tildes-community/tildes-parser
tildes-community
/
tildes-parser
1
Fork 0
Code Issues 1 Releases Activity

Compare commits

base: tildes-community:ba008a3ccaf5d1e4f4288429632c9d389e236675
Branches Tags
tildes-community:main
..
compare: tildes-community:a2b9efb94be83d5490bf260f36db3d0624199cfe
Branches Tags
tildes-community:main

No commits in common. "ba008a3ccaf5d1e4f4288429632c9d389e236675" and "a2b9efb94be83d5490bf260f36db3d0624199cfe" have entirely different histories.
ba008a3cca ... a2b9efb94b

13 changed files with 1 additions and 479 deletions
Show Stats Expand all files Collapse all files

3
source/from_str.rs
View File

@ -4,13 +4,12 @@ use std::str::FromStr;
use {duplicate::duplicate_item, scraper::Html};
use crate::{Group, GroupList, ParseError, Topic};
use crate::{Group, GroupList, ParseError};
#[duplicate_item(
_Struct;
[Group];
[GroupList];
[Topic];
)]
impl FromStr for _Struct {
type Err = ParseError;

2
source/lib.rs
View File

@ -28,11 +28,9 @@ pub(crate) mod error;
pub(crate) mod from_str;
pub(crate) mod group;
pub(crate) mod group_list;
pub(crate) mod topic;
pub use {
error::*,
group::{Group, GroupWikiLink},
group_list::{GroupList, GroupListSummary},
topic::*,
};

24
source/selectors.rs
View File

@ -28,28 +28,4 @@ lazy_static! {
/// Selector for group wiki links.
pub static ref GROUP_WIKI_LINKS: Selector = selector(r#"#sidebar [href*="/wiki/"]"#);
/// Selector for the topic comment count.
pub static ref TOPIC_COMMENT_COUNT: Selector = selector(".topic-comments-header h2");
/// Selector for the topic full byline.
pub static ref TOPIC_FULL_BYLINE: Selector = selector(".topic-full-byline");
/// Selector for a link topic's content.
pub static ref TOPIC_FULL_LINK: Selector = selector(".topic-full-link a");
/// Selector for the topic tag elements.
pub static ref TOPIC_FULL_TAGS: Selector = selector(".topic-full-tags a");
/// Selector for a text topic's content.
pub static ref TOPIC_FULL_TEXT: Selector = selector(".topic-full-text");
/// Selector for the main topic `<article>`.
pub static ref TOPIC_MAIN_ARTICLE: Selector = selector("main > .topic-full");
/// Selector for a topic toast warning.
pub static ref TOPIC_TOAST_WARNING: Selector = selector(".toast.toast-warning");
/// Selector for the topic vote count.
pub static ref TOPIC_VOTE_COUNT: Selector = selector(".topic-voting-votes");
}

170
source/topic.rs
View File

@ -1,170 +0,0 @@
//! Parsing for `/~<group>/<topic-id>`.
use scraper::Html;
use crate::{
regexes::DUPLICATE_WHITESPACE_RE,
selectors::{
TOPIC_COMMENT_COUNT, TOPIC_FULL_BYLINE, TOPIC_FULL_LINK, TOPIC_FULL_TAGS,
TOPIC_FULL_TEXT, TOPIC_MAIN_ARTICLE, TOPIC_TOAST_WARNING, TOPIC_VOTE_COUNT,
},
utilities::select_first_element_text,
ParseError,
};
/// A Tildes topic.
#[derive(Debug)]
pub struct Topic {
/// The name of the author.
pub author: TopicAuthor,
/// The amount of comments the topic has. Comments themselves have to be
/// parsed separately.
pub comment_total: i32,
/// The content of the topic.
pub content: TopicContent,
/// The unique ID of the topic.
pub id: String,
/// Whether the topic is locked.
pub is_locked: bool,
/// Whether the topic is official (not yet implemented, is always false).
///
/// TODO: Add is_official. This isn't possible right now because topics don't
/// have any indicator of being marked as official. The only place it's shown
/// is in the topic listing. See #787 in the Tildes issue tracker.
pub is_official: bool,
/// All tags applied to the topic.
pub tags: Vec<String>,
/// The amount of votes the topic has received.
pub vote_count: i32,
}
/// All the different ways a topic author can be represented.
#[derive(Debug)]
pub enum TopicAuthor {
/// The normal case, where the topic author is available.
Name(String),
/// The topic was posted by Tildes itself.
///
/// Technically the user for this is [Tildes](https://tildes.net/user/tildes)
/// but in the topic it says "Automatically posted <date>" where the username
/// normally goes, so may as well special-case it here too.
Scheduled,
/// The user was banned, deleted their account or disassociated the topic from
/// their account.
Unknown,
}
/// The different types of content a topic can have.
#[derive(Debug)]
pub enum TopicContent {
/// The topic is a link topic pointing to an external site.
Link(String),
/// The topic is a text topic with a HTML body.
Text(String),
/// The topic's content is no longer available.
Unknown,
}
impl Topic {
/// Parses a [`Topic`] from a [`scraper::Html`] tree.
pub fn from_html(html: &Html) -> Result<Self, ParseError> {
let topic_article_element = html
.select(&TOPIC_MAIN_ARTICLE)
.next()
.ok_or(ParseError::MissingExpectedHtml)?;
let topic_byline =
select_first_element_text(topic_article_element, &TOPIC_FULL_BYLINE)
.map(|byline| {
DUPLICATE_WHITESPACE_RE
.replace_all(&byline, " ")
.to_string()
})
.ok_or(ParseError::MissingExpectedHtml)?;
let author = if topic_byline.starts_with("Automatically posted") {
TopicAuthor::Scheduled
} else if topic_byline.ends_with("unknown user") {
TopicAuthor::Unknown
} else {
TopicAuthor::Name(
topic_byline
.split(" ")
.last()
.ok_or(ParseError::MissingExpectedHtml)?
.to_string(),
)
};
let comment_total = if let Some(comment_total) =
select_first_element_text(topic_article_element, &TOPIC_COMMENT_COUNT)
{
comment_total
.split(" ")
.next()
.map(|count| count.parse::<i32>())
.ok_or(ParseError::MissingExpectedHtml)?
.map_err(|_| ParseError::MissingExpectedHtml)?
} else {
0
};
let content = if let Some(link_content) =
topic_article_element.select(&TOPIC_FULL_LINK).next()
{
TopicContent::Link(link_content.text().collect::<String>())
} else if let Some(text_content) =
topic_article_element.select(&TOPIC_FULL_TEXT).next()
{
TopicContent::Text(text_content.inner_html().trim().to_string())
} else {
TopicContent::Unknown
};
let id = topic_article_element
.value()
.id()
.ok_or(ParseError::MissingExpectedHtml)?[6..]
.to_string();
let is_locked =
select_first_element_text(topic_article_element, &TOPIC_TOAST_WARNING)
.map(|toast| toast.contains("This topic is locked."))
.unwrap_or_default();
let tags = topic_article_element
.select(&TOPIC_FULL_TAGS)
.map(|tag| tag.text().collect::<String>())
.collect::<Vec<_>>();
let vote_count =
select_first_element_text(topic_article_element, &TOPIC_VOTE_COUNT)
.map(|vote_count| vote_count.parse::<i32>())
.ok_or(ParseError::MissingExpectedHtml)?
.map_err(|_| ParseError::MissingExpectedHtml)?;
let topic = Topic {
author,
comment_total,
content,
id,
is_locked,
is_official: false, // TODO: Implement this once it can be done.
tags,
vote_count,
};
Ok(topic)
}
}

41
tests/samples/topic_deleted.html
View File

@ -1,41 +0,0 @@
<!DOCTYPE html>
<html lang="en">
<head>
<title>Deleted topic sample for topic.rs</title>
</head>
<body class="theme-white">
<main>
<article id="topic-666" class="topic-full">
<header>
<div class="topic-voting">
<span class="topic-voting-votes">0</span>
<span class="topic-voting-label">votes</span>
</div>
<h1></h1>
<div class="topic-full-byline">
Posted
<time class="time-responsive" datetime="2018-04-26T21:22:47Z"
title="2018-04-26 21:22:47 UTC" data-abbreviated="Apr 26, 2018">
<span class="time-responsive-full">April 26, 2018</span>
</time> by unknown user
</div>
<div class="topic-full-tags"></div>
</header>
<div class="text-error">Topic deleted by author</div>
<section class="topic-comments">
<div class="divider"></div>
<header class="topic-comments-header">
<h2>2 comments</h2>
</header>
</section>
</article>
</main>
</body>
</html>

50
tests/samples/topic_link.html
View File

@ -1,50 +0,0 @@
<!DOCTYPE html>
<html lang="en">
<head>
<title>Link topic sample for topic.rs</title>
</head>
<body>
<main>
<article id="topic-456" class="topic-full">
<header>
<div class="topic-voting">
<span class="topic-voting-votes">10</span>
<span class="topic-voting-label">votes</span>
</div>
<h1>Link Topic Title</h1>
<div class="topic-full-byline">
Posted
<time class="time-responsive" datetime="2021-12-16T17:26:03Z"
title="2021-12-16 17:26:03 UTC" data-abbreviated="Dec 16, 2021">
<span class="time-responsive-full">December 16, 2021</span>
</time>
by
<a href="/user/AnotherUser" class="link-user">AnotherUser</a>
</div>
<div class="topic-full-tags">
Tags:
<a href="/~group?tag=four">four</a>,
<a href="/~group?tag=five">five</a>,
<a href="/~group?tag=six">six</a>
</div>
</header>
<div class="topic-full-link">
<a href="https://tildes.net/~test">https://tildes.net/~test</a>
</div>
<section class="topic-comments">
<div class="divider"></div>
<header class="topic-comments-header">
<h2>30 comments</h2>
</header>
</section>
</article>
</main>
</body>
</html>

46
tests/samples/topic_scheduled.html
View File

@ -1,46 +0,0 @@
<!DOCTYPE html>
<html lang="en">
<head>
<title>Text topic sample for topic.rs</title>
</head>
<body>
<main>
<article id="topic-789" class="topic-full">
<header>
<div class="topic-voting">
<span class="topic-voting-votes">50</span>
<span class="topic-voting-label">votes</span>
</div>
<h1>Scheduled Topic Title</h1>
<div class="topic-full-byline">
Automatically posted
<time class="time-responsive" datetime="2022-12-16T17:26:03Z"
title="2022-12-16 17:26:03 UTC" data-abbreviated="Dec 16, 2022">
<span class="time-responsive-full">December 16, 2022</span>
</time>
</div>
<div class="topic-full-tags">
Tags:
<a href="/~group?tag=recurring">recurring.weekly</a>
</div>
</header>
<div class="topic-full-text">
<p>Scheduled Topic Text</p>
</div>
<section class="topic-comments">
<div class="divider"></div>
<header class="topic-comments-header">
<h2>100 comments</h2>
</header>
</section>
</article>
</main>
</body>
</html>

54
tests/samples/topic_text.html
View File

@ -1,54 +0,0 @@
<!DOCTYPE html>
<html lang="en">
<head>
<title>Text topic sample for topic.rs</title>
</head>
<body>
<main>
<article id="topic-123" class="topic-full">
<header>
<div class="topic-voting">
<span class="topic-voting-votes">5</span>
<span class="topic-voting-label">votes</span>
</div>
<h1>Text Topic Title</h1>
<div class="topic-full-byline">
Posted
<time class="time-responsive" datetime="2022-12-16T17:26:03Z"
title="2022-12-16 17:26:03 UTC" data-abbreviated="Dec 16, 2022">
<span class="time-responsive-full">December 16, 2022</span>
</time>
by
<a href="/user/User" class="link-user">User</a>
</div>
<div class="topic-full-tags">
Tags:
<a href="/~group?tag=one">one</a>,
<a href="/~group?tag=two">two</a>,
<a href="/~group?tag=three">three</a>
</div>
</header>
<div class="topic-full-text">
<p>Topic Text</p>
</div>
<div class="toast toast-warning">
This topic is locked. New comments can not be posted.
</div>
<section class="topic-comments">
<div class="divider"></div>
<header class="topic-comments-header">
<h2>15 comments</h2>
</header>
</section>
</article>
</main>
</body>
</html>

14
tests/snapshots/topic__deleted.snap
View File

@ -1,14 +0,0 @@
---
source: tests/topic.rs
expression: topic
---
Topic {
author: Unknown,
comment_total: 2,
content: Unknown,
id: "666",
is_locked: false,
is_official: false,
tags: [],
vote_count: 0,
}

22
tests/snapshots/topic__link.snap
View File

@ -1,22 +0,0 @@
---
source: tests/topic.rs
expression: topic
---
Topic {
author: Name(
"AnotherUser",
),
comment_total: 30,
content: Link(
"https://tildes.net/~test",
),
id: "456",
is_locked: false,
is_official: false,
tags: [
"four",
"five",
"six",
],
vote_count: 10,
}

18
tests/snapshots/topic__scheduled.snap
View File

@ -1,18 +0,0 @@
---
source: tests/topic.rs
expression: topic
---
Topic {
author: Scheduled,
comment_total: 100,
content: Text(
"<p>Scheduled Topic Text</p>",
),
id: "789",
is_locked: false,
is_official: false,
tags: [
"recurring.weekly",
],
vote_count: 50,
}

22
tests/snapshots/topic__text.snap
View File

@ -1,22 +0,0 @@
---
source: tests/topic.rs
expression: topic
---
Topic {
author: Name(
"User",
),
comment_total: 15,
content: Text(
"<p>Topic Text</p>",
),
id: "123",
is_locked: true,
is_official: false,
tags: [
"one",
"two",
"three",
],
vote_count: 5,
}

14
tests/topic.rs
View File

@ -1,14 +0,0 @@
use std::fs::read_to_string;
use {insta::assert_debug_snapshot, tildes_parser::Topic};
#[test]
fn test_topic_parsing() {
let samples = ["link", "text", "deleted", "scheduled"];
for sample in samples {
let html =
read_to_string(format!("tests/samples/topic_{sample}.html")).unwrap();
let topic = &html.parse::<Topic>().unwrap();
assert_debug_snapshot!(sample, topic);
}
}