Download the PHP package actengage/talon without Composer
On this page you can find all versions of the php package actengage/talon. It is possible to download/install these versions without Composer. Possible dependencies are resolved automatically.
FAQ
Example:
In general, it is recommended to use always a project to download your libraries. In an application normally there is more than one library needed.
- Some hosting areas are not accessible by a terminal or SSH. Then it is not possible to use Composer.
- To use Composer is sometimes complicated. Especially for beginners.
- Composer needs much resources. Sometimes they are not available on a simple webspace.
- If you are using private repositories you don't need to share your credentials. You can set up everything on our site and then you provide a simple download link to your team member.
- Simplify your Composer build process. Use our own command line tool to download the vendor folder as binary. This makes your build process faster and you don't need to expose your credentials for private repositories.
Informations about the package talon
Talon (PHP)
Extract the original message from an email reply chain. A PHP port of mailgun/talon, validated against ~42K real-world Python talon outputs at 99.66% parity.
Installation
Requires PHP 8.1+, ext-dom, ext-mbstring. Auto-registers in Laravel via package discovery.
Usage
Facade
When $contentType is null (the default), the body is scanned for HTML block-level tags (<html>, <body>, <div>, <p>, <br>, <table>, <li>, etc.). If any are present the input is treated as HTML; otherwise as plain text. Pass an explicit content type to override.
Service
Lower-level API
For direct access to the text-mode primitives:
What it handles
| HTML | Plain text |
|---|---|
| Gmail, Outlook 2003–2013, Zimbra, Windows Mail | > quotation blocks (≥3 consecutive) |
Top-level <blockquote> |
On <date>, <person> wrote: in 9 languages |
From: / Date: header blocks (text and tail) |
-----Original Message----- and variants |
| Known quote-container IDs | Multi-line splitters (≤6 lines) |
| Two-pass for nested forwarded guards | Inline replies preserved; forwarded messages skipped |
UTF-8 throughout (mb_* for offsets); both \n and \r\n delimiters.
Behaviour & limits
- Plain-text extraction caps at the first 1,000 lines (
TextQuotations::MAX_LINES). - HTML extraction returns the original unchanged when
treeToText()produces more than 10,000 lines (large marketing emails are passed through, since they rarely contain reply chains). - The HTML pipeline runs twice (
Talon::extractFromHtmlcallsextractFromHtmlOncetwice) to mirror Pythontalon.batch. The second pass catches forwarded-message guards that block the first. - Inputs without recognisable HTML block-level tags are returned as-is to avoid misfires on plain-text bodies that happen to contain
<email@...>brackets.
Python parity
This port is intentionally close to mailgun/talon. The implementation preserves:
- lxml-style
el.text/el.tailtraversal inTalon::walkForText - Exact XPath strings for Outlook splitter detection
mg:tail()-equivalent matching incutFromBlockCase 2 viafollowing-sibling::node()[1][self::text()]- Mandatory-newline
[^\n]+\nper header field inRE_FROM_COLON_OR_DATE_COLON(PCRE backtracking equivalent of Python's[^\n$]+\n) - Splitter pattern check order and regex flags
- Checkpoint stamping order (append, not prepend) so markers land on the last line of multi-line text blocks
Validated against the full Active Engagement mailbox dataset (41,977 messages):
| Metric | Count | % |
|---|---|---|
| Compared | 41,977 | 100% |
| Matches | 41,835 | 99.66% |
| Mismatches | 142 | 0.34% |
Remaining mismatches are all pre-existing data artefacts (stored Python results computed against a previous version of the body, or invalid-UTF-8 sequences that PHP/libxml2 and Python/lxml decode differently).
Development
145 Pest tests covering the cutters, splitter detection, marker logic, multilingual splitters, multi-line splitters, inline replies, forwarded-message guards, and end-to-end extraction. The codebase is held to PHPStan level max (with Larastan extensions), formatted with Laravel Pint, and refactor-checked with Rector.
CI runs all four checks in parallel on every push and PR (.github/workflows/ci.yml), with composer / vendor / phpstan / rector caches keyed off composer.lock and the workflow concurrency-grouped per ref.
Releases
Versioning is managed with Changesets. Add a changeset whenever you make a user-facing change:
On push to main, .github/workflows/release.yml opens a "Version Packages" PR. Merging that PR bumps the version, writes CHANGELOG.md, and tags the release. The package is private: true in package.json — no npm publish happens.
Public API
These methods form the stability surface — anything else is internal.
Talon::extractFrom(string $body, ?string $contentType = null): stringTalon::detectContentType(string $body): stringTalon::extractFromHtml(string $html): stringTalon::extractFromPlain(string $text): stringTextQuotations::extract(string $text): stringTextQuotations::isSplitter(string $line): ?stringTextQuotations::markLines(array $lines): stringTextQuotations::processMarkedLines(array $lines, string $markers, array &$flags = []): array
The HtmlQuotations::cut* methods and checkpoint helpers are public for testing but should be considered internal.
License
MIT. Originally derived from mailgun/talon (Apache 2.0).
All versions of talon with dependencies
ext-dom Version *
ext-mbstring Version *
illuminate/support Version ^10.0|^11.0|^12.0|^13.0