<!DOCTYPE html><html lang="en"><head><meta charset="utf-8"><meta name="viewport" content="width=device-width, initial-scale=1.0"><meta name="generator" content="rustdoc"><meta name="description" content="Source of the Rust file `src/client.rs`."><title>client.rs - source</title><script>if(window.location.protocol!=="file:")document.head.insertAdjacentHTML("beforeend","SourceSerif4-Regular-6b053e98.ttf.woff2,FiraSans-Italic-81dc35de.woff2,FiraSans-Regular-0fe48ade.woff2,FiraSans-MediumItalic-ccf7e434.woff2,FiraSans-Medium-e1aa3f0a.woff2,SourceCodePro-Regular-8badfe75.ttf.woff2,SourceCodePro-Semibold-aa29a496.ttf.woff2".split(",").map(f=>`<link rel="preload" as="font" type="font/woff2"href="../../static.files/${f}">`).join(""))</script><link rel="stylesheet" href="../../static.files/normalize-9960930a.css"><link rel="stylesheet" href="../../static.files/rustdoc-b7b9f40b.css"><meta name="rustdoc-vars" data-root-path="../../" data-static-root-path="../../static.files/" data-current-crate="scrapfly_sdk" data-themes="" data-resource-suffix="" data-rustdoc-version="1.95.0 (59807616e 2026-04-14)" data-channel="1.95.0" data-search-js="search-63369b7b.js" data-stringdex-js="stringdex-b897f86f.js" data-settings-js="settings-170eb4bf.js" ><script src="../../static.files/storage-41dd4d93.js"></script><script defer src="../../static.files/src-script-813739b1.js"></script><script defer src="../../src-files.js"></script><script defer src="../../static.files/main-5013f961.js"></script><noscript><link rel="stylesheet" href="../../static.files/noscript-f7c3ffd8.css"></noscript><link rel="alternate icon" type="image/png" href="../../static.files/favicon-32x32-eab170b8.png"><link rel="icon" type="image/svg+xml" href="../../static.files/favicon-044be391.svg"></head><body class="rustdoc src"><a class="skip-main-content" href="#main-content">Skip to main content</a><!--[if lte IE 11]><div class="warning">This old browser is unsupported and will most likely display funky things.</div><![endif]--><nav class="sidebar"><div class="src-sidebar-title"><h2>Files</h2></div></nav><div class="sidebar-resizer" title="Drag to resize sidebar"></div><main><section id="main-content" class="content" tabindex="-1"><div class="main-heading"><h1><div class="sub-heading">scrapfly_sdk/</div>client.rs</h1><rustdoc-toolbar></rustdoc-toolbar></div><div class="example-wrap digits-4"><pre class="rust"><code><a href=#1 id=1 data-nosnippet>1</a><span class="doccomment">//! HTTP client for the Scrapfly API.
<a href=#2 id=2 data-nosnippet>2</a>//!
<a href=#3 id=3 data-nosnippet>3</a>//! Built on `reqwest` with `rustls`. Single shared [`reqwest::Client`]
<a href=#4 id=4 data-nosnippet>4</a>//! re-used across every call.
<a href=#5 id=5 data-nosnippet>5</a>
<a href=#6 id=6 data-nosnippet>6</a></span><span class="kw">use </span>std::collections::HashMap;
<a href=#7 id=7 data-nosnippet>7</a><span class="kw">use </span>std::sync::Arc;
<a href=#8 id=8 data-nosnippet>8</a><span class="kw">use </span>std::time::Duration;
<a href=#9 id=9 data-nosnippet>9</a>
<a href=#10 id=10 data-nosnippet>10</a><span class="kw">use </span>futures_util::stream::{Stream, StreamExt};
<a href=#11 id=11 data-nosnippet>11</a><span class="kw">use </span>reqwest::header::{HeaderMap, HeaderValue, ACCEPT, CONTENT_TYPE, USER_AGENT};
<a href=#12 id=12 data-nosnippet>12</a><span class="kw">use </span>reqwest::{Method, Response, Url};
<a href=#13 id=13 data-nosnippet>13</a>
<a href=#14 id=14 data-nosnippet>14</a><span class="kw">use </span><span class="kw">crate</span>::config::crawler::CrawlerConfig;
<a href=#15 id=15 data-nosnippet>15</a><span class="kw">use </span><span class="kw">crate</span>::config::extraction::ExtractionConfig;
<a href=#16 id=16 data-nosnippet>16</a><span class="kw">use </span><span class="kw">crate</span>::config::scrape::ScrapeConfig;
<a href=#17 id=17 data-nosnippet>17</a><span class="kw">use </span><span class="kw">crate</span>::config::screenshot::ScreenshotConfig;
<a href=#18 id=18 data-nosnippet>18</a><span class="kw">use </span><span class="kw">crate</span>::enums::HttpMethod;
<a href=#19 id=19 data-nosnippet>19</a><span class="kw">use </span><span class="kw">crate</span>::error::{from_response, parse_retry_after, ApiError, ScrapflyError};
<a href=#20 id=20 data-nosnippet>20</a><span class="kw">use </span><span class="kw">crate</span>::monitoring::{
<a href=#21 id=21 data-nosnippet>21</a> CloudBrowserMonitoringOptions, MonitoringDataFormat, MonitoringMetricsOptions,
<a href=#22 id=22 data-nosnippet>22</a> MonitoringTargetMetricsOptions,
<a href=#23 id=23 data-nosnippet>23</a>};
<a href=#24 id=24 data-nosnippet>24</a><span class="kw">use </span><span class="kw">crate</span>::result::account::{AccountData, VerifyApiKeyResult};
<a href=#25 id=25 data-nosnippet>25</a><span class="kw">use </span><span class="kw">crate</span>::result::classify::{ClassifyRequest, ClassifyResult};
<a href=#26 id=26 data-nosnippet>26</a><span class="kw">use </span><span class="kw">crate</span>::result::crawler::{
<a href=#27 id=27 data-nosnippet>27</a> CrawlerArtifact, CrawlerArtifactType, CrawlerContents, CrawlerStartResponse, CrawlerStatus,
<a href=#28 id=28 data-nosnippet>28</a> CrawlerUrls,
<a href=#29 id=29 data-nosnippet>29</a>};
<a href=#30 id=30 data-nosnippet>30</a><span class="kw">use </span><span class="kw">crate</span>::result::extraction::ExtractionResult;
<a href=#31 id=31 data-nosnippet>31</a><span class="kw">use </span><span class="kw">crate</span>::result::scrape::{ResultData, ScrapeResult};
<a href=#32 id=32 data-nosnippet>32</a><span class="kw">use </span><span class="kw">crate</span>::result::screenshot::{ScreenshotMetadata, ScreenshotResult};
<a href=#33 id=33 data-nosnippet>33</a>
<a href=#34 id=34 data-nosnippet>34</a><span class="kw">const </span>DEFAULT_HOST: <span class="kw-2">&</span>str = <span class="string">"https://api.scrapfly.io"</span>;
<a href=#35 id=35 data-nosnippet>35</a><span class="kw">const </span>DEFAULT_CLOUD_BROWSER_HOST: <span class="kw-2">&</span>str = <span class="string">"https://browser.scrapfly.io"</span>;
<a href=#36 id=36 data-nosnippet>36</a><span class="kw">const </span>SDK_USER_AGENT: <span class="kw-2">&</span>str = <span class="string">"Scrapfly-Rust-SDK"</span>;
<a href=#37 id=37 data-nosnippet>37</a><span class="kw">const </span>DEFAULT_RETRIES: usize = <span class="number">3</span>;
<a href=#38 id=38 data-nosnippet>38</a><span class="kw">const </span>DEFAULT_RETRY_DELAY: Duration = Duration::from_secs(<span class="number">1</span>);
<a href=#39 id=39 data-nosnippet>39</a><span class="kw">const </span>DEFAULT_TIMEOUT: Duration = Duration::from_secs(<span class="number">150</span>);
<a href=#40 id=40 data-nosnippet>40</a>
<a href=#41 id=41 data-nosnippet>41</a><span class="doccomment">/// Request-inspection callback. Fires right before `send()`.
<a href=#42 id=42 data-nosnippet>42</a>///
<a href=#43 id=43 data-nosnippet>43</a>/// Used by the integration harness to record the outgoing method/URL/headers
<a href=#44 id=44 data-nosnippet>44</a>/// without wrapping the `reqwest::Client` in a middleware layer.
<a href=#45 id=45 data-nosnippet>45</a></span><span class="kw">pub type </span>OnRequest = Arc<<span class="kw">dyn </span>Fn(<span class="kw-2">&</span>Method, <span class="kw-2">&</span>Url, <span class="kw-2">&</span>HeaderMap) + Send + Sync>;
<a href=#46 id=46 data-nosnippet>46</a>
<a href=#47 id=47 data-nosnippet>47</a><span class="doccomment">/// Scrapfly API client. Cheap to `Clone` (the inner `reqwest::Client` is
<a href=#48 id=48 data-nosnippet>48</a>/// `Arc`'d so all clones share one connection pool).
<a href=#49 id=49 data-nosnippet>49</a></span><span class="attr">#[derive(Clone)]
<a href=#50 id=50 data-nosnippet>50</a></span><span class="kw">pub struct </span>Client {
<a href=#51 id=51 data-nosnippet>51</a> http: reqwest::Client,
<a href=#52 id=52 data-nosnippet>52</a> key: String,
<a href=#53 id=53 data-nosnippet>53</a> host: String,
<a href=#54 id=54 data-nosnippet>54</a> cloud_browser_host: String,
<a href=#55 id=55 data-nosnippet>55</a> on_request: <span class="prelude-ty">Option</span><OnRequest>,
<a href=#56 id=56 data-nosnippet>56</a>}
<a href=#57 id=57 data-nosnippet>57</a>
<a href=#58 id=58 data-nosnippet>58</a><span class="kw">impl </span>std::fmt::Debug <span class="kw">for </span>Client {
<a href=#59 id=59 data-nosnippet>59</a> <span class="kw">fn </span>fmt(<span class="kw-2">&</span><span class="self">self</span>, f: <span class="kw-2">&mut </span>std::fmt::Formatter<<span class="lifetime">'_</span>>) -> std::fmt::Result {
<a href=#60 id=60 data-nosnippet>60</a> f.debug_struct(<span class="string">"Client"</span>)
<a href=#61 id=61 data-nosnippet>61</a> .field(<span class="string">"host"</span>, <span class="kw-2">&</span><span class="self">self</span>.host)
<a href=#62 id=62 data-nosnippet>62</a> .field(<span class="string">"cloud_browser_host"</span>, <span class="kw-2">&</span><span class="self">self</span>.cloud_browser_host)
<a href=#63 id=63 data-nosnippet>63</a> .finish()
<a href=#64 id=64 data-nosnippet>64</a> }
<a href=#65 id=65 data-nosnippet>65</a>}
<a href=#66 id=66 data-nosnippet>66</a>
<a href=#67 id=67 data-nosnippet>67</a><span class="doccomment">/// Builder for [`Client`].
<a href=#68 id=68 data-nosnippet>68</a></span><span class="attr">#[derive(Default)]
<a href=#69 id=69 data-nosnippet>69</a></span><span class="kw">pub struct </span>ClientBuilder {
<a href=#70 id=70 data-nosnippet>70</a> api_key: <span class="prelude-ty">Option</span><String>,
<a href=#71 id=71 data-nosnippet>71</a> host: <span class="prelude-ty">Option</span><String>,
<a href=#72 id=72 data-nosnippet>72</a> cloud_browser_host: <span class="prelude-ty">Option</span><String>,
<a href=#73 id=73 data-nosnippet>73</a> timeout: <span class="prelude-ty">Option</span><Duration>,
<a href=#74 id=74 data-nosnippet>74</a> danger_accept_invalid_certs: bool,
<a href=#75 id=75 data-nosnippet>75</a> http_client: <span class="prelude-ty">Option</span><reqwest::Client>,
<a href=#76 id=76 data-nosnippet>76</a> on_request: <span class="prelude-ty">Option</span><OnRequest>,
<a href=#77 id=77 data-nosnippet>77</a>}
<a href=#78 id=78 data-nosnippet>78</a>
<a href=#79 id=79 data-nosnippet>79</a><span class="kw">impl </span>ClientBuilder {
<a href=#80 id=80 data-nosnippet>80</a> <span class="doccomment">/// Set the API key (required).
<a href=#81 id=81 data-nosnippet>81</a> </span><span class="kw">pub fn </span>api_key(<span class="kw-2">mut </span><span class="self">self</span>, key: <span class="kw">impl </span>Into<String>) -> <span class="self">Self </span>{
<a href=#82 id=82 data-nosnippet>82</a> <span class="self">self</span>.api_key = <span class="prelude-val">Some</span>(key.into());
<a href=#83 id=83 data-nosnippet>83</a> <span class="self">self
<a href=#84 id=84 data-nosnippet>84</a> </span>}
<a href=#85 id=85 data-nosnippet>85</a> <span class="doccomment">/// Override the API host.
<a href=#86 id=86 data-nosnippet>86</a> </span><span class="kw">pub fn </span>host(<span class="kw-2">mut </span><span class="self">self</span>, host: <span class="kw">impl </span>Into<String>) -> <span class="self">Self </span>{
<a href=#87 id=87 data-nosnippet>87</a> <span class="self">self</span>.host = <span class="prelude-val">Some</span>(host.into());
<a href=#88 id=88 data-nosnippet>88</a> <span class="self">self
<a href=#89 id=89 data-nosnippet>89</a> </span>}
<a href=#90 id=90 data-nosnippet>90</a> <span class="doccomment">/// Override the Cloud Browser host (`https://browser.scrapfly.io`).
<a href=#91 id=91 data-nosnippet>91</a> </span><span class="kw">pub fn </span>cloud_browser_host(<span class="kw-2">mut </span><span class="self">self</span>, host: <span class="kw">impl </span>Into<String>) -> <span class="self">Self </span>{
<a href=#92 id=92 data-nosnippet>92</a> <span class="self">self</span>.cloud_browser_host = <span class="prelude-val">Some</span>(host.into());
<a href=#93 id=93 data-nosnippet>93</a> <span class="self">self
<a href=#94 id=94 data-nosnippet>94</a> </span>}
<a href=#95 id=95 data-nosnippet>95</a> <span class="doccomment">/// Override the HTTP timeout (default 150s).
<a href=#96 id=96 data-nosnippet>96</a> </span><span class="kw">pub fn </span>timeout(<span class="kw-2">mut </span><span class="self">self</span>, t: Duration) -> <span class="self">Self </span>{
<a href=#97 id=97 data-nosnippet>97</a> <span class="self">self</span>.timeout = <span class="prelude-val">Some</span>(t);
<a href=#98 id=98 data-nosnippet>98</a> <span class="self">self
<a href=#99 id=99 data-nosnippet>99</a> </span>}
<a href=#100 id=100 data-nosnippet>100</a> <span class="doccomment">/// Accept invalid TLS certificates (tests / self-signed dev hosts).
<a href=#101 id=101 data-nosnippet>101</a> </span><span class="kw">pub fn </span>danger_accept_invalid_certs(<span class="kw-2">mut </span><span class="self">self</span>, v: bool) -> <span class="self">Self </span>{
<a href=#102 id=102 data-nosnippet>102</a> <span class="self">self</span>.danger_accept_invalid_certs = v;
<a href=#103 id=103 data-nosnippet>103</a> <span class="self">self
<a href=#104 id=104 data-nosnippet>104</a> </span>}
<a href=#105 id=105 data-nosnippet>105</a> <span class="doccomment">/// Inject a pre-built `reqwest::Client`. Bypasses the timeout /
<a href=#106 id=106 data-nosnippet>106</a> /// TLS-verify options.
<a href=#107 id=107 data-nosnippet>107</a> </span><span class="kw">pub fn </span>http_client(<span class="kw-2">mut </span><span class="self">self</span>, client: reqwest::Client) -> <span class="self">Self </span>{
<a href=#108 id=108 data-nosnippet>108</a> <span class="self">self</span>.http_client = <span class="prelude-val">Some</span>(client);
<a href=#109 id=109 data-nosnippet>109</a> <span class="self">self
<a href=#110 id=110 data-nosnippet>110</a> </span>}
<a href=#111 id=111 data-nosnippet>111</a> <span class="doccomment">/// Install a pre-send request callback (used by the integration runner
<a href=#112 id=112 data-nosnippet>112</a> /// to capture SDK-layer attribution without installing middleware).
<a href=#113 id=113 data-nosnippet>113</a> </span><span class="kw">pub fn </span>on_request(<span class="kw-2">mut </span><span class="self">self</span>, cb: OnRequest) -> <span class="self">Self </span>{
<a href=#114 id=114 data-nosnippet>114</a> <span class="self">self</span>.on_request = <span class="prelude-val">Some</span>(cb);
<a href=#115 id=115 data-nosnippet>115</a> <span class="self">self
<a href=#116 id=116 data-nosnippet>116</a> </span>}
<a href=#117 id=117 data-nosnippet>117</a> <span class="doccomment">/// Build the client.
<a href=#118 id=118 data-nosnippet>118</a> </span><span class="kw">pub fn </span>build(<span class="self">self</span>) -> <span class="prelude-ty">Result</span><Client, ScrapflyError> {
<a href=#119 id=119 data-nosnippet>119</a> <span class="kw">let </span>key = <span class="self">self</span>.api_key.ok_or(ScrapflyError::BadApiKey)<span class="question-mark">?</span>;
<a href=#120 id=120 data-nosnippet>120</a> <span class="kw">if </span>key.is_empty() {
<a href=#121 id=121 data-nosnippet>121</a> <span class="kw">return </span><span class="prelude-val">Err</span>(ScrapflyError::BadApiKey);
<a href=#122 id=122 data-nosnippet>122</a> }
<a href=#123 id=123 data-nosnippet>123</a>
<a href=#124 id=124 data-nosnippet>124</a> <span class="kw">let </span>http = <span class="kw">if let </span><span class="prelude-val">Some</span>(c) = <span class="self">self</span>.http_client {
<a href=#125 id=125 data-nosnippet>125</a> c
<a href=#126 id=126 data-nosnippet>126</a> } <span class="kw">else </span>{
<a href=#127 id=127 data-nosnippet>127</a> <span class="kw">let </span><span class="kw-2">mut </span>builder = reqwest::Client::builder()
<a href=#128 id=128 data-nosnippet>128</a> .timeout(<span class="self">self</span>.timeout.unwrap_or(DEFAULT_TIMEOUT))
<a href=#129 id=129 data-nosnippet>129</a> .user_agent(SDK_USER_AGENT);
<a href=#130 id=130 data-nosnippet>130</a> <span class="kw">if </span><span class="self">self</span>.danger_accept_invalid_certs {
<a href=#131 id=131 data-nosnippet>131</a> builder = builder.danger_accept_invalid_certs(<span class="bool-val">true</span>);
<a href=#132 id=132 data-nosnippet>132</a> }
<a href=#133 id=133 data-nosnippet>133</a> builder.build().map_err(ScrapflyError::Transport)<span class="question-mark">?
<a href=#134 id=134 data-nosnippet>134</a> </span>};
<a href=#135 id=135 data-nosnippet>135</a>
<a href=#136 id=136 data-nosnippet>136</a> <span class="prelude-val">Ok</span>(Client {
<a href=#137 id=137 data-nosnippet>137</a> http,
<a href=#138 id=138 data-nosnippet>138</a> key,
<a href=#139 id=139 data-nosnippet>139</a> host: <span class="self">self</span>.host.unwrap_or_else(|| DEFAULT_HOST.to_string()),
<a href=#140 id=140 data-nosnippet>140</a> cloud_browser_host: <span class="self">self
<a href=#141 id=141 data-nosnippet>141</a> </span>.cloud_browser_host
<a href=#142 id=142 data-nosnippet>142</a> .unwrap_or_else(|| DEFAULT_CLOUD_BROWSER_HOST.to_string()),
<a href=#143 id=143 data-nosnippet>143</a> on_request: <span class="self">self</span>.on_request,
<a href=#144 id=144 data-nosnippet>144</a> })
<a href=#145 id=145 data-nosnippet>145</a> }
<a href=#146 id=146 data-nosnippet>146</a>}
<a href=#147 id=147 data-nosnippet>147</a>
<a href=#148 id=148 data-nosnippet>148</a><span class="kw">impl </span>Client {
<a href=#149 id=149 data-nosnippet>149</a> <span class="doccomment">/// Start a new [`ClientBuilder`].
<a href=#150 id=150 data-nosnippet>150</a> </span><span class="kw">pub fn </span>builder() -> ClientBuilder {
<a href=#151 id=151 data-nosnippet>151</a> ClientBuilder::default()
<a href=#152 id=152 data-nosnippet>152</a> }
<a href=#153 id=153 data-nosnippet>153</a>
<a href=#154 id=154 data-nosnippet>154</a> <span class="doccomment">/// Return the configured API key.
<a href=#155 id=155 data-nosnippet>155</a> </span><span class="kw">pub fn </span>api_key(<span class="kw-2">&</span><span class="self">self</span>) -> <span class="kw-2">&</span>str {
<a href=#156 id=156 data-nosnippet>156</a> <span class="kw-2">&</span><span class="self">self</span>.key
<a href=#157 id=157 data-nosnippet>157</a> }
<a href=#158 id=158 data-nosnippet>158</a>
<a href=#159 id=159 data-nosnippet>159</a> <span class="doccomment">/// Return the configured API host.
<a href=#160 id=160 data-nosnippet>160</a> </span><span class="kw">pub fn </span>host(<span class="kw-2">&</span><span class="self">self</span>) -> <span class="kw-2">&</span>str {
<a href=#161 id=161 data-nosnippet>161</a> <span class="kw-2">&</span><span class="self">self</span>.host
<a href=#162 id=162 data-nosnippet>162</a> }
<a href=#163 id=163 data-nosnippet>163</a>
<a href=#164 id=164 data-nosnippet>164</a> <span class="doccomment">/// Return the configured Cloud Browser host.
<a href=#165 id=165 data-nosnippet>165</a> </span><span class="kw">pub fn </span>cloud_browser_host(<span class="kw-2">&</span><span class="self">self</span>) -> <span class="kw-2">&</span>str {
<a href=#166 id=166 data-nosnippet>166</a> <span class="kw-2">&</span><span class="self">self</span>.cloud_browser_host
<a href=#167 id=167 data-nosnippet>167</a> }
<a href=#168 id=168 data-nosnippet>168</a>
<a href=#169 id=169 data-nosnippet>169</a> <span class="doccomment">/// Build a URL by joining `path` onto the configured host.
<a href=#170 id=170 data-nosnippet>170</a> /// Crate-internal shim over `build_url`, used by `schedule.rs` to share
<a href=#171 id=171 data-nosnippet>171</a> /// the same auth + host wiring as the rest of the SDK without exposing
<a href=#172 id=172 data-nosnippet>172</a> /// the helper publicly.
<a href=#173 id=173 data-nosnippet>173</a> </span><span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">fn </span>build_url_public(
<a href=#174 id=174 data-nosnippet>174</a> <span class="kw-2">&</span><span class="self">self</span>,
<a href=#175 id=175 data-nosnippet>175</a> path: <span class="kw-2">&</span>str,
<a href=#176 id=176 data-nosnippet>176</a> query: <span class="kw-2">&</span>[(String, String)],
<a href=#177 id=177 data-nosnippet>177</a> ) -> <span class="prelude-ty">Result</span><Url, ScrapflyError> {
<a href=#178 id=178 data-nosnippet>178</a> <span class="self">self</span>.build_url(path, query)
<a href=#179 id=179 data-nosnippet>179</a> }
<a href=#180 id=180 data-nosnippet>180</a>
<a href=#181 id=181 data-nosnippet>181</a> <span class="doccomment">/// Crate-internal shim over `send_simple` — same rationale as
<a href=#182 id=182 data-nosnippet>182</a> /// `build_url_public`.
<a href=#183 id=183 data-nosnippet>183</a> </span><span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">async fn </span>send_simple_public(
<a href=#184 id=184 data-nosnippet>184</a> <span class="kw-2">&</span><span class="self">self</span>,
<a href=#185 id=185 data-nosnippet>185</a> method: Method,
<a href=#186 id=186 data-nosnippet>186</a> url: Url,
<a href=#187 id=187 data-nosnippet>187</a> headers: <span class="prelude-ty">Option</span><HeaderMap>,
<a href=#188 id=188 data-nosnippet>188</a> body: <span class="prelude-ty">Option</span><Vec<u8>>,
<a href=#189 id=189 data-nosnippet>189</a> ) -> <span class="prelude-ty">Result</span><Response, ScrapflyError> {
<a href=#190 id=190 data-nosnippet>190</a> <span class="self">self</span>.send_simple(method, url, headers, body).<span class="kw">await
<a href=#191 id=191 data-nosnippet>191</a> </span>}
<a href=#192 id=192 data-nosnippet>192</a>
<a href=#193 id=193 data-nosnippet>193</a> <span class="kw">fn </span>build_url(<span class="kw-2">&</span><span class="self">self</span>, path: <span class="kw-2">&</span>str, query: <span class="kw-2">&</span>[(String, String)]) -> <span class="prelude-ty">Result</span><Url, ScrapflyError> {
<a href=#194 id=194 data-nosnippet>194</a> <span class="kw">let </span><span class="kw-2">mut </span>u = Url::parse(<span class="kw-2">&</span><span class="macro">format!</span>(<span class="string">"{}{}"</span>, <span class="self">self</span>.host, path))
<a href=#195 id=195 data-nosnippet>195</a> .map_err(|e| ScrapflyError::Config(<span class="macro">format!</span>(<span class="string">"invalid url: {}"</span>, e)))<span class="question-mark">?</span>;
<a href=#196 id=196 data-nosnippet>196</a> {
<a href=#197 id=197 data-nosnippet>197</a> <span class="kw">let </span><span class="kw-2">mut </span>pairs = u.query_pairs_mut();
<a href=#198 id=198 data-nosnippet>198</a> pairs.append_pair(<span class="string">"key"</span>, <span class="kw-2">&</span><span class="self">self</span>.key);
<a href=#199 id=199 data-nosnippet>199</a> <span class="kw">for </span>(k, v) <span class="kw">in </span>query {
<a href=#200 id=200 data-nosnippet>200</a> pairs.append_pair(k, v);
<a href=#201 id=201 data-nosnippet>201</a> }
<a href=#202 id=202 data-nosnippet>202</a> }
<a href=#203 id=203 data-nosnippet>203</a> <span class="prelude-val">Ok</span>(u)
<a href=#204 id=204 data-nosnippet>204</a> }
<a href=#205 id=205 data-nosnippet>205</a>
<a href=#206 id=206 data-nosnippet>206</a> <span class="doccomment">/// Verify the API key by hitting `/account`.
<a href=#207 id=207 data-nosnippet>207</a> </span><span class="kw">pub async fn </span>verify_api_key(<span class="kw-2">&</span><span class="self">self</span>) -> <span class="prelude-ty">Result</span><VerifyApiKeyResult, ScrapflyError> {
<a href=#208 id=208 data-nosnippet>208</a> <span class="kw">let </span>url = <span class="self">self</span>.build_url(<span class="string">"/account"</span>, <span class="kw-2">&</span>[])<span class="question-mark">?</span>;
<a href=#209 id=209 data-nosnippet>209</a> <span class="kw">let </span>resp = <span class="self">self</span>.send_simple(Method::GET, url, <span class="prelude-val">None</span>, <span class="prelude-val">None</span>).<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#210 id=210 data-nosnippet>210</a> <span class="prelude-val">Ok</span>(VerifyApiKeyResult {
<a href=#211 id=211 data-nosnippet>211</a> valid: resp.status().is_success(),
<a href=#212 id=212 data-nosnippet>212</a> })
<a href=#213 id=213 data-nosnippet>213</a> }
<a href=#214 id=214 data-nosnippet>214</a>
<a href=#215 id=215 data-nosnippet>215</a> <span class="doccomment">/// Fetch account info.
<a href=#216 id=216 data-nosnippet>216</a> </span><span class="kw">pub async fn </span>account(<span class="kw-2">&</span><span class="self">self</span>) -> <span class="prelude-ty">Result</span><AccountData, ScrapflyError> {
<a href=#217 id=217 data-nosnippet>217</a> <span class="kw">let </span>url = <span class="self">self</span>.build_url(<span class="string">"/account"</span>, <span class="kw-2">&</span>[])<span class="question-mark">?</span>;
<a href=#218 id=218 data-nosnippet>218</a> <span class="kw">let </span>resp = <span class="self">self</span>.send_simple(Method::GET, url, <span class="prelude-val">None</span>, <span class="prelude-val">None</span>).<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#219 id=219 data-nosnippet>219</a> <span class="kw">let </span>(status, _headers, body) = read_response(resp).<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#220 id=220 data-nosnippet>220</a> <span class="kw">if </span>status != <span class="number">200 </span>{
<a href=#221 id=221 data-nosnippet>221</a> <span class="kw">return </span><span class="prelude-val">Err</span>(from_response(status, <span class="kw-2">&</span>body, <span class="number">0</span>, <span class="bool-val">false</span>));
<a href=#222 id=222 data-nosnippet>222</a> }
<a href=#223 id=223 data-nosnippet>223</a> <span class="prelude-val">Ok</span>(serde_json::from_slice(<span class="kw-2">&</span>body)<span class="question-mark">?</span>)
<a href=#224 id=224 data-nosnippet>224</a> }
<a href=#225 id=225 data-nosnippet>225</a>
<a href=#226 id=226 data-nosnippet>226</a> <span class="doccomment">/// Classify an already-fetched HTTP response for anti-bot blocking.
<a href=#227 id=227 data-nosnippet>227</a> ///
<a href=#228 id=228 data-nosnippet>228</a> /// Runs the same detection pipeline used by every live Scrapfly scrape
<a href=#229 id=229 data-nosnippet>229</a> /// against a response you already have (from your own proxy, cache, etc).
<a href=#230 id=230 data-nosnippet>230</a> /// 1 API credit per successful call. See
<a href=#231 id=231 data-nosnippet>231</a> /// <https://scrapfly.io/docs/scrape-api/classify>.
<a href=#232 id=232 data-nosnippet>232</a> </span><span class="kw">pub async fn </span>classify(<span class="kw-2">&</span><span class="self">self</span>, req: <span class="kw-2">&</span>ClassifyRequest) -> <span class="prelude-ty">Result</span><ClassifyResult, ScrapflyError> {
<a href=#233 id=233 data-nosnippet>233</a> <span class="kw">if </span>req.url.is_empty() {
<a href=#234 id=234 data-nosnippet>234</a> <span class="kw">return </span><span class="prelude-val">Err</span>(ScrapflyError::Config(<span class="string">"classify: url is required"</span>.into()));
<a href=#235 id=235 data-nosnippet>235</a> }
<a href=#236 id=236 data-nosnippet>236</a> <span class="kw">if </span>!(<span class="number">100</span>..=<span class="number">599</span>).contains(<span class="kw-2">&</span>req.status_code) {
<a href=#237 id=237 data-nosnippet>237</a> <span class="kw">return </span><span class="prelude-val">Err</span>(ScrapflyError::Config(
<a href=#238 id=238 data-nosnippet>238</a> <span class="string">"classify: status_code must be in [100, 599]"</span>.into(),
<a href=#239 id=239 data-nosnippet>239</a> ));
<a href=#240 id=240 data-nosnippet>240</a> }
<a href=#241 id=241 data-nosnippet>241</a>
<a href=#242 id=242 data-nosnippet>242</a> <span class="kw">let </span>url = <span class="self">self</span>.build_url(<span class="string">"/classify"</span>, <span class="kw-2">&</span>[])<span class="question-mark">?</span>;
<a href=#243 id=243 data-nosnippet>243</a> <span class="kw">let </span>body = serde_json::to_vec(req)
<a href=#244 id=244 data-nosnippet>244</a> .map_err(|e| ScrapflyError::Config(<span class="macro">format!</span>(<span class="string">"marshal classify request: {}"</span>, e)))<span class="question-mark">?</span>;
<a href=#245 id=245 data-nosnippet>245</a>
<a href=#246 id=246 data-nosnippet>246</a> <span class="kw">let </span><span class="kw-2">mut </span>headers = HeaderMap::new();
<a href=#247 id=247 data-nosnippet>247</a> headers.insert(CONTENT_TYPE, HeaderValue::from_static(<span class="string">"application/json"</span>));
<a href=#248 id=248 data-nosnippet>248</a> headers.insert(ACCEPT, HeaderValue::from_static(<span class="string">"application/json"</span>));
<a href=#249 id=249 data-nosnippet>249</a>
<a href=#250 id=250 data-nosnippet>250</a> <span class="kw">let </span>resp = <span class="self">self
<a href=#251 id=251 data-nosnippet>251</a> </span>.send_simple(Method::POST, url, <span class="prelude-val">Some</span>(headers), <span class="prelude-val">Some</span>(body))
<a href=#252 id=252 data-nosnippet>252</a> .<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#253 id=253 data-nosnippet>253</a> <span class="kw">let </span>(status, _headers, bytes) = read_response(resp).<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#254 id=254 data-nosnippet>254</a> <span class="kw">if </span>status >= <span class="number">400 </span>{
<a href=#255 id=255 data-nosnippet>255</a> <span class="kw">return </span><span class="prelude-val">Err</span>(from_response(status, <span class="kw-2">&</span>bytes, <span class="number">0</span>, <span class="bool-val">false</span>));
<a href=#256 id=256 data-nosnippet>256</a> }
<a href=#257 id=257 data-nosnippet>257</a> <span class="kw">let </span>out: ClassifyResult = serde_json::from_slice(<span class="kw-2">&</span>bytes)
<a href=#258 id=258 data-nosnippet>258</a> .map_err(|e| ScrapflyError::Config(<span class="macro">format!</span>(<span class="string">"decode classify response: {}"</span>, e)))<span class="question-mark">?</span>;
<a href=#259 id=259 data-nosnippet>259</a> <span class="prelude-val">Ok</span>(out)
<a href=#260 id=260 data-nosnippet>260</a> }
<a href=#261 id=261 data-nosnippet>261</a>
<a href=#262 id=262 data-nosnippet>262</a> <span class="comment">// ── Monitoring API (Enterprise+ plan only) ──────────────────────
<a href=#263 id=263 data-nosnippet>263</a> // The Monitoring API exposes per-product aggregates and per-target
<a href=#264 id=264 data-nosnippet>264</a> // timeseries. Web Scraping / Screenshot / Extraction / Crawler share
<a href=#265 id=265 data-nosnippet>265</a> // one shape (request-based) but live under different URL prefixes;
<a href=#266 id=266 data-nosnippet>266</a> // Cloud Browser is session-based and exposes a distinct shape.
<a href=#267 id=267 data-nosnippet>267</a> // See <https://scrapfly.io/docs/monitoring#api>.
<a href=#268 id=268 data-nosnippet>268</a>
<a href=#269 id=269 data-nosnippet>269</a> </span><span class="kw">fn </span>build_metrics_pairs(opts: <span class="kw-2">&</span>MonitoringMetricsOptions) -> Vec<(String, String)> {
<a href=#270 id=270 data-nosnippet>270</a> <span class="kw">let </span><span class="kw-2">mut </span>pairs: Vec<(String, String)> = Vec::new();
<a href=#271 id=271 data-nosnippet>271</a> <span class="kw">let </span>format = opts.format.unwrap_or(MonitoringDataFormat::Structured);
<a href=#272 id=272 data-nosnippet>272</a> pairs.push((<span class="string">"format"</span>.into(), format.as_str().into()));
<a href=#273 id=273 data-nosnippet>273</a> <span class="kw">if let </span><span class="prelude-val">Some</span>(p) = opts.period {
<a href=#274 id=274 data-nosnippet>274</a> pairs.push((<span class="string">"period"</span>.into(), p.as_str().into()));
<a href=#275 id=275 data-nosnippet>275</a> }
<a href=#276 id=276 data-nosnippet>276</a> <span class="kw">if let </span><span class="prelude-val">Some</span>(<span class="kw-2">ref </span>aggs) = opts.aggregation {
<a href=#277 id=277 data-nosnippet>277</a> <span class="kw">if </span>!aggs.is_empty() {
<a href=#278 id=278 data-nosnippet>278</a> <span class="kw">let </span>joined = aggs
<a href=#279 id=279 data-nosnippet>279</a> .iter()
<a href=#280 id=280 data-nosnippet>280</a> .map(|a| a.as_str())
<a href=#281 id=281 data-nosnippet>281</a> .collect::<Vec<<span class="kw">_</span>>>()
<a href=#282 id=282 data-nosnippet>282</a> .join(<span class="string">","</span>);
<a href=#283 id=283 data-nosnippet>283</a> pairs.push((<span class="string">"aggregation"</span>.into(), joined));
<a href=#284 id=284 data-nosnippet>284</a> }
<a href=#285 id=285 data-nosnippet>285</a> }
<a href=#286 id=286 data-nosnippet>286</a> <span class="kw">if </span>opts.include_webhook {
<a href=#287 id=287 data-nosnippet>287</a> pairs.push((<span class="string">"include_webhook"</span>.into(), <span class="string">"true"</span>.into()));
<a href=#288 id=288 data-nosnippet>288</a> }
<a href=#289 id=289 data-nosnippet>289</a> pairs
<a href=#290 id=290 data-nosnippet>290</a> }
<a href=#291 id=291 data-nosnippet>291</a>
<a href=#292 id=292 data-nosnippet>292</a> <span class="kw">fn </span>build_target_pairs(
<a href=#293 id=293 data-nosnippet>293</a> opts: <span class="kw-2">&</span>MonitoringTargetMetricsOptions,
<a href=#294 id=294 data-nosnippet>294</a> ) -> <span class="prelude-ty">Result</span><Vec<(String, String)>, ScrapflyError> {
<a href=#295 id=295 data-nosnippet>295</a> <span class="kw">if </span>opts.domain.is_empty() {
<a href=#296 id=296 data-nosnippet>296</a> <span class="kw">return </span><span class="prelude-val">Err</span>(ScrapflyError::Config(
<a href=#297 id=297 data-nosnippet>297</a> <span class="string">"monitoring target metrics: domain is required"</span>.into(),
<a href=#298 id=298 data-nosnippet>298</a> ));
<a href=#299 id=299 data-nosnippet>299</a> }
<a href=#300 id=300 data-nosnippet>300</a> <span class="kw">if </span>opts.start.is_some() != opts.end.is_some() {
<a href=#301 id=301 data-nosnippet>301</a> <span class="kw">return </span><span class="prelude-val">Err</span>(ScrapflyError::Config(
<a href=#302 id=302 data-nosnippet>302</a> <span class="string">"monitoring target metrics: start and end must be provided together"</span>.into(),
<a href=#303 id=303 data-nosnippet>303</a> ));
<a href=#304 id=304 data-nosnippet>304</a> }
<a href=#305 id=305 data-nosnippet>305</a> <span class="kw">let </span><span class="kw-2">mut </span>pairs: Vec<(String, String)> = Vec::new();
<a href=#306 id=306 data-nosnippet>306</a> pairs.push((<span class="string">"domain"</span>.into(), opts.domain.clone()));
<a href=#307 id=307 data-nosnippet>307</a> pairs.push((<span class="string">"group_subdomain"</span>.into(), opts.group_subdomain.to_string()));
<a href=#308 id=308 data-nosnippet>308</a> <span class="kw">match </span>(<span class="kw-2">&</span>opts.start, <span class="kw-2">&</span>opts.end) {
<a href=#309 id=309 data-nosnippet>309</a> (<span class="prelude-val">Some</span>(s), <span class="prelude-val">Some</span>(e)) => {
<a href=#310 id=310 data-nosnippet>310</a> pairs.push((<span class="string">"start"</span>.into(), s.clone()));
<a href=#311 id=311 data-nosnippet>311</a> pairs.push((<span class="string">"end"</span>.into(), e.clone()));
<a href=#312 id=312 data-nosnippet>312</a> }
<a href=#313 id=313 data-nosnippet>313</a> <span class="kw">_ </span>=> {
<a href=#314 id=314 data-nosnippet>314</a> <span class="kw">let </span>period = opts
<a href=#315 id=315 data-nosnippet>315</a> .period
<a href=#316 id=316 data-nosnippet>316</a> .unwrap_or(<span class="kw">crate</span>::monitoring::MonitoringPeriod::Last24h);
<a href=#317 id=317 data-nosnippet>317</a> pairs.push((<span class="string">"period"</span>.into(), period.as_str().into()));
<a href=#318 id=318 data-nosnippet>318</a> }
<a href=#319 id=319 data-nosnippet>319</a> }
<a href=#320 id=320 data-nosnippet>320</a> <span class="kw">if </span>opts.include_webhook {
<a href=#321 id=321 data-nosnippet>321</a> pairs.push((<span class="string">"include_webhook"</span>.into(), <span class="string">"true"</span>.into()));
<a href=#322 id=322 data-nosnippet>322</a> }
<a href=#323 id=323 data-nosnippet>323</a> <span class="prelude-val">Ok</span>(pairs)
<a href=#324 id=324 data-nosnippet>324</a> }
<a href=#325 id=325 data-nosnippet>325</a>
<a href=#326 id=326 data-nosnippet>326</a> <span class="kw">async fn </span>fetch_monitoring_json(
<a href=#327 id=327 data-nosnippet>327</a> <span class="kw-2">&</span><span class="self">self</span>,
<a href=#328 id=328 data-nosnippet>328</a> path: <span class="kw-2">&</span>str,
<a href=#329 id=329 data-nosnippet>329</a> pairs: <span class="kw-2">&</span>[(String, String)],
<a href=#330 id=330 data-nosnippet>330</a> ) -> <span class="prelude-ty">Result</span><serde_json::Value, ScrapflyError> {
<a href=#331 id=331 data-nosnippet>331</a> <span class="kw">let </span>url = <span class="self">self</span>.build_url(path, pairs)<span class="question-mark">?</span>;
<a href=#332 id=332 data-nosnippet>332</a> <span class="kw">let </span>resp = <span class="self">self</span>.send_simple(Method::GET, url, <span class="prelude-val">None</span>, <span class="prelude-val">None</span>).<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#333 id=333 data-nosnippet>333</a> <span class="kw">let </span>(status, _headers, body) = read_response(resp).<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#334 id=334 data-nosnippet>334</a> <span class="kw">if </span>status != <span class="number">200 </span>{
<a href=#335 id=335 data-nosnippet>335</a> <span class="kw">return </span><span class="prelude-val">Err</span>(from_response(status, <span class="kw-2">&</span>body, <span class="number">0</span>, <span class="bool-val">false</span>));
<a href=#336 id=336 data-nosnippet>336</a> }
<a href=#337 id=337 data-nosnippet>337</a> <span class="prelude-val">Ok</span>(serde_json::from_slice(<span class="kw-2">&</span>body)<span class="question-mark">?</span>)
<a href=#338 id=338 data-nosnippet>338</a> }
<a href=#339 id=339 data-nosnippet>339</a>
<a href=#340 id=340 data-nosnippet>340</a> <span class="comment">// ── Web Scraping API ─────────────────────────────────────────────
<a href=#341 id=341 data-nosnippet>341</a>
<a href=#342 id=342 data-nosnippet>342</a> </span><span class="doccomment">/// Fetch aggregate monitoring metrics for the Web Scraping API.
<a href=#343 id=343 data-nosnippet>343</a> </span><span class="kw">pub async fn </span>get_monitoring_metrics(
<a href=#344 id=344 data-nosnippet>344</a> <span class="kw-2">&</span><span class="self">self</span>,
<a href=#345 id=345 data-nosnippet>345</a> opts: <span class="kw-2">&</span>MonitoringMetricsOptions,
<a href=#346 id=346 data-nosnippet>346</a> ) -> <span class="prelude-ty">Result</span><serde_json::Value, ScrapflyError> {
<a href=#347 id=347 data-nosnippet>347</a> <span class="self">self</span>.fetch_monitoring_json(
<a href=#348 id=348 data-nosnippet>348</a> <span class="string">"/scrape/monitoring/metrics"</span>,
<a href=#349 id=349 data-nosnippet>349</a> <span class="kw-2">&</span><span class="self">Self</span>::build_metrics_pairs(opts),
<a href=#350 id=350 data-nosnippet>350</a> )
<a href=#351 id=351 data-nosnippet>351</a> .<span class="kw">await
<a href=#352 id=352 data-nosnippet>352</a> </span>}
<a href=#353 id=353 data-nosnippet>353</a>
<a href=#354 id=354 data-nosnippet>354</a> <span class="doccomment">/// Fetch per-target monitoring metrics for the Web Scraping API.
<a href=#355 id=355 data-nosnippet>355</a> </span><span class="kw">pub async fn </span>get_monitoring_target_metrics(
<a href=#356 id=356 data-nosnippet>356</a> <span class="kw-2">&</span><span class="self">self</span>,
<a href=#357 id=357 data-nosnippet>357</a> opts: <span class="kw-2">&</span>MonitoringTargetMetricsOptions,
<a href=#358 id=358 data-nosnippet>358</a> ) -> <span class="prelude-ty">Result</span><serde_json::Value, ScrapflyError> {
<a href=#359 id=359 data-nosnippet>359</a> <span class="kw">let </span>pairs = <span class="self">Self</span>::build_target_pairs(opts)<span class="question-mark">?</span>;
<a href=#360 id=360 data-nosnippet>360</a> <span class="self">self</span>.fetch_monitoring_json(<span class="string">"/scrape/monitoring/metrics/target"</span>, <span class="kw-2">&</span>pairs)
<a href=#361 id=361 data-nosnippet>361</a> .<span class="kw">await
<a href=#362 id=362 data-nosnippet>362</a> </span>}
<a href=#363 id=363 data-nosnippet>363</a>
<a href=#364 id=364 data-nosnippet>364</a> <span class="comment">// ── Screenshot API ───────────────────────────────────────────────
<a href=#365 id=365 data-nosnippet>365</a>
<a href=#366 id=366 data-nosnippet>366</a> </span><span class="doccomment">/// Fetch aggregate monitoring metrics for the Screenshot API.
<a href=#367 id=367 data-nosnippet>367</a> </span><span class="kw">pub async fn </span>get_screenshot_monitoring_metrics(
<a href=#368 id=368 data-nosnippet>368</a> <span class="kw-2">&</span><span class="self">self</span>,
<a href=#369 id=369 data-nosnippet>369</a> opts: <span class="kw-2">&</span>MonitoringMetricsOptions,
<a href=#370 id=370 data-nosnippet>370</a> ) -> <span class="prelude-ty">Result</span><serde_json::Value, ScrapflyError> {
<a href=#371 id=371 data-nosnippet>371</a> <span class="self">self</span>.fetch_monitoring_json(
<a href=#372 id=372 data-nosnippet>372</a> <span class="string">"/screenshot/monitoring/metrics"</span>,
<a href=#373 id=373 data-nosnippet>373</a> <span class="kw-2">&</span><span class="self">Self</span>::build_metrics_pairs(opts),
<a href=#374 id=374 data-nosnippet>374</a> )
<a href=#375 id=375 data-nosnippet>375</a> .<span class="kw">await
<a href=#376 id=376 data-nosnippet>376</a> </span>}
<a href=#377 id=377 data-nosnippet>377</a>
<a href=#378 id=378 data-nosnippet>378</a> <span class="doccomment">/// Fetch per-target monitoring metrics for the Screenshot API.
<a href=#379 id=379 data-nosnippet>379</a> </span><span class="kw">pub async fn </span>get_screenshot_monitoring_target_metrics(
<a href=#380 id=380 data-nosnippet>380</a> <span class="kw-2">&</span><span class="self">self</span>,
<a href=#381 id=381 data-nosnippet>381</a> opts: <span class="kw-2">&</span>MonitoringTargetMetricsOptions,
<a href=#382 id=382 data-nosnippet>382</a> ) -> <span class="prelude-ty">Result</span><serde_json::Value, ScrapflyError> {
<a href=#383 id=383 data-nosnippet>383</a> <span class="kw">let </span>pairs = <span class="self">Self</span>::build_target_pairs(opts)<span class="question-mark">?</span>;
<a href=#384 id=384 data-nosnippet>384</a> <span class="self">self</span>.fetch_monitoring_json(<span class="string">"/screenshot/monitoring/metrics/target"</span>, <span class="kw-2">&</span>pairs)
<a href=#385 id=385 data-nosnippet>385</a> .<span class="kw">await
<a href=#386 id=386 data-nosnippet>386</a> </span>}
<a href=#387 id=387 data-nosnippet>387</a>
<a href=#388 id=388 data-nosnippet>388</a> <span class="comment">// ── Extraction API ───────────────────────────────────────────────
<a href=#389 id=389 data-nosnippet>389</a>
<a href=#390 id=390 data-nosnippet>390</a> </span><span class="doccomment">/// Fetch aggregate monitoring metrics for the Extraction API.
<a href=#391 id=391 data-nosnippet>391</a> </span><span class="kw">pub async fn </span>get_extraction_monitoring_metrics(
<a href=#392 id=392 data-nosnippet>392</a> <span class="kw-2">&</span><span class="self">self</span>,
<a href=#393 id=393 data-nosnippet>393</a> opts: <span class="kw-2">&</span>MonitoringMetricsOptions,
<a href=#394 id=394 data-nosnippet>394</a> ) -> <span class="prelude-ty">Result</span><serde_json::Value, ScrapflyError> {
<a href=#395 id=395 data-nosnippet>395</a> <span class="self">self</span>.fetch_monitoring_json(
<a href=#396 id=396 data-nosnippet>396</a> <span class="string">"/extraction/monitoring/metrics"</span>,
<a href=#397 id=397 data-nosnippet>397</a> <span class="kw-2">&</span><span class="self">Self</span>::build_metrics_pairs(opts),
<a href=#398 id=398 data-nosnippet>398</a> )
<a href=#399 id=399 data-nosnippet>399</a> .<span class="kw">await
<a href=#400 id=400 data-nosnippet>400</a> </span>}
<a href=#401 id=401 data-nosnippet>401</a>
<a href=#402 id=402 data-nosnippet>402</a> <span class="doccomment">/// Fetch per-target monitoring metrics for the Extraction API.
<a href=#403 id=403 data-nosnippet>403</a> </span><span class="kw">pub async fn </span>get_extraction_monitoring_target_metrics(
<a href=#404 id=404 data-nosnippet>404</a> <span class="kw-2">&</span><span class="self">self</span>,
<a href=#405 id=405 data-nosnippet>405</a> opts: <span class="kw-2">&</span>MonitoringTargetMetricsOptions,
<a href=#406 id=406 data-nosnippet>406</a> ) -> <span class="prelude-ty">Result</span><serde_json::Value, ScrapflyError> {
<a href=#407 id=407 data-nosnippet>407</a> <span class="kw">let </span>pairs = <span class="self">Self</span>::build_target_pairs(opts)<span class="question-mark">?</span>;
<a href=#408 id=408 data-nosnippet>408</a> <span class="self">self</span>.fetch_monitoring_json(<span class="string">"/extraction/monitoring/metrics/target"</span>, <span class="kw-2">&</span>pairs)
<a href=#409 id=409 data-nosnippet>409</a> .<span class="kw">await
<a href=#410 id=410 data-nosnippet>410</a> </span>}
<a href=#411 id=411 data-nosnippet>411</a>
<a href=#412 id=412 data-nosnippet>412</a> <span class="comment">// ── Crawler API ──────────────────────────────────────────────────
<a href=#413 id=413 data-nosnippet>413</a>
<a href=#414 id=414 data-nosnippet>414</a> </span><span class="doccomment">/// Fetch aggregate monitoring metrics for the Crawler API.
<a href=#415 id=415 data-nosnippet>415</a> </span><span class="kw">pub async fn </span>get_crawler_monitoring_metrics(
<a href=#416 id=416 data-nosnippet>416</a> <span class="kw-2">&</span><span class="self">self</span>,
<a href=#417 id=417 data-nosnippet>417</a> opts: <span class="kw-2">&</span>MonitoringMetricsOptions,
<a href=#418 id=418 data-nosnippet>418</a> ) -> <span class="prelude-ty">Result</span><serde_json::Value, ScrapflyError> {
<a href=#419 id=419 data-nosnippet>419</a> <span class="self">self</span>.fetch_monitoring_json(
<a href=#420 id=420 data-nosnippet>420</a> <span class="string">"/crawl/monitoring/metrics"</span>,
<a href=#421 id=421 data-nosnippet>421</a> <span class="kw-2">&</span><span class="self">Self</span>::build_metrics_pairs(opts),
<a href=#422 id=422 data-nosnippet>422</a> )
<a href=#423 id=423 data-nosnippet>423</a> .<span class="kw">await
<a href=#424 id=424 data-nosnippet>424</a> </span>}
<a href=#425 id=425 data-nosnippet>425</a>
<a href=#426 id=426 data-nosnippet>426</a> <span class="doccomment">/// Fetch per-target monitoring metrics for the Crawler API.
<a href=#427 id=427 data-nosnippet>427</a> </span><span class="kw">pub async fn </span>get_crawler_monitoring_target_metrics(
<a href=#428 id=428 data-nosnippet>428</a> <span class="kw-2">&</span><span class="self">self</span>,
<a href=#429 id=429 data-nosnippet>429</a> opts: <span class="kw-2">&</span>MonitoringTargetMetricsOptions,
<a href=#430 id=430 data-nosnippet>430</a> ) -> <span class="prelude-ty">Result</span><serde_json::Value, ScrapflyError> {
<a href=#431 id=431 data-nosnippet>431</a> <span class="kw">let </span>pairs = <span class="self">Self</span>::build_target_pairs(opts)<span class="question-mark">?</span>;
<a href=#432 id=432 data-nosnippet>432</a> <span class="self">self</span>.fetch_monitoring_json(<span class="string">"/crawl/monitoring/metrics/target"</span>, <span class="kw-2">&</span>pairs)
<a href=#433 id=433 data-nosnippet>433</a> .<span class="kw">await
<a href=#434 id=434 data-nosnippet>434</a> </span>}
<a href=#435 id=435 data-nosnippet>435</a>
<a href=#436 id=436 data-nosnippet>436</a> <span class="comment">// ── Cloud Browser API (session-based, distinct shape) ────────────
<a href=#437 id=437 data-nosnippet>437</a>
<a href=#438 id=438 data-nosnippet>438</a> </span><span class="doccomment">/// Fetch aggregate monitoring metrics for the Cloud Browser API.
<a href=#439 id=439 data-nosnippet>439</a> </span><span class="kw">pub async fn </span>get_browser_monitoring_metrics(
<a href=#440 id=440 data-nosnippet>440</a> <span class="kw-2">&</span><span class="self">self</span>,
<a href=#441 id=441 data-nosnippet>441</a> opts: <span class="kw-2">&</span>CloudBrowserMonitoringOptions,
<a href=#442 id=442 data-nosnippet>442</a> ) -> <span class="prelude-ty">Result</span><serde_json::Value, ScrapflyError> {
<a href=#443 id=443 data-nosnippet>443</a> <span class="kw">let </span>pairs = <span class="self">Self</span>::build_browser_pairs(opts)<span class="question-mark">?</span>;
<a href=#444 id=444 data-nosnippet>444</a> <span class="self">self</span>.fetch_monitoring_json(<span class="string">"/browser/monitoring/metrics"</span>, <span class="kw-2">&</span>pairs)
<a href=#445 id=445 data-nosnippet>445</a> .<span class="kw">await
<a href=#446 id=446 data-nosnippet>446</a> </span>}
<a href=#447 id=447 data-nosnippet>447</a>
<a href=#448 id=448 data-nosnippet>448</a> <span class="doccomment">/// Fetch monitoring time-series for the Cloud Browser API.
<a href=#449 id=449 data-nosnippet>449</a> </span><span class="kw">pub async fn </span>get_browser_monitoring_timeseries(
<a href=#450 id=450 data-nosnippet>450</a> <span class="kw-2">&</span><span class="self">self</span>,
<a href=#451 id=451 data-nosnippet>451</a> opts: <span class="kw-2">&</span>CloudBrowserMonitoringOptions,
<a href=#452 id=452 data-nosnippet>452</a> ) -> <span class="prelude-ty">Result</span><serde_json::Value, ScrapflyError> {
<a href=#453 id=453 data-nosnippet>453</a> <span class="kw">let </span>pairs = <span class="self">Self</span>::build_browser_pairs(opts)<span class="question-mark">?</span>;
<a href=#454 id=454 data-nosnippet>454</a> <span class="self">self</span>.fetch_monitoring_json(<span class="string">"/browser/monitoring/metrics/timeseries"</span>, <span class="kw-2">&</span>pairs)
<a href=#455 id=455 data-nosnippet>455</a> .<span class="kw">await
<a href=#456 id=456 data-nosnippet>456</a> </span>}
<a href=#457 id=457 data-nosnippet>457</a>
<a href=#458 id=458 data-nosnippet>458</a> <span class="kw">fn </span>build_browser_pairs(
<a href=#459 id=459 data-nosnippet>459</a> opts: <span class="kw-2">&</span>CloudBrowserMonitoringOptions,
<a href=#460 id=460 data-nosnippet>460</a> ) -> <span class="prelude-ty">Result</span><Vec<(String, String)>, ScrapflyError> {
<a href=#461 id=461 data-nosnippet>461</a> <span class="kw">if </span>opts.start.is_some() != opts.end.is_some() {
<a href=#462 id=462 data-nosnippet>462</a> <span class="kw">return </span><span class="prelude-val">Err</span>(ScrapflyError::Config(
<a href=#463 id=463 data-nosnippet>463</a> <span class="string">"cloud browser monitoring: start and end must be provided together"</span>.into(),
<a href=#464 id=464 data-nosnippet>464</a> ));
<a href=#465 id=465 data-nosnippet>465</a> }
<a href=#466 id=466 data-nosnippet>466</a> <span class="kw">let </span><span class="kw-2">mut </span>pairs: Vec<(String, String)> = Vec::new();
<a href=#467 id=467 data-nosnippet>467</a> <span class="kw">match </span>(<span class="kw-2">&</span>opts.start, <span class="kw-2">&</span>opts.end) {
<a href=#468 id=468 data-nosnippet>468</a> (<span class="prelude-val">Some</span>(s), <span class="prelude-val">Some</span>(e)) => {
<a href=#469 id=469 data-nosnippet>469</a> pairs.push((<span class="string">"start"</span>.into(), s.clone()));
<a href=#470 id=470 data-nosnippet>470</a> pairs.push((<span class="string">"end"</span>.into(), e.clone()));
<a href=#471 id=471 data-nosnippet>471</a> }
<a href=#472 id=472 data-nosnippet>472</a> <span class="kw">_ </span>=> {
<a href=#473 id=473 data-nosnippet>473</a> <span class="kw">if let </span><span class="prelude-val">Some</span>(p) = opts.period {
<a href=#474 id=474 data-nosnippet>474</a> pairs.push((<span class="string">"period"</span>.into(), p.as_str().into()));
<a href=#475 id=475 data-nosnippet>475</a> }
<a href=#476 id=476 data-nosnippet>476</a> }
<a href=#477 id=477 data-nosnippet>477</a> }
<a href=#478 id=478 data-nosnippet>478</a> <span class="kw">if let </span><span class="prelude-val">Some</span>(<span class="kw-2">ref </span>pool) = opts.proxy_pool {
<a href=#479 id=479 data-nosnippet>479</a> pairs.push((<span class="string">"proxy_pool"</span>.into(), pool.clone()));
<a href=#480 id=480 data-nosnippet>480</a> }
<a href=#481 id=481 data-nosnippet>481</a> <span class="prelude-val">Ok</span>(pairs)
<a href=#482 id=482 data-nosnippet>482</a> }
<a href=#483 id=483 data-nosnippet>483</a>
<a href=#484 id=484 data-nosnippet>484</a> <span class="doccomment">/// Scrape a URL.
<a href=#485 id=485 data-nosnippet>485</a> </span><span class="kw">pub async fn </span>scrape(<span class="kw-2">&</span><span class="self">self</span>, config: <span class="kw-2">&</span>ScrapeConfig) -> <span class="prelude-ty">Result</span><ScrapeResult, ScrapflyError> {
<a href=#486 id=486 data-nosnippet>486</a> <span class="kw">let </span>pairs = config.to_query_pairs()<span class="question-mark">?</span>;
<a href=#487 id=487 data-nosnippet>487</a> <span class="kw">let </span>url = <span class="self">self</span>.build_url(<span class="string">"/scrape"</span>, <span class="kw-2">&</span>pairs)<span class="question-mark">?</span>;
<a href=#488 id=488 data-nosnippet>488</a> <span class="kw">let </span>method = <span class="kw">match </span>config.method {
<a href=#489 id=489 data-nosnippet>489</a> <span class="prelude-val">Some</span>(m) => Method::from_bytes(m.as_str().as_bytes())
<a href=#490 id=490 data-nosnippet>490</a> .map_err(|e| ScrapflyError::Config(<span class="macro">format!</span>(<span class="string">"invalid method: {}"</span>, e)))<span class="question-mark">?</span>,
<a href=#491 id=491 data-nosnippet>491</a> <span class="prelude-val">None </span>=> Method::GET,
<a href=#492 id=492 data-nosnippet>492</a> };
<a href=#493 id=493 data-nosnippet>493</a> <span class="kw">let </span><span class="kw-2">mut </span>headers = HeaderMap::new();
<a href=#494 id=494 data-nosnippet>494</a> headers.insert(ACCEPT, HeaderValue::from_static(<span class="string">"application/json"</span>));
<a href=#495 id=495 data-nosnippet>495</a> <span class="kw">let </span>body = config.body.clone();
<a href=#496 id=496 data-nosnippet>496</a> <span class="kw">let </span>resp = <span class="self">self
<a href=#497 id=497 data-nosnippet>497</a> </span>.send_with_retry(method, url, <span class="prelude-val">Some</span>(headers), body.map(|b| b.into_bytes()))
<a href=#498 id=498 data-nosnippet>498</a> .<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#499 id=499 data-nosnippet>499</a> <span class="kw">let </span>(status, _h, body_bytes) = read_response(resp).<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#500 id=500 data-nosnippet>500</a> <span class="kw">if </span>status != <span class="number">200 </span>{
<a href=#501 id=501 data-nosnippet>501</a> <span class="kw">return </span><span class="prelude-val">Err</span>(from_response(status, <span class="kw-2">&</span>body_bytes, <span class="number">0</span>, <span class="bool-val">false</span>));
<a href=#502 id=502 data-nosnippet>502</a> }
<a href=#503 id=503 data-nosnippet>503</a> <span class="comment">// HEAD has no body per HTTP spec, so the Scrapfly API returns a 200
<a href=#504 id=504 data-nosnippet>504</a> // with an empty body — there's no JSON envelope to parse. Synthesize
<a href=#505 id=505 data-nosnippet>505</a> // a minimal ScrapeResult so callers still get a typed response with
<a href=#506 id=506 data-nosnippet>506</a> // status_code=200 and an empty content string. Matches Python SDK
<a href=#507 id=507 data-nosnippet>507</a> // behavior, which tolerates an empty body_handler read on HEAD.
<a href=#508 id=508 data-nosnippet>508</a> </span><span class="kw">if </span><span class="macro">matches!</span>(config.method, <span class="prelude-val">Some</span>(HttpMethod::Head)) && body_bytes.is_empty() {
<a href=#509 id=509 data-nosnippet>509</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(ScrapeResult {
<a href=#510 id=510 data-nosnippet>510</a> uuid: String::new(),
<a href=#511 id=511 data-nosnippet>511</a> config: serde_json::Value::Null,
<a href=#512 id=512 data-nosnippet>512</a> context: serde_json::Value::Null,
<a href=#513 id=513 data-nosnippet>513</a> result: ResultData {
<a href=#514 id=514 data-nosnippet>514</a> status_code: <span class="number">200</span>,
<a href=#515 id=515 data-nosnippet>515</a> success: <span class="bool-val">true</span>,
<a href=#516 id=516 data-nosnippet>516</a> ..Default::default()
<a href=#517 id=517 data-nosnippet>517</a> },
<a href=#518 id=518 data-nosnippet>518</a> });
<a href=#519 id=519 data-nosnippet>519</a> }
<a href=#520 id=520 data-nosnippet>520</a> <span class="kw">let </span><span class="kw-2">mut </span>result: ScrapeResult = serde_json::from_slice(<span class="kw-2">&</span>body_bytes)<span class="question-mark">?</span>;
<a href=#521 id=521 data-nosnippet>521</a> <span class="comment">// Upstream failure handling: the Scrapfly API call itself may succeed
<a href=#522 id=522 data-nosnippet>522</a> // (HTTP 200) while the *target* site returned a failure. In that case
<a href=#523 id=523 data-nosnippet>523</a> // result.result.success is false and we must surface it as an error
<a href=#524 id=524 data-nosnippet>524</a> // variant so callers can `match` on it. Mirrors the Go SDK behavior
<a href=#525 id=525 data-nosnippet>525</a> // in `sdk/go/client.go::checkResult` (4xx → UpstreamClient,
<a href=#526 id=526 data-nosnippet>526</a> // 5xx → UpstreamServer).
<a href=#527 id=527 data-nosnippet>527</a> </span><span class="kw">if </span>!result.result.success {
<a href=#528 id=528 data-nosnippet>528</a> <span class="kw">let </span>(err_code, err_message, err_doc) = <span class="kw">match </span><span class="kw-2">&</span>result.result.error {
<a href=#529 id=529 data-nosnippet>529</a> <span class="prelude-val">Some</span>(e) => (e.code.clone(), e.message.clone(), e.doc_url.clone()),
<a href=#530 id=530 data-nosnippet>530</a> <span class="prelude-val">None </span>=> (
<a href=#531 id=531 data-nosnippet>531</a> result.result.status.clone(),
<a href=#532 id=532 data-nosnippet>532</a> <span class="macro">format!</span>(
<a href=#533 id=533 data-nosnippet>533</a> <span class="string">"scrape failed with status_code={}"</span>,
<a href=#534 id=534 data-nosnippet>534</a> result.result.status_code
<a href=#535 id=535 data-nosnippet>535</a> ),
<a href=#536 id=536 data-nosnippet>536</a> String::new(),
<a href=#537 id=537 data-nosnippet>537</a> ),
<a href=#538 id=538 data-nosnippet>538</a> };
<a href=#539 id=539 data-nosnippet>539</a> <span class="kw">let </span>api_err = ApiError {
<a href=#540 id=540 data-nosnippet>540</a> code: err_code,
<a href=#541 id=541 data-nosnippet>541</a> message: err_message,
<a href=#542 id=542 data-nosnippet>542</a> http_status: result.result.status_code,
<a href=#543 id=543 data-nosnippet>543</a> documentation_url: err_doc,
<a href=#544 id=544 data-nosnippet>544</a> hint: String::new(),
<a href=#545 id=545 data-nosnippet>545</a> retry_after_ms: <span class="number">0</span>,
<a href=#546 id=546 data-nosnippet>546</a> };
<a href=#547 id=547 data-nosnippet>547</a> <span class="kw">let </span>sc = result.result.status_code;
<a href=#548 id=548 data-nosnippet>548</a> <span class="kw">if </span>(<span class="number">400</span>..<span class="number">500</span>).contains(<span class="kw-2">&</span>sc) {
<a href=#549 id=549 data-nosnippet>549</a> <span class="kw">return </span><span class="prelude-val">Err</span>(ScrapflyError::UpstreamClient(api_err));
<a href=#550 id=550 data-nosnippet>550</a> }
<a href=#551 id=551 data-nosnippet>551</a> <span class="kw">if </span>(<span class="number">500</span>..<span class="number">600</span>).contains(<span class="kw-2">&</span>sc) {
<a href=#552 id=552 data-nosnippet>552</a> <span class="kw">return </span><span class="prelude-val">Err</span>(ScrapflyError::UpstreamServer(api_err));
<a href=#553 id=553 data-nosnippet>553</a> }
<a href=#554 id=554 data-nosnippet>554</a> <span class="comment">// Unknown status code (e.g. 0, timeouts) — fall through to generic
<a href=#555 id=555 data-nosnippet>555</a> // Api error rather than silently returning a failed result.
<a href=#556 id=556 data-nosnippet>556</a> </span><span class="kw">return </span><span class="prelude-val">Err</span>(ScrapflyError::Api(api_err));
<a href=#557 id=557 data-nosnippet>557</a> }
<a href=#558 id=558 data-nosnippet>558</a> <span class="comment">// Transparent large-object handling: when a scrape response is too
<a href=#559 id=559 data-nosnippet>559</a> // large, the engine offloads the body to a signed URL and sets
<a href=#560 id=560 data-nosnippet>560</a> // `format=clob|blob`, stashing the URL in `content`. The SDK must
<a href=#561 id=561 data-nosnippet>561</a> // auto-fetch and surface the final bytes + a user-friendly format
<a href=#562 id=562 data-nosnippet>562</a> // marker (clob→text, blob→binary). Mirrors `sdk/go/client.go::handleLargeObjects`.
<a href=#563 id=563 data-nosnippet>563</a> </span><span class="kw">if </span>result.result.success && result.result.status == <span class="string">"DONE" </span>{
<a href=#564 id=564 data-nosnippet>564</a> <span class="kw">let </span>fmt = result.result.format.as_str();
<a href=#565 id=565 data-nosnippet>565</a> <span class="kw">if </span>fmt == <span class="string">"clob" </span>|| fmt == <span class="string">"blob" </span>{
<a href=#566 id=566 data-nosnippet>566</a> <span class="kw">let </span>(new_content, new_format) =
<a href=#567 id=567 data-nosnippet>567</a> <span class="self">self</span>.fetch_large_object(<span class="kw-2">&</span>result.result.content, fmt).<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#568 id=568 data-nosnippet>568</a> result.result.content = new_content;
<a href=#569 id=569 data-nosnippet>569</a> result.result.format = new_format;
<a href=#570 id=570 data-nosnippet>570</a> }
<a href=#571 id=571 data-nosnippet>571</a> }
<a href=#572 id=572 data-nosnippet>572</a> <span class="prelude-val">Ok</span>(result)
<a href=#573 id=573 data-nosnippet>573</a> }
<a href=#574 id=574 data-nosnippet>574</a>
<a href=#575 id=575 data-nosnippet>575</a> <span class="doccomment">/// Fetch an offloaded large-object body from its signed URL, re-attaching
<a href=#576 id=576 data-nosnippet>576</a> /// the API key as a query param. Returns `(content, format)`:
<a href=#577 id=577 data-nosnippet>577</a> /// `clob → ("…text…", "text")`, `blob → ("…bytes as lossy utf8…", "binary")`.
<a href=#578 id=578 data-nosnippet>578</a> </span><span class="kw">async fn </span>fetch_large_object(
<a href=#579 id=579 data-nosnippet>579</a> <span class="kw-2">&</span><span class="self">self</span>,
<a href=#580 id=580 data-nosnippet>580</a> content_url: <span class="kw-2">&</span>str,
<a href=#581 id=581 data-nosnippet>581</a> format: <span class="kw-2">&</span>str,
<a href=#582 id=582 data-nosnippet>582</a> ) -> <span class="prelude-ty">Result</span><(String, String), ScrapflyError> {
<a href=#583 id=583 data-nosnippet>583</a> <span class="kw">let </span><span class="kw-2">mut </span>url = Url::parse(content_url)
<a href=#584 id=584 data-nosnippet>584</a> .map_err(|e| ScrapflyError::Config(<span class="macro">format!</span>(<span class="string">"invalid large-object url: {}"</span>, e)))<span class="question-mark">?</span>;
<a href=#585 id=585 data-nosnippet>585</a> <span class="comment">// Append the API key without clobbering existing query params.
<a href=#586 id=586 data-nosnippet>586</a> </span>{
<a href=#587 id=587 data-nosnippet>587</a> <span class="kw">let </span>existing: Vec<(String, String)> = url
<a href=#588 id=588 data-nosnippet>588</a> .query_pairs()
<a href=#589 id=589 data-nosnippet>589</a> .filter(|(k, <span class="kw">_</span>)| k != <span class="string">"key"</span>)
<a href=#590 id=590 data-nosnippet>590</a> .map(|(k, v)| (k.into_owned(), v.into_owned()))
<a href=#591 id=591 data-nosnippet>591</a> .collect();
<a href=#592 id=592 data-nosnippet>592</a> <span class="kw">let </span><span class="kw-2">mut </span>qs = url.query_pairs_mut();
<a href=#593 id=593 data-nosnippet>593</a> qs.clear();
<a href=#594 id=594 data-nosnippet>594</a> <span class="kw">for </span>(k, v) <span class="kw">in </span>existing {
<a href=#595 id=595 data-nosnippet>595</a> qs.append_pair(<span class="kw-2">&</span>k, <span class="kw-2">&</span>v);
<a href=#596 id=596 data-nosnippet>596</a> }
<a href=#597 id=597 data-nosnippet>597</a> qs.append_pair(<span class="string">"key"</span>, <span class="self">self</span>.api_key());
<a href=#598 id=598 data-nosnippet>598</a> }
<a href=#599 id=599 data-nosnippet>599</a> <span class="kw">let </span><span class="kw-2">mut </span>headers = HeaderMap::new();
<a href=#600 id=600 data-nosnippet>600</a> headers.insert(ACCEPT, HeaderValue::from_static(<span class="string">"application/json"</span>));
<a href=#601 id=601 data-nosnippet>601</a> <span class="kw">let </span>resp = <span class="self">self
<a href=#602 id=602 data-nosnippet>602</a> </span>.send_with_retry(Method::GET, url, <span class="prelude-val">Some</span>(headers), <span class="prelude-val">None</span>)
<a href=#603 id=603 data-nosnippet>603</a> .<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#604 id=604 data-nosnippet>604</a> <span class="kw">let </span>(status, _h, body) = read_response(resp).<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#605 id=605 data-nosnippet>605</a> <span class="kw">if </span>status != <span class="number">200 </span>{
<a href=#606 id=606 data-nosnippet>606</a> <span class="kw">return </span><span class="prelude-val">Err</span>(from_response(status, <span class="kw-2">&</span>body, <span class="number">0</span>, <span class="bool-val">false</span>));
<a href=#607 id=607 data-nosnippet>607</a> }
<a href=#608 id=608 data-nosnippet>608</a> <span class="kw">let </span>new_format = <span class="kw">match </span>format {
<a href=#609 id=609 data-nosnippet>609</a> <span class="string">"clob" </span>=> <span class="string">"text"</span>,
<a href=#610 id=610 data-nosnippet>610</a> <span class="string">"blob" </span>=> <span class="string">"binary"</span>,
<a href=#611 id=611 data-nosnippet>611</a> <span class="kw">_ </span>=> {
<a href=#612 id=612 data-nosnippet>612</a> <span class="kw">return </span><span class="prelude-val">Err</span>(ScrapflyError::Config(<span class="macro">format!</span>(
<a href=#613 id=613 data-nosnippet>613</a> <span class="string">"unsupported large-object format: {}"</span>,
<a href=#614 id=614 data-nosnippet>614</a> format
<a href=#615 id=615 data-nosnippet>615</a> )))
<a href=#616 id=616 data-nosnippet>616</a> }
<a href=#617 id=617 data-nosnippet>617</a> };
<a href=#618 id=618 data-nosnippet>618</a> <span class="comment">// For blob (binary PDF, image, etc.) we use from_utf8_lossy to
<a href=#619 id=619 data-nosnippet>619</a> // preserve the raw bytes in the `content` string field, matching
<a href=#620 id=620 data-nosnippet>620</a> // the Go/Python SDKs' behavior.
<a href=#621 id=621 data-nosnippet>621</a> </span><span class="kw">let </span>content = String::from_utf8_lossy(<span class="kw-2">&</span>body).into_owned();
<a href=#622 id=622 data-nosnippet>622</a> <span class="prelude-val">Ok</span>((content, new_format.to_string()))
<a href=#623 id=623 data-nosnippet>623</a> }
<a href=#624 id=624 data-nosnippet>624</a>
<a href=#625 id=625 data-nosnippet>625</a> <span class="doccomment">/// Concurrent-scrape stream. Emits results in completion order.
<a href=#626 id=626 data-nosnippet>626</a> </span><span class="kw">pub fn </span>concurrent_scrape<<span class="lifetime">'a</span>, I>(
<a href=#627 id=627 data-nosnippet>627</a> <span class="kw-2">&</span><span class="lifetime">'a </span><span class="self">self</span>,
<a href=#628 id=628 data-nosnippet>628</a> configs: I,
<a href=#629 id=629 data-nosnippet>629</a> concurrency_limit: usize,
<a href=#630 id=630 data-nosnippet>630</a> ) -> <span class="kw">impl </span>Stream<Item = <span class="prelude-ty">Result</span><ScrapeResult, ScrapflyError>> + <span class="lifetime">'a
<a href=#631 id=631 data-nosnippet>631</a> </span><span class="kw">where
<a href=#632 id=632 data-nosnippet>632</a> </span>I: IntoIterator<Item = ScrapeConfig> + <span class="lifetime">'a</span>,
<a href=#633 id=633 data-nosnippet>633</a> <I <span class="kw">as </span>IntoIterator>::IntoIter: <span class="lifetime">'a</span>,
<a href=#634 id=634 data-nosnippet>634</a> {
<a href=#635 id=635 data-nosnippet>635</a> <span class="kw">let </span>limit = <span class="kw">if </span>concurrency_limit == <span class="number">0 </span>{
<a href=#636 id=636 data-nosnippet>636</a> <span class="number">5
<a href=#637 id=637 data-nosnippet>637</a> </span>} <span class="kw">else </span>{
<a href=#638 id=638 data-nosnippet>638</a> concurrency_limit
<a href=#639 id=639 data-nosnippet>639</a> };
<a href=#640 id=640 data-nosnippet>640</a> futures_util::stream::iter(
<a href=#641 id=641 data-nosnippet>641</a> configs
<a href=#642 id=642 data-nosnippet>642</a> .into_iter()
<a href=#643 id=643 data-nosnippet>643</a> .map(<span class="kw">move </span>|cfg| <span class="kw">async move </span>{ <span class="self">self</span>.scrape(<span class="kw-2">&</span>cfg).<span class="kw">await </span>}),
<a href=#644 id=644 data-nosnippet>644</a> )
<a href=#645 id=645 data-nosnippet>645</a> .buffer_unordered(limit)
<a href=#646 id=646 data-nosnippet>646</a> }
<a href=#647 id=647 data-nosnippet>647</a>
<a href=#648 id=648 data-nosnippet>648</a> <span class="doccomment">/// POST /scrape/batch: scrape up to 100 URLs and stream results
<a href=#649 id=649 data-nosnippet>649</a> /// back as each scrape completes. Returns an async stream where
<a href=#650 id=650 data-nosnippet>650</a> /// each item is `(correlation_id, Result<ScrapeResult, ScrapflyError>)`.
<a href=#651 id=651 data-nosnippet>651</a> ///
<a href=#652 id=652 data-nosnippet>652</a> /// Results arrive OUT OF ORDER — whichever scrape finishes first
<a href=#653 id=653 data-nosnippet>653</a> /// is yielded first. Every `ScrapeConfig` MUST carry a unique
<a href=#654 id=654 data-nosnippet>654</a> /// `correlation_id`; missing / duplicate values are caught
<a href=#655 id=655 data-nosnippet>655</a> /// client-side before the request is sent.
<a href=#656 id=656 data-nosnippet>656</a> ///
<a href=#657 id=657 data-nosnippet>657</a> /// Batch-level failures (plan gate, insufficient concurrency,
<a href=#658 id=658 data-nosnippet>658</a> /// validation) surface as the outer `Err(ScrapflyError)` returned
<a href=#659 id=659 data-nosnippet>659</a> /// from the `await` — the stream is only created after the
<a href=#660 id=660 data-nosnippet>660</a> /// batch request succeeds.
<a href=#661 id=661 data-nosnippet>661</a> </span><span class="kw">pub async fn </span>scrape_batch(
<a href=#662 id=662 data-nosnippet>662</a> <span class="kw-2">&</span><span class="self">self</span>,
<a href=#663 id=663 data-nosnippet>663</a> configs: <span class="kw-2">&</span>[ScrapeConfig],
<a href=#664 id=664 data-nosnippet>664</a> ) -> <span class="prelude-ty">Result</span><<span class="kw">impl </span>Stream<Item = (String, <span class="kw">crate</span>::batch::BatchOutcome)>, ScrapflyError> {
<a href=#665 id=665 data-nosnippet>665</a> <span class="self">self</span>.scrape_batch_with_options(configs, <span class="kw">crate</span>::batch::BatchOptions::default())
<a href=#666 id=666 data-nosnippet>666</a> .<span class="kw">await
<a href=#667 id=667 data-nosnippet>667</a> </span>}
<a href=#668 id=668 data-nosnippet>668</a>
<a href=#669 id=669 data-nosnippet>669</a> <span class="doccomment">/// Like `scrape_batch` but with explicit `BatchOptions`
<a href=#670 id=670 data-nosnippet>670</a> /// (msgpack wire format, etc.).
<a href=#671 id=671 data-nosnippet>671</a> </span><span class="kw">pub async fn </span>scrape_batch_with_options(
<a href=#672 id=672 data-nosnippet>672</a> <span class="kw-2">&</span><span class="self">self</span>,
<a href=#673 id=673 data-nosnippet>673</a> configs: <span class="kw-2">&</span>[ScrapeConfig],
<a href=#674 id=674 data-nosnippet>674</a> opts: <span class="kw">crate</span>::batch::BatchOptions,
<a href=#675 id=675 data-nosnippet>675</a> ) -> <span class="prelude-ty">Result</span><<span class="kw">impl </span>Stream<Item = (String, <span class="kw">crate</span>::batch::BatchOutcome)>, ScrapflyError> {
<a href=#676 id=676 data-nosnippet>676</a> <span class="kw">use </span><span class="kw">crate</span>::batch::{
<a href=#677 id=677 data-nosnippet>677</a> build_proxified_response, decode_part_body, parts_from_response, BatchOutcome,
<a href=#678 id=678 data-nosnippet>678</a> };
<a href=#679 id=679 data-nosnippet>679</a>
<a href=#680 id=680 data-nosnippet>680</a> <span class="kw">if </span>configs.is_empty() {
<a href=#681 id=681 data-nosnippet>681</a> <span class="kw">return </span><span class="prelude-val">Err</span>(ScrapflyError::Config(
<a href=#682 id=682 data-nosnippet>682</a> <span class="string">"scrape_batch: configs is empty"</span>.into(),
<a href=#683 id=683 data-nosnippet>683</a> ));
<a href=#684 id=684 data-nosnippet>684</a> }
<a href=#685 id=685 data-nosnippet>685</a>
<a href=#686 id=686 data-nosnippet>686</a> <span class="kw">if </span>configs.len() > <span class="number">100 </span>{
<a href=#687 id=687 data-nosnippet>687</a> <span class="kw">return </span><span class="prelude-val">Err</span>(ScrapflyError::Config(<span class="macro">format!</span>(
<a href=#688 id=688 data-nosnippet>688</a> <span class="string">"scrape_batch: max 100 configs per batch (got {})"</span>,
<a href=#689 id=689 data-nosnippet>689</a> configs.len()
<a href=#690 id=690 data-nosnippet>690</a> )));
<a href=#691 id=691 data-nosnippet>691</a> }
<a href=#692 id=692 data-nosnippet>692</a>
<a href=#693 id=693 data-nosnippet>693</a> <span class="kw">let </span><span class="kw-2">mut </span>seen: HashMap<String, usize> = HashMap::new();
<a href=#694 id=694 data-nosnippet>694</a> <span class="kw">let </span><span class="kw-2">mut </span>body_configs: Vec<HashMap<String, String>> = Vec::with_capacity(configs.len());
<a href=#695 id=695 data-nosnippet>695</a>
<a href=#696 id=696 data-nosnippet>696</a> <span class="kw">for </span>(i, cfg) <span class="kw">in </span>configs.iter().enumerate() {
<a href=#697 id=697 data-nosnippet>697</a> <span class="kw">let </span>correlation_id = cfg.correlation_id.clone().ok_or_else(|| {
<a href=#698 id=698 data-nosnippet>698</a> ScrapflyError::Config(<span class="macro">format!</span>(
<a href=#699 id=699 data-nosnippet>699</a> <span class="string">"scrape_batch: configs[{}] is missing correlation_id (required for matching streamed parts)"</span>,
<a href=#700 id=700 data-nosnippet>700</a> i
<a href=#701 id=701 data-nosnippet>701</a> ))
<a href=#702 id=702 data-nosnippet>702</a> })<span class="question-mark">?</span>;
<a href=#703 id=703 data-nosnippet>703</a>
<a href=#704 id=704 data-nosnippet>704</a> <span class="kw">if let </span><span class="prelude-val">Some</span>(prev) = seen.get(<span class="kw-2">&</span>correlation_id) {
<a href=#705 id=705 data-nosnippet>705</a> <span class="kw">return </span><span class="prelude-val">Err</span>(ScrapflyError::Config(<span class="macro">format!</span>(
<a href=#706 id=706 data-nosnippet>706</a> <span class="string">"scrape_batch: correlation_id {:?} reused by configs[{}] and configs[{}]"</span>,
<a href=#707 id=707 data-nosnippet>707</a> correlation_id, prev, i
<a href=#708 id=708 data-nosnippet>708</a> )));
<a href=#709 id=709 data-nosnippet>709</a> }
<a href=#710 id=710 data-nosnippet>710</a>
<a href=#711 id=711 data-nosnippet>711</a> seen.insert(correlation_id.clone(), i);
<a href=#712 id=712 data-nosnippet>712</a>
<a href=#713 id=713 data-nosnippet>713</a> <span class="kw">let </span>pairs = cfg.to_query_pairs()<span class="question-mark">?</span>;
<a href=#714 id=714 data-nosnippet>714</a> <span class="kw">let </span><span class="kw-2">mut </span>entry: HashMap<String, String> = HashMap::with_capacity(pairs.len());
<a href=#715 id=715 data-nosnippet>715</a>
<a href=#716 id=716 data-nosnippet>716</a> <span class="kw">for </span>(k, v) <span class="kw">in </span>pairs {
<a href=#717 id=717 data-nosnippet>717</a> <span class="kw">if </span>k == <span class="string">"key" </span>{
<a href=#718 id=718 data-nosnippet>718</a> <span class="kw">continue</span>;
<a href=#719 id=719 data-nosnippet>719</a> }
<a href=#720 id=720 data-nosnippet>720</a>
<a href=#721 id=721 data-nosnippet>721</a> entry.insert(k, v);
<a href=#722 id=722 data-nosnippet>722</a> }
<a href=#723 id=723 data-nosnippet>723</a>
<a href=#724 id=724 data-nosnippet>724</a> body_configs.push(entry);
<a href=#725 id=725 data-nosnippet>725</a> }
<a href=#726 id=726 data-nosnippet>726</a>
<a href=#727 id=727 data-nosnippet>727</a> <span class="kw">let </span>body = <span class="macro">serde_json::json!</span>({ <span class="string">"configs"</span>: body_configs });
<a href=#728 id=728 data-nosnippet>728</a> <span class="kw">let </span>body_bytes = serde_json::to_vec(<span class="kw-2">&</span>body)<span class="question-mark">?</span>;
<a href=#729 id=729 data-nosnippet>729</a>
<a href=#730 id=730 data-nosnippet>730</a> <span class="kw">let </span><span class="kw-2">mut </span>url = Url::parse(<span class="kw-2">&</span><span class="self">self</span>.host)
<a href=#731 id=731 data-nosnippet>731</a> .map_err(|e| ScrapflyError::Config(<span class="macro">format!</span>(<span class="string">"invalid host: {}"</span>, e)))<span class="question-mark">?</span>;
<a href=#732 id=732 data-nosnippet>732</a> url.set_path(<span class="string">"/scrape/batch"</span>);
<a href=#733 id=733 data-nosnippet>733</a> url.query_pairs_mut().append_pair(<span class="string">"key"</span>, <span class="kw-2">&</span><span class="self">self</span>.key);
<a href=#734 id=734 data-nosnippet>734</a>
<a href=#735 id=735 data-nosnippet>735</a> <span class="kw">let </span><span class="kw-2">mut </span>headers = HeaderMap::new();
<a href=#736 id=736 data-nosnippet>736</a> headers.insert(CONTENT_TYPE, HeaderValue::from_static(<span class="string">"application/json"</span>));
<a href=#737 id=737 data-nosnippet>737</a> headers.insert(
<a href=#738 id=738 data-nosnippet>738</a> ACCEPT,
<a href=#739 id=739 data-nosnippet>739</a> HeaderValue::from_static(opts.format.accept_header()),
<a href=#740 id=740 data-nosnippet>740</a> );
<a href=#741 id=741 data-nosnippet>741</a> headers.insert(USER_AGENT, HeaderValue::from_static(SDK_USER_AGENT));
<a href=#742 id=742 data-nosnippet>742</a>
<a href=#743 id=743 data-nosnippet>743</a> <span class="kw">let </span>method = Method::POST;
<a href=#744 id=744 data-nosnippet>744</a>
<a href=#745 id=745 data-nosnippet>745</a> <span class="kw">if let </span><span class="prelude-val">Some</span>(cb) = <span class="kw-2">&</span><span class="self">self</span>.on_request {
<a href=#746 id=746 data-nosnippet>746</a> cb(<span class="kw-2">&</span>method, <span class="kw-2">&</span>url, <span class="kw-2">&</span>headers);
<a href=#747 id=747 data-nosnippet>747</a> }
<a href=#748 id=748 data-nosnippet>748</a>
<a href=#749 id=749 data-nosnippet>749</a> <span class="kw">let </span>resp = <span class="self">self
<a href=#750 id=750 data-nosnippet>750</a> </span>.http
<a href=#751 id=751 data-nosnippet>751</a> .request(method, url)
<a href=#752 id=752 data-nosnippet>752</a> .headers(headers)
<a href=#753 id=753 data-nosnippet>753</a> .body(body_bytes)
<a href=#754 id=754 data-nosnippet>754</a> .send()
<a href=#755 id=755 data-nosnippet>755</a> .<span class="kw">await
<a href=#756 id=756 data-nosnippet>756</a> </span>.map_err(|e| ScrapflyError::Config(<span class="macro">format!</span>(<span class="string">"scrape_batch: send: {}"</span>, e)))<span class="question-mark">?</span>;
<a href=#757 id=757 data-nosnippet>757</a>
<a href=#758 id=758 data-nosnippet>758</a> <span class="kw">let </span>status = resp.status().as_u16();
<a href=#759 id=759 data-nosnippet>759</a>
<a href=#760 id=760 data-nosnippet>760</a> <span class="kw">if </span>status != <span class="number">200 </span>{
<a href=#761 id=761 data-nosnippet>761</a> <span class="kw">let </span>body_bytes = resp.bytes().<span class="kw">await</span>.unwrap_or_default();
<a href=#762 id=762 data-nosnippet>762</a>
<a href=#763 id=763 data-nosnippet>763</a> <span class="kw">return </span><span class="prelude-val">Err</span>(from_response(status, <span class="kw-2">&</span>body_bytes, <span class="number">0</span>, <span class="bool-val">false</span>));
<a href=#764 id=764 data-nosnippet>764</a> }
<a href=#765 id=765 data-nosnippet>765</a>
<a href=#766 id=766 data-nosnippet>766</a> <span class="kw">let </span>parts_stream = parts_from_response(resp)<span class="question-mark">?</span>;
<a href=#767 id=767 data-nosnippet>767</a>
<a href=#768 id=768 data-nosnippet>768</a> <span class="prelude-val">Ok</span>(parts_stream.map(|part_r| <span class="kw">match </span>part_r {
<a href=#769 id=769 data-nosnippet>769</a> <span class="prelude-val">Ok</span>(part) => {
<a href=#770 id=770 data-nosnippet>770</a> <span class="kw">let </span>correlation_id = part
<a href=#771 id=771 data-nosnippet>771</a> .headers
<a href=#772 id=772 data-nosnippet>772</a> .get(<span class="string">"x-scrapfly-correlation-id"</span>)
<a href=#773 id=773 data-nosnippet>773</a> .cloned()
<a href=#774 id=774 data-nosnippet>774</a> .unwrap_or_default();
<a href=#775 id=775 data-nosnippet>775</a>
<a href=#776 id=776 data-nosnippet>776</a> <span class="comment">// Proxified-response parts: the part body is the raw
<a href=#777 id=777 data-nosnippet>777</a> // upstream bytes, not a JSON envelope. Surface as a
<a href=#778 id=778 data-nosnippet>778</a> // BatchProxifiedResponse rather than attempting to
<a href=#779 id=779 data-nosnippet>779</a> // decode the body as JSON.
<a href=#780 id=780 data-nosnippet>780</a> </span><span class="kw">if </span>part
<a href=#781 id=781 data-nosnippet>781</a> .headers
<a href=#782 id=782 data-nosnippet>782</a> .get(<span class="string">"x-scrapfly-proxified"</span>)
<a href=#783 id=783 data-nosnippet>783</a> .map(|v| v == <span class="string">"true"</span>)
<a href=#784 id=784 data-nosnippet>784</a> .unwrap_or(<span class="bool-val">false</span>)
<a href=#785 id=785 data-nosnippet>785</a> {
<a href=#786 id=786 data-nosnippet>786</a> <span class="kw">let </span>prox = build_proxified_response(part);
<a href=#787 id=787 data-nosnippet>787</a> <span class="kw">return </span>(correlation_id, BatchOutcome::Proxified(prox));
<a href=#788 id=788 data-nosnippet>788</a> }
<a href=#789 id=789 data-nosnippet>789</a>
<a href=#790 id=790 data-nosnippet>790</a> <span class="kw">match </span>decode_part_body::<ScrapeResult>(<span class="kw-2">&</span>part) {
<a href=#791 id=791 data-nosnippet>791</a> <span class="prelude-val">Ok</span>(r) => (correlation_id, BatchOutcome::Scrape(r)),
<a href=#792 id=792 data-nosnippet>792</a> <span class="prelude-val">Err</span>(e) => (correlation_id, BatchOutcome::Err(e)),
<a href=#793 id=793 data-nosnippet>793</a> }
<a href=#794 id=794 data-nosnippet>794</a> }
<a href=#795 id=795 data-nosnippet>795</a> <span class="prelude-val">Err</span>(e) => (String::new(), BatchOutcome::Err(e)),
<a href=#796 id=796 data-nosnippet>796</a> }))
<a href=#797 id=797 data-nosnippet>797</a> }
<a href=#798 id=798 data-nosnippet>798</a>
<a href=#799 id=799 data-nosnippet>799</a> <span class="doccomment">/// Scrape a URL with `proxified_response=true`, returning the raw
<a href=#800 id=800 data-nosnippet>800</a> /// upstream `reqwest::Response` (target's status, headers, body).
<a href=#801 id=801 data-nosnippet>801</a> ///
<a href=#802 id=802 data-nosnippet>802</a> /// Unlike [`scrape()`], no JSON parsing occurs — the response body is
<a href=#803 id=803 data-nosnippet>803</a> /// the target page's raw content. Scrapfly metadata is available on
<a href=#804 id=804 data-nosnippet>804</a> /// the `X-Scrapfly-*` response headers (`Api-Cost`, `Content-Format`,
<a href=#805 id=805 data-nosnippet>805</a> /// `Log`, etc.).
<a href=#806 id=806 data-nosnippet>806</a> ///
<a href=#807 id=807 data-nosnippet>807</a> /// Automatically forces `proxified_response=true` regardless of the
<a href=#808 id=808 data-nosnippet>808</a> /// config's field value.
<a href=#809 id=809 data-nosnippet>809</a> </span><span class="kw">pub async fn </span>scrape_proxified(
<a href=#810 id=810 data-nosnippet>810</a> <span class="kw-2">&</span><span class="self">self</span>,
<a href=#811 id=811 data-nosnippet>811</a> config: <span class="kw-2">&</span>ScrapeConfig,
<a href=#812 id=812 data-nosnippet>812</a> ) -> <span class="prelude-ty">Result</span><reqwest::Response, ScrapflyError> {
<a href=#813 id=813 data-nosnippet>813</a> <span class="kw">let </span><span class="kw-2">mut </span>cfg = config.clone();
<a href=#814 id=814 data-nosnippet>814</a> cfg.proxified_response = <span class="bool-val">true</span>;
<a href=#815 id=815 data-nosnippet>815</a> <span class="kw">let </span>pairs = cfg.to_query_pairs()<span class="question-mark">?</span>;
<a href=#816 id=816 data-nosnippet>816</a> <span class="kw">let </span>url = <span class="self">self</span>.build_url(<span class="string">"/scrape"</span>, <span class="kw-2">&</span>pairs)<span class="question-mark">?</span>;
<a href=#817 id=817 data-nosnippet>817</a> <span class="kw">let </span>method = <span class="kw">match </span>cfg.method {
<a href=#818 id=818 data-nosnippet>818</a> <span class="prelude-val">Some</span>(m) => Method::from_bytes(m.as_str().as_bytes())
<a href=#819 id=819 data-nosnippet>819</a> .map_err(|e| ScrapflyError::Config(<span class="macro">format!</span>(<span class="string">"invalid method: {}"</span>, e)))<span class="question-mark">?</span>,
<a href=#820 id=820 data-nosnippet>820</a> <span class="prelude-val">None </span>=> Method::GET,
<a href=#821 id=821 data-nosnippet>821</a> };
<a href=#822 id=822 data-nosnippet>822</a> <span class="kw">let </span>body = cfg.body.clone();
<a href=#823 id=823 data-nosnippet>823</a> <span class="kw">let </span>resp = <span class="self">self
<a href=#824 id=824 data-nosnippet>824</a> </span>.send_with_retry(method, url, <span class="prelude-val">None</span>, body.map(|b| b.into_bytes()))
<a href=#825 id=825 data-nosnippet>825</a> .<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#826 id=826 data-nosnippet>826</a> <span class="comment">// Error restoration: if X-Scrapfly-Reject-Code is present, the
<a href=#827 id=827 data-nosnippet>827</a> // scrape failed. Return a typed error so callers get the same
<a href=#828 id=828 data-nosnippet>828</a> // interface as non-proxified mode.
<a href=#829 id=829 data-nosnippet>829</a> </span><span class="kw">if let </span><span class="prelude-val">Some</span>(reject_code) = resp.headers().get(<span class="string">"x-scrapfly-reject-code"</span>) {
<a href=#830 id=830 data-nosnippet>830</a> <span class="kw">let </span>code = reject_code.to_str().unwrap_or(<span class="string">""</span>).to_string();
<a href=#831 id=831 data-nosnippet>831</a> <span class="kw">let </span>desc = resp
<a href=#832 id=832 data-nosnippet>832</a> .headers()
<a href=#833 id=833 data-nosnippet>833</a> .get(<span class="string">"x-scrapfly-reject-description"</span>)
<a href=#834 id=834 data-nosnippet>834</a> .and_then(|v| v.to_str().ok())
<a href=#835 id=835 data-nosnippet>835</a> .unwrap_or(<span class="string">""</span>)
<a href=#836 id=836 data-nosnippet>836</a> .to_string();
<a href=#837 id=837 data-nosnippet>837</a> <span class="kw">let </span>retryable = resp
<a href=#838 id=838 data-nosnippet>838</a> .headers()
<a href=#839 id=839 data-nosnippet>839</a> .get(<span class="string">"x-scrapfly-reject-retryable"</span>)
<a href=#840 id=840 data-nosnippet>840</a> .and_then(|v| v.to_str().ok())
<a href=#841 id=841 data-nosnippet>841</a> .unwrap_or(<span class="string">"false"</span>)
<a href=#842 id=842 data-nosnippet>842</a> == <span class="string">"true"</span>;
<a href=#843 id=843 data-nosnippet>843</a> <span class="kw">let </span>retry_after_ms: u64 = <span class="kw">if </span>retryable {
<a href=#844 id=844 data-nosnippet>844</a> resp.headers()
<a href=#845 id=845 data-nosnippet>845</a> .get(<span class="string">"retry-after"</span>)
<a href=#846 id=846 data-nosnippet>846</a> .and_then(|v| v.to_str().ok())
<a href=#847 id=847 data-nosnippet>847</a> .and_then(|v| v.parse::<u64>().ok())
<a href=#848 id=848 data-nosnippet>848</a> .unwrap_or(<span class="number">0</span>)
<a href=#849 id=849 data-nosnippet>849</a> * <span class="number">1000 </span><span class="comment">// Retry-After header is in seconds
<a href=#850 id=850 data-nosnippet>850</a> </span>} <span class="kw">else </span>{
<a href=#851 id=851 data-nosnippet>851</a> <span class="number">0
<a href=#852 id=852 data-nosnippet>852</a> </span>};
<a href=#853 id=853 data-nosnippet>853</a> <span class="kw">let </span>status = resp.status().as_u16();
<a href=#854 id=854 data-nosnippet>854</a> <span class="kw">let </span>doc = resp
<a href=#855 id=855 data-nosnippet>855</a> .headers()
<a href=#856 id=856 data-nosnippet>856</a> .get(<span class="string">"x-scrapfly-reject-doc"</span>)
<a href=#857 id=857 data-nosnippet>857</a> .and_then(|v| v.to_str().ok())
<a href=#858 id=858 data-nosnippet>858</a> .unwrap_or(<span class="string">""</span>)
<a href=#859 id=859 data-nosnippet>859</a> .to_string();
<a href=#860 id=860 data-nosnippet>860</a> <span class="kw">return </span><span class="prelude-val">Err</span>(ScrapflyError::Api(<span class="kw">crate</span>::error::ApiError {
<a href=#861 id=861 data-nosnippet>861</a> code,
<a href=#862 id=862 data-nosnippet>862</a> message: <span class="macro">format!</span>(<span class="string">"Proxified scrape error: {}"</span>, desc),
<a href=#863 id=863 data-nosnippet>863</a> http_status: status,
<a href=#864 id=864 data-nosnippet>864</a> documentation_url: doc,
<a href=#865 id=865 data-nosnippet>865</a> hint: String::new(),
<a href=#866 id=866 data-nosnippet>866</a> retry_after_ms,
<a href=#867 id=867 data-nosnippet>867</a> }));
<a href=#868 id=868 data-nosnippet>868</a> }
<a href=#869 id=869 data-nosnippet>869</a> <span class="prelude-val">Ok</span>(resp)
<a href=#870 id=870 data-nosnippet>870</a> }
<a href=#871 id=871 data-nosnippet>871</a>
<a href=#872 id=872 data-nosnippet>872</a> <span class="doccomment">/// Screenshot a URL.
<a href=#873 id=873 data-nosnippet>873</a> </span><span class="kw">pub async fn </span>screenshot(
<a href=#874 id=874 data-nosnippet>874</a> <span class="kw-2">&</span><span class="self">self</span>,
<a href=#875 id=875 data-nosnippet>875</a> config: <span class="kw-2">&</span>ScreenshotConfig,
<a href=#876 id=876 data-nosnippet>876</a> ) -> <span class="prelude-ty">Result</span><ScreenshotResult, ScrapflyError> {
<a href=#877 id=877 data-nosnippet>877</a> <span class="kw">let </span>pairs = config.to_query_pairs()<span class="question-mark">?</span>;
<a href=#878 id=878 data-nosnippet>878</a> <span class="kw">let </span>url = <span class="self">self</span>.build_url(<span class="string">"/screenshot"</span>, <span class="kw-2">&</span>pairs)<span class="question-mark">?</span>;
<a href=#879 id=879 data-nosnippet>879</a> <span class="kw">let </span>resp = <span class="self">self</span>.send_with_retry(Method::GET, url, <span class="prelude-val">None</span>, <span class="prelude-val">None</span>).<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#880 id=880 data-nosnippet>880</a> <span class="kw">let </span>(status, headers, body) = read_response(resp).<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#881 id=881 data-nosnippet>881</a> <span class="kw">if </span>status != <span class="number">200 </span>{
<a href=#882 id=882 data-nosnippet>882</a> <span class="kw">return </span><span class="prelude-val">Err</span>(from_response(status, <span class="kw-2">&</span>body, <span class="number">0</span>, <span class="bool-val">false</span>));
<a href=#883 id=883 data-nosnippet>883</a> }
<a href=#884 id=884 data-nosnippet>884</a> <span class="kw">let </span>content_type = headers
<a href=#885 id=885 data-nosnippet>885</a> .get(CONTENT_TYPE)
<a href=#886 id=886 data-nosnippet>886</a> .and_then(|v| v.to_str().ok())
<a href=#887 id=887 data-nosnippet>887</a> .unwrap_or(<span class="string">"application/octet-stream"</span>);
<a href=#888 id=888 data-nosnippet>888</a> <span class="kw">let </span>ext = content_type
<a href=#889 id=889 data-nosnippet>889</a> .split(<span class="string">'/'</span>)
<a href=#890 id=890 data-nosnippet>890</a> .nth(<span class="number">1</span>)
<a href=#891 id=891 data-nosnippet>891</a> .and_then(|s| s.split(<span class="string">';'</span>).next())
<a href=#892 id=892 data-nosnippet>892</a> .unwrap_or(<span class="string">"bin"</span>)
<a href=#893 id=893 data-nosnippet>893</a> .to_string();
<a href=#894 id=894 data-nosnippet>894</a> <span class="kw">let </span>upstream_status_code: u16 = headers
<a href=#895 id=895 data-nosnippet>895</a> .get(<span class="string">"x-scrapfly-upstream-http-code"</span>)
<a href=#896 id=896 data-nosnippet>896</a> .and_then(|v| v.to_str().ok())
<a href=#897 id=897 data-nosnippet>897</a> .and_then(|s| s.parse().ok())
<a href=#898 id=898 data-nosnippet>898</a> .unwrap_or(<span class="number">0</span>);
<a href=#899 id=899 data-nosnippet>899</a> <span class="kw">let </span>upstream_url = headers
<a href=#900 id=900 data-nosnippet>900</a> .get(<span class="string">"x-scrapfly-upstream-url"</span>)
<a href=#901 id=901 data-nosnippet>901</a> .and_then(|v| v.to_str().ok())
<a href=#902 id=902 data-nosnippet>902</a> .unwrap_or(<span class="string">""</span>)
<a href=#903 id=903 data-nosnippet>903</a> .to_string();
<a href=#904 id=904 data-nosnippet>904</a> <span class="prelude-val">Ok</span>(ScreenshotResult {
<a href=#905 id=905 data-nosnippet>905</a> image: body,
<a href=#906 id=906 data-nosnippet>906</a> metadata: ScreenshotMetadata {
<a href=#907 id=907 data-nosnippet>907</a> extension_name: ext,
<a href=#908 id=908 data-nosnippet>908</a> upstream_status_code,
<a href=#909 id=909 data-nosnippet>909</a> upstream_url,
<a href=#910 id=910 data-nosnippet>910</a> },
<a href=#911 id=911 data-nosnippet>911</a> })
<a href=#912 id=912 data-nosnippet>912</a> }
<a href=#913 id=913 data-nosnippet>913</a>
<a href=#914 id=914 data-nosnippet>914</a> <span class="doccomment">/// Run AI extraction on a document.
<a href=#915 id=915 data-nosnippet>915</a> </span><span class="kw">pub async fn </span>extract(
<a href=#916 id=916 data-nosnippet>916</a> <span class="kw-2">&</span><span class="self">self</span>,
<a href=#917 id=917 data-nosnippet>917</a> config: <span class="kw-2">&</span>ExtractionConfig,
<a href=#918 id=918 data-nosnippet>918</a> ) -> <span class="prelude-ty">Result</span><ExtractionResult, ScrapflyError> {
<a href=#919 id=919 data-nosnippet>919</a> <span class="kw">let </span>pairs = config.to_query_pairs()<span class="question-mark">?</span>;
<a href=#920 id=920 data-nosnippet>920</a> <span class="kw">let </span>url = <span class="self">self</span>.build_url(<span class="string">"/extraction"</span>, <span class="kw-2">&</span>pairs)<span class="question-mark">?</span>;
<a href=#921 id=921 data-nosnippet>921</a> <span class="kw">let </span><span class="kw-2">mut </span>headers = HeaderMap::new();
<a href=#922 id=922 data-nosnippet>922</a> headers.insert(
<a href=#923 id=923 data-nosnippet>923</a> CONTENT_TYPE,
<a href=#924 id=924 data-nosnippet>924</a> HeaderValue::from_str(<span class="kw-2">&</span>config.content_type)
<a href=#925 id=925 data-nosnippet>925</a> .map_err(|e| ScrapflyError::Config(<span class="macro">format!</span>(<span class="string">"invalid content-type: {}"</span>, e)))<span class="question-mark">?</span>,
<a href=#926 id=926 data-nosnippet>926</a> );
<a href=#927 id=927 data-nosnippet>927</a> headers.insert(ACCEPT, HeaderValue::from_static(<span class="string">"application/json"</span>));
<a href=#928 id=928 data-nosnippet>928</a> <span class="kw">if let </span><span class="prelude-val">Some</span>(fmt) = config.document_compression_format {
<a href=#929 id=929 data-nosnippet>929</a> headers.insert(
<a href=#930 id=930 data-nosnippet>930</a> <span class="string">"content-encoding"</span>,
<a href=#931 id=931 data-nosnippet>931</a> HeaderValue::from_str(fmt.as_str())
<a href=#932 id=932 data-nosnippet>932</a> .map_err(|e| ScrapflyError::Config(<span class="macro">format!</span>(<span class="string">"invalid encoding: {}"</span>, e)))<span class="question-mark">?</span>,
<a href=#933 id=933 data-nosnippet>933</a> );
<a href=#934 id=934 data-nosnippet>934</a> }
<a href=#935 id=935 data-nosnippet>935</a> <span class="kw">let </span>resp = <span class="self">self
<a href=#936 id=936 data-nosnippet>936</a> </span>.send_with_retry(Method::POST, url, <span class="prelude-val">Some</span>(headers), <span class="prelude-val">Some</span>(config.body.clone()))
<a href=#937 id=937 data-nosnippet>937</a> .<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#938 id=938 data-nosnippet>938</a> <span class="kw">let </span>(status, _h, body_bytes) = read_response(resp).<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#939 id=939 data-nosnippet>939</a> <span class="kw">if </span>status != <span class="number">200 </span>{
<a href=#940 id=940 data-nosnippet>940</a> <span class="kw">return </span><span class="prelude-val">Err</span>(from_response(status, <span class="kw-2">&</span>body_bytes, <span class="number">0</span>, <span class="bool-val">false</span>));
<a href=#941 id=941 data-nosnippet>941</a> }
<a href=#942 id=942 data-nosnippet>942</a> <span class="prelude-val">Ok</span>(serde_json::from_slice(<span class="kw-2">&</span>body_bytes)<span class="question-mark">?</span>)
<a href=#943 id=943 data-nosnippet>943</a> }
<a href=#944 id=944 data-nosnippet>944</a>
<a href=#945 id=945 data-nosnippet>945</a> <span class="comment">// ==============================================================================
<a href=#946 id=946 data-nosnippet>946</a> // Crawler methods
<a href=#947 id=947 data-nosnippet>947</a> // ==============================================================================
<a href=#948 id=948 data-nosnippet>948</a>
<a href=#949 id=949 data-nosnippet>949</a> </span><span class="doccomment">/// Schedule a new crawler job.
<a href=#950 id=950 data-nosnippet>950</a> </span><span class="kw">pub async fn </span>start_crawl(
<a href=#951 id=951 data-nosnippet>951</a> <span class="kw-2">&</span><span class="self">self</span>,
<a href=#952 id=952 data-nosnippet>952</a> config: <span class="kw-2">&</span>CrawlerConfig,
<a href=#953 id=953 data-nosnippet>953</a> ) -> <span class="prelude-ty">Result</span><CrawlerStartResponse, ScrapflyError> {
<a href=#954 id=954 data-nosnippet>954</a> <span class="kw">let </span>body = config.to_json_body()<span class="question-mark">?</span>;
<a href=#955 id=955 data-nosnippet>955</a> <span class="kw">let </span>url = <span class="self">self</span>.build_url(<span class="string">"/crawl"</span>, <span class="kw-2">&</span>[])<span class="question-mark">?</span>;
<a href=#956 id=956 data-nosnippet>956</a> <span class="kw">let </span><span class="kw-2">mut </span>headers = HeaderMap::new();
<a href=#957 id=957 data-nosnippet>957</a> headers.insert(CONTENT_TYPE, HeaderValue::from_static(<span class="string">"application/json"</span>));
<a href=#958 id=958 data-nosnippet>958</a> headers.insert(ACCEPT, HeaderValue::from_static(<span class="string">"application/json"</span>));
<a href=#959 id=959 data-nosnippet>959</a> <span class="kw">let </span>resp = <span class="self">self
<a href=#960 id=960 data-nosnippet>960</a> </span>.send_with_retry(Method::POST, url, <span class="prelude-val">Some</span>(headers), <span class="prelude-val">Some</span>(body))
<a href=#961 id=961 data-nosnippet>961</a> .<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#962 id=962 data-nosnippet>962</a> <span class="kw">let </span>(status, _h, body_bytes) = read_response(resp).<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#963 id=963 data-nosnippet>963</a> <span class="kw">if </span>status != <span class="number">200 </span>&& status != <span class="number">201 </span>{
<a href=#964 id=964 data-nosnippet>964</a> <span class="kw">return </span><span class="prelude-val">Err</span>(from_response(status, <span class="kw-2">&</span>body_bytes, <span class="number">0</span>, <span class="bool-val">true</span>));
<a href=#965 id=965 data-nosnippet>965</a> }
<a href=#966 id=966 data-nosnippet>966</a> <span class="kw">let </span>parsed: CrawlerStartResponse = serde_json::from_slice(<span class="kw-2">&</span>body_bytes)<span class="question-mark">?</span>;
<a href=#967 id=967 data-nosnippet>967</a> <span class="kw">if </span>parsed.crawler_uuid.is_empty() {
<a href=#968 id=968 data-nosnippet>968</a> <span class="kw">return </span><span class="prelude-val">Err</span>(ScrapflyError::UnexpectedResponseFormat(
<a href=#969 id=969 data-nosnippet>969</a> <span class="string">"crawler start response missing crawler_uuid"</span>.into(),
<a href=#970 id=970 data-nosnippet>970</a> ));
<a href=#971 id=971 data-nosnippet>971</a> }
<a href=#972 id=972 data-nosnippet>972</a> <span class="prelude-val">Ok</span>(parsed)
<a href=#973 id=973 data-nosnippet>973</a> }
<a href=#974 id=974 data-nosnippet>974</a>
<a href=#975 id=975 data-nosnippet>975</a> <span class="doccomment">/// Fetch crawler status.
<a href=#976 id=976 data-nosnippet>976</a> </span><span class="kw">pub async fn </span>crawl_status(<span class="kw-2">&</span><span class="self">self</span>, uuid: <span class="kw-2">&</span>str) -> <span class="prelude-ty">Result</span><CrawlerStatus, ScrapflyError> {
<a href=#977 id=977 data-nosnippet>977</a> <span class="kw">if </span>uuid.is_empty() {
<a href=#978 id=978 data-nosnippet>978</a> <span class="kw">return </span><span class="prelude-val">Err</span>(ScrapflyError::Config(<span class="string">"uuid cannot be empty"</span>.into()));
<a href=#979 id=979 data-nosnippet>979</a> }
<a href=#980 id=980 data-nosnippet>980</a> <span class="kw">let </span>url = <span class="self">self</span>.build_url(<span class="kw-2">&</span><span class="macro">format!</span>(<span class="string">"/crawl/{}/status"</span>, uuid), <span class="kw-2">&</span>[])<span class="question-mark">?</span>;
<a href=#981 id=981 data-nosnippet>981</a> <span class="kw">let </span>resp = <span class="self">self</span>.send_with_retry(Method::GET, url, <span class="prelude-val">None</span>, <span class="prelude-val">None</span>).<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#982 id=982 data-nosnippet>982</a> <span class="kw">let </span>(status, _h, body) = read_response(resp).<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#983 id=983 data-nosnippet>983</a> <span class="kw">if </span>status != <span class="number">200 </span>{
<a href=#984 id=984 data-nosnippet>984</a> <span class="kw">return </span><span class="prelude-val">Err</span>(from_response(status, <span class="kw-2">&</span>body, <span class="number">0</span>, <span class="bool-val">true</span>));
<a href=#985 id=985 data-nosnippet>985</a> }
<a href=#986 id=986 data-nosnippet>986</a> <span class="prelude-val">Ok</span>(serde_json::from_slice(<span class="kw-2">&</span>body)<span class="question-mark">?</span>)
<a href=#987 id=987 data-nosnippet>987</a> }
<a href=#988 id=988 data-nosnippet>988</a>
<a href=#989 id=989 data-nosnippet>989</a> <span class="doccomment">/// List crawled URLs (streaming text endpoint).
<a href=#990 id=990 data-nosnippet>990</a> </span><span class="kw">pub async fn </span>crawl_urls(
<a href=#991 id=991 data-nosnippet>991</a> <span class="kw-2">&</span><span class="self">self</span>,
<a href=#992 id=992 data-nosnippet>992</a> uuid: <span class="kw-2">&</span>str,
<a href=#993 id=993 data-nosnippet>993</a> status_filter: <span class="prelude-ty">Option</span><<span class="kw-2">&</span>str>,
<a href=#994 id=994 data-nosnippet>994</a> page: u32,
<a href=#995 id=995 data-nosnippet>995</a> per_page: u32,
<a href=#996 id=996 data-nosnippet>996</a> ) -> <span class="prelude-ty">Result</span><CrawlerUrls, ScrapflyError> {
<a href=#997 id=997 data-nosnippet>997</a> <span class="kw">if </span>uuid.is_empty() {
<a href=#998 id=998 data-nosnippet>998</a> <span class="kw">return </span><span class="prelude-val">Err</span>(ScrapflyError::Config(<span class="string">"uuid cannot be empty"</span>.into()));
<a href=#999 id=999 data-nosnippet>999</a> }
<a href=#1000 id=1000 data-nosnippet>1000</a> <span class="kw">let </span>page = <span class="kw">if </span>page == <span class="number">0 </span>{ <span class="number">1 </span>} <span class="kw">else </span>{ page };
<a href=#1001 id=1001 data-nosnippet>1001</a> <span class="kw">let </span>per_page = <span class="kw">if </span>per_page == <span class="number">0 </span>{ <span class="number">100 </span>} <span class="kw">else </span>{ per_page };
<a href=#1002 id=1002 data-nosnippet>1002</a> <span class="kw">let </span>status_hint = status_filter.unwrap_or(<span class="string">"visited"</span>);
<a href=#1003 id=1003 data-nosnippet>1003</a> <span class="kw">let </span><span class="kw-2">mut </span>pairs: Vec<(String, String)> = <span class="macro">vec!</span>[
<a href=#1004 id=1004 data-nosnippet>1004</a> (<span class="string">"page"</span>.into(), page.to_string()),
<a href=#1005 id=1005 data-nosnippet>1005</a> (<span class="string">"per_page"</span>.into(), per_page.to_string()),
<a href=#1006 id=1006 data-nosnippet>1006</a> ];
<a href=#1007 id=1007 data-nosnippet>1007</a> <span class="kw">if let </span><span class="prelude-val">Some</span>(s) = status_filter {
<a href=#1008 id=1008 data-nosnippet>1008</a> pairs.push((<span class="string">"status"</span>.into(), s.to_string()));
<a href=#1009 id=1009 data-nosnippet>1009</a> }
<a href=#1010 id=1010 data-nosnippet>1010</a> <span class="kw">let </span>url = <span class="self">self</span>.build_url(<span class="kw-2">&</span><span class="macro">format!</span>(<span class="string">"/crawl/{}/urls"</span>, uuid), <span class="kw-2">&</span>pairs)<span class="question-mark">?</span>;
<a href=#1011 id=1011 data-nosnippet>1011</a> <span class="kw">let </span><span class="kw-2">mut </span>headers = HeaderMap::new();
<a href=#1012 id=1012 data-nosnippet>1012</a> headers.insert(
<a href=#1013 id=1013 data-nosnippet>1013</a> ACCEPT,
<a href=#1014 id=1014 data-nosnippet>1014</a> HeaderValue::from_static(<span class="string">"text/plain, application/json"</span>),
<a href=#1015 id=1015 data-nosnippet>1015</a> );
<a href=#1016 id=1016 data-nosnippet>1016</a> <span class="kw">let </span>resp = <span class="self">self
<a href=#1017 id=1017 data-nosnippet>1017</a> </span>.send_with_retry(Method::GET, url, <span class="prelude-val">Some</span>(headers), <span class="prelude-val">None</span>)
<a href=#1018 id=1018 data-nosnippet>1018</a> .<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#1019 id=1019 data-nosnippet>1019</a> <span class="kw">let </span>(status, resp_headers, body) = read_response(resp).<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#1020 id=1020 data-nosnippet>1020</a> <span class="kw">if </span>status != <span class="number">200 </span>{
<a href=#1021 id=1021 data-nosnippet>1021</a> <span class="kw">return </span><span class="prelude-val">Err</span>(from_response(status, <span class="kw-2">&</span>body, <span class="number">0</span>, <span class="bool-val">true</span>));
<a href=#1022 id=1022 data-nosnippet>1022</a> }
<a href=#1023 id=1023 data-nosnippet>1023</a> <span class="kw">let </span>ct = resp_headers
<a href=#1024 id=1024 data-nosnippet>1024</a> .get(CONTENT_TYPE)
<a href=#1025 id=1025 data-nosnippet>1025</a> .and_then(|v| v.to_str().ok())
<a href=#1026 id=1026 data-nosnippet>1026</a> .unwrap_or(<span class="string">""</span>);
<a href=#1027 id=1027 data-nosnippet>1027</a> <span class="kw">if </span>ct.contains(<span class="string">"application/json"</span>) {
<a href=#1028 id=1028 data-nosnippet>1028</a> <span class="kw">return </span><span class="prelude-val">Err</span>(ScrapflyError::UnexpectedResponseFormat(<span class="macro">format!</span>(
<a href=#1029 id=1029 data-nosnippet>1029</a> <span class="string">"GET /crawl/{}/urls returned JSON on a 200 response (expected text/plain)"</span>,
<a href=#1030 id=1030 data-nosnippet>1030</a> uuid
<a href=#1031 id=1031 data-nosnippet>1031</a> )));
<a href=#1032 id=1032 data-nosnippet>1032</a> }
<a href=#1033 id=1033 data-nosnippet>1033</a> <span class="kw">let </span>body_str = std::str::from_utf8(<span class="kw-2">&</span>body)
<a href=#1034 id=1034 data-nosnippet>1034</a> .map_err(|e| ScrapflyError::UnexpectedResponseFormat(<span class="macro">format!</span>(<span class="string">"invalid utf8: {}"</span>, e)))<span class="question-mark">?</span>;
<a href=#1035 id=1035 data-nosnippet>1035</a> <span class="prelude-val">Ok</span>(CrawlerUrls::from_text(
<a href=#1036 id=1036 data-nosnippet>1036</a> body_str,
<a href=#1037 id=1037 data-nosnippet>1037</a> status_hint,
<a href=#1038 id=1038 data-nosnippet>1038</a> page,
<a href=#1039 id=1039 data-nosnippet>1039</a> per_page,
<a href=#1040 id=1040 data-nosnippet>1040</a> ))
<a href=#1041 id=1041 data-nosnippet>1041</a> }
<a href=#1042 id=1042 data-nosnippet>1042</a>
<a href=#1043 id=1043 data-nosnippet>1043</a> <span class="doccomment">/// Bulk `GET /crawl/{uuid}/contents` in JSON mode.
<a href=#1044 id=1044 data-nosnippet>1044</a> </span><span class="kw">pub async fn </span>crawl_contents_json(
<a href=#1045 id=1045 data-nosnippet>1045</a> <span class="kw-2">&</span><span class="self">self</span>,
<a href=#1046 id=1046 data-nosnippet>1046</a> uuid: <span class="kw-2">&</span>str,
<a href=#1047 id=1047 data-nosnippet>1047</a> format: <span class="kw">crate</span>::enums::CrawlerContentFormat,
<a href=#1048 id=1048 data-nosnippet>1048</a> limit: <span class="prelude-ty">Option</span><u32>,
<a href=#1049 id=1049 data-nosnippet>1049</a> offset: <span class="prelude-ty">Option</span><u32>,
<a href=#1050 id=1050 data-nosnippet>1050</a> ) -> <span class="prelude-ty">Result</span><CrawlerContents, ScrapflyError> {
<a href=#1051 id=1051 data-nosnippet>1051</a> <span class="kw">if </span>uuid.is_empty() {
<a href=#1052 id=1052 data-nosnippet>1052</a> <span class="kw">return </span><span class="prelude-val">Err</span>(ScrapflyError::Config(<span class="string">"uuid cannot be empty"</span>.into()));
<a href=#1053 id=1053 data-nosnippet>1053</a> }
<a href=#1054 id=1054 data-nosnippet>1054</a> <span class="kw">let </span><span class="kw-2">mut </span>pairs: Vec<(String, String)> = <span class="macro">vec!</span>[(<span class="string">"formats"</span>.into(), format.as_str().into())];
<a href=#1055 id=1055 data-nosnippet>1055</a> <span class="kw">if let </span><span class="prelude-val">Some</span>(l) = limit {
<a href=#1056 id=1056 data-nosnippet>1056</a> pairs.push((<span class="string">"limit"</span>.into(), l.to_string()));
<a href=#1057 id=1057 data-nosnippet>1057</a> }
<a href=#1058 id=1058 data-nosnippet>1058</a> <span class="kw">if let </span><span class="prelude-val">Some</span>(o) = offset {
<a href=#1059 id=1059 data-nosnippet>1059</a> pairs.push((<span class="string">"offset"</span>.into(), o.to_string()));
<a href=#1060 id=1060 data-nosnippet>1060</a> }
<a href=#1061 id=1061 data-nosnippet>1061</a> <span class="kw">let </span>url = <span class="self">self</span>.build_url(<span class="kw-2">&</span><span class="macro">format!</span>(<span class="string">"/crawl/{}/contents"</span>, uuid), <span class="kw-2">&</span>pairs)<span class="question-mark">?</span>;
<a href=#1062 id=1062 data-nosnippet>1062</a> <span class="kw">let </span><span class="kw-2">mut </span>headers = HeaderMap::new();
<a href=#1063 id=1063 data-nosnippet>1063</a> headers.insert(ACCEPT, HeaderValue::from_static(<span class="string">"application/json"</span>));
<a href=#1064 id=1064 data-nosnippet>1064</a> <span class="kw">let </span>resp = <span class="self">self
<a href=#1065 id=1065 data-nosnippet>1065</a> </span>.send_with_retry(Method::GET, url, <span class="prelude-val">Some</span>(headers), <span class="prelude-val">None</span>)
<a href=#1066 id=1066 data-nosnippet>1066</a> .<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#1067 id=1067 data-nosnippet>1067</a> <span class="kw">let </span>(status, resp_headers, body) = read_response(resp).<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#1068 id=1068 data-nosnippet>1068</a> <span class="kw">if </span>status != <span class="number">200 </span>{
<a href=#1069 id=1069 data-nosnippet>1069</a> <span class="kw">return </span><span class="prelude-val">Err</span>(from_response(status, <span class="kw-2">&</span>body, <span class="number">0</span>, <span class="bool-val">true</span>));
<a href=#1070 id=1070 data-nosnippet>1070</a> }
<a href=#1071 id=1071 data-nosnippet>1071</a> <span class="kw">let </span>ct = resp_headers
<a href=#1072 id=1072 data-nosnippet>1072</a> .get(CONTENT_TYPE)
<a href=#1073 id=1073 data-nosnippet>1073</a> .and_then(|v| v.to_str().ok())
<a href=#1074 id=1074 data-nosnippet>1074</a> .unwrap_or(<span class="string">""</span>);
<a href=#1075 id=1075 data-nosnippet>1075</a> <span class="kw">if </span>!ct.contains(<span class="string">"application/json"</span>) {
<a href=#1076 id=1076 data-nosnippet>1076</a> <span class="kw">return </span><span class="prelude-val">Err</span>(ScrapflyError::UnexpectedResponseFormat(<span class="macro">format!</span>(
<a href=#1077 id=1077 data-nosnippet>1077</a> <span class="string">"expected JSON, got Content-Type={}"</span>,
<a href=#1078 id=1078 data-nosnippet>1078</a> ct
<a href=#1079 id=1079 data-nosnippet>1079</a> )));
<a href=#1080 id=1080 data-nosnippet>1080</a> }
<a href=#1081 id=1081 data-nosnippet>1081</a> <span class="prelude-val">Ok</span>(serde_json::from_slice(<span class="kw-2">&</span>body)<span class="question-mark">?</span>)
<a href=#1082 id=1082 data-nosnippet>1082</a> }
<a href=#1083 id=1083 data-nosnippet>1083</a>
<a href=#1084 id=1084 data-nosnippet>1084</a> <span class="doccomment">/// Plain single-URL `GET /crawl/{uuid}/contents?plain=true`.
<a href=#1085 id=1085 data-nosnippet>1085</a> </span><span class="kw">pub async fn </span>crawl_contents_plain(
<a href=#1086 id=1086 data-nosnippet>1086</a> <span class="kw-2">&</span><span class="self">self</span>,
<a href=#1087 id=1087 data-nosnippet>1087</a> uuid: <span class="kw-2">&</span>str,
<a href=#1088 id=1088 data-nosnippet>1088</a> target_url: <span class="kw-2">&</span>str,
<a href=#1089 id=1089 data-nosnippet>1089</a> format: <span class="kw">crate</span>::enums::CrawlerContentFormat,
<a href=#1090 id=1090 data-nosnippet>1090</a> ) -> <span class="prelude-ty">Result</span><String, ScrapflyError> {
<a href=#1091 id=1091 data-nosnippet>1091</a> <span class="kw">if </span>uuid.is_empty() {
<a href=#1092 id=1092 data-nosnippet>1092</a> <span class="kw">return </span><span class="prelude-val">Err</span>(ScrapflyError::Config(<span class="string">"uuid cannot be empty"</span>.into()));
<a href=#1093 id=1093 data-nosnippet>1093</a> }
<a href=#1094 id=1094 data-nosnippet>1094</a> <span class="kw">if </span>target_url.is_empty() {
<a href=#1095 id=1095 data-nosnippet>1095</a> <span class="kw">return </span><span class="prelude-val">Err</span>(ScrapflyError::Config(
<a href=#1096 id=1096 data-nosnippet>1096</a> <span class="string">"plain mode requires a single url argument"</span>.into(),
<a href=#1097 id=1097 data-nosnippet>1097</a> ));
<a href=#1098 id=1098 data-nosnippet>1098</a> }
<a href=#1099 id=1099 data-nosnippet>1099</a> <span class="kw">let </span>pairs: Vec<(String, String)> = <span class="macro">vec!</span>[
<a href=#1100 id=1100 data-nosnippet>1100</a> (<span class="string">"formats"</span>.into(), format.as_str().into()),
<a href=#1101 id=1101 data-nosnippet>1101</a> (<span class="string">"url"</span>.into(), target_url.into()),
<a href=#1102 id=1102 data-nosnippet>1102</a> (<span class="string">"plain"</span>.into(), <span class="string">"true"</span>.into()),
<a href=#1103 id=1103 data-nosnippet>1103</a> ];
<a href=#1104 id=1104 data-nosnippet>1104</a> <span class="kw">let </span>url = <span class="self">self</span>.build_url(<span class="kw-2">&</span><span class="macro">format!</span>(<span class="string">"/crawl/{}/contents"</span>, uuid), <span class="kw-2">&</span>pairs)<span class="question-mark">?</span>;
<a href=#1105 id=1105 data-nosnippet>1105</a> <span class="kw">let </span><span class="kw-2">mut </span>headers = HeaderMap::new();
<a href=#1106 id=1106 data-nosnippet>1106</a> headers.insert(ACCEPT, HeaderValue::from_static(<span class="string">"*/*"</span>));
<a href=#1107 id=1107 data-nosnippet>1107</a> <span class="kw">let </span>resp = <span class="self">self
<a href=#1108 id=1108 data-nosnippet>1108</a> </span>.send_with_retry(Method::GET, url, <span class="prelude-val">Some</span>(headers), <span class="prelude-val">None</span>)
<a href=#1109 id=1109 data-nosnippet>1109</a> .<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#1110 id=1110 data-nosnippet>1110</a> <span class="kw">let </span>(status, _h, body) = read_response(resp).<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#1111 id=1111 data-nosnippet>1111</a> <span class="kw">if </span>status != <span class="number">200 </span>{
<a href=#1112 id=1112 data-nosnippet>1112</a> <span class="kw">return </span><span class="prelude-val">Err</span>(from_response(status, <span class="kw-2">&</span>body, <span class="number">0</span>, <span class="bool-val">true</span>));
<a href=#1113 id=1113 data-nosnippet>1113</a> }
<a href=#1114 id=1114 data-nosnippet>1114</a> <span class="prelude-val">Ok</span>(String::from_utf8_lossy(<span class="kw-2">&</span>body).into_owned())
<a href=#1115 id=1115 data-nosnippet>1115</a> }
<a href=#1116 id=1116 data-nosnippet>1116</a>
<a href=#1117 id=1117 data-nosnippet>1117</a> <span class="doccomment">/// Bulk-batch `POST /crawl/{uuid}/contents/batch`.
<a href=#1118 id=1118 data-nosnippet>1118</a> /// Returns `url → format → content` (multipart/related response).
<a href=#1119 id=1119 data-nosnippet>1119</a> </span><span class="kw">pub async fn </span>crawl_contents_batch(
<a href=#1120 id=1120 data-nosnippet>1120</a> <span class="kw-2">&</span><span class="self">self</span>,
<a href=#1121 id=1121 data-nosnippet>1121</a> uuid: <span class="kw-2">&</span>str,
<a href=#1122 id=1122 data-nosnippet>1122</a> urls: <span class="kw-2">&</span>[String],
<a href=#1123 id=1123 data-nosnippet>1123</a> formats: <span class="kw-2">&</span>[<span class="kw">crate</span>::enums::CrawlerContentFormat],
<a href=#1124 id=1124 data-nosnippet>1124</a> ) -> <span class="prelude-ty">Result</span><
<a href=#1125 id=1125 data-nosnippet>1125</a> std::collections::BTreeMap<String, std::collections::BTreeMap<String, String>>,
<a href=#1126 id=1126 data-nosnippet>1126</a> ScrapflyError,
<a href=#1127 id=1127 data-nosnippet>1127</a> > {
<a href=#1128 id=1128 data-nosnippet>1128</a> <span class="kw">if </span>uuid.is_empty() {
<a href=#1129 id=1129 data-nosnippet>1129</a> <span class="kw">return </span><span class="prelude-val">Err</span>(ScrapflyError::Config(<span class="string">"uuid cannot be empty"</span>.into()));
<a href=#1130 id=1130 data-nosnippet>1130</a> }
<a href=#1131 id=1131 data-nosnippet>1131</a> <span class="kw">if </span>urls.is_empty() {
<a href=#1132 id=1132 data-nosnippet>1132</a> <span class="kw">return </span><span class="prelude-val">Err</span>(ScrapflyError::Config(<span class="string">"at least one URL is required"</span>.into()));
<a href=#1133 id=1133 data-nosnippet>1133</a> }
<a href=#1134 id=1134 data-nosnippet>1134</a> <span class="kw">if </span>urls.len() > <span class="number">100 </span>{
<a href=#1135 id=1135 data-nosnippet>1135</a> <span class="kw">return </span><span class="prelude-val">Err</span>(ScrapflyError::Config(<span class="macro">format!</span>(
<a href=#1136 id=1136 data-nosnippet>1136</a> <span class="string">"batch is limited to 100 URLs per request, got {}"</span>,
<a href=#1137 id=1137 data-nosnippet>1137</a> urls.len()
<a href=#1138 id=1138 data-nosnippet>1138</a> )));
<a href=#1139 id=1139 data-nosnippet>1139</a> }
<a href=#1140 id=1140 data-nosnippet>1140</a> <span class="kw">if </span>formats.is_empty() {
<a href=#1141 id=1141 data-nosnippet>1141</a> <span class="kw">return </span><span class="prelude-val">Err</span>(ScrapflyError::Config(
<a href=#1142 id=1142 data-nosnippet>1142</a> <span class="string">"at least one format is required"</span>.into(),
<a href=#1143 id=1143 data-nosnippet>1143</a> ));
<a href=#1144 id=1144 data-nosnippet>1144</a> }
<a href=#1145 id=1145 data-nosnippet>1145</a> <span class="kw">let </span>format_strs: Vec<<span class="kw-2">&</span><span class="lifetime">'static </span>str> = formats.iter().map(|f| f.as_str()).collect();
<a href=#1146 id=1146 data-nosnippet>1146</a> <span class="kw">let </span>pairs: Vec<(String, String)> = <span class="macro">vec!</span>[(<span class="string">"formats"</span>.into(), format_strs.join(<span class="string">","</span>))];
<a href=#1147 id=1147 data-nosnippet>1147</a> <span class="kw">let </span>url = <span class="self">self</span>.build_url(<span class="kw-2">&</span><span class="macro">format!</span>(<span class="string">"/crawl/{}/contents/batch"</span>, uuid), <span class="kw-2">&</span>pairs)<span class="question-mark">?</span>;
<a href=#1148 id=1148 data-nosnippet>1148</a> <span class="kw">let </span>body = urls.join(<span class="string">"\n"</span>).into_bytes();
<a href=#1149 id=1149 data-nosnippet>1149</a> <span class="kw">let </span><span class="kw-2">mut </span>headers = HeaderMap::new();
<a href=#1150 id=1150 data-nosnippet>1150</a> headers.insert(CONTENT_TYPE, HeaderValue::from_static(<span class="string">"text/plain"</span>));
<a href=#1151 id=1151 data-nosnippet>1151</a> headers.insert(
<a href=#1152 id=1152 data-nosnippet>1152</a> ACCEPT,
<a href=#1153 id=1153 data-nosnippet>1153</a> HeaderValue::from_static(<span class="string">"multipart/related, application/json"</span>),
<a href=#1154 id=1154 data-nosnippet>1154</a> );
<a href=#1155 id=1155 data-nosnippet>1155</a> <span class="kw">let </span>resp = <span class="self">self
<a href=#1156 id=1156 data-nosnippet>1156</a> </span>.send_with_retry(Method::POST, url, <span class="prelude-val">Some</span>(headers), <span class="prelude-val">Some</span>(body))
<a href=#1157 id=1157 data-nosnippet>1157</a> .<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#1158 id=1158 data-nosnippet>1158</a> <span class="kw">let </span>(status, resp_headers, body_bytes) = read_response(resp).<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#1159 id=1159 data-nosnippet>1159</a> <span class="kw">if </span>status != <span class="number">200 </span>{
<a href=#1160 id=1160 data-nosnippet>1160</a> <span class="kw">return </span><span class="prelude-val">Err</span>(from_response(status, <span class="kw-2">&</span>body_bytes, <span class="number">0</span>, <span class="bool-val">true</span>));
<a href=#1161 id=1161 data-nosnippet>1161</a> }
<a href=#1162 id=1162 data-nosnippet>1162</a> <span class="kw">let </span>ct = resp_headers
<a href=#1163 id=1163 data-nosnippet>1163</a> .get(CONTENT_TYPE)
<a href=#1164 id=1164 data-nosnippet>1164</a> .and_then(|v| v.to_str().ok())
<a href=#1165 id=1165 data-nosnippet>1165</a> .unwrap_or(<span class="string">""</span>);
<a href=#1166 id=1166 data-nosnippet>1166</a> <span class="kw">if </span>ct.contains(<span class="string">"application/json"</span>) {
<a href=#1167 id=1167 data-nosnippet>1167</a> <span class="kw">return </span><span class="prelude-val">Err</span>(ScrapflyError::UnexpectedResponseFormat(
<a href=#1168 id=1168 data-nosnippet>1168</a> <span class="string">"CrawlContentsBatch expected multipart/related, got JSON"</span>.into(),
<a href=#1169 id=1169 data-nosnippet>1169</a> ));
<a href=#1170 id=1170 data-nosnippet>1170</a> }
<a href=#1171 id=1171 data-nosnippet>1171</a> parse_multipart_related(
<a href=#1172 id=1172 data-nosnippet>1172</a> std::str::from_utf8(<span class="kw-2">&</span>body_bytes).unwrap_or(<span class="string">""</span>),
<a href=#1173 id=1173 data-nosnippet>1173</a> ct,
<a href=#1174 id=1174 data-nosnippet>1174</a> <span class="kw-2">&</span>format_strs,
<a href=#1175 id=1175 data-nosnippet>1175</a> )
<a href=#1176 id=1176 data-nosnippet>1176</a> }
<a href=#1177 id=1177 data-nosnippet>1177</a>
<a href=#1178 id=1178 data-nosnippet>1178</a> <span class="doccomment">/// Cancel a crawler job.
<a href=#1179 id=1179 data-nosnippet>1179</a> </span><span class="kw">pub async fn </span>crawl_cancel(<span class="kw-2">&</span><span class="self">self</span>, uuid: <span class="kw-2">&</span>str) -> <span class="prelude-ty">Result</span><(), ScrapflyError> {
<a href=#1180 id=1180 data-nosnippet>1180</a> <span class="kw">if </span>uuid.is_empty() {
<a href=#1181 id=1181 data-nosnippet>1181</a> <span class="kw">return </span><span class="prelude-val">Err</span>(ScrapflyError::Config(<span class="string">"uuid cannot be empty"</span>.into()));
<a href=#1182 id=1182 data-nosnippet>1182</a> }
<a href=#1183 id=1183 data-nosnippet>1183</a> <span class="kw">let </span>url = <span class="self">self</span>.build_url(<span class="kw-2">&</span><span class="macro">format!</span>(<span class="string">"/crawl/{}/cancel"</span>, uuid), <span class="kw-2">&</span>[])<span class="question-mark">?</span>;
<a href=#1184 id=1184 data-nosnippet>1184</a> <span class="kw">let </span><span class="kw-2">mut </span>headers = HeaderMap::new();
<a href=#1185 id=1185 data-nosnippet>1185</a> headers.insert(ACCEPT, HeaderValue::from_static(<span class="string">"application/json"</span>));
<a href=#1186 id=1186 data-nosnippet>1186</a> <span class="kw">let </span>resp = <span class="self">self
<a href=#1187 id=1187 data-nosnippet>1187</a> </span>.send_with_retry(Method::POST, url, <span class="prelude-val">Some</span>(headers), <span class="prelude-val">None</span>)
<a href=#1188 id=1188 data-nosnippet>1188</a> .<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#1189 id=1189 data-nosnippet>1189</a> <span class="kw">let </span>(status, _h, body) = read_response(resp).<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#1190 id=1190 data-nosnippet>1190</a> <span class="kw">if </span>status != <span class="number">200 </span>&& status != <span class="number">202 </span>{
<a href=#1191 id=1191 data-nosnippet>1191</a> <span class="kw">return </span><span class="prelude-val">Err</span>(from_response(status, <span class="kw-2">&</span>body, <span class="number">0</span>, <span class="bool-val">true</span>));
<a href=#1192 id=1192 data-nosnippet>1192</a> }
<a href=#1193 id=1193 data-nosnippet>1193</a> <span class="prelude-val">Ok</span>(())
<a href=#1194 id=1194 data-nosnippet>1194</a> }
<a href=#1195 id=1195 data-nosnippet>1195</a>
<a href=#1196 id=1196 data-nosnippet>1196</a> <span class="doccomment">/// Download a crawler artifact (WARC or HAR).
<a href=#1197 id=1197 data-nosnippet>1197</a> </span><span class="kw">pub async fn </span>crawl_artifact(
<a href=#1198 id=1198 data-nosnippet>1198</a> <span class="kw-2">&</span><span class="self">self</span>,
<a href=#1199 id=1199 data-nosnippet>1199</a> uuid: <span class="kw-2">&</span>str,
<a href=#1200 id=1200 data-nosnippet>1200</a> artifact_type: CrawlerArtifactType,
<a href=#1201 id=1201 data-nosnippet>1201</a> ) -> <span class="prelude-ty">Result</span><CrawlerArtifact, ScrapflyError> {
<a href=#1202 id=1202 data-nosnippet>1202</a> <span class="kw">if </span>uuid.is_empty() {
<a href=#1203 id=1203 data-nosnippet>1203</a> <span class="kw">return </span><span class="prelude-val">Err</span>(ScrapflyError::Config(<span class="string">"uuid cannot be empty"</span>.into()));
<a href=#1204 id=1204 data-nosnippet>1204</a> }
<a href=#1205 id=1205 data-nosnippet>1205</a> <span class="kw">let </span>pairs: Vec<(String, String)> = <span class="macro">vec!</span>[(<span class="string">"type"</span>.into(), artifact_type.as_str().into())];
<a href=#1206 id=1206 data-nosnippet>1206</a> <span class="kw">let </span>url = <span class="self">self</span>.build_url(<span class="kw-2">&</span><span class="macro">format!</span>(<span class="string">"/crawl/{}/artifact"</span>, uuid), <span class="kw-2">&</span>pairs)<span class="question-mark">?</span>;
<a href=#1207 id=1207 data-nosnippet>1207</a> <span class="kw">let </span><span class="kw-2">mut </span>headers = HeaderMap::new();
<a href=#1208 id=1208 data-nosnippet>1208</a> <span class="comment">// HAR is plain JSON — asking for `application/gzip` makes the server
<a href=#1209 id=1209 data-nosnippet>1209</a> // gzip-wrap it, and reqwest can't auto-decode it without a matching
<a href=#1210 id=1210 data-nosnippet>1210</a> // `Content-Encoding` header. Match `sdk/go/crawler.go::CrawlArtifact`
<a href=#1211 id=1211 data-nosnippet>1211</a> // which sends different Accept per artifact type.
<a href=#1212 id=1212 data-nosnippet>1212</a> </span><span class="kw">let </span>accept = <span class="kw">match </span>artifact_type {
<a href=#1213 id=1213 data-nosnippet>1213</a> CrawlerArtifactType::Har => <span class="string">"application/json, application/octet-stream"</span>,
<a href=#1214 id=1214 data-nosnippet>1214</a> CrawlerArtifactType::Warc => {
<a href=#1215 id=1215 data-nosnippet>1215</a> <span class="string">"application/gzip, application/octet-stream, application/json"
<a href=#1216 id=1216 data-nosnippet>1216</a> </span>}
<a href=#1217 id=1217 data-nosnippet>1217</a> };
<a href=#1218 id=1218 data-nosnippet>1218</a> headers.insert(ACCEPT, HeaderValue::from_static(accept));
<a href=#1219 id=1219 data-nosnippet>1219</a> <span class="kw">let </span>resp = <span class="self">self
<a href=#1220 id=1220 data-nosnippet>1220</a> </span>.send_with_retry(Method::GET, url, <span class="prelude-val">Some</span>(headers), <span class="prelude-val">None</span>)
<a href=#1221 id=1221 data-nosnippet>1221</a> .<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#1222 id=1222 data-nosnippet>1222</a> <span class="kw">let </span>(status, _h, body) = read_response(resp).<span class="kw">await</span><span class="question-mark">?</span>;
<a href=#1223 id=1223 data-nosnippet>1223</a> <span class="kw">if </span>status != <span class="number">200 </span>{
<a href=#1224 id=1224 data-nosnippet>1224</a> <span class="kw">return </span><span class="prelude-val">Err</span>(from_response(status, <span class="kw-2">&</span>body, <span class="number">0</span>, <span class="bool-val">true</span>));
<a href=#1225 id=1225 data-nosnippet>1225</a> }
<a href=#1226 id=1226 data-nosnippet>1226</a> <span class="prelude-val">Ok</span>(CrawlerArtifact {
<a href=#1227 id=1227 data-nosnippet>1227</a> artifact_type,
<a href=#1228 id=1228 data-nosnippet>1228</a> data: body,
<a href=#1229 id=1229 data-nosnippet>1229</a> })
<a href=#1230 id=1230 data-nosnippet>1230</a> }
<a href=#1231 id=1231 data-nosnippet>1231</a>
<a href=#1232 id=1232 data-nosnippet>1232</a> <span class="comment">// ==============================================================================
<a href=#1233 id=1233 data-nosnippet>1233</a> // Cloud browser methods (implementations in cloud_browser.rs)
<a href=#1234 id=1234 data-nosnippet>1234</a> // ==============================================================================
<a href=#1235 id=1235 data-nosnippet>1235</a>
<a href=#1236 id=1236 data-nosnippet>1236</a> </span><span class="doccomment">/// Fire a request through the retry loop.
<a href=#1237 id=1237 data-nosnippet>1237</a> </span><span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">async fn </span>send_with_retry(
<a href=#1238 id=1238 data-nosnippet>1238</a> <span class="kw-2">&</span><span class="self">self</span>,
<a href=#1239 id=1239 data-nosnippet>1239</a> method: Method,
<a href=#1240 id=1240 data-nosnippet>1240</a> url: Url,
<a href=#1241 id=1241 data-nosnippet>1241</a> headers: <span class="prelude-ty">Option</span><HeaderMap>,
<a href=#1242 id=1242 data-nosnippet>1242</a> body: <span class="prelude-ty">Option</span><Vec<u8>>,
<a href=#1243 id=1243 data-nosnippet>1243</a> ) -> <span class="prelude-ty">Result</span><Response, ScrapflyError> {
<a href=#1244 id=1244 data-nosnippet>1244</a> <span class="kw">let </span><span class="kw-2">mut </span>last_err: <span class="prelude-ty">Option</span><ScrapflyError> = <span class="prelude-val">None</span>;
<a href=#1245 id=1245 data-nosnippet>1245</a> <span class="kw">for </span>attempt <span class="kw">in </span><span class="number">0</span>..DEFAULT_RETRIES {
<a href=#1246 id=1246 data-nosnippet>1246</a> <span class="kw">let </span><span class="kw-2">mut </span>req = <span class="self">self</span>.http.request(method.clone(), url.clone());
<a href=#1247 id=1247 data-nosnippet>1247</a> <span class="kw">let </span><span class="kw-2">mut </span>hmap = headers.clone().unwrap_or_default();
<a href=#1248 id=1248 data-nosnippet>1248</a> <span class="kw">if </span>!hmap.contains_key(USER_AGENT) {
<a href=#1249 id=1249 data-nosnippet>1249</a> hmap.insert(USER_AGENT, HeaderValue::from_static(SDK_USER_AGENT));
<a href=#1250 id=1250 data-nosnippet>1250</a> }
<a href=#1251 id=1251 data-nosnippet>1251</a> <span class="kw">if let </span><span class="prelude-val">Some</span>(cb) = <span class="kw-2">&</span><span class="self">self</span>.on_request {
<a href=#1252 id=1252 data-nosnippet>1252</a> cb(<span class="kw-2">&</span>method, <span class="kw-2">&</span>url, <span class="kw-2">&</span>hmap);
<a href=#1253 id=1253 data-nosnippet>1253</a> }
<a href=#1254 id=1254 data-nosnippet>1254</a> req = req.headers(hmap);
<a href=#1255 id=1255 data-nosnippet>1255</a> <span class="kw">if let </span><span class="prelude-val">Some</span>(b) = <span class="kw-2">&</span>body {
<a href=#1256 id=1256 data-nosnippet>1256</a> req = req.body(b.clone());
<a href=#1257 id=1257 data-nosnippet>1257</a> }
<a href=#1258 id=1258 data-nosnippet>1258</a> <span class="kw">match </span>req.send().<span class="kw">await </span>{
<a href=#1259 id=1259 data-nosnippet>1259</a> <span class="prelude-val">Ok</span>(resp) => {
<a href=#1260 id=1260 data-nosnippet>1260</a> <span class="kw">let </span>status = resp.status().as_u16();
<a href=#1261 id=1261 data-nosnippet>1261</a> <span class="kw">if </span>(<span class="number">500</span>..<span class="number">600</span>).contains(<span class="kw-2">&</span>status) && attempt + <span class="number">1 </span>< DEFAULT_RETRIES {
<a href=#1262 id=1262 data-nosnippet>1262</a> last_err = <span class="prelude-val">Some</span>(ScrapflyError::ApiServer(<span class="kw">crate</span>::error::ApiError {
<a href=#1263 id=1263 data-nosnippet>1263</a> message: <span class="string">"server error"</span>.into(),
<a href=#1264 id=1264 data-nosnippet>1264</a> http_status: status,
<a href=#1265 id=1265 data-nosnippet>1265</a> ..Default::default()
<a href=#1266 id=1266 data-nosnippet>1266</a> }));
<a href=#1267 id=1267 data-nosnippet>1267</a> tokio::time::sleep(DEFAULT_RETRY_DELAY).<span class="kw">await</span>;
<a href=#1268 id=1268 data-nosnippet>1268</a> <span class="kw">continue</span>;
<a href=#1269 id=1269 data-nosnippet>1269</a> }
<a href=#1270 id=1270 data-nosnippet>1270</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(resp);
<a href=#1271 id=1271 data-nosnippet>1271</a> }
<a href=#1272 id=1272 data-nosnippet>1272</a> <span class="prelude-val">Err</span>(e) => {
<a href=#1273 id=1273 data-nosnippet>1273</a> last_err = <span class="prelude-val">Some</span>(ScrapflyError::Transport(e));
<a href=#1274 id=1274 data-nosnippet>1274</a> <span class="kw">if </span>attempt + <span class="number">1 </span>< DEFAULT_RETRIES {
<a href=#1275 id=1275 data-nosnippet>1275</a> tokio::time::sleep(DEFAULT_RETRY_DELAY).<span class="kw">await</span>;
<a href=#1276 id=1276 data-nosnippet>1276</a> <span class="kw">continue</span>;
<a href=#1277 id=1277 data-nosnippet>1277</a> }
<a href=#1278 id=1278 data-nosnippet>1278</a> }
<a href=#1279 id=1279 data-nosnippet>1279</a> }
<a href=#1280 id=1280 data-nosnippet>1280</a> }
<a href=#1281 id=1281 data-nosnippet>1281</a> <span class="prelude-val">Err</span>(last_err.unwrap_or_else(|| ScrapflyError::Config(<span class="string">"retry loop exhausted"</span>.into())))
<a href=#1282 id=1282 data-nosnippet>1282</a> }
<a href=#1283 id=1283 data-nosnippet>1283</a>
<a href=#1284 id=1284 data-nosnippet>1284</a> <span class="doccomment">/// Single-shot send, no retry (for `verify_api_key`/`account` style calls).
<a href=#1285 id=1285 data-nosnippet>1285</a> </span><span class="kw">async fn </span>send_simple(
<a href=#1286 id=1286 data-nosnippet>1286</a> <span class="kw-2">&</span><span class="self">self</span>,
<a href=#1287 id=1287 data-nosnippet>1287</a> method: Method,
<a href=#1288 id=1288 data-nosnippet>1288</a> url: Url,
<a href=#1289 id=1289 data-nosnippet>1289</a> headers: <span class="prelude-ty">Option</span><HeaderMap>,
<a href=#1290 id=1290 data-nosnippet>1290</a> body: <span class="prelude-ty">Option</span><Vec<u8>>,
<a href=#1291 id=1291 data-nosnippet>1291</a> ) -> <span class="prelude-ty">Result</span><Response, ScrapflyError> {
<a href=#1292 id=1292 data-nosnippet>1292</a> <span class="kw">let </span><span class="kw-2">mut </span>req = <span class="self">self</span>.http.request(method.clone(), url.clone());
<a href=#1293 id=1293 data-nosnippet>1293</a> <span class="kw">let </span><span class="kw-2">mut </span>hmap = headers.unwrap_or_default();
<a href=#1294 id=1294 data-nosnippet>1294</a> <span class="kw">if </span>!hmap.contains_key(USER_AGENT) {
<a href=#1295 id=1295 data-nosnippet>1295</a> hmap.insert(USER_AGENT, HeaderValue::from_static(SDK_USER_AGENT));
<a href=#1296 id=1296 data-nosnippet>1296</a> }
<a href=#1297 id=1297 data-nosnippet>1297</a> <span class="kw">if let </span><span class="prelude-val">Some</span>(cb) = <span class="kw-2">&</span><span class="self">self</span>.on_request {
<a href=#1298 id=1298 data-nosnippet>1298</a> cb(<span class="kw-2">&</span>method, <span class="kw-2">&</span>url, <span class="kw-2">&</span>hmap);
<a href=#1299 id=1299 data-nosnippet>1299</a> }
<a href=#1300 id=1300 data-nosnippet>1300</a> req = req.headers(hmap);
<a href=#1301 id=1301 data-nosnippet>1301</a> <span class="kw">if let </span><span class="prelude-val">Some</span>(b) = body {
<a href=#1302 id=1302 data-nosnippet>1302</a> req = req.body(b);
<a href=#1303 id=1303 data-nosnippet>1303</a> }
<a href=#1304 id=1304 data-nosnippet>1304</a> req.send().<span class="kw">await</span>.map_err(ScrapflyError::Transport)
<a href=#1305 id=1305 data-nosnippet>1305</a> }
<a href=#1306 id=1306 data-nosnippet>1306</a>}
<a href=#1307 id=1307 data-nosnippet>1307</a>
<a href=#1308 id=1308 data-nosnippet>1308</a><span class="doccomment">/// Drain a response into (status, headers, body bytes) and propagate
<a href=#1309 id=1309 data-nosnippet>1309</a>/// `Retry-After` into the retry-ms field when present.
<a href=#1310 id=1310 data-nosnippet>1310</a></span><span class="kw">async fn </span>read_response(resp: Response) -> <span class="prelude-ty">Result</span><(u16, HeaderMap, bytes::Bytes), ScrapflyError> {
<a href=#1311 id=1311 data-nosnippet>1311</a> <span class="kw">let </span>status = resp.status().as_u16();
<a href=#1312 id=1312 data-nosnippet>1312</a> <span class="kw">let </span>headers = resp.headers().clone();
<a href=#1313 id=1313 data-nosnippet>1313</a> <span class="kw">let </span>body = resp.bytes().<span class="kw">await</span>.map_err(ScrapflyError::Transport)<span class="question-mark">?</span>;
<a href=#1314 id=1314 data-nosnippet>1314</a> <span class="kw">let _ </span>= parse_retry_after(headers.get(<span class="string">"retry-after"</span>).and_then(|v| v.to_str().ok()));
<a href=#1315 id=1315 data-nosnippet>1315</a> <span class="prelude-val">Ok</span>((status, headers, body))
<a href=#1316 id=1316 data-nosnippet>1316</a>}
<a href=#1317 id=1317 data-nosnippet>1317</a>
<a href=#1318 id=1318 data-nosnippet>1318</a><span class="doccomment">/// Minimal RFC 2387 multipart/related parser — ported from
<a href=#1319 id=1319 data-nosnippet>1319</a>/// `sdk/go/crawler.go::parseMultipartRelated`.
<a href=#1320 id=1320 data-nosnippet>1320</a></span><span class="kw">fn </span>parse_multipart_related(
<a href=#1321 id=1321 data-nosnippet>1321</a> body: <span class="kw-2">&</span>str,
<a href=#1322 id=1322 data-nosnippet>1322</a> content_type: <span class="kw-2">&</span>str,
<a href=#1323 id=1323 data-nosnippet>1323</a> formats: <span class="kw-2">&</span>[<span class="kw-2">&</span>str],
<a href=#1324 id=1324 data-nosnippet>1324</a>) -> <span class="prelude-ty">Result</span><
<a href=#1325 id=1325 data-nosnippet>1325</a> std::collections::BTreeMap<String, std::collections::BTreeMap<String, String>>,
<a href=#1326 id=1326 data-nosnippet>1326</a> ScrapflyError,
<a href=#1327 id=1327 data-nosnippet>1327</a>> {
<a href=#1328 id=1328 data-nosnippet>1328</a> <span class="kw">let </span><span class="kw-2">mut </span>boundary = String::new();
<a href=#1329 id=1329 data-nosnippet>1329</a> <span class="kw">for </span>part <span class="kw">in </span>content_type.split(<span class="string">';'</span>) {
<a href=#1330 id=1330 data-nosnippet>1330</a> <span class="kw">let </span>p = part.trim();
<a href=#1331 id=1331 data-nosnippet>1331</a> <span class="kw">if let </span><span class="prelude-val">Some</span>(stripped) = p.strip_prefix(<span class="string">"boundary="</span>) {
<a href=#1332 id=1332 data-nosnippet>1332</a> boundary = stripped.trim_matches(<span class="string">'"'</span>).to_string();
<a href=#1333 id=1333 data-nosnippet>1333</a> <span class="kw">break</span>;
<a href=#1334 id=1334 data-nosnippet>1334</a> }
<a href=#1335 id=1335 data-nosnippet>1335</a> }
<a href=#1336 id=1336 data-nosnippet>1336</a> <span class="kw">if </span>boundary.is_empty() {
<a href=#1337 id=1337 data-nosnippet>1337</a> <span class="kw">return </span><span class="prelude-val">Err</span>(ScrapflyError::UnexpectedResponseFormat(<span class="macro">format!</span>(
<a href=#1338 id=1338 data-nosnippet>1338</a> <span class="string">"multipart response has no boundary in Content-Type: {}"</span>,
<a href=#1339 id=1339 data-nosnippet>1339</a> content_type
<a href=#1340 id=1340 data-nosnippet>1340</a> )));
<a href=#1341 id=1341 data-nosnippet>1341</a> }
<a href=#1342 id=1342 data-nosnippet>1342</a> <span class="kw">let </span>delimiter = <span class="macro">format!</span>(<span class="string">"--{}"</span>, boundary);
<a href=#1343 id=1343 data-nosnippet>1343</a> <span class="kw">let </span><span class="kw-2">mut </span>result: std::collections::BTreeMap<String, std::collections::BTreeMap<String, String>> =
<a href=#1344 id=1344 data-nosnippet>1344</a> std::collections::BTreeMap::new();
<a href=#1345 id=1345 data-nosnippet>1345</a> <span class="kw">let </span>segments: Vec<<span class="kw-2">&</span>str> = body.split(<span class="kw-2">&</span>delimiter <span class="kw">as </span><span class="kw-2">&</span>str).collect();
<a href=#1346 id=1346 data-nosnippet>1346</a> <span class="kw">for </span>segment <span class="kw">in </span>segments.iter().skip(<span class="number">1</span>) {
<a href=#1347 id=1347 data-nosnippet>1347</a> <span class="kw">let </span><span class="kw-2">mut </span>seg = <span class="kw-2">*</span>segment;
<a href=#1348 id=1348 data-nosnippet>1348</a> seg = seg.trim_start_matches(<span class="string">"\r\n"</span>).trim_start_matches(<span class="string">'\n'</span>);
<a href=#1349 id=1349 data-nosnippet>1349</a> <span class="kw">if </span>seg.starts_with(<span class="string">"--"</span>) {
<a href=#1350 id=1350 data-nosnippet>1350</a> <span class="kw">break</span>;
<a href=#1351 id=1351 data-nosnippet>1351</a> }
<a href=#1352 id=1352 data-nosnippet>1352</a> seg = seg.trim_end_matches(<span class="string">"\r\n"</span>).trim_end_matches(<span class="string">'\n'</span>);
<a href=#1353 id=1353 data-nosnippet>1353</a> <span class="kw">let </span>(headers_raw, part_body) = <span class="kw">if let </span><span class="prelude-val">Some</span>(idx) = seg.find(<span class="string">"\r\n\r\n"</span>) {
<a href=#1354 id=1354 data-nosnippet>1354</a> (<span class="kw-2">&</span>seg[..idx], <span class="kw-2">&</span>seg[idx + <span class="number">4</span>..])
<a href=#1355 id=1355 data-nosnippet>1355</a> } <span class="kw">else if let </span><span class="prelude-val">Some</span>(idx) = seg.find(<span class="string">"\n\n"</span>) {
<a href=#1356 id=1356 data-nosnippet>1356</a> (<span class="kw-2">&</span>seg[..idx], <span class="kw-2">&</span>seg[idx + <span class="number">2</span>..])
<a href=#1357 id=1357 data-nosnippet>1357</a> } <span class="kw">else </span>{
<a href=#1358 id=1358 data-nosnippet>1358</a> <span class="kw">continue</span>;
<a href=#1359 id=1359 data-nosnippet>1359</a> };
<a href=#1360 id=1360 data-nosnippet>1360</a> <span class="kw">let </span><span class="kw-2">mut </span>part_url = String::new();
<a href=#1361 id=1361 data-nosnippet>1361</a> <span class="kw">let </span><span class="kw-2">mut </span>part_format = String::new();
<a href=#1362 id=1362 data-nosnippet>1362</a> <span class="kw">for </span>line <span class="kw">in </span>headers_raw.split(<span class="string">'\n'</span>) {
<a href=#1363 id=1363 data-nosnippet>1363</a> <span class="kw">let </span>line = line.trim_end_matches(<span class="string">'\r'</span>);
<a href=#1364 id=1364 data-nosnippet>1364</a> <span class="kw">if let </span><span class="prelude-val">Some</span>(colon) = line.find(<span class="string">':'</span>) {
<a href=#1365 id=1365 data-nosnippet>1365</a> <span class="kw">let </span>name = line[..colon].trim().to_ascii_lowercase();
<a href=#1366 id=1366 data-nosnippet>1366</a> <span class="kw">let </span>value = line[colon + <span class="number">1</span>..].trim().to_string();
<a href=#1367 id=1367 data-nosnippet>1367</a> <span class="kw">match </span>name.as_str() {
<a href=#1368 id=1368 data-nosnippet>1368</a> <span class="string">"content-location" </span>=> part_url = value,
<a href=#1369 id=1369 data-nosnippet>1369</a> <span class="string">"content-type" </span>=> part_format = infer_format_from_content_type(<span class="kw-2">&</span>value),
<a href=#1370 id=1370 data-nosnippet>1370</a> <span class="kw">_ </span>=> {}
<a href=#1371 id=1371 data-nosnippet>1371</a> }
<a href=#1372 id=1372 data-nosnippet>1372</a> }
<a href=#1373 id=1373 data-nosnippet>1373</a> }
<a href=#1374 id=1374 data-nosnippet>1374</a> <span class="kw">if </span>part_url.is_empty() {
<a href=#1375 id=1375 data-nosnippet>1375</a> <span class="kw">continue</span>;
<a href=#1376 id=1376 data-nosnippet>1376</a> }
<a href=#1377 id=1377 data-nosnippet>1377</a> <span class="kw">if </span>part_format.is_empty() {
<a href=#1378 id=1378 data-nosnippet>1378</a> part_format = formats.first().copied().unwrap_or(<span class="string">"html"</span>).to_string();
<a href=#1379 id=1379 data-nosnippet>1379</a> }
<a href=#1380 id=1380 data-nosnippet>1380</a> result
<a href=#1381 id=1381 data-nosnippet>1381</a> .entry(part_url)
<a href=#1382 id=1382 data-nosnippet>1382</a> .or_default()
<a href=#1383 id=1383 data-nosnippet>1383</a> .insert(part_format, part_body.to_string());
<a href=#1384 id=1384 data-nosnippet>1384</a> }
<a href=#1385 id=1385 data-nosnippet>1385</a> <span class="prelude-val">Ok</span>(result)
<a href=#1386 id=1386 data-nosnippet>1386</a>}
<a href=#1387 id=1387 data-nosnippet>1387</a>
<a href=#1388 id=1388 data-nosnippet>1388</a><span class="kw">fn </span>infer_format_from_content_type(ct: <span class="kw-2">&</span>str) -> String {
<a href=#1389 id=1389 data-nosnippet>1389</a> <span class="kw">let </span>lc = ct
<a href=#1390 id=1390 data-nosnippet>1390</a> .split(<span class="string">';'</span>)
<a href=#1391 id=1391 data-nosnippet>1391</a> .next()
<a href=#1392 id=1392 data-nosnippet>1392</a> .unwrap_or(<span class="string">""</span>)
<a href=#1393 id=1393 data-nosnippet>1393</a> .trim()
<a href=#1394 id=1394 data-nosnippet>1394</a> .to_ascii_lowercase();
<a href=#1395 id=1395 data-nosnippet>1395</a> <span class="kw">match </span>lc.as_str() {
<a href=#1396 id=1396 data-nosnippet>1396</a> <span class="string">"text/html" </span>=> <span class="string">"html"</span>.into(),
<a href=#1397 id=1397 data-nosnippet>1397</a> <span class="string">"text/markdown" </span>=> <span class="string">"markdown"</span>.into(),
<a href=#1398 id=1398 data-nosnippet>1398</a> <span class="string">"text/plain" </span>=> <span class="string">"text"</span>.into(),
<a href=#1399 id=1399 data-nosnippet>1399</a> <span class="string">"application/json" </span>=> <span class="string">"json"</span>.into(),
<a href=#1400 id=1400 data-nosnippet>1400</a> <span class="kw">_ </span>=> String::new(),
<a href=#1401 id=1401 data-nosnippet>1401</a> }
<a href=#1402 id=1402 data-nosnippet>1402</a>}
</code></pre></div></section></main></body></html>