MensBeam
/
lax
3
0
Fork 0
Code Issues 3 Pull Requests Releases Activity
Browse Source

People fetching for entries

master
J. King 4 years ago
parent
f37de9a5c2
commit
6c7db5e762
7 changed files with 287 additions and 5 deletions
Whitespace
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Split View
Diff Options
Show Stats Download Patch File Download Diff File
  1. 8
      lib/Parser/XML/Construct.php
  2. 29
      lib/Parser/XML/Entry.php
  3. 102
      tests/cases/XML/entry-atom.yaml
  4. 54
      tests/cases/XML/entry-mixed.yaml
  5. 54
      tests/cases/XML/entry-other.yaml
  6. 26
      tests/cases/XML/entry-rss1.yaml
  7. 19
      tests/cases/XML/entry-rss2.yaml

8
lib/Parser/XML/Construct.php
View File

@ -159,9 +159,9 @@ abstract class Construct {
* - Full Name <user@example.com>
* - Full Name <mailto:user@example.com>
*/
protected function fetchPeople(string $query, string $role): ?PersonCollection {
protected function fetchPeople(string $query, string $role, ?\DOMNode $context = null): ?PersonCollection {
$out = new PersonCollection;
foreach ($this->fetchString($query, ".+", true) ?? [] as $person) {
foreach ($this->fetchString($query, ".+", true, $context) ?? [] as $person) {
$p = new Person;
if (preg_match("/^([^@\s]+@\S+) \((.+?)\)$/", $person, $match)) { // tests "user@example.com (Full Name)" form
if ($this->validateMail($match[1])) {
@ -304,8 +304,8 @@ abstract class Construct {
}
/** Finds and parses Atom person-constructs, and returns a collection of Person objects */
protected function fetchAtomPeople(string $query, string $role): ?PersonCollection {
$nodes = $this->xpath->query($query, $this->subject);
protected function fetchAtomPeople(string $query, string $role, \DOMNode $context = null): ?PersonCollection {
$nodes = $this->xpath->query($query, $context ?? $this->subject);
$out = new PersonCollection;
foreach ($nodes as $node) {
$p = new Person;

29
lib/Parser/XML/Entry.php
View File

@ -123,7 +123,21 @@ class Entry extends Construct implements \MensBeam\Lax\Parser\Entry {
}
public function getPeople(): PersonCollection {
return new PersonCollection;
// first try getting authors and contributors in the entry itself
$authors = $this->getAuthors($this->subject);
$contributors = $this->getContributors($this->subject) ?? new PersonCollection;
// if there are no authors but there is an Atom <source> element, get both authors and contributors from the source
if (!$authors) {
$src = $this->fetchElement("atom:source");
if ($src) {
$authors = $this->getAuthors($src) ?? new PersonCollection;
$srcContributors = $this->getContributors($src) ?? new PersonCollection;
} else {
$authors = new PersonCollection;
}
}
// merge all three lists
return $authors->merge($contributors, $srcContributors ?? new PersonCollection);
}
public function getCategories(): CategoryCollection {
@ -166,4 +180,17 @@ class Entry extends Construct implements \MensBeam\Lax\Parser\Entry {
}
return null;
}
protected function getAuthors(\DOMNode $context): ?PersonCollection {
return $this->fetchAtomPeople("atom:author", "author", $context) // Atom authors
?? $this->fetchPeople("dc:creator|dct:creator", "author", $context) // Dublin Core creators
?? $this->fetchPeople("rss2:author", "author", $context) // RSS 2.0 authors
?? $this->fetchPeople("gplay:author", "author", $context) // Google Play authors
?? $this->fetchPeople("apple:author", "author", $context); // iTunes authors
}
protected function getContributors(\DOMNode $context): ?PersonCollection {
return $this->fetchAtomPeople("atom:contributor", "contributor", $context) // Atom contributors
?? $this->fetchPeople("dc:contributor|dct:contributor", "contributor", $context); // Dublin Core contributors
}
}

102
tests/cases/XML/entry-atom.yaml
View File

@ -147,3 +147,105 @@ Content and summary:
summary: 'Plain summary'
- content: 'Plain content'
- summary: 'Plain summary'
Authors and contributors:
input: >
<feed xmlns="http://www.w3.org/2005/Atom">
<entry>
<author><name>John Doe</name></author>
</entry>
<entry>
<author><name>John Doe</name></author>
<author><name>Jane Doe</name></author>
</entry>
<entry>
<author><name>John Doe</name></author>
<contributor><name>Larry</name></contributor>
</entry>
<entry>
<author><name>John Doe</name></author>
<author><name>Jane Doe</name></author>
<contributor><name>Larry</name></contributor>
<contributor><name>Curly</name></contributor>
<contributor><name>Moe</name></contributor>
</entry>
<entry>
<author><name>John Doe</name></author>
<source>
<author><name>Jane Doe</name></author>
</source>
</entry>
<entry>
<author><name>John Doe</name></author>
<source>
<contributor><name>Larry</name></contributor>
</source>
</entry>
<entry>
<source>
<author><name>Jane Doe</name></author>
</source>
</entry>
<entry>
<source>
<author><name>Jane Doe</name></author>
<contributor><name>Curly</name></contributor>
</source>
</entry>
<entry>
<source>
<author><name>Jane Doe</name></author>
<contributor><name>Curly</name></contributor>
</source>
<contributor><name>Larry</name></contributor>
</entry>
</feed>
output:
format: atom
version: '1.0'
entries:
- people:
- name: 'John Doe'
role: author
- people:
- name: 'John Doe'
role: author
- name: 'Jane Doe'
role: author
- people:
- name: 'John Doe'
role: author
- name: Larry
role: contributor
- people:
- name: 'John Doe'
role: author
- name: 'Jane Doe'
role: author
- name: Larry
role: contributor
- name: Curly
role: contributor
- name: Moe
role: contributor
- people:
- name: 'John Doe'
role: author
- people:
- name: 'John Doe'
role: author
- people:
- name: 'Jane Doe'
role: author
- people:
- name: 'Jane Doe'
role: author
- name: Curly
role: contributor
- people:
- name: 'Jane Doe'
role: author
- name: Larry
role: contributor
- name: Curly
role: contributor

54
tests/cases/XML/entry-mixed.yaml
View File

@ -39,3 +39,57 @@ Update and creation dates:
- dateCreated: '2020-03-03T00:00:00Z'
dateModified: '2020-01-01T00:00:00Z'
- dateModified: '2020-03-03T01:00:00+01:00'
Entry author:
input: >
<rss><channel xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
<item>
<dc:creator>jane.doe@example.com (Jane Doe)</dc:creator>
<author>john.doe@example.com (John Doe)</author>
</item>
<item>
<dc:contributor>Larry</dc:contributor>
<author>john.doe@example.com (John Doe)</author>
</item>
<item>
<dc:contributor>Larry</dc:contributor>
<dc:contributor>Curly</dc:contributor>
<atom:source>
<author>jane.doe@example.com (Jane Doe)</author>
<dc:contributor>Moe</dc:contributor>
</atom:source>
</item>
<item>
<author>Jane Doe</author>
<atom:source>
<atom:author><atom:name>John Doe</atom:name></atom:author>
<dc:contributor>Moe</dc:contributor>
</atom:source>
</item>
</channel></rss>
output:
format: rss
entries:
- people:
- name: 'Jane Doe'
mail: 'jane.doe@example.com'
role: author
- people:
- name: 'John Doe'
mail: 'john.doe@example.com'
role: author
- name: Larry
role: contributor
- people:
- name: 'Jane Doe'
mail: 'jane.doe@example.com'
role: author
- name: Larry
role: contributor
- name: Curly
role: contributor
- name: Moe
role: contributor
- people:
- name: 'Jane Doe'
role: author

54
tests/cases/XML/entry-other.yaml
View File

@ -48,3 +48,57 @@ Google Play summary:
format: rss
entries:
- summary: 'Plain summary'
iTunes authors:
input: >
<rss><channel xmlns:itunes="http://www.itunes.com/dtds/podcast-1.0.dtd">
<item>
<itunes:author>jane.doe@example.com (Jane Doe)</itunes:author>
<itunes:author>john.doe@example.com (John Doe)</itunes:author>
</item>
</channel></rss>
output:
format: rss
entries:
- people:
- name: 'Jane Doe'
mail: 'jane.doe@example.com'
role: author
- name: 'John Doe'
mail: 'john.doe@example.com'
role: author
Google Play authors:
input: >
<rss><channel xmlns:play="http://www.google.com/schemas/play-podcasts/1.0">
<item>
<play:author>jane.doe@example.com (Jane Doe)</play:author>
<play:author>john.doe@example.com (John Doe)</play:author>
</item>
</channel></rss>
output:
format: rss
entries:
- people:
- name: 'Jane Doe'
mail: 'jane.doe@example.com'
role: author
- name: 'John Doe'
mail: 'john.doe@example.com'
role: author
Mixed podcast authors: # Google Play is arbitrarily preferred
input: >
<rss><channel xmlns:itunes="http://www.itunes.com/dtds/podcast-1.0.dtd" xmlns:play="http://www.google.com/schemas/play-podcasts/1.0">
<item>
<itunes:author>jane.doe@example.com (Jane Doe)</itunes:author>
<play:author>john.doe@example.com (John Doe)</play:author>
</item>
</channel></rss>
output:
format: rss
entries:
- people:
- name: 'John Doe'
mail: 'john.doe@example.com'
role: author

26
tests/cases/XML/entry-rss1.yaml
View File

@ -234,3 +234,29 @@ Dublin Core summary:
- summary: 'Plain summary'
- summary: 'Plain summary'
- summary: 'Abstract'
Dublin Core creators and contributors:
input: >
<rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns="http://purl.org/rss/1.0/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:term="http://purl.org/dc/terms/">
<item>
<dc:creator>john.doe@example.com (John Doe)</dc:creator>
<term:creator>jane.doe@example.com (Jane Doe)</term:creator>
<dc:contributor>Larry</dc:contributor>
<term:contributor>Curly</term:contributor>
</item>
</rdf:RDF>
output:
format: rdf
version: '1.0'
entries:
- people:
- name: 'John Doe'
mail: 'john.doe@example.com'
role: author
- name: 'Jane Doe'
mail: 'jane.doe@example.com'
role: author
- name: Larry
role: contributor
- name: Curly
role: contributor

19
tests/cases/XML/entry-rss2.yaml
View File

@ -140,3 +140,22 @@ Entry content:
format: rss
entries:
- content: {loose: 'Loose content'}
Entry author:
input: >
<rss><channel>
<item>
<author>jane.doe@example.com (Jane Doe)</author>
<author>john.doe@example.com (John Doe)</author>
</item>
</channel></rss>
output:
format: rss
entries:
- people:
- name: 'Jane Doe'
mail: 'jane.doe@example.com'
role: author
- name: 'John Doe'
mail: 'john.doe@example.com'
role: author

Write Preview
Loading…
Cancel
Save