From 30cb6ab69ca8f0a9bd036c1ae2523b3712d9e1eb Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Tom=C3=A1=C5=A1=20Ml=C3=A1dek?= <t@mldk.cz>
Date: Sat, 27 May 2023 23:05:22 +0200
Subject: [PATCH] feat: web extractor adds LBLs

---
 cli/src/extractors/web.rs | 25 ++++++++++++++++++++++---
 1 file changed, 22 insertions(+), 3 deletions(-)

diff --git a/cli/src/extractors/web.rs b/cli/src/extractors/web.rs
index 84dbabf..2baabb1 100644
--- a/cli/src/extractors/web.rs
+++ b/cli/src/extractors/web.rs
@@ -4,6 +4,7 @@ use super::Extractor;
 use crate::common::REQWEST_CLIENT;
 use anyhow::anyhow;
 use anyhow::Result;
+use upend::database::constants::LABEL_ATTR;
 use upend::{
     addressing::Address,
     database::{entry::Entry, stores::UpStore, UpEndConnection},
@@ -33,9 +34,16 @@ impl Extractor for WebExtractor {
                 let _ = job_handle.update_progress(50.0);
 
                 let mut entries = vec![
-                    html.title.map(|html_title| Entry {
+                    html.title.as_ref().map(|html_title| Entry {
                         entity: address.clone(),
                         attribute: "HTML_TITLE".to_string(),
+                        value: html_title.clone().into(),
+                        provenance: "SYSTEM EXTRACTOR".to_string(),
+                        timestamp: chrono::Utc::now().naive_utc(),
+                    }),
+                    html.title.map(|html_title| Entry {
+                        entity: address.clone(),
+                        attribute: LABEL_ATTR.to_string(),
                         value: html_title.into(),
                         provenance: "SYSTEM EXTRACTOR".to_string(),
                         timestamp: chrono::Utc::now().naive_utc(),
@@ -49,13 +57,24 @@ impl Extractor for WebExtractor {
                     }),
                 ];
                 for (key, value) in html.opengraph.properties {
+                    let attribute = format!("OG_{}", key.to_uppercase());
+                    if attribute == "OG_TITLE" {
+                        entries.push(Some(Entry {
+                            entity: address.clone(),
+                            attribute: LABEL_ATTR.to_string(),
+                            value: value.clone().into(),
+                            provenance: "SYSTEM EXTRACTOR".to_string(),
+                            timestamp: chrono::Utc::now().naive_utc(),
+                        }));
+                    }
+
                     entries.push(Some(Entry {
                         entity: address.clone(),
-                        attribute: format!("OG_{}", key.to_uppercase()),
+                        attribute,
                         value: value.into(),
                         provenance: "SYSTEM EXTRACTOR".to_string(),
                         timestamp: chrono::Utc::now().naive_utc(),
-                    }))
+                    }));
                 }
                 for image in html.opengraph.images {
                     entries.push(Some(Entry {