IIIF v3 manifest support (#29)

wikimedia/executors/downloader.py

@@ -56,9 +56,10 @@ def download(self, source, destination):
             destination, size = self._save_to_s3(source=source, bucket=bucket, key=key)
             self.tracker.increment(Result.DOWNLOADED, size=size)
             return destination, size
-        except Exception as exec:
+        except Exception as err:
             self.tracker.increment(Result.FAILED)
-            raise DownloadException(f"Failed download {source} - {str(exec)}") from exec
+            raise err
+            # DownloadException(f"Failed download {source} - {str(exec)}") from exec
 
     # TODO This maybe better in the FileSystem class
     def save_to_local(self, source, file):
@@ -73,10 +74,23 @@ def save_to_local(self, source, file):
         """
         try:
             response = requests.get(source, timeout=30)
+            invalid_types = [
+                "text/html",
+                "application/json",
+                "application/xml",
+                "text/plain",
+            ]
+            for invalid_type in invalid_types:
+                if invalid_type in response.headers["content-type"]:
+                    raise DownloadException(
+                        f"Invalid content-type: {response.headers['content-type']}"
+                    )
             with open(file, "wb") as f:
                 f.write(response.content)
             file_size = os.path.getsize(file)
             return file, file_size
+        except DownloadException as de:
+            raise de
         except Exception as exec:
             raise DownloadException(f"Failed saving to local {str(exec)}") from exec
 

wikimedia/utilities/iiif.py

@@ -25,15 +25,38 @@ def __init__(self):
     def iiif_v2_urls(self, iiif):
         """
         Extracts image URLs from IIIF manfiest and returns them as a list
-        # TODO
         """
+        urls = []
+        sequences = iiif.get("sequences", [])
+        sequence = sequences[0:1] if len(sequences) == 1 else None
+        canvases = sequence[0].get("canvases", []) if sequence else []
+
+        for canvase in canvases:
+            for image in canvase.get("images", []):
+                url = image.get("resource", {}).get("@id", None)
+                if url:
+                    urls.append(url)
+        return urls
 
     def iiif__v3_urls(self, iiif):
-        """
-        Needs to be implemented for Georgia uploads to Wikimedia Commons
-        To be done by October 2023
-        # TODO
-        """
+        """ """
+        # items[0] \ items[x] \ items[0] \ body \ id
+        resolution = "/full/full/0/default.jpg"
+        urls = []
+        for item in iiif.get("items", []):
+            try:
+                url = item["items"][0]["items"][0].get("body", {}).get("id", None)
+                # This is a hack to get around that v3 presumes the user supplies the
+                # resolution in the URL
+                if url:
+                    # This condition may not be necessary but I'm leaving it in for now
+                    if url.endswith("default.jpg"):
+                        urls.append(url)
+                    else:
+                        urls.append(f"{url}{resolution}")
+            except (IndexError, TypeError, KeyError):
+                return []
+        return urls
 
     def get_iiif_urls(self, iiif):
         """
@@ -46,16 +69,19 @@ def get_iiif_urls(self, iiif):
 
         manifest = self._get_iiif_manifest(iiif)
 
-        urls = []
-        sequences = manifest.get("sequences", [])
-        sequence = sequences[0:1] if len(sequences) == 1 else None
-        canvases = sequence[0].get("canvases", []) if sequence else []
-        for canvase in canvases:
-            for image in canvase.get("images", []):
-                url = image.get("resource", {}).get("@id", None)
-                if url:
-                    urls.append(url)
-        return urls
+        # v2 or v3?
+        if (
+            manifest.get("@context", None)
+            == "http://iiif.io/api/presentation/3/context.json"
+        ):
+            return self.iiif__v3_urls(manifest)
+        elif (
+            manifest.get("@context", None)
+            == "http://iiif.io/api/presentation/2/context.json"
+        ):
+            return self.iiif_v2_urls(manifest)
+        else:
+            raise IIIFException("Unknown IIIF version")
 
     def _get_iiif_manifest(self, url):
         """
@@ -64,18 +90,16 @@ def _get_iiif_manifest(self, url):
         if not validators.url(url):
             raise IIIFException(f"Invalid url {url}")
         try:
-            request = requests.get(url, timeout=30, headers=self.HEADERS)
+            request = requests.get(
+                url, timeout=30, allow_redirects=True, headers=self.HEADERS
+            )
             if request.status_code not in [200, 301, 302]:
-                raise IIIFException(
-                    f"Unable to request: {url} - \
-                                    Status code {request.status_code}"
-                )
+                raise IIIFException(f"Invalid response code: {request.status_code}")
             data = request.content
             return json.loads(data)
         except json.decoder.JSONDecodeError as jdex:
-            raise IIIFException(
-                f"Unable to decode JSON: {url} - \
-                                {str(jdex)}"
-            ) from jdex
+            raise IIIFException(f"Unable to decode JSON: {url} - {str(jdex)}") from jdex
         except requests.exceptions.RequestException as re:
-            raise IIIFException(f"Unable to request: {url} - {str(re)}") from re
+            raise IIIFException(f"{str(re)}") from re
+        except Exception as ex:
+            raise Exception(f"Unknown error: {str(ex)}") from ex