Add sha256: search criterion

weasyl/search.py

@@ -41,8 +41,9 @@ def __init__(self):
         self.required_user_excludes = set()
         self.ratings = set()
         self.find = None
+        self.sha256: str | None = None
 
-    def add_criterion(self, criterion):
+    def add_criterion(self, criterion: str):
         def add_nonempty(s, item):
             if item:
                 s.add(item)
@@ -65,6 +66,9 @@ def add_nonempty(s, item):
         elif criterion.startswith("-user:"):
             user = d.get_sysname(criterion.split(":", 1)[1])
             add_nonempty(self.required_user_excludes, user)
+        elif criterion.startswith("sha256:"):
+            sha256 = criterion.split(":", 1)[1]
+            self.sha256 = sha256 or None
         elif criterion.startswith("+"):
             tag = d.get_search_tag(criterion[1:])
             add_nonempty(self.required_includes, tag)
@@ -85,6 +89,7 @@ def __bool__(self):
             self.required_excludes or
             self.required_user_includes or
             self.required_user_excludes or
+            self.sha256 or
             self.ratings)
 
     @classmethod
@@ -135,7 +140,7 @@ def select_users(q):
 
 
 def _find_without_media(userid, rating, limit,
-                        search, within, cat, subcat, backid, nextid):
+                        search: Query, within, cat, subcat, backid, nextid):
     type_code, type_letter, table, select, subtype = _TABLE_INFORMATION[search.find]
 
     # Begin statement
@@ -163,6 +168,12 @@ def _find_without_media(userid, rating, limit,
         elif cat:
             statement_where.append("AND content.subtype >= %(category)s AND content.subtype < %(category)s + 1000")
 
+    if search.find == "submit" and search.sha256:
+        statement_from.append("INNER JOIN submission_media_links ON submission_media_links.submitid = content.{select}")
+        statement_from.append("INNER JOIN media ON media.mediaid = submission_media_links.mediaid")
+        statement_where.append("AND submission_media_links.link_type = 'submission'")
+        statement_where.append("AND media.sha256 = %(sha256)s")
+
     if userid:
         if within == "notify":
             # Search within notifications
@@ -265,6 +276,7 @@ def make_statement(statement_select, statement_additional_where, statement_order
         ]).format(
             table=table,
             find=search.find,
+            sha256=search.sha256,
             select=select,
             subtype=subtype,
             title_field="char_name" if search.find == "char" else "title",
@@ -302,6 +314,7 @@ def get_ids(names):
         "required_excludes": get_ids(search.required_excludes),
         "required_user_includes": list(search.required_user_includes),
         "required_user_excludes": list(search.required_user_excludes),
+        "sha256": search.sha256,
         "type": type_letter,
         "userid": userid,
         "rating": rating,

weasyl/templates/help/searching.html

@@ -46,6 +46,15 @@ <h4>Search modifiers (can only be used with at least one search term above)</h4>
           <td>Limits results to the <em>content</em> level specified. <em>Use as many of these per search as you wish.</em></td>
         </tr>
       </table>
+
+      <h4>Advanced search types</h4>
+
+      <table>
+        <tr>
+          <th><pre>sha256:8c474ca2ebfa54b1439a5c6f025f58ab443e6e8ba8a0ed885046d6375fbf4e8a</pre></th>
+          <td>Search for a submission whose media matches the given SHA-256 hash. Given a piece of artwork, this can be used to find the original submission on Weasyl, in a process known as <em>reverse image search</em>.</td>
+        </tr>
+      </table>
     </div>
   </div>
 </div>

weasyl/test/test_search.py

@@ -3,7 +3,7 @@
 import pytest
 
 from libweasyl.models.helpers import CharSettings
-from libweasyl import ratings
+from libweasyl import media, ratings
 from weasyl.test import db_utils
 from weasyl import search
 
@@ -217,3 +217,50 @@ def _select(userid: int):
 
     assert len(_select(user1_id)) == 1
     assert len(_select(user1_id)) == 1
+
+
+def test_search_sha256(db):
+    userid = db_utils.create_user()
+    tagid = db_utils.create_tag('test')
+
+    def _select(term: str):
+        results, _, _ = search.select(
+            search=search.Query.parse(term, 'submit'),
+            userid=userid, rating=ratings.EXPLICIT.code, limit=100,
+            cat=None, subcat=None, within='', backid=None, nextid=None,
+        )
+
+        return results
+
+    submitid_foo = db_utils.create_submission(userid, 'Foo')
+    db_utils.create_submission_tag(tagid, submitid_foo)
+    with open('libweasyl/test/data/1x70.jpg', 'rb') as file_foo:
+        data_foo = file_foo.read()
+    media_foo = media.MediaItem.fetch_or_create(data_foo, file_type='jpg')
+    media.SubmissionMediaLink.make_or_replace_link(submitid_foo, 'submission', media_foo)
+
+    submitid_bar = db_utils.create_submission(userid, 'Bar')
+    db_utils.create_submission_tag(tagid, submitid_bar)
+    with open('libweasyl/test/data/2x233.gif', 'rb') as file_bar:
+        data_bar = file_bar.read()
+    media_bar = media.MediaItem.fetch_or_create(data_bar, file_type='gif')
+    media.SubmissionMediaLink.make_or_replace_link(submitid_bar, 'submission', media_bar)
+
+    submitid_baz = db_utils.create_submission(userid, 'Baz')
+    db_utils.create_submission_tag(tagid, submitid_baz)
+    with open('libweasyl/test/data/2x233.gif', 'rb') as file_baz:
+        data_baz = file_baz.read()
+    media_baz = media.MediaItem.fetch_or_create(data_baz, file_type='gif')
+    media.SubmissionMediaLink.make_or_replace_link(submitid_baz, 'submission', media_baz)
+
+    results_foo = _select('sha256:a4e6ff301f671e8afb2397b4967c6a7004657447a5877a08d2dc29af58fccf3b')
+    results_bar = _select('sha256:fa6bd1dacf7b3d351172051826bb6d123113cfb7314322732a3ee1a917531672')
+    results_unknown = _select('sha256:0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f')
+
+    assert len(results_foo) == 1
+    assert results_foo[0]['title'] == 'Foo'
+
+    assert len(results_bar) == 2
+    assert sorted(map(lambda x: x['title'], results_bar)) == ['Bar', 'Baz']
+
+    assert len(results_unknown) == 0