be preferable to be as explicit as possible.

Hi Arthur,

I recently happen to need a php parser for a bookmark file and stumped upon Shaarli's repository.
I cleaned up as much code as I could without introducing any backward compatibility breaks so that Shaarli community can benefit from improvements as well.

But, I plan to make the parser more compatible and easier to use in my Symfony project, and I'm afraid this will introduce a lot of BC breaking changes.

Would you be willing to take hold of the significant workload needed for a hypothetical 4.0 integration into Shaarli project, or do you believe I should create and maintain my own fork ?

Here are the changes I am planning to introduce:

• Rename class to "NetscapeBookmarkDecoder" for nomenclature consistency.
• 'parseBoolean' method should always return a boolean value / refactor tests accordingly.
• All methods from parser should be private except for entrypoint / remove tests accordingly.
• Remove dependency on Katzgrau\KLogger and remove logging altogether.
• Remove useless "parseFile" method which violates Single Responsibility Principle / update documentation accordingly.
• Replace parameters from constructor with "$context" array and class constants: This will drastically change class signature but will allow for Dependency Injection and Symfony autowiring.

And finally:

• Make parser implement "Symfony\Component\Serializer\Encoder\DecoderInterface". I understand this change would create an unwanted dependency on Symfony Serializer but I am willing to make a version for Shaarli if necessary.
• Rename "parseString" method to "decode" for nomenclature consistency and Inferface conformity.
• Create "supportsDecoding" method for Inferface conformity and Symfony services integration.

New features (maybe):

• Use Data Transer Object design pattern for bookmark.
• Include "ICON" property.
• Create "NetscapeBookmarkEncoder" to return bookmark file from array.
• Rename properties to be (mostly) compatible with "https://schema.org" structured data vocabulary:
  • 'uri' => 'url'
  • 'title' => 'name'
  • 'icon' => 'image'
  • 'note' => 'description'
  • 'tags' => 'tags'
  • 'time' => 'dateCreated'
  • 'pub' => 'published'

Issue

When parsing a bookmark whose description contains tabs (tabulation character), each one is replaced by a single space.

This could break formatting if the description contains formatted text, code examples, or terminal output.

I had a problem importing descriptions from an old scuttle installation.

Is it possible that the regexp pattern is too specific? I had to change the pattern in line 148
https://github.com/shaarli/netscape-bookmark-parser/blob/master/NetscapeBookmarkParser.php#L148

from
/note="(.*?)"<\/a>/i
to
/note="(.*?)"/i

ie. removing the requirement for the description to be immediately followed by the closing anchor-tag. (my export had attributes like the tags and the date etc in between the description and the closing anchor.

also, the description in the scuttle export is called "description", instead of "note". so maybe the parser could check for both versions?