From 8359f78e2e6a0ba56f069e89884905f9682d1805 Mon Sep 17 00:00:00 2001
From: Niels Simenon <niels.simenon@gmail.com>
Date: Sun, 23 Feb 2020 22:01:12 +0100
Subject: [PATCH] Fixed RK scraper returning dick size as bust size.

---
 src/actors.js            |  2 +-
 src/media.js             | 11 ++++++++---
 src/scrapers/mindgeek.js |  8 +++++---
 3 files changed, 14 insertions(+), 7 deletions(-)

diff --git a/src/actors.js b/src/actors.js
index c5f9b793..8da740d3 100644
--- a/src/actors.js
+++ b/src/actors.js
@@ -315,7 +315,7 @@ async function mergeProfiles(profiles, actor) {
             residencePlace: prevProfile.residencePlace || profile.residencePlace,
             nationality: prevProfile.nationality || profile.nationality, // used to derive country when not available
             ethnicity: prevProfile.ethnicity || profile.ethnicity,
-            bust: prevProfile.bust || profile.bust,
+            bust: prevProfile.bust || (/\d+\w+/.test(profile.bust) && profile.bust),
             waist: prevProfile.waist || profile.waist,
             hip: prevProfile.hip || profile.hip,
             naturalBoobs: prevProfile.naturalBoobs === undefined ? profile.naturalBoobs : prevProfile.naturalBoobs,
diff --git a/src/media.js b/src/media.js
index 73bea226..e2eabb49 100644
--- a/src/media.js
+++ b/src/media.js
@@ -115,7 +115,7 @@ async function fetchItem(source, index, existingItemsBySource, domain, role, att
             return null;
         }
 
-        logger.verbose(`Fetching media item from ${source.src || source}`);
+        logger.verbose(`Fetching ${domain} ${role} from ${source.src || source}`);
 
         // const res = await bhttp.get(source.src || source);
         const res = await get(source.src || source);
@@ -199,8 +199,13 @@ async function saveItems(items, domain, role) {
             logger.verbose(`Saved ${domain} ${role} to ${filepath}`);
 
             return {
-                ...item,
                 filepath,
+                mimetype: item.mimetype,
+                extension: item.extension,
+                hash: item.hash,
+                entropy: item.entropy,
+                quality: item.quality,
+                source: item.source,
             };
         } catch (error) {
             logger.error(`Failed to store ${domain} ${role} from ${item.source}: ${error.message}`);
@@ -250,7 +255,7 @@ async function storeMedia(sources, domain, role, { entropyFilter = 2.5 } = {}) {
             const { hash: fetchedItemsByHash } = groupItems(fetchedItems);
 
             // find hash duplicates that don't need to be re-saved
-            const uniqueFetchedItems = Object.values(fetchedItemsByHash).filter(item => !entropyFilter || item.entropy >= entropyFilter);
+            const uniqueFetchedItems = Object.values(fetchedItemsByHash).filter(item => !entropyFilter || item.entropy === null || item.entropy >= entropyFilter);
             const existingHashItems = await knex('media').whereIn('hash', uniqueFetchedItems.map(item => item.hash));
             const { hash: existingHashItemsByHash } = groupItems(existingHashItems);
 
diff --git a/src/scrapers/mindgeek.js b/src/scrapers/mindgeek.js
index fac69001..c2f45c2a 100644
--- a/src/scrapers/mindgeek.js
+++ b/src/scrapers/mindgeek.js
@@ -150,9 +150,11 @@ function scrapeProfile(data, html, releases = [], networkName) {
 
     profile.gender = data.gender === 'other' ? 'transsexual' : data.gender;
 
-    if (bust) profile.bust = bust.toUpperCase();
-    if (waist) profile.waist = waist;
-    if (hip) profile.hip = hip;
+    if (profile.gender === 'female') {
+        if (bust) profile.bust = bust.toUpperCase();
+        if (waist) profile.waist = waist;
+        if (hip) profile.hip = hip;
+    }
 
     if (data.birthPlace) profile.birthPlace = data.birthPlace;
     if (data.height) profile.height = inchesToCm(data.height);