Switch from nimquery to q

This commit is contained in:
Zed 2019-06-26 18:51:21 +02:00
parent b51f1cbbd4
commit 02fcd7b880
4 changed files with 60 additions and 47 deletions

View File

@ -11,4 +11,8 @@ bin = @["nitter"]
# Dependencies # Dependencies
requires "nim >= 0.19.9" requires "nim >= 0.19.9"
requires "regex", "nimcrypto", "norm", "jester", "nimquery#head" requires "norm >= 1.0.11"
requires "jester >= 0.4.1"
requires "regex >= 0.11.2"
requires "q >= 0.0.7"
requires "nimcrypto >= 0.3.9"

View File

@ -1,8 +1,8 @@
import httpclient, asyncdispatch, htmlparser, times import httpclient, asyncdispatch, htmlparser, times
import sequtils, strutils, strformat, json, xmltree, uri import sequtils, strutils, strformat, json, xmltree, uri
import nimquery, regex import regex
import ./types, ./parser import ./types, ./parser, ./parserutils
const const
agent = "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/71.0.3578.98 Safari/537.36" agent = "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/71.0.3578.98 Safari/537.36"
@ -162,7 +162,7 @@ proc getProfile*(username: string): Future[Profile] {.async.} =
if html.isNil: if html.isNil:
return Profile() return Profile()
if not html.querySelector(".ProfileCard-sensitiveWarningContainer").isNil: if not html.select(".ProfileCard-sensitiveWarningContainer").isNil:
return await getProfileFallback(username, headers) return await getProfileFallback(username, headers)
result = parsePopupProfile(html) result = parsePopupProfile(html)

View File

@ -1,10 +1,9 @@
import xmltree, sequtils, strtabs, strutils, strformat, json import xmltree, sequtils, strtabs, strutils, strformat, json
import nimquery
import ./types, ./parserutils, ./formatters import ./types, ./parserutils, ./formatters
proc parsePopupProfile*(node: XmlNode): Profile = proc parsePopupProfile*(node: XmlNode): Profile =
let profile = node.querySelector(".profile-card") let profile = node.select(".profile-card")
if profile.isNil: return if profile.isNil: return
result = Profile( result = Profile(
@ -24,9 +23,9 @@ proc parseIntentProfile*(profile: XmlNode): Profile =
fullname: profile.getName("a.fn.url.alternate-context"), fullname: profile.getName("a.fn.url.alternate-context"),
username: profile.getUsername(".nickname"), username: profile.getUsername(".nickname"),
bio: profile.getBio("p.note"), bio: profile.getBio("p.note"),
userpic: profile.querySelector(".profile.summary").getAvatar("img.photo"), userpic: profile.select(".profile.summary").getAvatar("img.photo"),
verified: not profile.querySelector("li.verified").isNil, verified: not profile.select("li.verified").isNil,
protected: not profile.querySelector("li.protected").isNil, protected: not profile.select("li.protected").isNil,
banner: getBanner(profile) banner: getBanner(profile)
) )
@ -55,7 +54,11 @@ proc parseQuote*(quote: XmlNode): Quote =
result.getQuoteMedia(quote) result.getQuoteMedia(quote)
proc parseTweet*(tweet: XmlNode): Tweet = proc parseTweet*(node: XmlNode): Tweet =
let tweet = node.select(".tweet")
if tweet.isNil():
return Tweet()
result = Tweet( result = Tweet(
id: tweet.getAttr("data-item-id"), id: tweet.getAttr("data-item-id"),
link: tweet.getAttr("data-permalink-path"), link: tweet.getAttr("data-permalink-path"),
@ -74,29 +77,28 @@ proc parseTweet*(tweet: XmlNode): Tweet =
result.retweetBy = some(by.stripText()) result.retweetBy = some(by.stripText())
result.retweetId = some(tweet.getAttr("data-retweet-id")) result.retweetId = some(tweet.getAttr("data-retweet-id"))
let quote = tweet.querySelector(".QuoteTweet-innerContainer") let quote = tweet.select(".QuoteTweet-innerContainer")
if not quote.isNil: if not quote.isNil:
result.quote = some(parseQuote(quote)) result.quote = some(parseQuote(quote))
proc parseTweets*(node: XmlNode): Tweets = proc parseTweets*(node: XmlNode): Tweets =
if node.isNil: return if node.isNil or node.kind == xnText: return
node.querySelectorAll(".tweet").map(parseTweet) for n in node.selectAll(".stream-item"):
result.add parseTweet(n)
proc parseConversation*(node: XmlNode): Conversation = proc parseConversation*(node: XmlNode): Conversation =
result = Conversation( result = Conversation(
tweet: parseTweet(node.querySelector(".permalink-tweet-container > .tweet")), tweet: parseTweet(node.select(".permalink-tweet-container")),
before: parseTweets(node.querySelector(".in-reply-to")) before: parseTweets(node.select(".in-reply-to"))
) )
let replies = node.querySelector(".replies-to") let replies = node.select(".replies-to")
if replies.isNil: return if replies.isNil: return
result.after = parseTweets(replies.querySelector(".ThreadedConversation--selfThread")) result.after = parseTweets(replies.select(".ThreadedConversation--selfThread"))
for reply in replies.querySelectorAll("li > .stream-items"): for reply in replies.select(".stream-items"):
let thread = parseTweets(reply) result.replies.add parseTweets(reply)
if not thread.anyIt(it in result.after):
result.replies.add thread
proc parseVideo*(node: JsonNode): Video = proc parseVideo*(node: JsonNode): Video =
let track = node{"track"} let track = node{"track"}

View File

@ -1,30 +1,39 @@
import xmltree, htmlparser, strtabs, strformat, times import xmltree, htmlparser, strtabs, strformat, times
import nimquery, regex import regex
import ./types, ./formatters, ./api import ./types, ./formatters, ./api
from q import nil
const const
thumbRegex = re".+:url\('([^']+)'\)" thumbRegex = re".+:url\('([^']+)'\)"
gifRegex = re".+thumb/([^\.']+)\.jpg.*" gifRegex = re".+thumb/([^\.']+)\.jpg.*"
proc selectAll*(node: XmlNode; selector: string): seq[XmlNode] =
q.select(node, selector)
proc select*(node: XmlNode; selector: string): XmlNode =
let nodes = node.selectAll(selector)
if nodes.len > 0: nodes[0] else: nil
proc getAttr*(node: XmlNode; attr: string; default=""): string = proc getAttr*(node: XmlNode; attr: string; default=""): string =
if node.isNil or node.attrs.isNil: return default if node.isNil or node.attrs.isNil: return default
return node.attrs.getOrDefault(attr) return node.attrs.getOrDefault(attr)
proc selectAttr*(node: XmlNode; selector: string; attr: string; default=""): string = proc selectAttr*(node: XmlNode; selector: string; attr: string; default=""): string =
let res = node.querySelector(selector) let res = node.select(selector)
if res == nil: "" else: res.getAttr(attr, default) if res == nil: "" else: res.getAttr(attr, default)
proc selectText*(node: XmlNode; selector: string): string = proc selectText*(node: XmlNode; selector: string): string =
let res = node.querySelector(selector) let res = node.select(selector)
result = if res == nil: "" else: res.innerText() result = if res == nil: "" else: res.innerText()
proc getHeader(profile: XmlNode): XmlNode = proc getHeader(profile: XmlNode): XmlNode =
result = profile.querySelector(".permalink-header") result = profile.select(".permalink-header")
if result.isNil: if result.isNil:
result = profile.querySelector(".stream-item-header") result = profile.select(".stream-item-header")
if result.isNil: if result.isNil:
result = profile.querySelector(".ProfileCard-userFields") result = profile.select(".ProfileCard-userFields")
proc isVerified*(profile: XmlNode): bool = proc isVerified*(profile: XmlNode): bool =
getHeader(profile).selectText(".Icon.Icon--verified").len > 0 getHeader(profile).selectText(".Icon.Icon--verified").len > 0
@ -39,25 +48,23 @@ proc getUsername*(profile: XmlNode; selector: string): string =
profile.selectText(selector).strip(chars={'@', ' '}) profile.selectText(selector).strip(chars={'@', ' '})
proc emojify*(node: XmlNode) = proc emojify*(node: XmlNode) =
for i in node.querySelectorAll(".Emoji"): for i in node.selectAll(".Emoji"):
i.add newText(i.getAttr("alt")) i.add newText(i.getAttr("alt"))
proc getQuoteText*(tweet: XmlNode): string = proc getQuoteText*(tweet: XmlNode): string =
let text = tweet.querySelector(".QuoteTweet-text") let text = tweet.select(".QuoteTweet-text")
emojify(text) emojify(text)
result = stripText(selectText(text, ".tweet-text")) result = stripText(text.innerText())
result = stripTwitterUrls(result) result = stripTwitterUrls(result)
proc getTweetText*(tweet: XmlNode): string = proc getTweetText*(tweet: XmlNode): string =
let let
selector = ".tweet-text > a.twitter-timeline-link.u-hidden" quote = tweet.select(".QuoteTweet")
link = tweet.selectAttr(selector, "data-expanded-url") text = tweet.select(".tweet-text")
quote = tweet.querySelector(".QuoteTweet") link = text.selectAttr("a.twitter-timeline-link.u-hidden", "data-expanded-url")
text = tweet.querySelector(".tweet-text")
emojify(text) emojify(text)
result = stripText(selectText(text, ".tweet-text")) result = stripText(text.innerText())
if not quote.isNil and link.len > 0: if not quote.isNil and link.len > 0:
result = result.replace(link, "") result = result.replace(link, "")
@ -65,7 +72,7 @@ proc getTweetText*(tweet: XmlNode): string =
result = stripTwitterUrls(result) result = stripTwitterUrls(result)
proc getTime(tweet: XmlNode): XmlNode = proc getTime(tweet: XmlNode): XmlNode =
tweet.querySelector(".js-short-timestamp") tweet.select(".js-short-timestamp")
proc getTimestamp*(tweet: XmlNode): Time = proc getTimestamp*(tweet: XmlNode): Time =
let time = getTime(tweet).getAttr("data-time", "0") let time = getTime(tweet).getAttr("data-time", "0")
@ -92,7 +99,7 @@ proc getBanner*(tweet: XmlNode): string =
result = "background-color: #161616" result = "background-color: #161616"
proc getPopupStats*(profile: var Profile; node: XmlNode) = proc getPopupStats*(profile: var Profile; node: XmlNode) =
for s in node.querySelectorAll( ".ProfileCardStats-statLink"): for s in node.selectAll( ".ProfileCardStats-statLink"):
let text = s.getAttr("title").split(" ")[0] let text = s.getAttr("title").split(" ")[0]
case s.getAttr("href").split("/")[^1] case s.getAttr("href").split("/")[^1]
of "followers": profile.followers = text of "followers": profile.followers = text
@ -101,7 +108,7 @@ proc getPopupStats*(profile: var Profile; node: XmlNode) =
proc getIntentStats*(profile: var Profile; node: XmlNode) = proc getIntentStats*(profile: var Profile; node: XmlNode) =
profile.tweets = "?" profile.tweets = "?"
for s in node.querySelectorAll( "dd.count > a"): for s in node.selectAll( "dd.count > a"):
let text = s.innerText() let text = s.innerText()
case s.getAttr("href").split("/")[^1] case s.getAttr("href").split("/")[^1]
of "followers": profile.followers = text of "followers": profile.followers = text
@ -111,7 +118,7 @@ proc getTweetStats*(tweet: Tweet; node: XmlNode) =
tweet.replies = "0" tweet.replies = "0"
tweet.retweets = "0" tweet.retweets = "0"
tweet.likes = "0" tweet.likes = "0"
for action in node.querySelectorAll(".ProfileTweet-actionCountForAria"): for action in node.selectAll(".ProfileTweet-actionCountForAria"):
let text = action.innerText.split() let text = action.innerText.split()
case text[1] case text[1]
of "replies": tweet.replies = text[0] of "replies": tweet.replies = text[0]
@ -126,30 +133,30 @@ proc getGif(player: XmlNode): Gif =
Gif(url: url, thumb: thumb) Gif(url: url, thumb: thumb)
proc getTweetMedia*(tweet: Tweet; node: XmlNode) = proc getTweetMedia*(tweet: Tweet; node: XmlNode) =
for photo in node.querySelectorAll(".AdaptiveMedia-photoContainer"): for photo in node.selectAll(".AdaptiveMedia-photoContainer"):
tweet.photos.add photo.attrs["data-image-url"] tweet.photos.add photo.attrs["data-image-url"]
let player = node.querySelector(".PlayableMedia") let player = node.select(".PlayableMedia")
if player.isNil: if player.isNil:
return return
if "gif" in player.getAttr("class"): if "gif" in player.getAttr("class"):
tweet.gif = some(getGif(player.querySelector(".PlayableMedia-player"))) tweet.gif = some(getGif(player.select(".PlayableMedia-player")))
elif "video" in player.getAttr("class"): elif "video" in player.getAttr("class"):
tweet.video = some(Video()) tweet.video = some(Video())
proc getQuoteMedia*(quote: var Quote; node: XmlNode) = proc getQuoteMedia*(quote: var Quote; node: XmlNode) =
let sensitive = node.querySelector(".QuoteTweet--sensitive") let sensitive = node.select(".QuoteTweet--sensitive")
if not sensitive.isNil: if not sensitive.isNil:
quote.sensitive = true quote.sensitive = true
return return
let media = node.querySelector(".QuoteMedia") let media = node.select(".QuoteMedia")
if not media.isNil: if not media.isNil:
quote.thumb = some(media.selectAttr("img", "src")) quote.thumb = some(media.selectAttr("img", "src"))
let badge = node.querySelector(".AdaptiveMedia-badgeText") let badge = node.select(".AdaptiveMedia-badgeText")
let gifBadge = node.querySelector(".Icon--gifBadge") let gifBadge = node.select(".Icon--gifBadge")
if not badge.isNil: if not badge.isNil:
quote.badge = some(badge.innerText()) quote.badge = some(badge.innerText())