Merge pull request #127 from internetarchive/issue-21-fixes

glenrobson · web-flow · commit d71989de50f7 · 2025-05-27T18:54:00.000+01:00
IIIF search fixes
diff --git a/iiify/app.py b/iiify/app.py
@@ -19,7 +19,10 @@
 app = Flask(__name__)
 # disabling sorting of the output json
 app.config['JSON_SORT_KEYS'] = False
-app.config['CACHE_TYPE'] = "FileSystemCache" if os.environ.get("FLASK_CACHE_DISABLE", None) != "true" else "NullCache"
+# To disable the cache you must call 
+# app.config['CACHE_TYPE'] = "NullCache"
+# as this code is run before the test.setUp 
+app.config['CACHE_TYPE'] = "FileSystemCache" 
 app.config['CACHE_DIR'] = "cache"
 cors = CORS(app) if cors else None
 cache = Cache(app)
@@ -28,7 +31,6 @@
 
 # cache.init_app(app)
 
-
 def cache_bust():
     return request.args.get("recache", "") in ["True", "true", "1"]
 
diff --git a/iiify/resolver.py b/iiify/resolver.py
@@ -646,7 +646,7 @@ def create_manifest3(identifier, domain=None, page=None):
             # Add search service
             service = {
                 "@context": "http://iiif.io/api/search/1/context.json",
-                "@id": f"{domain.replace("https","http")}search/{identifier}",
+                "@id": f"{domain}search/{identifier}",
                 "@type": "SearchService1",
                 "profile": "http://iiif.io/api/search/1/search"
             }
diff --git a/iiify/search.py b/iiify/search.py
@@ -41,6 +41,9 @@ def iiif_search(identifier, query):
             largeResults = False
             soup = BeautifulSoup(match['text'], 'html.parser')
             match = [tag.text for tag in soup.find_all('ia_fts_match')]
+            if len(match) != len(paragraph['boxes']):
+                for i in range(len(paragraph['boxes']) - len(match)):
+                    match.append(query)
         else:
             text = query    
 
@@ -49,8 +52,19 @@ def iiif_search(identifier, query):
         for box in paragraph['boxes']:
             x = int(box['l'])
             y = int (box['t'])
-            width = int(box['r']) - x
+            right = 0
+            # If r is missing then use the paragraph
+            if 'r' in box:
+                right = int(box['r'])
+            else:
+                right = paragraph['r'] 
+
+            width = right - x
             height = int(box['b']) - y
+            page = int(paragraph['page']) - 1
+            if "leaf0_missing" in ia_response and ia_response['leaf0_missing'] == False:
+                page = int(paragraph['page'])
+
             searchResponse['resources'].append({
                 "@id": f"{URI_PRIFIX}/{identifier}/annotation/anno{count}",
                 "@type": "oa:Annotation",
@@ -59,7 +73,7 @@ def iiif_search(identifier, query):
                     "@type": "cnt:ContentAsText",
                     "chars": text if largeResults else match[matchNo] 
                 },
-                "on": f"{URI_PRIFIX}/{identifier}${int(paragraph['page']) - 1}/canvas#xywh={x},{y},{width},{height}",
+                "on": f"{URI_PRIFIX}/{identifier}${page}/canvas#xywh={x},{y},{width},{height}",
                 "within": {
                     "@id": f"{URI_PRIFIX}/{identifier}/manifest.json",
                     "type": "sc:Manifest"
diff --git a/tests/fixtures/search/missmatch.json b/tests/fixtures/search/missmatch.json
@@ -0,0 +1,55 @@
+{
+    "ia": "mrr_401",
+    "q": "top",
+    "indexed": true,
+    "leaf0_missing": false,
+    "page_count": 136,
+    "matches": [
+        {
+            "text": "SONYA: I have a lot going on right now. I just decided to retire all the jeans. Now I am working on a cut-and-sew line but it’s <IA_FTS_MATCH>TOP</IA_FTS_MATCH> secret so don’t tell anyone. I am currently making a jacket that is an interpretation of a Vito Acconci piece.",
+            "par": [
+                {
+                    "t": 1661,
+                    "b": 2351,
+                    "l": 3632,
+                    "r": 4694,
+                    "boxes": [
+                        {
+                            "t": 1927,
+                            "b": 1998,
+                            "l": 4586,
+                            "page": 78,
+                            "r": 4678
+                        },
+                        {
+                            "t": 2015,
+                            "b": 2086,
+                            "l": 3651,
+                            "r": 4678
+                        },
+                        {
+                            "t": 2104,
+                            "b": 2176,
+                            "l": 3651,
+                            "r": 4678
+                        },
+                        {
+                            "t": 2192,
+                            "b": 2263,
+                            "l": 3650,
+                            "r": 4679
+                        },
+                        {
+                            "t": 2284,
+                            "b": 2351,
+                            "l": 3650
+                        }
+                    ],
+                    "page": 78,
+                    "page_width": 5105,
+                    "page_height": 6492
+                }
+            ]
+        }
+    ]
+}
diff --git a/tests/test_annotations.py b/tests/test_annotations.py
@@ -7,7 +7,7 @@
 class TestAnnotations(unittest.TestCase):
 
     def setUp(self) -> None:
-        os.environ["FLASK_CACHE_DISABLE"] = "true"
+        app.config['CACHE_TYPE'] = "NullCache"
         self.test_app = FlaskClient(app)
 
     def test_v3_manifest_has_annotations(self):
diff --git a/tests/test_audio.py b/tests/test_audio.py
@@ -8,7 +8,7 @@
 class TestAudio(unittest.TestCase):
 
     def setUp(self) -> None:
-        os.environ["FLASK_CACHE_DISABLE"] = "true"
+        app.config['CACHE_TYPE'] = "NullCache"
         self.test_app = FlaskClient(app)
 
     def test_audio_no_derivatives(self):
diff --git a/tests/test_basic.py b/tests/test_basic.py
@@ -8,10 +8,9 @@
 class TestBasic(unittest.TestCase):
 
     def setUp(self) -> None:
-        os.environ["FLASK_CACHE_DISABLE"] = "true"
+        app.config['CACHE_TYPE'] = "NullCache"
         self.test_app = FlaskClient(app)
 
-
     def test_documentation(self):
         resp = self.test_app.get("/iiif/documentation")
         self.assertEqual(resp.status_code, 200)
diff --git a/tests/test_cantaloupe_resolver.py b/tests/test_cantaloupe_resolver.py
@@ -2,11 +2,12 @@
 
 import unittest
 from iiify.resolver import cantaloupe_resolver
+from iiify.app import app
 
 class TestCantaloupeResolver(unittest.TestCase):
 
     def setUp(self) -> None:
-        os.environ["FLASK_CACHE_DISABLE"] = "true"
+        app.config['CACHE_TYPE'] = "NullCache"
 
     def test_single_image(self):
         cid = cantaloupe_resolver("img-8664_202009")
diff --git a/tests/test_collections.py b/tests/test_collections.py
@@ -6,7 +6,7 @@
 class TestCollections(unittest.TestCase):
 
     def setUp(self) -> None:
-        os.environ["FLASK_CACHE_DISABLE"] = "true"
+        app.config['CACHE_TYPE'] = "NullCache"
         self.test_app = FlaskClient(app)
 
     def test_v3_collection(self):
diff --git a/tests/test_helper.py b/tests/test_helper.py
@@ -8,6 +8,7 @@
 class TestHelper(unittest.TestCase):
 
     def setUp(self) -> None:
+        app.config['CACHE_TYPE'] = "NullCache"
         self.test_app = FlaskClient(app)
 
     def test_single_image(self):
diff --git a/tests/test_images.py b/tests/test_images.py
@@ -7,7 +7,7 @@
 class TestImages(unittest.TestCase):
 
     def setUp(self) -> None:
-        os.environ["FLASK_CACHE_DISABLE"] = "true"
+        app.config['CACHE_TYPE'] = "NullCache"
         self.test_app = FlaskClient(app)
 
     def test_v3_resolving(self):
diff --git a/tests/test_linking.py b/tests/test_linking.py
@@ -7,7 +7,7 @@
 class TestLinking(unittest.TestCase):
 
     def setUp(self) -> None:
-        os.environ["FLASK_CACHE_DISABLE"] = "true"
+        app.config['CACHE_TYPE'] = "NullCache"
         self.test_app = FlaskClient(app)
 
     def convertListToHash(self, items):
diff --git a/tests/test_manifests.py b/tests/test_manifests.py
@@ -7,7 +7,7 @@
 class TestManifests(unittest.TestCase):
 
     def setUp(self) -> None:
-        os.environ["FLASK_CACHE_DISABLE"] = "true"
+        app.config['CACHE_TYPE'] = "NullCache"
         self.test_app = FlaskClient(app)
 
     def test_no_version(self):
diff --git a/tests/test_manifests_v2.py b/tests/test_manifests_v2.py
@@ -7,7 +7,7 @@
 class TestManifests(unittest.TestCase):
 
     def setUp(self) -> None:
-        os.environ["FLASK_CACHE_DISABLE"] = "true"
+        app.config['CACHE_TYPE'] = "NullCache"
         self.test_app = FlaskClient(app)
 
     def test_v2_image_manifest(self):
diff --git a/tests/test_resolver.py b/tests/test_resolver.py
@@ -2,10 +2,11 @@
 
 import unittest
 from iiify.resolver import purify_domain, collection, manifest_page
+from iiify.app import app
 
 class TestResolver(unittest.TestCase):
     def setUp(self) -> None:
-        os.environ["FLASK_CACHE_DISABLE"] = "true"
+        app.config['CACHE_TYPE'] = "NullCache"
 
     def test_purify(self):
         domain = purify_domain("https://example.org/iiif/")
diff --git a/tests/test_search.py b/tests/test_search.py
@@ -23,7 +23,7 @@ def raise_for_status(self):
 class TestSearch(unittest.TestCase):
 
     def setUp(self) -> None:
-        os.environ["FLASK_CACHE_DISABLE"] = "true"
+        app.config['CACHE_TYPE'] = "NullCache"
         self.test_app = FlaskClient(app)
 
     def test_live(self):
@@ -33,6 +33,19 @@ def test_live(self):
 
         self.assertEqual(len(results["resources"]),2, "Expected two results for Brunswick")
 
+    def test_search_in_manifest(self):    
+        resp = self.test_app.get("/iiif/3/journalofexpedit00ford/manifest.json?recache=true")
+        self.assertEqual(resp.status_code, 200)
+        manifest = resp.json    
+
+        self.assertTrue("service" in manifest, "Failed to find search service in manifest")
+        self.assertEqual(len(manifest['service']), 1, "Expected a single search service")
+        service = manifest['service'][0]
+        self.assertEqual(service['@id'], "https://localhost/iiif/search/journalofexpedit00ford")
+        self.assertEqual(service['@type'], "SearchService1")
+        self.assertEqual(service['profile'], "http://iiif.io/api/search/1/search")
+
+
     @patch("requests.get")
     def test_fixture(self, searchPatch):
             # Define mock response for the specific URLs
@@ -116,6 +129,34 @@ def mock_response(url, *args, **kwargs):
         self.assertEqual(results[2]["resource"]["chars"], "to", "Unexpected search match")
         self.assertEqual(results[3]["resource"]["chars"], "observe", "Unexpected search match")
 
+    @patch("requests.get")
+    def multi_box(self, searchPatch):
+        # Define mock response for the specific URLs
+        def mock_response(url, *args, **kwargs):
+            if "fulltext/inside.php" in url:
+                return mockResponse("tests/fixtures/search/missmatch.json")
+            elif "metadata" in url:
+                # Although this doesn't match the item used in the 
+                # search above it shouldn't matter for this test
+                return mockResponse("tests/fixtures/metadata/journalofexpedit00ford.json")
+
+        searchPatch.side_effect = mock_response    
+
+        resp = self.test_app.get("/iiif/search/mrr_401/?q=top")
+        self.assertEqual(resp.status_code, 200)
+        results = resp.json        
+
+        # this doesn't have the r field in the box:
+        anno5 = results["resources"][4]
+        self.assertEqual(anno5['on'], "https://iiif.archive.org/iiif/mrr_401$78/canvas#xywh=3650,2284,1044,67")
+        anno1 = results["resources"][0]
+        # First anno should use the IA_FTS_MATCH
+        self.assertEqual(anno1["resource"]["chars"], "TOP")
+        # The rest should use the query 
+        for anno in results["resources"][1:]:
+            self.assertEqual(anno["resource"]["chars"], "top")
+
+
     def test_matching_characters_not_empty(self):
         resp = self.test_app.get("/iiif/search/journalofexpedit00ford/?q=Brunswick")
         results = resp.json
diff --git a/tests/test_video.py b/tests/test_video.py
@@ -8,7 +8,7 @@
 class TestVideo(unittest.TestCase):
 
     def setUp(self) -> None:
-        os.environ["FLASK_CACHE_DISABLE"] = "true"
+        app.config['CACHE_TYPE'] = "NullCache"
         self.test_app = FlaskClient(app)
 
     def test_v3_single_video_manifest(self):

Original file line number	Diff line number	Diff line change
`@@ -646,7 +646,7 @@ def create_manifest3(identifier, domain=None, page=None):`
`646`	`646`	`# Add search service`
`647`	`647`	`service = {`
`648`	`648`	`"@context": "http://iiif.io/api/search/1/context.json",`
`649`		`- "@id": f"{domain.replace("https","http")}search/{identifier}",`
	`649`	`+ "@id": f"{domain}search/{identifier}",`
`650`	`650`	`"@type": "SearchService1",`
`651`	`651`	`"profile": "http://iiif.io/api/search/1/search"`
`652`	`652`	`}`